Proceedings of Microwave Update 2010

PROCEEDINGS OF

MICROWAVE

2010

Cerritos, California October 21-24

Published by:

ARRL AMATEUR RADIO The national association for

Copyright  2010 by The American Radio Relay League, Inc. Copyright secured under the Pan-American Convention International Copyright secured All rights reserved. No part of this work may be reproduced in any form except by written permission of the publisher. All rights of translation reserved. Printed in USA Quedan reservados todos los derechos

First Edition

ii

Welcome to Microwave Update 2010 Hosted by San Bernardino Microwave Society And San Diego Microwave Group

Welcome to all the microwavers around the world, 25 years of M.U.D. This year the San Bernardino Microwave Society And San Diego Microwave Group bring you the world’s most prominent Microwave conference, there will be Talks ,antenna range, Swap meet and vendors We have a very good line up for you and hope you enjoy I would like to give a special thanks to DICK KOLBLY K6HIJ who we lost just recently Dick was not much into operating, but was a most appreciated asset to the SBMS and the microwave world, Dick was always willing to go out of his way to help anyone who wanted it, and never asking for anything in return, DICK you will be missed The following people have made this conference and proceedings outstanding N6RMJ K6JEY WB6CWN WA6CGR WA6JBD Phyllis Kolbly KH6WZ KC6QHP WB6DNX W6OYJ N6IZN

Chairman CO - Chairman Speakers \ Proceedings Prizes \ Testing Surplus Tour \ Testing Registration Publicity JPL Tour Vendors Antenna Testing Antenna Testing

A special thanks to the ladies for putting on the family program There is a lot more people who have helped to put this on, Thank You SBMS and SDMG Pat N6RMJ

iii

History of Microwave Update by Al Ward W5LUA August 2010 Edition In 1985, Don Hilliard, WØPW, felt the need to organize a conference dedicated to microwave equipment design, construction, and operation. At the time of its conception, many microwave terrestrial and EME firsts were occurring on the microwave bands and it appeared that microwave needed a dedicated conference. Don held the first conference which he named “The 1296 and 2304 MHz Conference.” It was held at the Holiday Inn in Estes Park, Colorado. 66 people were in attendance. It sure seemed like Don was on the right track with his idea and he was right. In 1986, Don held the second conference which he rightfully named “Microwave Update 86.” 64 people were in attendance. The 1987 and 1988 “Microwave Update” conferences were again held in Estes Park, CO. and chaired by Don Hilliard. After putting on 4 fine conferences in Colorado, Don decided to take a break from all of the work. Don turned over the responsibility of coordinating the event to the North Texas Microwave Society (NTMS). In 1989, WB5LUA and WA5VJB of the NTMS hosted the 5th “Microwave Update” in Arlington, Texas where 94 people were in attendance. The 1990 “Microwave Update” was to go back to Colorado where Keith Ericson, KØKE and Don Lund , WAØIQN, were to head up the event. Unfortunately, Don Lund passed away during the year and Keith decided to postpone the 1990 Update. WB5LUA and WA5VJB of the NTMS hosted “Microwave Update” 91 in Arlington, Texas. “Microwave Update” ’92 was held in Rochester, New York and sponsored by the Rochester VHF Group. The conference was chaired by Frank Pollino, K2OS and Dave Hallidy, K2DH (x KD5RO/2). “Microwave Update” ’93 was held in Atlanta, Georgia. The conference was organized by Jim Davey, WA8NLC, and assisted by Rick Campbell, KK7B and Charles Osborne, WD4MBK. “Microwave Update” ’94 was brought back to Estes Park, Colorado where it was chaired by Bill McCaa, KØRZ. Bill was assisted by Al Ward, WB5LUA, Jim Davey, WA8NLC, Jim Starkey, WØKJY, Phil Gabriel, AAØBR, and other local area amateurs. “Microwave Update” ’95 was brought back to Arlington, Texas and was chaired by Al Ward, WB5LUA and Kent Britain, WA5VJB of the NTMS. The ’96 “Microwave Update” was held in Phoenix, Arizona and was chaired by Jim Vogler, WA7CJO. The ’97 “Microwave Update” was held in Sandusky, Ohio and sponsored by Tom Whitted, WA8WZG, with the assistance of Tony Emanuele, WA8RJF. The 1998 “Microwave Update” was held in Colorado under the guidance of Bill McCae, KØRZ, and John Anderson, WD4MUO. The 1999 “Microwave Update” was held in Plano, Texas with Al Ward, W5LUA and Kent Britain, WA5VJB hosting the event. The 2000 Microwave Update was held in the Philadelphia area with John Sortor, KB3XG, and Paul Drexler, W2PED hosting the event. The 2001 Microwave Update was hosted by Jim Moss, N9JIM and Will Jensby, WØEOM, in the Sunnyvale, California area. The 2002 conference was held in conjunction with the Eastern VHF/UHF Conference in Enfield, CT. The conference was hosted by Paul Wade, W1GHZ, Matt Reilly, KB1VC, Tom Williams, WA1MBA and Bruce Wood, N2LIV. The 2003 conference moved across country to Seattle, WA where Rick Beatty, NU7Z and the PNWVHFS hosted the event. Rick’s committee consisted of John N7MWV as the co-chairman along with Jim K7ND, Eric N7EPD, Jim, W7DHC, Jimmy, K7NQ, and Lynn, N7CFO. The 2004 conference was held in Dallas, Texas where Al Ward, W5LUA, Bob Gormley, WA5YWC, Kent Britain, WA5VJB, and the North Texas Microwave Society hosted the event.

iv

The 2005 conference was held in Cerritos, CA. The event was hosted by Pat Coker, N6RMJ and Chip Angle, N6CA ,along with the San Bernardino Microwave Society and the Western States Weak Signal Society. The 2006 conference was held in Dayton, Ohio and was hosted by Tom Holmes, N8ZM, and Gerd Schrick, WB8IFM of the Midwest VHF/UHF Society. The 2007 conference was held in Valley Forge, PA at the Dolce Valley Forge. The conference was hosted by Phil Theis, K3TUF, David Fleming, KB3HCL, Rick Rosen, K1DS, and Paul Drexler, W2PED of the Mt Airy VHF Radio Club. The 2008 conference was hosted by Donn Baker, WA2VOI, Barry Malowanchuk, VE4MA, Jon Platt, W0ZQ, Bruce Richardson, W9FZ, Bob Wesslund, WØAUS, of the Northern Lights Radio Society and was held in Bloomington, MN. The 2009 conference was held in Dallas, Texas and was hosted by Steve Hicks, N5AC, Al Ward, W5LUA, Bob Gormley, WA5YWC, and Kent Britain, WA5VJB, of the North Texas Microwave Society. In 2009, The Don Hilliard Technical Achievement Award was created in honor of our founding father Don Hilliard, WØPW. The first recipient was Paul Wade, W1GHZ, in recognition of his many years of service to the amateur microwave community. The 2010 conference is being hosted by the San Bernardino Microwave Society in Southern California. The 2011 conference will be hosted by the North East Weak Signal Group. Those that are interested in sponsoring a conference may contact myself, Al Ward, W5LUA or Kent Britain, WA5VJB. Respectfully Submitted, Al Ward, W5LUA 08-30-2010

v

The Don Hilliard Technical Achievement Award Don Hilliard, WØPW, (exWØEYE) an early VHF pioneer was involved with the formation of the Central States VHF Society back in 1967. The Central States VHF Society was and still is very instrumental in promoting VHF and above activity. Back in 1985, Don realized that there was a considerable thrust in new microwave technology above 902 MHz. As a result, Don felt the need to have a conference devoted to the higher frequencies. The conference would be devoted to microwave equipment design, construction, and operation. At the time of its conception, many microwave terrestrial and EME firsts were occurring on the microwave bands and it truly appeared that microwave needed a dedicated conference. Don organized the first conference which he named “The 1296 and 2304 MHz Conference”. It was held at the Holiday Inn in Estes Park, Colorado. 66 people were in attendance. It sure seemed like Don was on the right track with his idea and he was right. In 1986, Don held the second conference which he rightfully named “Microwave Update 86.” 64 people were in attendance. The 1987 and 1988 “Microwave Update” conferences were again held in Estes Park, Co. and chaired by Don Hilliard. After putting on 4 fine conferences in Colorado, Don decided to take a break from all of the work. Don turned over the responsibility of coordinating the event to the North Texas Microwave Society (NTMS). The rest is history. With the exception of one year where one of the organizers, Don Lund passed away, Microwave Update has been held every year. To this date including the 2009 conference being held in Irving, Texas, Microwave Update has been hosted 24 times. The conference has been successfully organized and run by various local VHF and microwave clubs and groups around the US. In tribute to Don Hilliard and his tremendous contributions to VHF and microwave technology and for appreciation of his forward looking into the fascinating world of “microwaves,” the North Texas Microwave Society on behalf of Microwave Update would like to create “The Don Hilliard Technical Achievement Award” presented each year to an amateur radio operator who has made significant contributions to amateur microwave operation and technology. The NTMS proposes that this award be presented to a deserving amateur each year by each sponsoring organization. Respectfully submitted Al Ward W5LUA Kent Britain WA5VJB Steve Hicks N5AC September 9, 2009

vi

Table of Contents Welcome ....................................................................................................................................... iii History of Microwave Update ...................................................................................................... iv Don Hillard Award ....................................................................................................................... vi Introduction to Hot Rod Microwave Radios; Rick Campbell, KK7B ............................................1 Hot Rod Radio for 5760; Rick Campbell, KK7B ...........................................................................6 A 432/1296 MHz SSB/CW Direction Conversion Transceiver; Jim Davey, K8RZ ....................15 How to Increase 23 cm Power to 250W with 2 x XRF 286, Some Modifications to the W6PQL Kit; Dominique Faessler, HB9BBD ...................................................................22 2010 Observations on Phase Noise from Local Oscillator Strings; Gerald Johnson, KØCQ .......28 Safe Tapping in Soft Metals; Gerald Johnson, KØCQ .................................................................30 Taming Phase Noise at EHF; Brian Justin, WA1ZMS/4 ..............................................................34 Ka-Band Integrated-Circuit Interferometer for Sensing; Seok-Tae Kim and Cam Nguyen .........39 A Novel Approach to a Multiband Transverter Design; Jeff Kruth, WA3ZKR ...........................42 A YIG Filter Primer & Simple Driver Circuit for HAM Projects; Jeff Kruth, WA3ZKR ...........56 LO Phase Noise Effects on MDS; Gary Lauterbach, AD6FP ......................................................64 A Modern 47 GHz Transverter; Tony Long, KC6QHP ...............................................................76 NJR2145J 10 GHz Pre-amplifier Adaptation and Construction; Gary Lopes, WA6MEM ..........87 Propagation Observations with the 10 & 24 GHz VE4 Beacons; Barry Malowanchuk. VE4MA ..........................................................................................94 PIC’n on the ThunderBolt; John Maetta, N6VMO .....................................................................109 Development of an UWB CMOS Transmitter-Antenna Module; Meng Miao and Cam Nguyen .........................................................................................112 Frequency Stability Measurement: Technologies, Trends, and Tricks; John Miles, KE5FX .....116 Compass Basics And Some Representative Types; Doug Millar, K6JEY .................................134

vii

Signal Level Meter Throw Down; Doug Millar, K6JEY ...........................................................140 C-Band LNB to LNA Conversion; Christian Shoaff, N9RIN ....................................................144 A Personal Beacon for 10 GHz (That Can’t Possibly Work); Paul Wade, W1GHZ ..................149 High-Power Directional Couplers with Excellent Performance; That You Can Build; Paul Wade, W1GHZ .......................................................................................................152 Analysis of the WA1MBA 78 GHz Low Noise Amplifiers; Al Ward, W5LUA .......................167 Moving Ahead with the 78 GHz Low Noise Amplifier Project; Tom Williams, WA1MBA .....173 An Improved 2 x MRF286 Power Amplifier for 1296 MHz; Darrell Ward, VE1ALQ .............175 Modifying a DMC Dielectric Resonator Oscillator for Amateur 10 GHz Use; Brian Yee ........184 Connectorize Your IF Radio!; Wayne Yoshida, KH6WZ ..........................................................190 Working on the Microwaves: Seeing is Believing; Gene Zimmerman, W3ZZ, and David Mertz, WA3OFF ...........................................................................................191 Physical Optics Demonstrations with Microwave ......................................................................204 Noise Figure Measurements 2009 ..............................................................................................215

viii

Introduction to Hot Rod Microwave Radios Rick Campbell KK7B The following essays and projects illustrate a different approach to Amateur Microwave Radio, an undisciplined, enthusiasticly attention-deficit creative process loosely defined as “Stuff we do because it’s Cool.” These projects don’t increase frequency, reduce noise temperature, or chop signals to bits and reassemble them in some alternate domain. They are unconstrained by rules for coloring in squares on a map or counting the distant nerds one can greet in a weekend. These aren’t the microphones to grab in an emergency or broadband pipelines for 3-D real-time holographic video. But as design exercises they have stretched our limits and as construction projects they have forced us to learn new skills and refresh old ones. College engineering students find them irresistable. Have fun, and don’t take this stuff too seriously. Hot Rod Radios My friends and I find many attractions in Microwave Amateur Radio: pushing upper frequency limits, competition, radio science, and a love affair with the technology. Other papers in this digest are devoted to extending the state-of-the-art into micron-dimension waveguide bands, and we apply radio science every time we scatter a signal off some hard object or atmospheric anomaly to add a far-off grid to our cumulative total or contest score. This set of papers addresses a life-long love affair with the technology. Lifting the covers to see what’s inside a radio and figure out how it works was our original attraction, and it has stayed with us for half a century. As such, it provides a key to attracting and retaining the next generation of radio amateurs, scientists, technologists, and general technical problem solvers. Society needs other contributors, so if your natural tendencies involve compassion, fixing cuts and bruises, caring for animals, or bossing around other people based on your interpretation of the fine print in some law-don’t feel bad. Society has a place for you too. That fine print gives us access to our slices of spectrum. Jim Davey and I both have roots in Michigan. My grandfather took me to the Ford Rotunda for the world premier auto show every year as a child, and at an early age I knew the difference between a concept car, a test track vehicle, and the family car we drove to the Great Scot store. My father walked the family through the Edison Institute and Henry Ford Museum more times than I can count. We wandered through the evolution of ideas and products as they drifted off on bizarre styling tangents and offered new approaches to a changing national landscape. Along with the other boys of the time, we viewed the family car as a collection of parts that could be reassembled into something Really Cool if only our dads would let us. I have no idea what girls at the time thought-I still don’t. It’s a mystery. We played with Erector Sets and Knight-Kit 12-in-1 labs, following our imaginations along ten impractical paths for every one that actually led somewhere. An Amateur Radio License allows us to design and build our own gear, which has always seemed to me to be more significant than making contacts. We lured

1

unsuspecting Heathkits into dad’s shop and hacked and modified them into creations that sometimes worked almost as well as the originals. Our high-school classmates drove Hot Rod cars, and we squeezed 100 watts out of a single 6146. Briefly. It has been with some dismay that I have observed a flood of nostalgia for the unmodified radios of the past. They retain some of the romance generated by glossy ads in the back of the 1962 ARRL Handbook--but most of their personal appeal is still as a collection of parts that could be transformed into something Really Cool. My personal definition of Really Cool has often included operating on a higher frequency, hence my decade of contributions to the early years of this conference. Others have alternate definitions. My good friend Wes Hayward thinks Really Cool is a collection of individual components from the Tektronix Surplus Store, assembled on scraps of unetched circuit board, spread over the bench, and outperforming the $1000 appliance pushed to the back of the operating table/work bench. We’ll use that diversity as the first premise of Hot Rod Radio: 1. Hot Rot Radios are Really Cool--but acknowledge that the appeal may be limited. You may think my creations are merely strange. The second premise of Hot Rod Radios is that they exhibit individual creative contributions of the designer/builder/owner/operator. A simple hack isn’t enough--particularly a non-invasive, fully recoverable mod that can easily be reversed so that the rig appears original. I admit that years ago I caught a moderate case of Vintage Disease, and I have trouble drilling a hole in the front panel of a rare 1960 era radio. But I have no such qualms about ripping out all the guts and adding good connectors to the rear deck. So the next premise is: 2. Hot Rod Radios are exhibit enthusiastic, individualistic modifications. My goal is for the KK7B Hot Rod version of the family radio to have more appeal and higher street value than an unmodified stock example in good condition...at least, to me. Mint condition examples of even common, low cost radios should be left alone. There is some unwritten rule about that, and it is a good one to follow. Trade the mint example to a collector for two of the same model in modest condition. If basic performance is adequate, additional hardware can be added to enhance performance, or interface to microwave transverters. Recently I’ve been gutting simple radios with appealing cases and linear mechanical tuning mechanisms, and building a high performance analog radio in the box. Then I interface the radio with some additional gear--usually vintage homebrew--and use it on the air. My Hot Rod creations aren’t just art objects, they are street legal and operate well enough to be fun. That’s the third premise: 3. Hot Rod Radios perform well for a specific, often challenging technical function. In fact, all of my Hod Rod versions outperform anything commercially available on the specific bands and modes for which they were conceived, designed, and built.

2

Transforming a 1950s family sedan into a 1960s Hot Rod involved the sacrifice of some family sedan features: beige paint, automatic transmission, back seat, muffler--to gain performance in a particular area, primarily attracting the attention of girls. Since guys have no idea how to attract girls, they settle for the next best thing: intimidating other guys. That makes them feel cool. Guys who think they are cool radiate some kind of gas that lowers common sense. So by default, pretty girls end up sitting in the coolest cars. Briefly.

Photo 1. A drool-inducing package enclosing a truly marginal receiver. But appearance is deceiving. Under this mild-mannered exterior purrs a high-end R2pro receiver used as a high performance microwave IF.

Photo 2. A peek under the hood at the new high-stability VFO

3

Since electronics radiate some kind of consciousness-raising gas, radios have limited use in attracting girls. From observing my daughters I suspect that the mere existence of a radio as the object of a guy’s attention has an anti aphrodisiac effect. This is perhaps universal across many species. One could observe moose with radio collars... That is a fundamental difference between Hot Rod cars and Hot Rod radios. The designer/builder of a Hot Rod car expects some of the coolness to rub off on him. The Hot Rod radio designer has no such illusions. The radio is Cool, all by itself. That is enough. That is the final premise of Hot Rod radios: 4. There is no ulterior motive. A Hot Rod Radio is complete, in itself, creating its own context. It just sits there, being Really Cool. It is Art.

Photo 3. A Hot-Rod 6m SSB transmitter-Receiver in a Heathkit Q-Multiplier case, with styling cues from the E F Johnson Ranger. This radio takes 6m about as seriously as I do, but even non-hams think it is cute. After a decade of using an Eddystone Dial for tuning the home microwave station, I have reverted to analog, mechanical dials whenever practical. I use an analog mechanism to steer my car too. I won’t be replacing it with a computer anytime soon. I can’t quite put my finger on the appeal of this radio, but a quote from the poet Nelson Bentley comes to mind: “I have this sneaking suspicion that not everything is always happening in the present tense.”

4

My creative thoughts originate outside engineering. On my writing desk are collections of boat designs by the late Phil Bolger and poems by the late Nelson Bentley. The Hot Rod creations of the late Ed “Big Daddy” Roth inspired a generation of kids to question convention. A Hot Rod Radio inspires kids and old men to dig the old dusty shortwave set out of the garage and fill the notebook with ideas and sketches. That is enough. Sudoku for nerds. If it gets built, fine--but we develop technical problem solving skills as much by practicing the art of design as by cutting metal. Make ten sketches for every complete design, and design ten for every one you build. Build one a year. That has been my habit. This essay was inspired by friends who mess around with Radio Frequency Electronics, in particular my close collaborators for decades: Jim Davey, Wes Hayward and Merle Cox. Although some of the text has been gender-specific, I’d like to also acknowledge the influence on my work of two women as fluent with Smith Charts as any of my male professional colleagues: Allison Parent and Lorene Samoska. Their creations in the radio arts are Totally Cool.

5

Hot Rod Radio for 5760 Some Thoughts on Packaging Amateur Microwave Systems Rick Campbell KK7B

Photo 1. Classic-Deluxe 5760 Station, set up on the picnic table in the back yard at KK7B. The modular black-box 144 MHz rig on top of the SX-140mkII contains a T2 exciter, LM2 VXO system, R2pro receiver, and CDS Cell based Audio AGC system. The 1296 IF transverter is visible behind the microphone, on top of the 5760 transverter. For anyone experienced with HF operation using an old SX-140 HF receiver, this appears to be a rather marginal microwave station. Looks can be deceiving...

6

The only original equipment inside the SX-140 case is the slide rule dial and a set of air-variable VFO capacitors. A frequency compensated JFET Hartley VFO is inside the small black die-cast box. After a brief warmup, it drifts less than 100 Hz per hour, and tuning with the large dial is silky smooth.

Photo 2. Inside the SX-140 case, showing all the room available for IF converters. This set of projects was assembled after Jim Davey and I started kicking around the idea of Hot Rod Radios--gear that displays a bit of whimsy and more than a little Art. For this particular application, the SX-140 tunable IF is used as the receive portion of a microwave system with most of the microwave hardware at the feed point. The trasmitter is a VXO controlled phasing system operating directly on 144.1 MHz, with a second 144 MHz R2pro receiver slaved to the transmitter. A switch selects either receiver. For portable operation with space limitations, the SX-140 may be left at home, and the rest of the system is still fully operational. A coax relay selects the active receiver, controlled by one of the red switches on the front panel of the SX-140. The 144 MHz receive converter (a Kanga Rcx2 module not shown in this photo) is fastened to the top of the chassis with double-sided tape. Power is 12 volts DC, and power supplies are remote. Although there is space, mounting the power supply in the receiver limits flexibility and introduces hum in the sensitive audio electronics.

7

Photo 3. Under the chassis is a set of R2pro receiver modules, also available from Kanga US. After selecting the appropriate sideband using jumpers and tweaking the alignment, the receiver will work forever without adjustment. Other modules in the system are designed for similar set-and-forget utility. Reconfiguring the receiver for 6m, Straight Key Night on 40m, adding the 7 MHz SSB exciter etc. involves simply swapping modules above the chassis. For several decades the approach to microwave operation at KK7B has focused on portable operation with modular equipment. As a receiver designer, I have never been satisfied with the performance compromises in commercial VHF gear. After designing and building my first serious HF receivers in the early 1990s, that discomfort extended to HF equipment as well. I sold the Collins gear, relegated the Racal RA-6790/gm to the lab bench, and sketched receiver designs for my bands of interest. Although an HF receiver might not qualify as Microwave Gear, it really is the core of my microwave station. Filters, noise floor, gain distribution, dynamic range, and stability are all designed specifically to meet my weak-signal microwave IF needs. Once I have selected the receiver core, I add an assortment of modules to meet performance requirements on a particular band. One lesson I have to keep re-learning is not to put too many functions into any one module. This receiver gets used all the time, specifically because it is flexible.

8

Creative packaging, or Thinking Inside the Box. Some very interesting microwave paths are far from my home QTH in Portland Oregon...far enough that I might have to fly on a commercial airplane to get there. Since anything cylindrical and electric looks suspicious on a Committee for State Security X-Ray machine, it is better to ship the microwave gear by alternate means. After a decade of carefully packing and mailing my electronic gear, I realized that Flat-Rate Priority Mail boxes make nice transverter boxes. If the Flat-Rate box IS the radio, I don’t even have to unpack the gear.

Photo 4. An entirely conventional 1296 IF transverter. TUF-15 mixer on the left, a pair of 1296 Ace Hardware filters, two SMA relays, and two VNA-25 amplifiers. The DC electronics on the board convinces the latching relays to switch. It works, but not well enough for me to want anyone to duplicate the circuit. The Priority Mail box in the background is the packaged 5760 Transverter, all set to drop in the mail to Annapolis, MD. Not shown is a second Priorty Mail box with foam inserts that contains the 1296 transverter and 144 MHz VXO IF rig. For less than the price of a checked bag, the complete microwave station is waiting for me when I step off the plane.

9

Photo 5. Here’s the inside of the Medium Sized USPS Flat-Rate box containing the 5760 transverter and horn antenna. Some of the packing foam has been removed to take the photo. In normal operation the box is sealed shut, with only the 12 DC power and 1296 IF cable leaving through the upper left corner. The 17 dBi horn antenna shines out the lower right corner of the box, and may be used alone or to illuminate an offset dish or flyswatter. Out to 4 miles or so the barefoot horn is plenty. The 5760 transverter was as described in QST over 20 years ago, and stability enhancement was described last year at Microwave Update 2009.. Long ago I adopted the practice of building up complete portable stations for single bands, packaging them up in a cardboard box with a cover, and leaving them on a shelf in the garage. Some of them weren’t used for ten years, as my children navigated high school and then college. But having them boxed up all ready to go meant that I could grab a 3456.1 SSB-CW station off the shelf, toss it in the car trunk, and not even look at it until I opened the box on the deck overlooking Pickering Passage at the W7YOZ QTH. Then I discovered that I had forgotten the cable adapter to connect the antenna. The above system addresses that issue. This rig has been ideal for tracking down 5.7 GHz noise and interference sources.

10

After a decade of absense from the 6cm amateur band, it was a rude awakening to discover all of the interference from computers, cordless phones, and unmentionable personal devices. Even electric outboard motors communicate with on-board energy management systems using microwave links. Microwave operation on over-water paths accessible only by shallow draft boat is becoming more and more attractive. Some lakes and reservoirs discourage power boats with electronic ignition systems, and even some that don’t are large enough that it is possible to arrive at a location a mile from the nearest wall wart connected to a digital noise generator. Wind Power is The Latest Thing. Rather than convert it to electricity, I decided to use it to power my transportation. The fuel I save in the Honda Outboard could power my station, in principle. Here is a photo of KK7B heading out across Timothy Lake to scout a microwave site with a clear shot to the summit of Mount Hood.

Photo 6. The boat is home built, a Phil Bolger Nymph. The energy conversion system is my own design. She is not easy to sail, but she is a lot of fun--rather like making maritime mobile contacts on 5760.1 MHz. Photo by KB8FCZ. Operating on 5760.1 MHz SSB or CW from a small boat is interesting, to say the least. The Pacific Northwest has a plethora of knock-your-socks-off gorgeous over-water paths, some of them line-of-sight to mountain peak reflectors or well-equipped fixed stations. The bad news is that when the weather is that nice and you are out in a sailboat, the rig gets in the way... I’ve had a great time, but contacts have been sparse.

11

Photo 7. A Nautical 5760.1 SSB-CW station tucked into the stern sheets of a small Gaff rigged sloop. The technical term is “Darned Cute.” The rig sits in the stern of the boat, tucked under the tiller. A horn antenna fits in the mount for the boom crutch, leaving the boom and gaff-rigged main sail flopping around the other side of the boat. This needs more work. I designed the packaging before I added the wind power system to the boat. Also note that the Horn antenna is vertical polarized in this photo. The rig has been completely described in QST, and is as much a Classic as the SX-140. It has been working without adjustment, since it was first built in 1990. It is one of the few pieces of electronic equipment I own that stands up well to operation in a bumpy, wet portable environment. The two 5760 rigs described here have approximately 10 dB noise figure and 0 dBm power output, which is enough for marginal SSB or reliable CW over a 10 mile overwater path. In the maritime mobile environment, the wide beamwidth of the +17 dBi horn antenna is a necessary feature. If more gain is needed, it belongs at the fixed end of the path.

12

Photo 8. The view over the stern. Note the use of lashings instead of rigid fasteners. Everything flexes in a wooden boat, particularly anything you expect to be stiff. Bits of string and double-sided tape are usually superior to screws. The horn is also homebrew, and I wouldn’t build it the same way again. This is an old designer’s trick: provide just enough information to encourage someone to work the problem, but not enough that they will duplicate your mistakes. Sometimes it pays off and the next generation comes up with something much better. But you have to turn a deaf ear to the din of lesser talents demanding more construction detail. Most of the microwave bits are vintage KK7B no-tune transverters. They work well at sea-level, line-of-sight to another station, and miles away from the nearest RFI plagued urban hilltop. More recent gear handles the proliferation of commercial microwave energy better than quarter-century old first generation MMIC and bare PC board technology. Plans for the boat, a Phil Bolger designed Nymph are available from Phil Bolger and Friends. The wooden boat community is as focused and skilled as the amateur microwave community. Bob Larkin W7PUA also built and sails a Phil Bolger designed boat, the much larger Birdwatcher. You can find details on the web.

13

Photo 9. A detail shot of the maritime mobile 5760.1 SSB-CW station. This station probably qualifies as Vintage Gear. The gray and black box in the lower left corner contains the original prototype KK7B 1296 No-TuneTransverter, as pictured on the cover of April 1993 QST and featured in several ARRL Handbooks. Invisible on top of the 1296 IF transverter is the original 5760 No-Tune Transverter, featured in QST October 1990. The gray and black box with all the switches and lights houses the original prototypes of the R2 and T2 rigs from January and March 1993 QST. Maybe this stuff should be in a museum instead of getting wet in the back of a boat. Just visible to the right of the premixed 144 MHz VFO (April 1993 QST) is the Navy Knob on an EF Johnson Key. I haven’t touched this stuff since it was built, and it still meets all the original specs. Few pieces of commercial gear from the early 1990s can make that claim. This has been a rather light-hearted look at microwave station packaging, with a bit of Pacific Northwest whimsy and art tossed into the mix. As amateurs, we have the luxury of following a different path--enclosing our microwave gear in a free cardboard box or a unique wooden sailboat. If you get bored with contests and grid collecting, try a quirky station package. These have been a lot of fun, and generated considerable interest.

14

         
   ­€‚ƒ „

…†  ‡ ˆ ‰Š Š  … ˆ ˆ  … …   Š ‹ Œ 

†‰ …    Ž ˆ ‘ ’ ˆŠ“ Ž“†Šˆ“”•–—––ˆ …  ˆ ‹  ˆ  †Ž  †ˆ …

ˆ †   …†

    † ‘ „“   …  ˆ˜     ˜“ ‘  ˆ ‹˜ Ž ‹‹˜ … ‹ Ž    Ž …  ˆ    ˜†…“  ˆ † “   ‰  …‹…†    ˆ“Ž  ‹ …Ž ˆ ˜   ˜˜    ‹ ˆ“ ˜ 

˜†…“Š “   ‹‰ …  ˜ “  ‘™…‹  ‚… ‰ ˜˜‹† ˜ … ˆ …    

†  Ž š†Š   …„Š ˜†˜    “ ˆ … ‰••”••–•–•••  †˜˜“  ˆ    Š ˆ … ˜ Š ‘ „‹ †…ˆ“  “  Š˜ ‹ ˆ‹  ”•—–… ‰   ŠŠ †  ˜ ‰‹   “ ˆ Ž †˜…… Š  ‘„–• †Š … Š Š˜ Žˆ˜   Ž  “˜† …  ˆ ‹˜  ‰  ˜   ˆ ‹ €† ‹ ›  ‘‚ ”•œ–‰

‰… “  ‹ ˆˆ  

   Žˆ ˆ  ˜…

 …  ˆ  †˜…“   „ ž‹  Ž „›‘„ ˜˜ˆ†  ˆ    …  ˆ ‹ †…… ˜… ‹‹…˜ Š 

ˆ “ˆ   †  … Ž ‰ ‹ ‘ „ˆ…ˆ  †    ‹  ˜ š†Š  ”•—– ˆ‰ †ŠŠ  ˜ ˜˜†   •–‘‹ Ž … Ž Š †˜ ˜

    ˜ Ž ˆ Ž  ‰ ˜ •– •–•”• –”• –Ÿ–¡ „˜ ‰‰   ŽŠ  ‹

ˆ 

  Ž  ˆ …Š Ž  ˜ š†˜   ˆ    … … ˜ …† Ž ˆ ‰ ˆ ‹ †˜   ‹ ˆ   ……   Š ‘ †  †  ˜  ˆ ˆ  Š Ž †…  ˆ ˆ … ‹ ”•””–  –”•
”•••”•–•–
”• ˜    ‹ ˆ  

˜ ˆ … ˜•
 Ÿ•”•• –•••–––Ÿ–
 ”•”” ˆ †  Ž     Ž †˜…   †˜˜Š Š     “   ‹ ˜  ˆ  ˆ   ˜  Š     Žˆ  ˆ  ˜    †˜…ˆ ˜

˜ 

† † ‘†“ †ŠŽ ˆ ˆ ‹ ˜˜ Ž“ “  ‡  ‡›   ˆ“ˆ ˜†… ˜ ““ š†Š  ‹ Žˆˆ„ˆ  ˜

    ••• ••Ÿ•• –••”•”Ÿ” •––”–•” ˆ   ‹ “ …… “‘ ˜ ˆ ‚  ˆ“ˆ ˆ “ ‘ ¢†

“ †‡ Ž 

  “ˆ  ¢  ŠŠ £  ˆ‰ Ž Žˆ  ”• ••–‘ “ˆ ˜˜“‡   ˜… ˆ       š†Š  ˜‰ › †  †˜…‰  “ …  “Š  ‹  ˜ … ‘¤†  ˆ … …†  ˆ ˜‰ ˆ˜…  †˜ …˜‘€ 

15

           
   ­­€‚ƒ  ‚
  „…  †‡­­…ˆ‚  ‰Š… ˆ   ‹   †Œ  Œ  ‹„    Š Œ      ‚‡Ž‘‘Ž’‘ŽŽŽ‘Ž‘ŽŽ ‘“

†‘Ž Ž

Š”  Ž‘Ž•Ž Ž   ‹„ ‚       – „­‡  –   —˜ ‘•‘‘ ••Ž Ž‘Ž ŽŽ
‹      –    –  ‚  •ŽŽ‘ –    ŽŽ•Ž ’Ž
  – –      ƒŒ   – ­             ˆ         •†­ŽŽŽ ‘™Ž    

16

                           


         
     
   
         
             
  
  
­        
  


         

           
     

  

               
             €‚ ƒ
        „‚ …†‡‡ˆ  ­    €­  
   
    ‰   
      ˆ     ­Š‹      

†ŒŽ ‘
       

17

             
 
      ­€‚ƒ„ …†‡ˆ‰‰‡‡‰ ‡‰ ‰‰‰Š‹‹  ‰†‰…  ‰Œ…‡‰‰‡
ˆ‰††‡‰‰‰ ‰ˆ‰ € Ž‘’  
   “   ”‡ˆ‡‰‡ˆ‰‡‰•† ‡ ’ ”  “Œ   

    €‚‡‰
 –—‘ ˜ € €‰ ‡ ‡‰™‰‡ˆ‰‰   ‘

 ­ ‘€‘     ‘’

€”š›
     Œ    

  
  “­™ … œ    € €ˆ‡‰€‰‚‹ ‰‰™„ Œž  €

 
 €€  €Ÿ        

€€“—–—  “¡—€˜   €    ‘    € ™ ˜ Ÿ  €Ÿ        €Ÿ     ‘         €˜      “€ —       Ž €    ˜„™ƒ € 

18

            
        
­€‚ƒ „   …†‡ˆ‚     †       
€‚ƒ ‰ ‡Š          
€‚ƒ ‹Œ ‡ ­Ž       ‘ Ž  ‡                ‡     …  ‡‚      ­ ’‡   ‡     “          ƒ      …†     …†„‹Š    ”­ŽŠ    ‰     ˆŠ“‡„‹Š   ‹“ •

      
        ­  €    €

19

          

    

) !&!"&+2&!3 &! $$           =) &" )& &'" &!& >81­ $&!         !! 1&&"&&!2'&+3&)& ­$+!($"$+- !  )& $        = €    ‚= !& <81 ƒ81  - && & &'$  € „  !'1$­ &. +&$!&     !$ !)- 1(!& &"&&! . …- $+! 1&" & ­.  ‚    = „ ‚†‡
     „ ˆ   ˆ     ‚% ‰    ‡Š
  ‹  %ŒŽ     

20

                  

 
        

21

 

How to Increase 23 cm Power to 250W with 2 x XRF 286 Some Modifications to the W6PQL kit By Dominique HB9BBD  

Observations • The PQL layout may be ideal for some XRF286 but was not for mine • Unfortunately matched pairs cannot be figured out because of soldering/desoldering issues • The board did not fit to some used transistors I got so I had to fit the boards to the transistors • I built 10 double boards (pcb version 7.2) and try to summarize the findings  

22 1 

 

At a Glance – Visible Imbalances Thermal analysis of the original board at 94 W out (due to imbalance of hybrid at input, one transistor seems to be driven harder?)

Output capacitor is thermally the hottest part Difference in outputs results as unbalanced power in the termination  

Addition of Matching Flaps Gives Better Output & Heat Distribution Now at 10W in 180 W out

In

First try of improvement by mobile flaps

The hybrid is still lossy but now symmetrically loaded Out

Less power loss at 50 Ohm load!!!  

23 2 

 

Parts to be Replaced • Replace the tiny ceramic trimmer TC1 and TC2 with High Q piston capacitor 1-5pF (muRata)* or similar It offers higher Q, thus rewards with less loss • Replace C2 and C12 by ATC 800A 22pF *Thanks

to Mike, JH1KRC I have these on my bench..!

 

How to find out what to do? As mentioned earlier, each transistor is unique in it‘s capacitance etc. This is why I recommend you find out what is needed on your Board, by doing the following: • Prepare a teflon stick with little flap at it‘s end • At reduced drive (5W) carefully position the flap to the board (without shorting DC !) • At position where positive effect is detected, lay down a flap and so on • Once no further improvement can be seen, solder all mobile flaps to the board • Beware of excessive RF exposure to your eyes, man! Remain at max. distance from the amp when unshielded. Avoid long periods of exposure.  

24 3 

 

Observations • The PQL layout may be ideal for some XRF286 but was not for mine • Unfortunately matched pairs cannot be figured out because of soldering/desoldering issues • The board did not fit to some used transistors I got so I had to fit the boards to the transistors • I built 10 double boards (pcb version 7.2) and try to summarize the findings  

These Flaps Helped a lot to Balance the Amplifier in Power and Phase

Input termination Output termination muRata 1-5pF

 

25 4 

 

These Flaps Helped a lot to Balance the Amplifier in Power and Phase • Output lines remarkably different.. • Due to high current, generously solder the Drain & Gate to the board! Thicker is better here

 

These Flaps Helped a lot to Balance the Amplifier in Power and Phase • The hybrid also needed some mods (Note C2 still original size, Changed Later)

 

26 5 

 

Final Results • At an input power of 14W the gain is 14dB • At 250W Output and 28VDC 16A, key down for 1 minute is possible with moderate heatsink without fan • Enjoy these fine transistors and call me off the moon! • 73 HB9BBD Dominique JN47ee

 

27 6 

2010 observations on phase noise from local oscillator strings. By KØCQ 10 MHz reference oscillators have improved. Now OCXO (Oven Controlled Crystal Oscillator) are available that will do nearly as good as a rubidium oscillator for frequency stability and better for phase noise. In several recent magazines (High Frequency Electronics, MicroWaves and RF, Military Microwaves 2010, and Microwave Journal, July and August 2010 issues) Vectron (www.vectron.com) advertises an OCXO that has phase noise at -165 dBc/Hz at 10 kHz offset and a 10 MHz oscillator. Similarly, Valpey Fisher Corp offers their VFOV600 OCXO with -165 dBc/Hz at 10 kHz offset. Details on line at www.valpeyfisher.com. Going higher in frequency, Holzworth (www.holzworth.com) offers a synthesizer rated -151 dBc/Hz for 10 kHz offset at 100 MHz carrier frequency. Pascall (www.pascall.co.uk) advertised a 100 MHz oscillator with phase noise -178 dB/c at 10 kHz spacing. That's clean! -182 dBc/Hz at 100 kHz offset. 20 dB quieter than the rest of the world of oscillators. Probably a variation on the Driscoll circuit.. At 1 GHz, Synergy (www.synergymwave.com) offers a VCO with -170 dBc/Hz phase noise at 100 kHz and greater spacings. A low noise floor. They call it Ultra Low Noise. At 10 GHz, Microwave Lambda Wireless (www.microlambdawireless.com) offers a YIG oscillator that free running does -105 dBc/Hz at 10 kHz offset. Giga-tronics (www.gigatronics.com) offers their model SNP synthesizer giving -115 dBc/Hz at 10 kHz offset also at 10 GHz carrier frequency.. Miteq (www.miteq.com) offers their model DLCRO synthesizer with phase noise -115 dB/c at 10 kHz offset, also at 10 GHz carrier frequency.. Nexyn (www.nexyn.com) offers a phase locked DRO (Dielectric Resonator Oscillator) with -120 dBc/Hz at 10 kHz offset, also at 10 GHz carrier frequency. Phase Matrix (www.phasematrix.com) offers their QuickSyn synthesizer that shows phase noise 121 dBc/Hz at 10 kHz offset and 10 GHz carrier frequency. Wenzel (www.wenzel.com) offers an oscillator multiplier chain, their model MXO-10000-33, with 129 dBc/Hz phase noise at 10 kHz offset and 10 GHz output. Oewaves (www.oewaves.com) using an electro-optical oscillator claims -145 dBc/Hz phase noise for 10 kHz offset and 10 GHz output. This is new technology using a modulated light wave in a length of fiber optic, then detecting the RF modulation. The longer fiber optic, the better the stability and phase noise because the more rapid the phase change with frequency. And the fiber has much less loss than coax or waveguide. Hittite (www.hittite.com) offers a PLL + VCO that does -100 dBc/Hz for 10 kHz offset and 11.5 – 12.5 GHz

28

Herley (www.herley.com) didn't advertise a specific product in these magazines, but their PDRO line of military oscillator multipliers shows -120 dBc/Hz at 10 kHz offset for a carrier frequency of 13.5 GHz. The phase locked DRO is a good and not expensive technique because the DRO has good phase noise by itself. The August 2010 issue of Microwaves & RF has an article titled “Shrinking Sources Aim for Lower Noise” that begins on page 29. On line its link is http://www.mwrf.com/Articles/ArticleID/22869/22869.html Its a survey of some other oscillators that weren't advertised in these magazines. Several recent years of articles are on line at www.mwrf.com and free. Similarly, Microwave Journal, one of the oldest microwave publications is on line at www.mwjournal.com with many back issues as well as the articles from the current issue. As is High Frequency Electronics at www.highfrequencyelectronics.com. Sometimes these industry journals have articles that are truly state of the art and sometimes they show research done so far from mainstream that the state of the article is 25 years poorer than common ham gear performance.

29

Safe Tapping in Soft Metals By KØCQ Tapping small holes, like 4-40 in aluminum often leads to the nightmare of the broken tap. Copper is worse. Conventional wisdom says use a tap lubricant, and back up the tap often to break off the chip so it can't bind in the flutes. Tap Magic from the Steco Corporation has a good reputation. I have a bottle, but I've not yet tested it. Mistic Metal Mover now in version II from Mistic Metal Mover, Princeton, IL also has a great reputation. I've not yet tested it either. I found it at my local tool and fastener store. In steel I've been using pipe threading oil, and for aluminum the text books say kerosene is good. Probably we don't back up often enough, probably backing up a full turn of the tap for each quarter turn of cutting isn't too often. Even then, in gooey copper or 4003 aluminum, the chip doesn't always break off. I'm of the opinion that the tap drill charts are incorrect. The conventional tap drill charts drill a hole the size to cause the tap to cut 75% thread depth. This is a fine size for tapping cast iron and brass, especially free machining brass where there is no extrusion of the thread. But in soft steel, most aluminum and copper alloys, I believe there is significant extrusion of the thread leading to the tap binding on the threads, not from chips. For half a century I've followed a simple rule of using a tap drill 3 or 4 numbers (when using number drills) larger than the tap tables. Now I've decided that using a 50% tap drill from the few tap drill charts that show those sizes is better. The larger drill does a couple things, first it reduces the depth of cut, and then it leaves more room for the extruded part of the thread. It also significantly reduces the torque required for tapping. Is it strong enough? Some references admit that if the tapped hole deeper than the major diameter of the screw that the threads probably won't strip. Many times we are tapping a 4-40 thread in a 3/4” thick slab of aluminum. There will be more than an adequate number of threads. Back in the 1980s when the Story County ARC mounted a Cablewave Station Master antenna on a water tower, I sent a 50% tap drill up the water tower. The ham on top thought the tap sure turned easy. I had assembled a three legged T to be bolted to the top sheet of the water tank to hold that 20+ foot antenna. I think it had 2 bolts per leg, 3/8-24. They would be still holding if the antenna wasn't taken down for tower painting.

30

tap size (major dia. threads / inch) #0-80 #1-64 #2-56 #3-48 #4-40 #5-40 #6-32 #8-32 #10-24 #12-24 1/4-20 5/16-18 3/8-16

tap size (major dia. threads / inch) #1-72 #2-64 #3-56 #4-48 #5-44 #6-40 #8-36 #10-32 #12-28 1/4-28 5/16-24 3/8-24

Some tap drill data: Coarse Threads - English screw tap drill size tap drill size clearance drill major dia. for 75% .dia for 50% .dia 0.060 0.073 0.086 0.099 0.112 0.125 0.138 0.164 0.190 0.216 .2500 .3125 .3750

3/64 (.0469) 53 (.0595) 50 (.0700) 47 (.0785) 43 (.0890) 38 (.1015) 36 (.1065) 29 (.1360) 25 (.1495) 16 (.1770) 7 (.2010) F (.2570) 5/16 (.3125)

55 (.0520) 1/16 (.0625) 49 (.0730) 44 (.0860) 41 (.0960) 7/64 (.1094) 32 (.1160) 27 (.1440) 20 (.1610) 12 (.1890) 7/32 (.2188) J (.2770) Q (.3320)

50 (.0700) 46 (.0810) 41 (.0960) 35 (.1100) 30 (.1285) 29 (.1360) 25 (.1495) 16 (.1770) 7 (.2010) 1 (.2280) H (.2660) Q (.3320) X (.3970)

Fine Threads - English screw tap drill size tap drill size clearance drill major dia. for 75% .dia for 50% .dia 0.073 0.086 0.099 0.112 0.125 0.138 0.164 0.190 0.216 .2500 .3125 .3750

53 (.0595) 50 (.0700) 45 (.0820) 42 (.0935) 37 (.1040) 33 (.1130) 29 (.1360) 21 (.1590) 14 (.1820) 3 (.2130) I (.2720) Q (.3320)

52 (.0635) 48 (.0760) 43 (.0890) 40 (.0980) 35 (.1100) 31 (.1200) 26 (.1470) 18 (.1695) 10 (.1935) 1 (.2280) 9/32 (.2812) S (.3480)

46 (.0810) 41 (.0960) 35 (.1100) 30 (.1285) 29 (.1360) 25 (.1495) 16 (.1770) 7 (.2010) 1 (.2280) H (.2660) Q (.3320) X (.3970)

These came from a web page by Curious Inventor L.L.C. They included metric tables but only for 75% threads. I've read that concentrated nitric acid will remove the remains of steel taps from aluminum. I've not tried

31

it, remembering how the fumes from nitric acid turned chemistry labs to rust even with the containers closed. In my library the machinist handbooks differ in their opinions on tap drill sizes. On page 1381 of the wartime supplement section of my American Machinists' Handbook, Eighth edition, copyright 1945, it says:

Update: Between writing this for CSVHF and the conference, I did some experiments to prove (?) my point. I bought some really soft aluminum alloy 1100. I cut two sample pieces and clamped them together in the milling machine vice. I milled the top perfectly flat and then drilled and tapped. The first hole on the right in this picture I drilled dry. This is 4-40 thread and that right hand hole is with a 75% tap drill. The drill plugged up with aluminum which had to be pried out of the flutes several times. Then I tapped dry backing up a full turn for each quarter turn forward. By the time I was 7 or 8 turns in, the tap was binding and springing as if I could snap it off. So I shot it with a handy lubricant and it finished the full depth of the tap cleanly without any more binding. The hole on the left was drilled with a 50% tap drill and tapped with that same lubricant. You can see the shallower thread, but the absence of hogging out the hole and threads.

32

Then as a check I did a lubricated drill 75% and tap in this picture. Its neat and nice, full length of the 4-40 tap with no torn threads. That lubricant was a JD universal spray “oil” their number TY6350. Its almost as oily as WD-40 and not much of a lubricant for the application I purchased it for. Its MSDS says it contains a carcinogen and several components hazardous to your health whether breathed or absorbed through the skin. That Mistic Metal Mover Company (1160 N. Sixth St, Princeton, IL 61356) makes a product they call Alumicut. Its MSDS lists no ingredients, just says none are toxic. At CSVHF 2010, several bragged on the high quality of Mistic Metal Mover but none had tried Alumicut. I have a can but I've not tried it yet. When tapping small sizes, its a benefit to the tap to remove the cross bar on the tap wrench. There's plenty torque available with thumb and forefinger on the collet nut, twisting with the cross bar just snaps the tap. In model making circles they often use a tap guide. Sometimes its a little miniature hand spun drill press to guide the tap wrench and tap, some times its a lump of metal with a hole perpendicular to the base with turning clearance for the shank of the tap. It can be turned on a lathe. Its held against the metal and keeps the tap from being bent or from being started crooked. Its likely with small taps that bending or cutting crooked is as much a cause of breaking as of drilling to small a hole to begin with. Its a good for the work side of the guide to be opened up to allow for burs and rises of the metal at the threads. I often hold a shaft in the chuck on my milling machine that fits in the back of the tap wrench to hold it straight while starting the tap. I still turn the tap by hand, not by power. Another hint: I'm finding the 135º split point drills are much nicer for all drilling applications. Their point doesn't wander and even large drills like 3/4” in 3/4” thick steel don't need pilot holes. A couple years back drilling 21/32” holes in 1/2” and 3/4” steel plate I drilled more than 50 holes with the same drill before it needed a touch up on the cutting edges. The first 40 holes were for 5/8” bolts to mate after market rims with JD cast wheel centers and the 1/32” diameter tolerance was enough that all 40 lined up on assembly. 1/32” diameter was typical clearance used at Collins in the early 1960s pretty much no matter what the bolt size. When the holes were made with numerically controlled machines, that was probably excessive clearance, but when we drill by hand without a drill press or milling machine, that may not be enough clearance, but that's what rat tailed files were made for, adjusting hole positions. Otherwise called the “tolerance tool.” There are several ways to grind that split point and they all seem to work. Ace Hardware stores now carry them so they are not hard to find. Most of mine were bought from McMaster-Carr. For sheet metal work, I much prefer my Whitney number 5 junior and number XX hole punches. I also have a number 2 and a number 10 for steel work.

33

Taming Phase Noise at EHF -Brian Justin, WA1ZMS/4

1) Abstract Besides generating real RF power or establishing a low receiver noise figure, one of the fundamental hurdles of mm-Wave operation is creating a local oscillator chain with acceptable low phase noise. This paper will focus on the impact to phase noise that is a direct result of frequency multiplication in the LO chain. General background will be reviewed and a specific example of LO noise at 78GHz will be reviewed. 2) Background For the purposes of this paper, phase noise is defined as incidental FM modulation of a local oscillator signal. When defining or specifying phase noise the amount of noise is commonly described as a signal level in dB below the carrier power which is measured in a given bandwidth at a specific offset frequency from the carrier. That definition can be intimidating to some but it is the only way to quantify phase noise of a signal when comparisons between signal sources are being made. As an example, a given oscillator at a frequency of 10MHz might have a table of specified phase noise such as: Offset Noise 100Hz

125dBc/Hz

1KHz

140dBc/Hz

10KHz

147dBc/Hz

Table 1 – Example of phase noise specifications for a hypothetical 10MHz crystal reference oscillator. The question of whether or not this oscillator is “good enough” cannot be directly answered until the final LO frequency has been arrived at by frequency multiplication of this example 10MHz oscillator, but more on that later. 3) Minimum acceptable noise Before any oscillator specifications can be determined to be acceptable, one must first define what level of phase noise at a specific offset frequency can be tolerated by the type of modulation used in a communications link. Perhaps fortunately, most all of the

34

modulations used by amateurs on the mm-wave bands are either CW or SSB modes that all occupy bandwidths below a kHz or two. This fact helps us focus on the region of phase noise within 1 or 2 KHz of the carrier frequency. This region of interest will help determine how well a signal “sounds” particularly when it comes to the quality of the CW note being heard. Phase noise that is much farther away from the carrier (i.e.: 100 KHz) has no direct impact on the tonal quality of a CW signal but can limit the receiver’s dynamic range by causing intermodulation products and reciprocal mixing of noise. But this is often not an issue on the mm-Wave bands until such time that bands become as crowded as 20 meters might be on a contest weekend! The author’s presentation at Microwave Update 2009 included the results of empirical phase noise tests and the resulting effect on signal quality and demodulation of CW by ear.[1] The key parameter that was the result of those tests was a metric for phase noise of a CW signal where the signal quality was just starting to degrade the “CW note”. A value of -72dBc/Hz @ 1KHz offset was determined to just cause a perceptible degradation of the CW carrier quality. The tests also showed that if one is willing to give-up about 2dB worth of minimum detectable sensitivity (MDS), then the same carrier could have even poorer phase noise of -48dBc/Hz @ 1KHz offset. While the limiting range of phase noise anywhere between -72 and -48dBc/Hz @1KHz (a 24dB window) is rather large, it does give us some bounds to begin to explore what range quality of phase noise one would need for CW work on the mm-Wave bands. For the examples used in the paper, the target goal of -72dBc/Hz @ 1kHz will be used. In this way we allow for as much noise as we can, yet obtain a clean signal and not give up any MDS performance. 4) Main sources of LO noise When considering the impact that phase noise can have on mm-Wave communications system, one must not only be concerned about RX LO phase noise but also the phase noise of the transmitted signal that one is receiving. In such a communications system, the noise of both the TX and the RX LO contribute to the total noise of the received carrier. Using the value of -72dBc/Hz @ 1kHz from above, that should be the goal for both the TX LO and RX LO chains of both stations. Since phase noise is noise power, the composite of both noise sources will add to degrade the final signal as heard in the IF radio. Therefore in a purest sense, the LO chains of both the RX and TX should be 3dB better to arrive at a final value of -72dBc/Hz @ 1kHz. But since -72dBc is a rather conservative value, little harm to the signal will take place if both the RX and TX LO chains deliver such noise specifications. Regardless of which LO chain we are designing (TX or RX) the phase noise at the final and highest frequency in the chain must meet our phase noise goal of -72dBc/Hz @ 1kHz. With that phase noise goal in mind, we can now work backwards through our LO chain in order for us to determine just how “good” our base frequency reference must be to ensure that our final mm-Wave carrier meets our minimum specification.

35

In the simplest of calculations, assuming we are starting with a 10MHz crystal oscillator and our final carrier frequency is to be 78,000MHz. We can use the equation below to calculate what the phase noise of our 10MHz reference must be. dB = 20Log(N), where N is the ratio of input to output frequency of the stage. In this example: dB = 20* Log (78000MHz/10MHz) dB = 20 *Log(7800) dB= 20 * (3.892) dB= 77.8dB The value of 77.8dB is how much better our phase noise of the 10MHz reference must be than our desired carrier phase noise of -72dBc/Hz @ 1kHz. The result requires the 10MHz crystal reference oscillator to deliver -149.8dBc/Hz @ 1kHz. In contrast, if 2dB of MDS performance is willing to be sacrificed, then the 10MHz reference need only be as good as -125.8dBc/Hz @ 1kHz. It should be noted that there is a difference in between the two phase noise values of some 24dB. Since the effect of phase noise on the MDS of a CW signal is really a result total integrated noise, the range of 24dB gives moderate leverage in the quality of the reference oscillator used. Notice that in all of these calculations, no mention is made of the topology of the LO chain whether it be direct multiplication, direct frequency synthesis, or phase locked loop. Each method has its inherent benefits and drawbacks but when looking at the LO chain as a whole, the close-in or phase noise at 1KHz offset follows the 20Log(n) rule. Noise at offsets of 10’s or 100’s of kHz are often negatively impacted with LO chain topologies that include phase locked loops. The next section of this paper will look at that specific case and how it pertains to the LO chain. 5) The Unique PLL case When a phase locked loop or PLL is used as part of an LO chain it becomes imperative that the designer know what the closed loop bandwidth (BW) of the PLL is. Here is why: for within the bandwidth of the loop, the phase noise of the PLL’s output directly follows the phase noise of the reference signal by the factor of 20Log(n). Right at the frequency of the BW of the loop, the phase noise is a complex combination of both reference noise as well as the noise of the voltage controlled oscillator (VCO) if it were left to free-run at the desired output frequency. For frequency offsets greater than the loop BW, the phase noise is dominated by the noise of the VCO if it were free running on its own. A simple design rule must be followed if your mm-Wave LO chain includes a PLL. The loop bandwidth must be greater than about 5KHz if your goal is to control phase noise at our previously assumed offset of 1kHz. The value of >5kHz will help insure that all of the resulting noise at the output frequency of the PLL is truly a function of 20Log(n) of the applied reference frequency.

36

Another fortunate outcome of using the ever common Frequency West or California Microwave PLL blocks as part of a mm-Wave LO chain is that their closed loop bandwidths are often several 10’s of kHz. This is very helpful in that it can be safely presumed that all of the phase noise at the output frequency of the PLL block for many kHz anyway from the carrier is directly following the 20Log(n) scaling factor. 6) 78GHz Example In this section an example LO chain for 78GHz is presented as shown in Figure 1 below. The figure denotes the overall LO chain as the reference oscillator (in this case, a 10MHz OCXO) signal is multiplied and increased in frequency towards the final LO frequency of 78.000GHz.

Figure 1 – Example of 78.000GHz LO chain showing the degradation of phase noise from each multiplier stage.

As the frequency is increased by a given multiplier stage, the effective increase in dB of phase noise is noted. For example, in the first multiplier stage the 10MHz reference is directly multiplied to 50MHz. This results in a phase noise degradation of 13.97dB which is a result of the 20Log (n) formula. As the LO frequency increases with each successive multiplier stage the phase noise continues to degrade. The result of this particular LO chain starting at 10MHz and progressing to 78.000GHz has a total of impact of 77.84dB on the phase noise of the frequency reference. This means that the phase noise of the 10MHz reference oscillator must be 77.84dB better than our minimum signal specification of -72dBc/Hz @ 1KHz offset. The required phase noise of the reference must then be (72 + 77.84) or -149.84dBc/Hz @ 1KHz. In the above case, the Frequency West PLL assembly has no worse impact on the phase noise of the LO so long as the bandwidth of the PLL is greater than the frequency offset range that we are interested in. Since our target offset frequency of interest is 1KHz (as

37

noted in section 3) and the PLL bandwidth (as noted in section 5) is at least an order of magnitude greater, there is no concern and the PLL will act just like a direct multiplier.

7) Conclusions From the example above it can be concluded that any 10MHz reference oscillator that has a phase noise specification between -149.84dBc/Hz and -125.84dBc/Hz at a 1kHz offset frequency would result in a 78GHz radio that has between 0 and 2dB of MDS impact respectively. It can also be concluded that the theoretical crystal oscillator specified in Table 1 above would in fact be quite usable as a reference oscillator in a 78GHz station. This oscillator would result in almost no detectable impact to ear-copy of a CW signal on the band. Keep in mind that narrower bandwidth modes (PSK-31, WSJT, QRSS, etc.) will require even lower values of phase noise and are purely dependant on the particular modulation mode in question. 8) References [1] – B. Justin, WA1ZMS, Microwave Update presentation, Dallas, TX, 2009.

38

Ka-Band Integrated-Circuit Interferometer for Sensing Seok-Tae Kim and Cam Nguyen Department of Electrical and Computer Engineering Texas A&M University College Station, TX 77843 Abstract A multi-function millimeter-wave integrated-circuit sensor operating at 35.6 GHz has been developed and demonstrated for monitoring of displacement and low velocity. Measured displacement results show an unprecedented resolution of only 10 m, approximately equivalent to 0/840 in terms of free-space wavelength 0, with a maximum error of only 27 m. The sensor can measure speed as low as 27.7 mm/s, corresponding to 6.6 Hz in Doppler frequency, with an estimated velocity resolution of 2.7 mm/s. 1. Introduction Microwave and millimeter-wave interferometry has been widely used for various applications such as position sensing [1], velocity profile [2]-[3], and displacement measurement [3]-[4]. Interferometry is basically a phase-sensitive detection process, capable of resolving any measured physical quantity within a fraction of the operating wavelength. Interferometric sensors also have relatively faster system response time than other sensors due to the fact that they are generally operated with single-frequency sources. Millimeter-wave interferometer is thus an attractive instrument for various engineering applications requiring fine resolution and fast response. In this paper, we report on the development of a multi-function millimeter-wave integrated-circuit sensor capable of measuring both displacement and velocity (particularly low velocity), based on phase detection, for potential industrial applications. For displacement sensing, the sensor achieves a resolution and maximum error of only 10 and 27 m at 35.6 GHz, respectively. The attained resolution, approximately equal to 0/840, is the best reported resolution in terms of wavelength. The sensor can measure speed as low as 27.7 mm/s, corresponding to 6.6 Hz in Doppler frequency, with an estimated velocity resolution of 2.7 mm/s. 2. System Principle The overall system configuration is shown in Fig. 1. The system is divided into three parts: a millimeterwave subsystem for processing millimeter-wave signal, an intermediate-signal subsystem for processing signals at intermediate frequencies, and a digital signal processor. The sensor transmits a millimeter-wave signal toward a target. The signal reflected from the target is captured and directed to the receiver, and down-converted to a low-frequency signal, namely the measurement-channel signal vM(t), which contains information on the phase or phase change over time generated by the target displacement or movement, respectively. For displacement measurement, the measured phase of vM(t) is compared with that of the reference-channel signal, vR(t), coming from the direct digital synthesizer (DDS). If the target is in motion, the frequency of vM(t) is shifted by the Doppler frequency. In velocity measurement, the phase change over time is detected in the signal processing and only measurement-channel signal is processed to extract the Doppler frequency shift. The sensor’s signal processing consists of two distinct parts: one for detecting the phase difference needed for measuring the displacement and another one for estimating the Doppler frequency used for calculating the velocity.

39

PLO-1

1.5m fEXT=17.8 GHz

× 2 Frequency Doubler

Directional Lens Horn Coupler Antenna fC= 35.6 GHz

PA Power Amp.

Target

XYZ axis Stage

Power Divider

Conveyor

Down Converter

LNA Up Converter

BPF LNA

RF_OUT

MMW Subsystem

RF_IN

AMP

AMP

PLO-2 fIF1=1.8GHz

Fig. 1 Overall system block diagram. The target sits either on the XYZ axis (for displacement sensing) or on the conveyor (for velocity measurement). The Reference Channel is not needed for velocity measurement.

Mea. Ch. AMP

Digital Signal Processor

Down Converter AMP

Quadrature Up Converter

Intermediate Subsystem

Ref. Ch.

DDS fIF2= 50 kHz

3. Fabrication and Test

(a)

(b)

Fig. 2 Photograph of the fabricated millimeter-wave (a) and intermediate-signal (b) subsystems. The millimeter-wave and intermediate-signal subsystems, shown in Fig. 2, were realized using both MICs and MMICs. We have tested the developed sensor for measuring the displacement of a metal plate mounted on a XYZ axis stage. Fig. 3 shows the measured displacement along with error. The result indicates that a resolution of only 10 m, equivalent to about 0/840, is attained. We have tested the velocity of a closing metal-plate target. The experimental results are shown in Fig. 4. The average measured velocities are 27.7, 32.6 and 38.6 mm/s. The corresponding standard deviation of the Doppler frequency estimates are inferred as 0.50, 0.61 and 0.64 Hz, respectively.

40

0.008

16

45

0.007

15

40

0.25

14

0.005 0.004

0.15 0.003 0.10

Error (mm)

Measured (mm)

0.20

0.002

0.05

0.10

0.15

0.20

0.25

30

12 11

25

10

20

9

15

8 7

0.000

6

-0.001 0.30

35

13

0.001 0.05

0.00 0.00

Doppler frequency (Hz)

0.006

10 5

5

0 1

2

Displacement (mm)

Fig. 3 Measured displacement every 10 m.

Velocity (mm/s)

0.30

3

4

5

Measurement index

Fig. 4 Measured velocity of a closing target. 4. Conclusion

A multi-function millimeter-wave integrated-circuit sensor operating at 35.6 GHz has been developed and demonstrated for displacement sensing, with micron resolution and accuracy, and for high-resolution lowvelocity measurement. Displacement measurement results indicate that the sensor can resolve displacement within 10 m or 0/840, which represents the best-reported resolution in terms of wavelength in the millimeter wave range. Velocity as low as 27.7 mm/s, equivalent to 6.6 Hz in terms of Doppler frequency, has been measured at 35.6 GHz for a moving target. The developed sensor demonstrates that displacement sensing with micron resolution and accuracy and high-resolution lowvelocity measurement are feasible using millimeter-wave interferometer, which is attractive not only for displacement and velocity measurement, but also for other industrial sensing applications requiring very fine resolution and accuracy. Acknowledgement This work was supported in part by the National Science Foundation and in part by the National Academy of Sciences. References [1] A. Stelzer, C.G. Diskus, K. Lubke, H.W. Thim, “Microwave Position Sensor with Submillimeter Accuracy,” IEEE Trans. Microwave Theory Tech., vol. 47, no. 12, pp. 2621–2624, Dec.1999. [2] A Benlarbi, J.C Van De Velde, D. Matton, Leroy, Y., “Position, Velocity Profile Measurement of a Moving Body by Microwave Interferometry,” IEEE Trans. Instrum. Meas., vol. 39, no. 4, pp. 632636, Aug. 1990. [3] Seoktae Kim and Cam Nguyen, “On the Development of a Multifunction Millimeter-Wave Sensor for Displacement Sensing and Low-velocity Measurement,” IEEE Trans. Microwave Theory Tech., vol. 52, no. 6, pp. 1503-1512, Nov. 2004. [4] Seoktae Kim and Cam Nguyen, “A Displacement Measurement Technique Using Millimeter Wave Interferometry,” IEEE Trans. Microwave Theory Tech., vol. 51, no. 6, pp. 1724 -1728, June 2003.

41

9/15/2010

A Novel Approach to a Multiband Transverter Design Jeff Kruth WA3ZKR Presented to the MUD 2010 Conference

Why a new transverter design? Are not the old ones good enough? • Yes, but our nature is to experiment! • New system level components offer greater flexibility (synthesizers!) • Multiband operation is costly, yet desirable (Rovers, etc)! • Some still like to homebrew….

42 1

9/15/2010

Basis for Conventional Single Band TransverterApproach • All communications is about filtering & noise i rejection. j ti • Single band approaches minimize filter design/implementation difficulties. • Clever use of hairpin/pipecap filter designs on FR-4 boards meet all requirement in a single band design. • Can require significant real estate.

Local Oscillators: Problematic! • LO is key component, used to be difficult, simpler i l with ith “b “building ildi bl blocks”. k ” • Single LO Frequency – Easy to do w/ surplus PLO or Custom XO/Multiplier. • Each band required a solution for the LO issue many times not trivial to meet issue, stability desired, etc.

43 2

9/15/2010

Conventional RF Approach • Basic RF Converter. • No Amplifiers, therefore Bi-directional Bi directional in signal path. • Filters usually considered a necessity in RF & IF path for noise & image rejection. RF Filter

IF Filter

Narrowband Mixer & Associated LO

Typical M/W Amateur Transverter • Features added for utility: – IF Attenuator – T/R Amplifiers – IF Filter usually not needed, IF radio suffices – Can remote LNA/PA, add line amp

44 3

9/15/2010

Multi-Band Approaches • Desirable to cover at least 2.3-10.368 GHz in one box (4 bands) bands). • Potential for significant size reduction. • Front ends could be in box or remoted up tower. • Use modern technologies to solve old problem! • Cost savings is a possibility. • Drawbacks include – Higher complexity – Single point failure, all bands off the air!

Multiband Issues • Broadband mixer required, many types available 2-18 GHz, 1-15 Ghz, etc. • Need Multiple LO’s LO’s, multipole switch switch. • Multipole switch and multiple filters needed for RF side. • Can be built up over time, but bulky, and LO’s may not be lockable to common reference.

45 4

9/15/2010

Multi-band LO Requirements • Previously a severe constraint! • Prior schemes involved either PLO bricks or crystal multiplier schemes with different multiplication ratios. • Availability of modern frequency agile synthesizers can change this! • Currently L band .9-2 Ghz in sub-bands.

Multiband Design Improvements • Use a synthesizer locked to a reference f stability for t bilit issues. i • Use multipliers from old PLO blocks for LO multipliers for ease of implementation. • Use a single electronically tunable filter for all band RF image reject task. task • Integrate a broadband amp and reversing switch for RF driver/line loss comp.

46 5

9/15/2010

Multi-band Block Diagram • • • • •

Simple! LO’ can b LO’s be added dd d llater t YIG filter is fixed tuned SMA n-pole relays, cheap WJ, RHG, Add. Labs Anzac M/A-COM Mxrs Anzac, • Multiplier Sections from old broken PLO’s

What’s this YIG filter thing? • YIG tuned filters (YTF’s) key to M/W wideband receiver systems systems. • Provide stable, easily tuned passband over multi-octave range. • Made by wide variety of vendors! • Real cheap at hamfests ($10-$100), if thrifty shopper! (I found 3 at M/W update) • Need a “special” DC driver circuit? Not really!

47 6

9/15/2010

More on YTF’s • YIG material provides a magnetically tunable resonance at M/W frequencies. frequencies • Magnetic field created by electromagnet in form of solenoidal pair. • Current sets field hence frequency, so current source should be clean (low noise as possible) and stable (low DC drift). • YIG sphere kept stable by small 24 VDC heater (150 mA)-not always needed. • YIG magnet current typical 0-1A or less.

YTF’s, All Shapes & Sizes! • • • • • •

Made by lots of folks for the last 50 years! Inside of all kinds of old M/W stuff! HP 8445A preselectors HP 8441 preselectors AILTech 707 SA RF Black boxes…

48 7

9/15/2010

Typical YTF Response • YTF circuits inherently broadband. • Yigs marked as octave typically broader • Example is YTF used as 4-8 GHz, found to be 2-18 GHz.

Powering Up Your YIG! • Many hams shy away as these seem too exotic. ti • Driver circuit seems to be a stumbling block. • Driver design sought that was very simple yet worked well well. • Decided on a cheap power OP amp!

49 8

9/15/2010

YIG Driver Schematic • Very simple circuit, needs good heatsink! • Part is L165 5 terminal power op-amp. • 5 watt low ohm stable R needed. • Bi-polar supply, neg. is low current, DC-DC conv.

Datasheet for L165 • Really nice power op-amp! • Capable of 3 Amps! • Less than 1A needed for us! • Made by ST Microelectronics.

50 9

9/15/2010

Breadboard YIG Driver • Obviously non-critical construction! • Key K was heatsinking h t i ki both device and power resistor! • Current can be sourced by either polarity supply by inverting drive voltage polarity.

Test Setup to Test Driver • Simple to align, tune for peak output. • Use power meter or crystal detector. • DC voltage tuning approx. 0-3 V.

51 10

9/15/2010

Filter Responses 2.3 & 3.45 GHz • Simple, stable, easy to get filter shape!

Filter Responses 5.7 & 10.4 GHz • Typical bandpass, approx 30 MHz wide

52 11

9/15/2010

Converted Multipliers • Bricks are cheap, esp. broken ones! • Multipliers are simple, really don’t fail much. • Old brick PLO is what dies. • Simple to excise mult. • Add SMA connector.

Performance of Multipliers • Variety of bad bricks in Junque box. • 3.3 & 5.6 were easy! Low drive requirements, high output power. • 10 GHz from White box LO! (At last, its good for something…) • +21 dBm drives from 1 GHz surplus amps.

53 12

9/15/2010

Power In/Out for Multipliers • 3312 MHz - +7 to +13 dBm out, 1104 in@ 15 20 dB 15-20 dBm. • 2 types for 5616 MHz: x4 & x5, x4 gave 9.5 dBm out for +20 in @ 1404, x5 gave +12 out for 1123.2 @+20 dBm. • White box converted mult mult. X6 X6, gave 10224 MHz @12.5 dBm for 1704 Mhz @20 dBm. • No 2160 MHz Doubler tested, DBM?

The Guts of a 4 Bander • • • •

Spread out on bench approx. 14 “ square Will pack up much smaller. Key is A32 synth. Can add multipliers as you develop them! • Lock to a Rube? XO?

54 13

9/15/2010

Final Thoughts! • Presented to provoke experimentation with new (to us) approach. approach • 2nd YTF could be used with untuned SRD Multiplier to make tunable LO as well. • System would make a nice Noise Figure Meter front end for conferences….. • Has p potential to educate and p provide utility! y • I recognize that this will not supplant traditional transverter approaches, merely complement them.

Questions?

• & Thank You for listening!

55 14

A YIG Filter Primer & Simple Driver Circuit for HAM Projects Jeff Kruth, WA3ZKR I have always been fascinated by YIG tuned microwave components for many reasons. One of the foremost is that they revolutionized RF test equipment, allowing broadband sweep generators and spectrum analyzers, without the need for the bulky and problematic high voltage power supplies needed for backward wave oscillator tubes (1). Since YIG frequency changes are linear with current, the need for breakpoint linearizers, required in BWO and varactor circuits, was eliminated. This simplified the drive circuits and made the frequency scales linear with ease. Another reason is that they made airborne electronic warfare wideband receiver systems smaller, cheaper and very flexible, and this was another area of my interest, both professional and hobby. YIG tuned filters (YTF) and oscillators (YTO) are simple devices, using the fact that YIG (yttrium-irongarnet, a ferrite) material, if properly made and shaped, will exhibit a low loss microwave resonance when magnetically biased. This resonance frequency can be tuned by varying a magnetic field imposed on the YIG material, which is usually shaped in the form of a sphere, although rods and thin film sheets have been used on occasion for special purposes. The unloaded Q of the resonators can be quite high, in the 5000 to 8000 range, resulting in narrow band tunable responses. The bandpass remains narrow even when loaded by microwave coupling loops attached to the real world. The magnetic field is developed by a solenoidal electromagnet whose field is at right angles to the coupling loops. The current needed to tune a device over an octave or better is usually less than 1 ampere maximum, and sometimes quite a bit less. These resonators can be used as elements of a filter or in the feedback circuit of a semiconductor oscillator. These oscillators typically use transistors (both bipolar and FET, although FET’s have been replaced by bipolars as the phase noise is better (2)). In the past, at frequencies above 8 GHz, Gunn diodes were sometimes used, but these had their own problems, and are not used very much anymore, except at millimeter wave frequencies. Ham use of YIG oscillators and filters is usually limited to what is built into their commercial bench test equipment. However, a few hardy souls have made their own sweepers and spectrum analyzers using these devices (See the 1994 Proceeding of the Microwave Update Conference for some examples). I also did this in the early days, before I had built up my test bench. One project was to substitute a YIG tuned oscillator for the 2-4 GHz BWO in the early HP8551-851 spectrum analyzer, a very rewarding project (3) at the time. YIG devices are ridiculously easy to employ from a standpoint of the RF parts end, but many shun their use. This is because they do not understand how easy YIGs are to use, and they do not have a ready solution for the driver circuit that controls the current in the field coil. The goal of this paper is to encourage experimentation in the ham community with these useful devices by making the analog driver circuit easy to implement. One project I had long thought of which would benefit from YIG technology is a multiband ham microwave transverter for 2.3 through 10.3 GHz. This was based on my knowledge of wideband receiver systems and how YIGs are used in them. When I attended the 2009 SVHF Conference I heard several talks in which the authors presented ideas for a multi-band transverter. One issue, then as always, is the need for multiple filters for LO & RF portions of the transverter, and this was part of the discussion at these presentations. Naturally, it proves non-trivial to implement highly selective filter/amplifier chains for a multiband design. It was natural for me to think of a YIG filter for this

56 1

portion of the application. I did not consider a YIG oscillator for the LO as it would have to cover 2.1610.224 GHz, and while these wideband devices do exist, they are harder to find, the typical octave band devices being more common. In addition, a synthesizer loop would be needed to lock the LO up, and then there would be lots of discussion on phase noise performance, size and weight, etc. However, YIG filters are relatively abundant (I bought three for 10.00 each at Microwave Update in Dallas in 2009), can be broadband and still very selective, and are easy to use. For the LO, I have several thoughts based on the A32 LO synthesizer for this purpose and have bought several from Down East MW to learn with. The actual multiband transverter implementation is the subject of another paper. Another interesting application is a broadband panoramic receiver, made by using a YIG filter hooked to a crystal detector! Combined with the appropriate driver, sweep circuit, an old low frequency oscilloscope and a little time, a useful broadband spectrum analyzer can be made with approximately -45 dBm sensitivity and multi-octave-in-one-sweep frequency coverage! Many companies over the years have built and offered such products, which are very useful for harmonic testing, etc. Testing YIG Filters I measure the YIG filters I get my hands on in a setup that includes a laboratory grade YIG current driver box, and a scalar network analyzer covering 2-20 GHz. I usually generate a plot of frequency coverage and current requirements to tune the device. A typical device plot and data table is shown as Figure 1. It is interesting to note that almost all YIG devices work far outside their stated bandpasses. This is true for oscillators as well as filters. I have had filters marked 3.7-8.2 GHz tune from 3.3 to 18.6 GHz, and 2-4 GHz filters tune from 2-12.4 GHz or higher! So often, those junk box items are useful beyond their marked range.

.

Figure 1. Typical plot of YIG Filter Bandpass & Current

A YIG filter is a simple device as far as DC connections go, usually having only four terminals, two for the main coil, easily identified by using an ohm-meter and measuring for the low resistance value (< 25 Ω) across the appropriate pair of terminals). The other pair of terminals is connected to a positive

57 2

temperature co-efficient (PTC) thermistor that is used as the YIG sphere heater. Controlling/stabilizing the temperature of the tiny YIG ball to the correct value generally provides for the lowest insertion loss across the operating range. This voltage is usually 24-28 VDC at 150 mA or less at steady state. A small current spike is observed at turn on, rapidly decreasing to the steady state value, so the heater supply must accommodate this. Years of experience has shown polarity is not critical, although if marked, you might as well conform. I have also run these devices without using the heater, and have generally found little or no change in performance, but also recognize that for these devices to perform well in mil-spec temperature range environments that the heater is probably needed (good for rovers to know…). My test setup offers considerable ease in test, but YIG devices can be tested at home, without an elaborate setup. What is needed is a tunable signal source covering the frequency range of interest, a device to detect the RF (a crystal detector or power meter is OK, but a spectrum analyzer is nice…) and a stable DC power supply capable of supplying up to 1A or so, with a series resistor (such as a 5-10 Ω, 5 W type). Your testing will take a little more time, but is very simple: you wish to “find” the current that gives the minimum insertion loss at the frequency of interest. If, for example, you wished to use the filter on the RF side of the system at 13 cm, you might set the generator to 2304 MHz, verifying with your detection system that you have sufficient RF amplitude to cause a noticeable response, then, insert the filter between the generator and detector. Adjusting the current (by either adjusting voltage or current depending on your supply) carefully from 0 to 1 A you should see a peak in RF output power. If a crystal detector is used with an oscilloscope, the generator could be amplitude modulated (1 KHz is typical, usually a feature on most generators) and the AM waveform peaked up on the scope. The current for this setting is noted (a digital ammeter in line is nice…), and the process continued until the bounds of the filter are found. The Driver Circuit To use the filter, I went to my file of YIG driver circuits and looked for a reasonably modern, relatively simple driver approach that would also be of sufficient performance to be stable enough to make the rest of the electronics simple (4,5). I discarded many designs, including an earlier one of my own (6) and finally went about rolling my own based on a power op-amp device made by ST Microelectronics, called the L165 (7). While not a recent part, this device is newer than some of the discrete implementations I am familiar with. Additionally, it makes a functional design that met my goals and did not require any exceptional effort or creative genius on my part, which is good! It did require some attention to thermal stability issues, as I did not want the circuit to “drift” as it warmed up, detuning the filter off the frequency of interest. The input voltage was roughly 0-2.5 volts, so the output current was negative, so the negative supply was the more heavily loaded. This is not consistent with most ham designs, where minus voltage supplies are usually capable of only modest current sinking. Usually, there is plenty of current available from the +12 VDC supply, as in mobile applications. So, if a negative polarity input voltage is used, the positive supply to the op amp will be the more heavily loaded, supplying the YIG coil current. If a positive drive voltage is desired, based on other considerations, then an inverting amp may be used before the power op-amp. This would also allow for scaling and offsetting, if required. A small DC-DC converter could be used to supply the negative rail. It might be possible to work up a single supply design, but this requires more skill and several tricks, so I stuck to basics. A note on the datasheet for this part: There is an error in Figure 2 of the L165 data sheet, the basic circuit of my driver, and that is that the pin numbers for the non-inverting and inverting inputs are switched. The circuit drawing is correct in that the input is to the inverting pin, but it should be labeled “2” not “1”.

58 3

The basic circuit is presented in Figure 2. The layout for the circuit is non-critical. I hand-cut a circuit board out of a scrap of FR-4 (G-10) material for quick mounting of the parts for the test needed to write this paper. The L165 comes in a 5 leaded TO-220 outline package called a “Pentawatt Package”, which I assume means 5 watts continuous dissipation. The leads are staggered, making hookup a bit tough, but I used a pair of needle-nose pliers to flatten and reform them to a single plane in order to simplify my breadboarding. It is important to use high quality components for the driver circuit, such as 1% metal film resistors, in order to minimize any drift. It is also important to heat sink the L165 power amplifier IC adequately to prevent thermal drift in this device as well. Wire leads could even be attached to the device in order to ease the mounting/heatsinking requirements. The tab is not at ground potential so the tab must be either isolated from the heatsink, or the heatsink must “float”. It is also convenient to mount the high power feedback resistor on the same heatsink. I used one of the gold anodized “Dale” types, as I have these. Other types could be used as well as long as they are thermally stable types and can handle the power. I used rather large tantalum caps for the bypass function on the + & - power supply legs, a smaller value would suffice. Not indicated on the diagram is the power supply voltages (plus & minus 12 volts), and the input is the left-most 10K resistor. The “snubber” circuit on the op-amp output, consisting of a .22uF film cap and a 10 Ω resistor, is recommended by the manufacturer of the op-amp, in order to increase stability of the amplifier when driving inductive loads, such as a YIG main tuning coil.

Figure 2. YIG Driver Circuit using L165 Figure 3 shows a picture of the bench test lash-up. The filter chosen for the tests was a YIG-TEK 183 series filter (whose data sheet is shown as Figure 1). These are broadband 2-18 GHz designs and are found in old spectrum analyzers like the AILTECH 707-727-757 series. They are also found in the HP 8445 preselector for the 141-8555-8552 analyzer system. I chose this device purposely as the tuning

59 4

coil current is less than 500 mA at the maximum value. Higher current can be supplied from the L165 device, up to 3 amps, but I wanted to keep the heat dissipation and heat sink size smaller for this test. This device performed well. Initially, the bandpass showed a bit of drift as the circuit warmed up, but this was my fault, as I was in a hurry to see it work, and did not use enough heatsink. This drift was greatly reduced, really virtually eliminated, by using a larger heatsink. Devices from other manufacturers could also be used from vendors such as Ferretec, Trak, Watkins Johnson, Avantek, Varian and others, with good results. The old preselector for the HP 8551/851, called a 8441, has a nice WJ YTF which covers 1.7-12.4 GHz.

Figure 3. Test Setup w/Driver & YIG Connected to Scalar Analyzer Figures 4a-d are the swept bandpasses at the four RF frequencies of interest, which are 2304, 3456, 5760 and 10368 MHz. Notice the rejection for low inject LO frequencies for 144 MHz IF (2160, 3312, 5616 and 10224 MHz respectively). There would be no difficulty whatsoever with image and LO rejection! The bandpass of the filter is observed to be approximately 30 to 40 MHz across this range. The 3 dB down point and the pass band in between shows good symmetry and is relatively well behaved. The upper slope of the pass band shows the typical “bumpiness” associated with the usual spurious resonances associated with YIG devices. This is of no importance to our needs, and is simply mentioned to explain the non-symmetry of the pass band shape. This particular filter shows a small “double-hump” response at 5760 MHz, not common, but considered no difficulty for the project. The shape and selectivity of this filter is very good for our uses.

60 5

Figure 4a. Passband at 2304 MHz

61 6

Figure 4b. Passband at 3456 MHz

Figure 4c. Passband at 5760 MHz

Figure 4d. Passband at 10368 MHz

62 7

In conclusion, YIG tuned components provide the opportunity to make multiband systems for amateur radio and test bench use. The ability to realize all the filters required for a multiband transverter by using one or two YTF’s and simply changing a voltage is quite interesting and should be explored by the innovators among us. Surplus store, hamfests and eBay yield a surprising number of these devices at low cost, if looked for. Best thing is, the filter engineering has been done for you, and they are simple to use as well. 1. 2. 3. 4. 5. 6. 7.

References “Magnetically Tunable Microwave Filters Using Single Crystal Yittrium Iron Garnet Resonators”, P. Carter, Microwave Theory and Transactions, IEEE Press, May 1961 “Low Noise Bipolars Silence Noise in 18 GHz YIG Source” Microwaves & RF Magazine, November 1988 “HP 8551B YIG Conversion”, J. Kruth, 1991, self published “Design a Stable Current Source for YIG Filters”, B. Taher, Microwaves & RF Magazine, February 1988 “YIG Drivers” Application Note 99-002, Micro Lambda Inc. “HP8551 YIG Driver Circuit”, J. Kruth, Proceeding of the 1994 MUD Conference Datasheet, L165 3A Power Operational Amplifier, ST Microelectronics, July, 2003

63 8

LO Phase Noise effects on MDS Gary Lauterbach AD6FP

How could LO phase noise effect MDS? • MDS is determined by Signal/Noise ratio – Only three scenarios are possible

• Noise floor increase:

– In the presence of strong interfering signals – With only a single weak signal

• Signal level decrease

– Spreading of signal energy beyond discernable bandwidth

• Both: signal decrease and noise increase

64 1

History of amateur LO phase noise MDS effects • MUD 1996,1997 W8MQW, WA6KBL

– Noise floor increases – Theory and negative confirmation with measurements

• MUD 2008 W1GHZ

– Observations on field MDS tests, no theory of signal or noise effects

• MUD 2008 K0CQ

– Noise floor increase – Theory with no experimental data

• MUD 2004,2009 WA1ZMS

– Signal decrease due to spreading – 2004: Low BW sub-mmw needs low phase noise – 2009: Experimental “confirmation” of W1GHZ observations

What is LO phase Noise? • Two views: Spectral and Temporal

– LO energy spread over spectrum surrounding the LO center frequency – Time jitter of LO waveform zero crossing

• Both views are valid and measurements can be translated between them • Can be both deterministic and random • Modulates the LO: AM and FM – FM phase noise creates no additional LO energy

65 2

Reproducing WA1ZMS results • Brian well documented his experimental setup – But no phase noise plots were included 

• Random noise FM modulating a laboratory signal generator – Brian used:

• Homemade noise source • HP 8640a signal generator

– I wound up using:

• HP 3561a random noise source • HP 8662a signal generator

– 8640a is not sufficiently stable for very close-in phase noise measurements

Brians measured spectrum • FM noise modulation producing 2db MDS degradation

66 3

My measured spectrum of WA1ZMS -2db MDS experiment • 8662a @ 144 MHz FM modulated with random noise

Phase Noise Measurement • Pair of HP 8662a signal generators

– First 8662a is EFC locked through a 0.1 Hz BW PLL to the DUT for PN measurements • Switchable 40 db gain baseband LNA

– Second 8662a is a noise modulated source

• • • •

HP 70210a Spectrum analyzer HP 3561a random noise source KE5FX USB baseband digitizer KE5FX TimeLab measurement software

67 4

WA1ZMS -2db MDS phase noise plot • Note the 20db/decade phase noise increase – Unlike any real world LO – F^-2 over entire PN range of interest

• Integrated power over 1-100 Hz is a significant % of LO total power

Here’s the visible MDS effect • SDR-IQ 10 MHz IF of 154 MHz signal and 144 MHz LO: – Left: clean LO – Right: -52dbc PN @ 1 KHz WA1ZMS noisy LO

• Signal spreading causes S/N decrease of >1.5 db in 27 Hz BW • Noise floor didn’t go up

68 5

Why 1 KHz offset PN is a bad metric • 1 KHz offset PN is not a predictor of sub 100 Hz PN • Sub 100 Hz offset PN effects “purity” of narrow band weak signals – CW: 30 Hz “ear” BW, <100Hz PN matters – JT: 2.5, 5, 10 Hz BW, <20Hz PN matters

What about the K0CQ, WA6KBL Noise floor increase theory? • HP 70210a capture of 1 GHz noise source down converted to 100 MHz IF – Top: clean LO – Bottom: WA1ZMS -52dbc @ 1 KHz LO

• Noise powers match to <0.5 db

69 6

Why doesn’t the noise floor increase? • Think of the temporal jitter model: • LO Spectral energy is not simultaneous at all frequencies but is a probability density function (PDF) over time • As the LO jitters its energy is spread over a range of frequencies at different times • The input noise to the down converter has a completely level spectral distribution ergo as the LO jitters in frequency it always down converts a constant input energy

N5AC synthesizer PN measurements • Measured with several reference sources: – Cheap SMD CTS crystal oscillator – FOX801BE Downeast recommended source – Vectron 718Y precision OCXO

• >20 Hz offset PN the reference doesn’t matter, the PLL loop noise dominates • <-75 dbc @ 1 KHz with all sources

70 7

N5AC synthesizer PN results

N5AC with CTS reference

71 8

N5AC synthesizer PN on HP E5052 with CTS reference

MicroLO PN measurement

72 9

MicroLO, N5AC and WA1ZMS overlayed PN

100 MHz DFS with Vectron 718Y reference

73 10

AD6IW 1152 MHz PLL

Bare Vectron 718Y OCXO

74 11

Conclusions • Low LO phase noise is most important in the presence of strong signals

– PN at offsets from 1KHz to 1 MHz can be important with strong interfering signals

• Don’t use 1 KHz offset PN measurements to determine narrow band MDS effects – Measure your LO you may be surprised – Close-in PN of synthesizers can be better than crystals

• Even when multiplied to 10 GHz the N5AC synthesizer is clean enough for no CW MDS degradation (as long as there are no strong adjacent channel signals)

Acknowledgements • Thanks to: – KE5FX for loaning the baseband USB analyzer and hacking Timelab to support it, it’s a fantastic baseband PN measurement tool – WA1ZMS for answering questions regarding his prior measurements – AD6IW for E5052a results

75 12

A Modern 47 GHz Transverter Tony Long KC6QHP

Figure 1: Local Oscillator schematic diagram

Abstract: A 47 GHz transverter has been constructed, tested, and operated. Bare MMIC die are used for the RF front end. The local oscillator is generated with a surplus commercial phase locked DRO mixed with a traditional VCO based PLL synthesizer. Topics covered include the stages of development of the transverter, working with bare die, wire bonding, special machining techniques, a different method for biasing, and real world performance of the transverter.

1.0

Introduction/History

Having built radios for both the 10 and 24 GHz bands, I decided to build a 47 GHz radio in 2004. At the time, members of the San Diego Microwave group were experimenting with homebuilt sub-harmonic mixers on 47 Ghz. I decided to follow their lead and built up a very basic 47GHz radio. The mixers are easy to make and well documented [1], but coming up with a 23.472 GHz LO (144 MHz IF for a 47.088 GHz mixing product) is not as straightforward. Luckily I came across a surplus phase locked DRO oscillator which I could double to 24 Ghz. Unfortunately, 24 GHz is not what I needed so I modified a Qualcomm “1152” synthesizer board to put out a 1056 MHz signal. I fed this into a divide-by-two prescalar chip stripped off another Qualcomm board and had myself a 528 MHz signal which I could now mix with the 24 Ghz signal and obtain the magical 23.472 GHz halfLO. The business of mixing, filtering, and amplification were taken care of by modified surplus PCOM 23 GHz modules (Figure 1). The first version of the radio (Figure 2) was used to make several 1 km contacts during a couple ARRL 10 GHz and Up Contests. After making these contacts, I wanted some DX. This meant adding some receive gain and some transmit power. By mid-2007, I began devising a method for doing just that. I chose to use bare MMIC dies over packaged parts mainly because I

76 1

was concerned about fabricating 5 mil PTFE boards of sufficient quality for 47 GHz circuits. It seemed easier to just use bare MMICs with 50 ohm substrates and the timely availability of surplus wedge bonding machines helped considerably. Little did I know I was getting myself into a huge undertaking. Three years later, the radio is complete and has given me the DX I was looking for (best at the time of writing is a 100 mile (160 km) contact between myself and Gary, AD6FP). Figure 2: First version of my 47 GHz radio

2.0 The New Front End One challenge in microwave radio construction is the rejection of undesired mixing products. In a receiver the presence of the image sideband effectively results in a 3 dB increase in noise figure. On transmit, the result is wasted transmit power (predominantly in the leakage of the LO and first undesired mixing product. A common approach to the elimination of undesired mixing spurs and LO leakage is to simply filter them out. Unfortunately, when the IF frequency is very small with respect to the final RF frequency, filtering becomes exceedingly difficult. Since I wanted to reuse the LO source that I had previously built I was stuck with a 144 MHz IF, which meant that I would need filters with a 0.3% bandwidth! The solution came in the form of a really excellent mixer made by UMS (CHM1294). The mixer in question is an image reject type, with low LO leakage and an on-chip LO doubler amplifier that is self-biased from a single +4V supply. So this mixer became the heart of each of the transmit and receive modules. On the receive side, all that was left was to add some gain, which was achieved with two stages of a nice balanced LNA from Hittite (Figure 3). At the time of writing, the receive module is only configured to use one of the channels of the mixer and thus does not take advantage of the image rejection. Luckily I built a third housing which will soon house an improved receiver, using both the I and Q outputs feeding a hybrid.

amplifier die. (Figure 4) This 3-stage part requires individual gate and drain biasing, which greatly complicated the module design. Prone to oscillation, this part required the delicate addition of extra drain capacitors during final testing. Figure 4: Transmit Module Schematic

Common to the receive and transmit modules is the microstrip-waveguide transition. This particular portion of the module construction represents the single biggest challenge. I have gone through three iterations of the basic design which is based on E-plane transition designs published by S. Weinreb [2]. Figure 5: Waveguide Probe Variations

Figure 3: Receive Module Schematic

On transmit, I used another one of the Hittite LNAs as a post-mixer driver powering another Hittite part, a 100 mW medium power

All of the probes were designed and modeled using Ansoft’s HFSS 3-D EM simulator. The first iteration of this transition consisted of a 5 mil thick alumina substrate with a tiny tab of gold plated metal epoxied to the end of a 50 ohm gold line, hanging out into the waveguide. After the first tests of the receive module showed poor performance, I abandoned this highly variable and difficult to model style. Lacking the ability to pattern gold onto

77 2

existing alumina substrates, I did the next best thing in version 2, which was to smear on silver epoxy and trim to the right shape. The results were slightly better and both the transmit and receive modules use this transition. Later on, I decided I needed an even better transition and came up with a very easy to build, repeatable transition requiring no patterning. I did a much more careful job of modeling this transition which is simply a 50 ohm line (on Quartz this time) that protrudes into the waveguide 62 mils. Simulation (including material losses, non-zero radius corners, mousehole geometries, etc.) gives about 17 dB return loss.

3.0 Transmit and Receive Module Design My first big design decision in this front end was how best to segment the RF functional blocks. Should I use a single mixer and switch between TX and RX gain stages? Would I be better off with a single low noise, high power gain-stage and switch that in and out (and reverse for TX/RX)? Or should I have separate mixers in individual upconverter / downconverter modules?

plumbing. Since I would be working with multiple die in each housing, it made little sense to use a single mixer. So the front end is really two small modules attached to a custom-made WR-22 rotary waveguide switch. (Figure 6) In keeping with my desire to have as little waveguide as possible, I settled on the previously discussed E-plane transition. This topology allows the modules to be butted up right against the switch (Figure 6) and allows for a flat single-piece surface on which to mount the die, substrates, capacitors, and so on. The assembly of the modules called for a split-block design, where the baseplate would form the waveguide flange and component mounting area, and the upper half would form the backshort for the waveguide, provide protective walls and coaxial connector mounts. During the wire bonding process the absence of obstructions such as walls makes life a lot easier! Figure 7: Transmit and receive modules after final machining (plus a spare)

Figure 6: Transmit and receive module attached to waveguide switch

I ended up going with the last option for a few reasons, but mostly to avoid WR-22

78 3

Since there are no unique mechanical features between the transmit and receive

modules, they were built identically. (Figure 7) Subsequent modifications were made however to the transmit module as it needed an additional 6 bias pins. I built a third module as a spare.

4.0 Module Construction Construction of the receive module began first as it was the simplest and I needed to learn how to wire bond, mount die, etc. I didn’t want to learn all of that on the much more complex transmit module. Figure 8 shows the mostly completed receive module with the first version of my waveguide transition probe. Figure 8: Nearly complete receive module

The construction of the module appears straightforward, however there are a number of lessons I learned along the way, and through the assistance of some knowledgable friends. First off, spacing is critical. At 47 GHz, a wire bond, even a well done bond, has considerable inductance, made much worse if parts are placed excessively far apart. “Excessive” meaning more than 5 mils! As you can see, with a cascade of 9 separate pieces, maintaining a <5 mill spacing between parts is tough. Typically the MMICs themselves will be well within 1 mil of their specified length, so the real trick then comes in the accurate cutting of substrates. I bought substrates from a vendor in 500 mil lengths and then cut them down to size using a special saw that I built specifically for this project. [3] The saw uses a commercial diamond

wafer dicing blade and is attached to the spindle of an old hard drive. The hard drive has a motor of sufficient speed and bearings of sufficient accuracy for this project. I mounted that to a sturdy column, and then added a precision X-Y table with a vacuum chuck to hold my workpiece. (Figure 9) The 500 mil substrates are held down onto a large glass slide with a special thermoplastic adhesive called Crystalbond. I was lucky enough to find some of this on eBay, enough to last me many lifetimes! After ‘gluing’ the substrate to the glass slide, it is held onto the X-Y stage with the vacuum chuck. The saw blade is brought down to the piece just like a woodworking chop saw with a micrometer movement. At 5 mils thick, it is easy to cut too quickly and send pieces flying into oblivion. Water cooling is provided by a hobby airbrush aimed at the cutting interface. The entire setup for dicing these substrates really requires a lot of equipment! While it is possible to scribe and break these substrates, I have not tried this, and suspect that it requires more practice and skill than I have. Using a digital gauge and compensating for the width of the cut (about 12 mils), I am able to reliably cut substrates to length with an accuracy of just a couple mils. I am also able to make angled cuts which helps in cases where parts are not oriented in a straight line. Figure 9: Dicing saw I built for cutting substrates

79 4

Figure 10: Test substrates after dicing.

After the substrates are cut to the right length, the next step is to glue the parts down to the housing. The housing acts as an electrical ground plane as well as a thermal path from each die to a heat sink. In order to realize a good electrical and thermal connection to the housing, silver loaded epoxy is used. This is the far simpler alternative to a eutectic solder bond which requires costly tooling and equipment as well as a solderable housing material such as gold, brass, or copper. Figure 11: Partially complete transmit module

There are a large number of silver epoxies available for purchase, of different viscosities, different thermal and electrical properties, different curing schedules, etc. In this project I used two different epoxies with some distinct difference in their performance. The first is a twopart epoxy made by Circuit Works (CW2400) which has been used by a few hams over the years. The best part about CW2400 is that it is cheap. Other than that I wasn't impressed with it. It has a rather high viscosity which makes it tough

80 5

to work with, as well as a very short pot life (5 minutes). This means that a lot of epoxy is used as the mounting of each die, cap, or substrate may require a new mix. Additionally it doesn't seem to cure very well, resulting in a slightly rubbery consistency. The other epoxy I used is sold by McMaster Carr (part number 7661A13), and comes in a small segmented pouch which is squeezed together to mix the two pre-measured parts. This seemed to work better and has a 3 hour pot life, at a cost of just over $20 per packet. Better epoxies exist but they are very expensive, some require freezing at -40C, and some require careful mixing of two parts. Gluing die to the housing is a delicate art! The first issue is spreading out just enough epoxy to ensure complete coverage of the bottom of the chip, while avoiding excessive squeeze-out which can leak over and ground exposed metal on the die. A quick way to create smoke on power-up! I typically use a needle to deposit a tiny bit of epoxy where the chip will go, and then use a fine paintbrush with just a few hairs, or a very new Exacto blade to smear the epoxy into a uniform, smooth layer, hopefully just under 1 mil thick. After the epoxy is in place, the delicate task of placing and dropping the die comes next. A really good pair of tweezers helps considerably here, with the two most important features being extremely sharp points, and relatively light spring force. The dies are just a few mllimeters across, weigh practically nothing, and will shatter if you look at them wrong. I've found that placing the waffle or gel pack (these are the containers the die come in from the factory) as near to the mounting location as possible results in fewer accidental drops. Once the die is placed, it needs to be pushed down into the epoxy. I used Q-tips and sliced the handles into sharp points. These became fairly soft-tipped pokers with which I use to press the die down. This is of course an extremely delicate task that requires a good stereo microscope. If you push down onto the wrong place you could crush an airbidge, smash a delicate gate, or collapse some other structure. Assuring the flatness of the mounted die is crucial

as any gaps beneath the die, or tilt, can result in a broken die when it comes time to wire bond! (I learned this the hard way) After the die is placed and pressed into the epoxy there is frequently some epoxy that has squeezed out from underneath the die. I have found that this is not the best time for cleanup, as you are likely to move the chip. Instead, I do a partial cure of the epoxy so that it is still easily removed, but the part is fairly well locked in place. Cleaning squeeze out is critical at the RF ports of the die, as the spacing between dies and substrates is critical as mentioned earlier. Remember that the squeeze-out from an adjacent substrate/chip cannot be cleaned out once the first part is in place! Once all the pieces are in place, it is time for a final cure (typically at 150 degrees C) for about 1 hour. The final step is wire bonding. I used a thermosonic compression wedge bonder for this project. This specialized tool allows the welding of tiny gold wire between gold plated pads on MMICs to caps, substrate, and other MMICs. The gold wire is hard to come by, but I managed to do a group buy with a couple other hams, and I now have a supply of gold wire that will last longer than its shelf life (about 2 years). It turns out that gold bonding wire loses some of its key performance characteristics with age for reasons I do not completely understand. After the wirebonding was completed, the housings were put together and tested. The receiver ended up with 10 dB of gain and a 9 dB noise figure (this will be improved considerably in my updated receiver as I take advantage of the image rejection, and use the greatly improved 3rd iteration waveguide transition). The transmitter testing was not completed fully as I only ever tested it with a single mixer input. In this configuration I measured +10 dBm output power. With the second mixer input in use (as it is in the radio) I should gain another 3 dB. This module is capable of substantially more output power, evidenced by the inability to saturate the final amplifier. I believe that my early attempts at wire bonding and inefficient 2nd iteration transition leave some performance on the table. Luckily the PA module seems to work with this power level

and I have a much more potent transmit signal (unmeasured at this point).

5.0

Module Mechanical Construction

Machining small waveguides is tricky, and there are a few special techniques I used in this project which allowed me great flexibility in the design of the housings. Observe in Figure 8, the rectangular waveguide hole in the housing. Using a standard milling machine to make a rectangular hole in metal with sharp corners is hard, and I haven't figured out a way to do it. Instead, I enlisted the help of a good friend from another hobby who works at an EDM shop. He was kind enough to cut these waveguides using a CNC Electrical Discharge Machining (EDM) tool. EDM operates by dumping a ton of current into the tiny contact area of a cutting tool (a wire or a small metal or graphite electrode) and workpiece. As this current arcs over it erodes the less conductive workpiece and very slowly eats away at it. This same process is repeated many times until the shape is formed. So next time you create a pitted surface on something that has arced over, consider that you have just done a tiny EDM job! My waveguide holes ended up with a fairly smooth (better than 1000 grit sandpaper) hole with corners that have a radius almost the same as the thin (a few mils) wire used to cut it! I used the same technique to cut the waveguide holes in the rotary waveguide switch described later. For the upper half of the housing, (Figure 12) I enlisted the help of my father-in-law who is a die maker and has a die-sinker EDM machine (the type that uses a small copper or graphite electrode). The main housing cavity was formed using this machine. Final machining of the modules was performed using a standard milling machine. The backshort cavities were made using a 3/32” end mill, of which I only managed to break one! A final detail on the housings is the use of registration pins used to precisely mate the top and bottom halves. At 47 GHz, tiny mismatches add

81 6

up. As the waveguide is only 224 mils by 112 mils, an error of just 10 mils becomes an error of 5-10% Figure 12: Upper half of housing with EDM’d main cavity (covered with Eccosorb) and machined waveguide backshort

Controlling the servo motor is an Arduino microcontroller board. (Figure 14) This easy to program controller includes a PWM output and several A/D inputs. I use a pair of 10-turn trimpots connected to these A/D inputs to set the transmit and receive positions for the switch. These in turn set the PWM levels which control the servo. Since I have a complete microcontroller, I also use it to sequence the transmit and receive modules. An additional position indicator switch on the waveguide switch functions with the microcontroller to protect the transmitter from putting power out until the switch is verified to be in the right position. Figure 14: Arduino microcontroller module

.

6.0

Waveguide Switch Design and Construction

Commercial waveguide switches are available for WR-22 in very small numbers and for a very high price! I decided to build my own, based in part on the success I had building a WR42 switch for my 24 Ghz radio. The switch is a simple rotary type switch made from aluminum. I chose to use a hobby servo control motor to position the switch. (Figure 13) There are a number of advantages to this technique including low power consumption and electrically adjustable stops. Figure 13: Waveguide switch with servo

82 7

7.0 Antenna I chose to use a surplus 39 GHz 12 inch diameter dish antenna for this project. It uses a splash plate style feed and has been used by others on 47 GHz with good results. Unfortunately, the feed uses round waveguide and all of my hardware is based on rectangular waveguide.

Figure 15: Round to rectangular adapter

Instead of buying a large and costly tapered transition (round to rectangular) I built a ¼ wave transformer style transition. Modeled in HFSS, this transition is simple to make with a milling machine and gives reasonably good return loss (~10 dB) when integrated with the antenna and waveguide switch.

For those devices needing negative gate bias (the output stage and the PA) I use a circuit that not only sequences the voltage (assuring gate bias exists before drain voltage) but also protects the drains from excessive current and overvoltage conditions. Typically this circuit is realized using a feedback system, automatically adjusting the gate voltage for sensed drain current. For large MMIC devices this topology can be very difficult to construct, as the low frequency gain of the transistors can be enormous. This can lead to an oscillation condition if the design does not account for the reactive components in the feedback loop. Figure 17: DC Biasing board with 6 channels of overcurrent/overvoltage protection circuits and gate bias potentiometers

Figure 16: T/R module with waveguide switch and feed waveguide

8.0 DC-DC Converter and Bias Control Biasing up GaAs PHEMT devices can be tricky. This is especially so for devices with large periphery. In this project the majority of the MMICs are self-biased. This makes biasing much easier as only a single positive supply is used, and no sequencing needs to take place. For those devices I use a simple linear regulator to clean up the output of a switching converter.

Another downside of the typical feedback system is that it offers little protection if for some reason the bond wire leading to the gate breaks, or some other open condition occurs. Unless designed otherwise, the circuit will simply reduce the gate voltage to it’s minimum while drain current flows unlimited. I use the Linear Technologies LT4361 chip for the overvoltage/overcurrent protection. In addition to the listed features, it also has a softstart function, which ramps up the drain current in a smooth manner rather than the abrupt pulse of a switch, which could lead to voltage spikes if lead lengths are long enough. Acting as a fuse, the LT4361 will shut down the drain current if it excess a pre-programmed (via a sense reissotr)

83 8

level.

All of the MMICs I am using run off either +4 or+5v, and converting this from the 12V input is the job of a two-stage regulator. I use a switching converter to drop down to 6.5 volts (primarily for efficiency reasons) and then low drop out linear regulators to get to the final 4.0 or 5.0 volts. The switcher (Linear LT1074 with 5A switch) is enabled only in the presence of the negative voltage. Figure 18: Power supply board for MMIC section including a switching converter (right) and three linear low-dropout regulators (left)

9.0 Putting it All Together Most of the microwave radios I have built have been constructed in some pre-existing chassis or box. This time I built a custom chassis that features a great deal of mounting flexibility and because every panel can be taken off, makes troubleshooting and repair fast and easy. The box consists of ½ inch square profile edges machined with a recess to allow flush mounting of sheet aluminum sides. The front and back are 1/4 inch thick aluminum plates. Holding the whole thing together are a billion 4-40 screws. The box took a LOT of time to build, mostly in drilling and tapping all those holes. Figure 20: Top view of the radio showing modular construction and chassis

Gate voltages are set using 10-turn trim pots. I used a specially made breakout box to set the gate biases by monitoring the drain current. Figure 19: Drain current breakout box inline with the bias board and MMIC modules

The two circuit boards described above were fabricated by PCB Express. If you would like schematic diagrams and layouts of the boards, email me and I will send them to you.

84 9

Every circuit board or piece of electronics is further enclosed in its own machined housing. In the case of the DC-DC converters, the LO sources and certain RF components, this is essential for good thermal conduction to the outside. The small cases also provide RF shielding and mechanical robustness. In a radio this complex it is too easy to pull on a wire or short something together and cause some expensive or time consuming part to release smoke.

Figure 21: Partially populated chassis (LO components shown)

As this radio is fairly heavy, I added handles to the front panel and a large handle on the top. A sighting scope is attached to the left side along with a bubble level. Figure 2210: Completed radio with antenna, handles, and scope (on the backside)

10.0

Real World Performance

This radio was completed in July 2010 and saw its first test in the driveway of K6JEY. We made a short contact and did a basic checkout.

Figure 23: Testing the radio in Doug K6JEY's driveway

The next set of tests were done with Gary AD6FP. On Saturday July 25th 2010, I set up my radios on top of Frazier Mountain about 100 miles north of Los Angeles, CA. Gary drove out to Mojave, about 60 miles away to a spot where four grid squares come together. We made contact at this range and I collected four grid squares. He then drove up into central California for a 100 mile contact. This gave me my 5th grid square and in a period of a few hours I had made enough contacts on 47 GHz to qualify for DXCC! In addition to the DXCC, I broke my personal best DX by a very wide margin! Signal reports from Gary were excellent and showed that there was plenty of margin for longer contacts. All of my information was sent using SSB and was easily copied. Gary’s CW (coming from a multiplier output) was loud enough to copy without much effort. During the first weekend of the 10 GHz and Up contest we made an attempt at the world record for 47 GHz. I was again stationed at Frazier Mountain and Gary operated from two positions at 256 miles (412 km) and 327 miles (526 km). Over the weekend I worked Gary on 10 Ghz at both locations and at the closer location on 24 GHz. Frank, WB6CWN made contact with Gary on 24 GHz at both locations, just a few miles short of a new 24 GHz record! The 47 GHz band did not give us similar success. After hours of trying no signals were ever heard from either end. Gary drained two large batteries sending his 30 watt 47 Ghz signal my way.

85 10

Figure 11: The author copying CW from AD6FP during a 60 mile QSO atop Frazier Mountain (47 Ghz radio on the far side, not visible)

12.0 References and Resources [1] http://www.hamradio.com/sbms/sd/47ghzndx.htm [2] Full Band Waveguide-Microstrip Probe Transitions, Yoke-Choy Leong; Weinreb, S.; Microwave Symposium Digest, 1999 IEEE MTTS International pages 1435-1438 vol. 4 [3] Tony’s Diamond Chop Saw http://mightyohm.com/blog/2009/07/tonysdiamond-chop-saw-part-1/ Resources for wire bonding, etc.: Wedge Bonding Introduction: http://www.gaisertool.com/catalog/4.pdf

11.0 Conclusions While there is work left to do on the radio before I can truly call it ‘high performance,’ I have met some of my personal goals including much better DX, and learning how to build with bare MMIC dies, wire bonding, and precision machining. My near term plans include replacing the receive module to improve noise figure and gain, and modify the transmit module (or just build another one) to get closer to the output power I expect. After working solid for 3 years on this radio I’m really glad to see it working well. It could not have been completed without help though, and I would like to thank the following people for their assistance in this project: Frank Kelly WB6CWN, Gary Lauterbach AD6FP, Kerry Banke N6IZW, Mike Aust WB6DJI, Derrick Yamauchi KE6QXR, Doug Millar K6JEY, Dave Glawson WA6CGR, Luis Cupido CT1DMK, Ernesto Casco Mansoor Siddiqui, David Brunone, Ben Gorospe, Sam Esparza, and Rich Katz, Thomas Spretke

86 11

Wikipedia article on wire bonding: http://en.wikipedia.org/wiki/Wire_bonding Another introduction to wire bonding: http://extra.ivf.se/ngl/AWireBonding/ChapterA1.htm Toshiba’s guide to wire bonding FETs: http://ni.toshiba.co.jp/snis/des/micro/technical/AP P-1.pdf California Fine Wire (sells wire bonding wire) http://www.calfinewire.com/ Bonding Source (sells everything you need for wire bonding, in small quantities) http://www.bondingsource.com/ Some details on wire bonding wire metallurgy http://www.williams-adv.com/documents/goldbonding-wire.pdf Richardson Electronics (distributors for several MMIC brands) http://www.rell.com/Pages/home.aspx Advanced Thin Films (Alumina and Quartz substrates) http://thinfilm.com/

NJR2145J 10 GHz pre-amplifier adaptation and construction Gary Lopes – WA6MEM [email protected]

The pursuit of perfection is a never ending process for the microwave enthusiast. The very notion that you can eek another dB of gain or reduce the noise figure 0.1 dB keeps all of us looking for the perfect design. The intent of this paper is to provide guidance for the conversion of the popular NJR2145J TV LNB to the 10 GHz amateur band to be used as a receiver pre-amplifier and the adaptation of the amplifier from wave guide input to an SMA connector interface. The methods used for this work employ basic hand tools and a drill press. I hope to show that most anyone can enjoy and be successful building and modifying microwave equipment. I built my present 10 GHz radio a several years ago and had employed a receiver pre-amplifier that was less than optimal. The gain was ok but the noise figure was approaching 2 dB and I knew something better was available that would allow me to hear that elusive carrier that was just out of the noise. I had acquired a surplus NJR2145J unit and based on the published performance specifications it seem to be the perfect replacement for my old pre-amplifier. The modifications to the circuit seemed simple but it used a wave guide interface. Because my 10 GHz radio was rather compact and used coax interconnects I had no room for a “large” WR42 transition. I had to have a slim, coax connected, preamplifier.

To begin the conversion and modification process, the following steps were completed: 1. Remove nut on the F connector and slide the metal case off.

87 1

2. Remove the die cast lid. Save the lid! You will need it for the finished product. Unsolder, remove and discard the rectangular DRO can on the backside. Remove the mounting screws for the small DC power PCB on the other side of the unit and lift out of the way. This will allow removal of the DRO.

3. Remove the PC board and attached DC power PCB from the main housing.

Prepare an aluminum 2.75” x 1.25 x 0.125” plate to be used for mounting the amplifier.

88 2

4. Position the main pre-amplifier board on the aluminum and mark the four mounting screws locations using the die cast lid as guide. 5. Drill and tap the four 4-40 size holes that will be used for mounting the board and the die cast lid to the aluminum plate. Also, drill two 4-40 holes for the power supply board to be mounted to the side of the amplifier. 6. There are two areas of protruding wires on the backside of the pre-amplifier board that require a corresponding “well” on the aluminum plate to prevent these connections from shorting. I wish I owned a mill to make this step a bit more elegant with a corresponding aesthetically better result but I don’t. I used a Dremmel tool with a small carving bit to hog out the appropriate areas required to allow clearance of the connections on the board. 7. Mount the pre-amplifier board to the aluminum plate with screws used for the die cast cover. 8. Remove the three terminal component marked R1E from the board. Remove the circuit traces highlighted in green in the photo.

9. Drill two 0.052” (#55 drill) holes using the two holes already on the board as guides. These will be used for the SMA connector center pins.

89 3

10. Mount the board on to the aluminum plate. Mark the position of the two #55 drill holes on the aluminum plate. 11. Remove the board from the aluminum plate and drill two 0.1610” (#20 drill) holes through the aluminum plate on the previously marked spots.

12 Insert two SMA female connectors in the two holes. The connectors should be the type with the extended Teflon insulator body. Mark the positions of the connector mounting holes. Drill and tap two 4-40 holes for each connector as shown.

90 4

13. After the connectors are mounted to the aluminum plate with the mounting screws, use a flat razor blade to trim the excess Teflon around the center pin of the connector. The top of the Teflon should now be flat with the surface of the aluminum plate.

14. Mount the pre-amplifier to the aluminum plate by aligning the two connector pins to the two #55 holes. Carefully slide the SMA center pins through the holes and seat the pre amplifier flat on the aluminum plate. 15. Attach the pre amplifier to the aluminum plate with the four mounting screws to insure the amplifier is flat against the aluminum plate. 16. solder the two SMA cent pins to the amplifier input and output connections. Remove the excess center pin sections above the soldered connection point. 17. DC power for the amplifier is supplied to the connection where the F connector attached to the board. You can drill a hole and use a DC feed through pin as I did or simply solder a wire to the DC

91 5

point. And bring it out under the die cast lid. Be sure and cut a notch in the lid so the DC feed wire is not pinched. 18. Remove the screws from the amplifier and install the die cast lid using the four assembly screws. Attach the power supply board using the two mounting holes.

92 6

That’s all there is to it! When you’re done you have a small and clean 10.368 GHz receiver pre amplifier. The preamp has an internal regulator that will take from 11V up to 30V. It works nicely on 12VDC without an external regulator. The measured performance was as expected. A measured 32 dB of gain and a 0.94 dB noise figure was achieved. That should give me just what I need to hear that weak one just out of the noise. This is a great project that can be completed in a few hours and simple tools. I would like to recognize Frank Kelly, WB6CWN, for his encouragement and guidance in perusing this project.

93 7

Propagation Observations with the 10 & 24 GHz VE4 Beacons By Barry Malowanchuk VE4MA Introduction The centre of Canada has not been known as a hotbed of microwave ham activity. I am located approximately 50 deg N by 97 deg W and about 60 miles north of the boundary between Canada and the US states of Minnesota and North Dakota. The location is dominated by the Red River valley ( of the North) and I am about 350 miles from Minneapolis – St Paul the nearest centre of activity in the US. None the less this location offers some extremes in weather that have long been known to produce some interesting propagation in the VHF and UHF spectrum. This paper describes the establishment of 10 & 24 GHz beacon transmitters in the fall of 2009 and the propagation observed over the following fall, winter and spring seasons Propagation in the Red River Valley The Red River Valley is the bottom of an ancient lake bed about 300mi (530 km) long, 50 miles (80 km) wide, extending from south of Fargo, ND to Lake Winnipeg in Manitoba as shown in the yellow ( N-S running) region of Figure 1.

  Figure 1 Red River Valley of the North Within the valley the terrain is very flat as exemplified in Figure 2 which is a typical rural road in the valley. The valley is “Shallow” and “Very Flat” but does have a sharp 400m step on the western edge, while the eastern edge is less distinct.

94

As I have operated on 432 MHz and above in this area since 1966 this area has a long history of producing enhance propagation within the valley. This has been supported by the operations of VE4MA, W0PHD, K0AWU and NT0V on both 432 & 1296 MHz. The big question was whether the valley supported extended enhance propagation to 10 GHz & Beyond? We know from basic propagation theory that 24 GHz wants low humidity situations for good results, yet we know that W5LUA and WW2R hold a North American 24 GHz record of over 500 km resulting from a tropospheric enhancement. Similarly in Europe there has been some long “coastal” openings where water vapour is clearly evident in the equation. This region is relatively unique as it is noted for very low winter temperatures with ample “dry” snow and a relatively dry climate which can produce low dew point temperatures that are essential for good microwave propagation on the highest frequencies. Yet at the peak of summer, high temperatures, humidity Figure 2 Typical Red River Valley Terrain and heavy thunderstorms are also experienced, so that the area offers an ideal location for propagation study, offering flat terrain, and all combinations of weather.

95

VE4 Microwave Beacon Site The only thing missing is steady signals to monitor on a regular basis. There are unfortunately very few UHF and microwave operators within a reasonable distance to provide signals even on an occasional basis let alone continuously. Fortunately access was secured to a communications site at Letellier, MB about 12 km North of USA and right at MN/ ND border line (EN19id). The site is 80 km / 50 miles from VE4MA, 113 km/ 70 mi from W0PHD, and 160 km/ 99.5 mi to NT0V and is ideally situated near the middle of these stations ( see Figure 3), which is ideal for monitoring rain scatter or tropo conditions. In fact in July 2008 a 500 mW 24 GHz beacon was located near this site for one (1) day and a big tropo enhancement was observed

Figure 3 Location of Letellier MB Beacon Site The site does have a microwave tower, but with the limitations of antenna cables and wanting convenient access, the antennas were located at a low elevation. 10 GHz Beacon Details

96

•

Frequency 10368.300 +/- 5 kHz from a Frequency West brick

•

Equipment mounted in climate controlled building

•

Transmitter is 9 W with CW ID (A1)

•

Antenna is a 2 x 16 horizontal slot (omni-directional) at 15 ft (see Figure 4) mounted on tower cable bridge. This clears the local building but is partially obstructed by large tower structure. It is connected by 25 ft of EW-90 Elliptical Waveguide.

Figure 4 Location of 10 GHz Antenna                      Figure 5 Location of 24 GHz Equipment 24 GHz Beacon Details •

Frequency 24192.000 +/- 30 kHz with a Frequency West brick

•

Equipment mounted outside in a weather proof, ventilated but unheated box at 10 ft above ground atop the tower cable bridge (see Figure 5).

•

Antenna is 17 dB gain 90 degree sector horn (90 deg wide but only a few degrees high) and pointed at Winnipeg (EN19lu)

•

Transmitter 1/2 Watt with a CW ID (A1)

•

Antenna is looking between Building & Tower which is not ideal for a 90 degree sector antenna but unavoidable (see Figure 6)

97

  Figure 5 Narrow 24 GHz Antenna Window VE4MA QTH 10/ 24 GHz Systems Unlike many microwave amateurs my location is clear of high trees and buildings so that I have been able to install my 10 & 24 GHz antennas on my tower and have an unobstructed view of the horizon. Because of the extreme weather and the need to operate EME as well as terrestrially, I have chosen to locate all electronic equipment inside and connect it to the antenna with good “low loss” elliptical waveguide. Because the 10/ 24 GHz antenna is mounted at only 50 ft, even with the relatively high loss of elliptical waveguide (6 dB/ 100 ft for EW180) the losses are acceptable. Of course the losses can be easily overcome by upsizing the antennas slightly. As shown in Figure 6 I have a 30 inch offset dish, with an inverted feedhorn location to avoid water accumulation and bird interventions with feedhorn windows. The feed is an early version of the W1GHZ dual band feed. The beamwidth at 24 GHz is approximately one (1) degree and he dish has elevation control with a digital readout in order to ensure that the dish is on the horizon in all directions. Unfortunately my tower is not perfectly vertical! 10 & 24 GHz Beacon Observations The 10 GHz beacon was installed on September 10, 2009 and the 24 GHz beacon on November 31, 2009 (the last nice day!). The 10 GHz beacon is normally 15 dB/ noise at VE4MA, detectable at S0 at NT0V, but not easily found at W0PHD (trees near QTH). The 24 GHz beacon is not normally heard as it’s about 10- 15 dB below 10 GHz signal (winter time conditions). This path to VE4MA has about 43dB of Scatter Loss above the “Line of Sight” losses.

98

  Figure 6 VE4MA 30 Inch Dish for 10 & 24 GHz It is known that 24 GHz is severely impacted by the Water Vapour absorption peak at 23 GHz. Mike KM0T has had a lot of experience with 24 GHz, including rain and snow scatter on a spot QSO basis, but perhaps we can learn more from more continuous monitoring? For the six (6) month monitoring period from December 1, 2009 to May 9, 2010 it is amazing that FREQUENT enhancements are seen on both bands and both bands are most often enhanced similarly! Enhanced conditions were seen on the following dates: •

Dec 1, 5, 9, 12, 20, 24, 25, 26 (VE4MA Vacation Jan 1-9)

•

Jan 21, 22, 23, 25,26

•

Feb 1, 2, 6, 9, 10, 16, 17

•

VE4MA Vacation Feb 20 –April 3

•

April 5, 6, 9, 10, 17, 18, 19, 22, 23, 26

•

May 1, 9

•

Note Bold shows ~43 dB enhancement (i.e. flat earth conditions), while Bold Italics represents Super Refractive conditions with gain above flat earth conditions.

As indicated earlier the normal 10 GHz Signal is about 15 dB above the noise. This is quite consistent across winter, spring and on average in summer the signal level does rise with the increase in refractive index of the atmosphere, and it’s difficult to identify an average signal level but it probably rises to 20-25 dB above the noise. For my main transceiver I am using a Flex SDR-1000 and the Panadapter display

99

allows an excellent view of signal conditions and characteristics as we will see in displays further in this paper. The normal winter time 10 GHz signal level is shown in Figure 7.

  Figure 7 Normal Winter 10 GHz Beacon Signal There are many displays that can be shown but in order to limit them I will show examples of “Light Snow Scatter”, “Heavy Snow”, “Ice Fog” and “Super Refraction” enhancements. Light Snow Scatter Figure 8 shows the local Environment Canada radar trace for December 12, 2009. It shows some very light precipitation traces in the general area of the beacon site. The circles around the radar site are in 50 km increments, and Environment Canada uses “anti-clutter” reduction techniques to clean up the display within a 150 km radius of the site (first 3 rings). Light snow was known to exist over the whole area. The temperature at the time was -22 C ( -8F) and the winds were light at 11 km/hr (7 mph). The 10 and 24 GHz signals were up slightly by about 5-10 dB with 10 G being 15-20 dB/ N and 24 GHz about 8 dB/N. There did appear to be a slight spreading of the signals.

100

  Figure 8 Radar Display of Light Snow Heavy Snow On December 25, 2009 there was a heavy snow storm moving North into the area with very strong winds. The signals were strongly enhanced initially but dropped as the storm moved further into the path (see the radar trace in Figure 9). The signals were heavily distorted (spread) although 24 GHz was not spread as much as 10 GHz (see Figure 10). This may be related to the differences in antenna beamwidth. Ice Fog On February 1, 2010 there was an event that from long term memory is fairly common in this part of the world. When it gets “really cold” there is no wind and whatever water vapour is in the atmosphere tends to stratify in the atmosphere and even sublimate (condense) out to frost on ground based objects. The temperature on this day was -28C (-18C) and virtually no wind (5kph or 3mph). Signals were enhanced by 40 dB or more on both bands.

101

Figure 9 Heavy Snow Storm Moving into Area on December 25, 2009 The Canadian weather radar shown in Figure 11 showed nothing going on but of course the anti-clutter software was probably eliminating any trace of the atmospheric conditions. It does show some activity North-West of Winnipeg which I know is related to the rise in terrain leading to the “mountains” in Western Manitoba…this is ground clutter just outside the 150 km suppression zone. Figure 12 shows the signals on 10 & 24 GHz. Both signals are clean and free of spreading, with the 10 GHz signal peaking at 60 dB/ N on 10 GHz and about 40 dB/ N on 24 GHz. The bands were subject to independent fading as might be expected, so that one band might fade while the other remained unchanged. Super Refraction Event February 16, 2010 There are occasions where the signal levels exceed that which would occur if there was no curvature of the Earth and this occurred on February 16. This event was similar to many experienced over the years and observed on VHF/ UHF. Figure 13 shows an infra-red satellite picture that is showing low cloud/ fog over the Red River valley. Notably the difference between 10 & 24 GHz signals shown in Figure 14 has dropped to only 10 dB. The slight spreading on 24 is not propagation related. 

102

  Figure 10 Heavy Snow Scatter on 10 & 24 GHz Super Refraction Event April 9, 2010 Spring has arrived just as I have returned from a vacation in Arizona. The temperature is +9.6C (49 F)…a nice spring day! Notably the Relative humidity is low at 29% with a low Dew Point temperature. As per usual the Canadian radar does not show much and neither does the USA radar due to anticlutter software. Again the difference between the 10 and 24 GHz signals is only 10 dB but the 10 GHZ signals are 50 dB above normal at 65 dB/ N as shown in Figure 15a & b. Note that the spurs shown in the 24 GHz picture of Figure 15b are internal to the SDR-1000.

103

  Figure 11 Radar Display of Ice Fog Conditions Although the Canadian and US radar did not show anything to suggest enhanced propagation on the evening of April 9, the next morning the enhancement had disappeared from the beacon signals, however upon examination of the radar from Grand Forks ND in the US and Dryden ON in Canada clearly show enhancement in the Red River Valley and outside the 150 km radius around Dryden. Please see Figures 16a & b. The indications of this event are very similar to the February 1, 2010 Ice Fog event. Affect of Summer Weather Previously during the dry winter months we had seen the 24 GHz signals about 10-15 dB lower than 10 GHz but after about April 23 the 24 GHz signals dropped by about 20 dB! This coincided with a rapid and dramatic increase in the humidity levels and the disappearance of the almost daily night time inversions. Unfortunately monitoring of 24 GHz had to stop in order to reconfigure the station to have a separate 24048 MHz station for EME and another for 24192 MHz terrestrial operations. Monitoring resumed about August 21st but no signal has been found to date (September 1st).The status of the 24 GHz beacon will be confirmed in short order.

104

Summary of Propagation Encountered with 10 & 24 GHz VE4 Beacons There is a lot of data that has been collected in the six (6) month monitoring period but in summary I have found: 1. An unexpected high frequency of enhanced conditions 2. Subsidence type inversions are frequent with: –

“Dry” winter air with Low Dew Point temperatures

–

Calm winds in late evening, after a “Warm” Day leading to a “Cool” night

3. Rain & snow scatter works well on 24 GHz –

But only when there is not too much moisture to be travelled through

4. Often 24 GHz is enhanced a little more than 10 GHz 5. 24 GHz is severely impacted by summer month humidity levels (even in Canada “eh!”) 6. The lower power of the 24 GHz transmitter and more directional 24 GHz antenna makes direct comparisons difficult for all situations. Possible Future Work Expansion to include a 47 GHz beacon would be very interesting propagation wise! It certainly is possible however the technical challenges are considerable. Given the increased path loss and lower transmitter powers (~100 mW) available this would have to be offset with higher gain antennas. I would have to install another dish at home but not on the existing tower. The equipment would have to be mounted at the dish at both locations. Site access would be more difficult as tower mounting would be necessary and alignment of the antennas at both ends would be a big challenge without signals available to peak on. The frequency control of the beacons needs to be improved on all bands with a higher stability reference of whatever form. This is especially critical for equipment mounted out of doors that is subjected to such a wide range of temperatures but would be a welcome addition even for the indoor mounted 10 GHz equipment.

   

105

  Figure 12 Signal Levels During "Ice Fog" Event

  Figure 13 Super Refraction Event on Feb 16

106

  Figure 14 Signal Levels for Feb 16 Super Refraction Event

  Figure 15a 10 GHz Signals with April l9, 2010 Super Refraction Event

107

  Figure 15b 24 GHz Signals with April 9, 2010 Super Refraction Event

  Figure 16 a & b Radar Clutter after April 9, 2010 Super Refraction Event

108

PIC’n on the ThunderBolt By John Maetta N6VMO

More radio and test equipment manufacturers are adding external reference frequency inputs to their products. As a consequence, many amateurs are incorporating GPS Disciplined Oscillators (GPSDO) into their shacks. Microwave enthusiasts and many VHF, UHF amateurs have been snatching up Trimble ThunderBolt GPS Disciplined Oscillators from the surplus market. These once very expensive and accurate time pieces are now within the budget range of most amateurs and experimenters. My ThunderBolt ‘kit’ came with a power supply, antenna, coax, serial cable and the GPSDO receiver. A GPSDO provides highly accurate 10MHz and 1PPS (Pulse per Second) signals to synchronize your radios and test equipment. My ThunderBolt GPSDO is used to provide a 10MHz reference frequency to my Flex-1500 Software Defined Radio and my VHF/UHF/uW transverters. Frequency accuracy is better than 1 ppB (Parts per Billion) or ± 1.4Hz on the 2 Meter band. The ThunderBolt is equipped with a 9600 baud RS-232 port to obtain status and to control several aspects of its operation. Trimble has developed a Windows GUI interface, called TBoltMon.exe, which will enable you to control or obtain status to and from the ThunderBolt via a PC com port. TBoltMon.exe is available on the internet for free. The problem… Although the TBoltMon software is easy to use, amateurs have found that they only need to know if the ThunderBolt has acquired enough GPS satellite data to achieve a phase lock of its 10MHz and 1PPS outputs. Many don’t like having to connect a serial cable and then run TBoltMon to see if the GPSDO is ready to use. Also, having a PC constantly connected to the ThunderBolt is not always practical. Monitoring and reporting ThunderBolt status seemed like the perfect job for a PIC microcontroller. This article will only detail the PIC status module design, software description and circuitry. Your construction method, installation and the PIC programming interface is a personal preference. There are many simple homebrew and inexpensive PIC programming hardware solutions available on the internet. Or, I can also program your PIC and get it back to you for return postage if you choose. The solution… I mounted my ThunderBolt receiver and its power supply in an appropriate enclosure, with access ports for its various outputs and power. The PIC microcontroller chosen for this function is the Microchip PIC16F627 (IC1). It is powerful, fast and cheap. It has 1K bytes of RAM and a built-in USART (Universal Synchronous Asynchronous Receiver Transmitter), also known as a serial port. The PIC’s USART can be easily programmed to transmit and receive the 9600 baud ThunderBolt commands and status packets. The results can then be displayed by using just two LEDs. What data and how much?… The ThunderBolt can provide more status data than this project requires. Unfortunately, it does not output standard NEMA data, it communicates in ASCII HEX. You can reference the Trimble Standard Interface Protocol section A.10.31 of the ThunderBolt User Guide for more information on command and status data.

109

When the ThunderBolt reports that it does not have any Critical or Minor alarms, the oscillator is phase locked and is useable. All the PIC needs to do is send the ThunderBolt the request for its Supplemental Timing Packet (8F-AC), then read the data and check for any alarms. The circuit… As powerful as it is, the PIC can’t perform all these tasks alone, requires some additional components to perform this task. You will need a 4 MHz crystal(Q1), a MAX233 RS-232 level converter integrated circuit (IC2), LEDs, resistors, capacitors and something to mount these components on. All these components are detailed in the Bill of Materials at the end of the article. You can use any method you prefer to build this circuit. I used a PIC prototyping assembly board to develop the software, hardware selection and schematic. Once the circuit and software were stable, I used Eagle Layout Editor from CadSoft to develop a printed circuit board. The “Light” version is available free on the internet. Power requirements are +3.3 to +5 volts DC. Current draw is ~100ma. You can use the ThunderBolt’s external +5 volt DC supply to power this project. The MAX233 (IC2) converts the PIC’s USART RS-232 levels, at RB1 and RB2 into the standard RS232 levels required by the ThunderBolt at X1. PIC outputs RB4 and RB5 drive the two status LEDs (LED1, LED2). The program… The PIC will start to execute the program upon power up and will transmit the status request to the ThunderBolt. The ThunderBolt will then respond back with the 68 byte serial Supplemental Timing Packet data. The PIC will store each of the 68 bytes into its internal memory. After all 68 bytes have been transferred and stored, the PIC will read bytes 8 and 9 (Critical alarm data) and add them together. If the sum of these two bytes is not equal to zero, the PIC brings RB4 high to light LED1 indicating that one or more Critical alarms are present. The PIC then reads bytes 10 and 11 (Minor alarm data) and adds these two bytes together. Once again, if the sum is not equal to zero, the PIC brings RB5 high to light LED2 indicating that one or more Minor alarms are present. If no alarms are detected then LED1 and LED2 remain extinguished. The process repeats every 6 seconds. Preparation and Operation… The following instructions assume you have just received your ThunderBolt and installed the correct outdoor GPS antenna, downloaded the Users Guide and familiarized yourself with the operation of TBoltMon. More than likely, your ThunderBolt may have arrived factory configured to continuously output all data packets every second. Or, its last user my have altered its configuration. So, we must prepare the it to stop automatically transmitting status and cooperate with our PIC status module. We’ll instead have it wait for our PIC to ask it for status. To do this, connect the ThunderBolt serial port to a PC com port and run TBoltMon.exe. Pull down the “Setup” menu and select “Packet Masks and Options…”. Uncheck all boxes in the “Packet 8E-A5 Masks” window. Click “Set Masks” button and then “Save Segment” button. Click “Close” and exit TBoltMon. Connect the PIC status module to the ThunderBolt serial port. Power them both up. Depending on how long the ThunderBolt has been powered off, it can take up to one hour for it to acquire satellites, clear all of its alarms and turn off both LEDs.

110

Conclusion… Now all you need to do is glance over at the ThunderBolt to see that it is ready to use. Hmmm…it may be time to enter this year’s ARRL Frequency Measurement Test!... The PIC was programmed using PICBasicPro 2.60A. All program source files, schematic, PCB files, and a bill of materials can be downloaded at www.n6vmo.com\PIC\TBOLTVMOn.zip TBoltMon.exe http://www.trimble.com/support.shtml ThunderBolt Users Guide http://www.trimble.com/support.shtml Eagle Layout Editor http://www.cadsoftusa.com 73, John Maetta, N6VMO 460 Milky Way Lompoc, CA. 93436 [email protected]

111

Development of an UWB CMOS Transmitter-Antenna Module Meng Miao and Cam Nguyen Department of Electrical and Computer Engineering Texas A&M University College Station, TX 77843-3128 Abstract A low-cost, low-power fully integrated CMOS transmitter-antenna was designed, built and tested for UWB systems. The CMOS pulse generator can generate monocycle and impulse signals with tunable pulse duration. The UWB antenna is integrated directly with the CMOS chip. The antenna has less than 18-dB return loss and can transmit and receive UWB impulse signals over the entire UWB bandwidth of 3.1–10.6 GHz. The UWB transmit-antenna module can generate and transmit both monocycle pulses from 140 to 350 ps and impulses from 100 to 300 ps. Introduction Impulse ultra-wide band (UWB) technique implementing CMOS radio-frequency integrated circuits (RFICs) is attractive for high-data-rate, short-range communication, radar and sensor systems because of low cost, low power consumption, and easy integration with digital ICs. The pulse generator and antenna are two key components in both the transmitter and receiver in UWB impulse systems. Monocycle pulse has band-limited characteristics without DC component, facilitating its transmission through practical antenna, and is normally preferred. Meanwhile, pulse with tunable duration has both advantages of increased penetration or range and fine range resolution and is attractive for UWB systems [1], [2]. Tunable pulse duration is also useful for compensating variations caused by typical CMOS processes. CMOS tunable pulse generators should thus be attractive for UWB systems. The transmitted and received signals of UWB systems require antennas not only radiating energy efficiently but also having linear phase response. Current commercial UWB antennas [3] have relatively large size. There is a great need for low-cost, compact, easy-to-manufacture UWB antennas that are omni-directional, radiation-efficient, and have low distortion. The antenna should also facilitate integration with CMOS chips. In this paper, we present an UWB transmit module, integrating a CMOS tunable pulse generator and a uniplanar UWB antenna. The CMOS-based UWB transmitter-antenna subsystem is capable of radiating both tunable monocycle pulse (140-350 ps) and impulses (100-300 ps) for UWB impulse systems [4].

112

Design and Performance of the CMOS Tunable Pulse Generators Figure 1 shows the CMOS tunable monocycle pulse generator. It integrates a tuning delay circuit, a square-wave generator, an impulse-forming circuit, and a pulse-shaping circuit in a single chip. The fabricated chip using TSMC 0.25-�m process is shown in Fig. 2. The tuning delay component includes a pair of parallel tunable delay and reference cell using shunt-capacitor delay elements. Control voltage is applied to the tunable delay cell to produce continuous delay variation, while the reference cell is fixed to provide a reference. A series of inverters with increasing size form the square-wave generator to increase the drive capabilities and produce a square wave with fast rising/falling edges when a sinusoidal clock signal is fed. The impulse-forming block provides driving capability to the next stage and evokes the impulse response of the succeeding component. As the pulse-shaping circuit functions approximately like a differentiator, a monocycle pulse signal with tunable pulse duration is produced when the tunable impulselike signal is fed to the circuit. Tuning delay

Square wave generation

Impulse forming

Pulse shaping

Tunable delay cell

A NOR Reference cell

C

B

D

Output

Fig. 1. CMOS UWB tunable monocycle pulse generator.

Fig. 2. CMOS tunable pulse generator chip.

A separate chip without the pulse-shaping circuitry, generating tunable impulses, was first measured to verify the design concept and the results are shown in Fig. 3, which have 0.5–1.3 V with 100–300 ps duration. The measured tunable monocycle pulse signals are shown in Fig. 4. Symmetric monocycle pulses with 0.3–0.6 V and 140–350 ps duration were achieved. It is noted that the tunable narrow impulse generated at node C consists of three parts: rising edge, tuning delay, and falling edge. For pulses with very narrow width, only part of the rising and falling edges is involved, resulting in impulse with much smaller amplitudes. Consequently, a tunable monocycle pulse is achieved at node D. 0.4

1.5 V 2.0 V 2.3 V

Output Voltage (V)

0.3 0.2 0.1 0 -0.1 -0.2 -0.3 -0.4 1000

Fig. 3. Impulse signals with tunable pulse duration.

1500

2000 Time (ps)

2500

3000

Fig. 4. Measured monocycle-pulse signals.

113

Design and Performance of the UWB Uniplanar Antenna A compact uniplanar UWB antenna, which facilitates integration with the CMOS chip, is shown in Fig. 5. The area occupied by the antenna aperture is only 1.2 in  1.4 in. The operation of the antenna is based on the principles of non-uniform transmission lines and well-known traveling-wave antennas. In the coplanar waveguide (CPW) section to the antenna center, most of the energy is confined within the transmission line until it reaches the center, where the energy is coupled from the CPW to the two parallel 100- slot lines. Tapered slot lines are used to simulate an impedance transformer from 100  to about 377 . The tapered sections were optimized to produce minimum internal reflections for the antenna input signal over the UWB frequency range of 3.1 to 10.6 GHz. 0

Measured Simulation

-5

Return Loss (dB)

-10 -15 -20 -25 -30 -35 -40 2

Fig. 5. UWB transmitter-antenna module. 1

4

6 8 Frequency (GHz)

10

12

Fig. 6. Return loss of the uniplanar UWB antenna.

To verify the antenna performance, a separate antenna including SMA 0.8 Input impulse connector and uniform CPW section was 0.6 fabricated and measured. Fig. 6 shows the measured and simulated return loss in the SMA response 0.4 frequency domain. The corresponding TDR response results in time domain for a 0.2 50-ps input impulse signal is shown in Fig. 0 7. The response after 0.5 ns is caused by 0 0.5 1 1.5 2 2.5 Time (ns) the designed antenna aperture and, as can Fig. 7. TDR responses of the uniplanar UWB antenna. be seen, the measured result matches very well with that simulated, which confirms the antenna design. The TDR performance also demonstrates excellent time-domain behavior of the designed antenna, which is crucial for UWB time-domain impulse applications. The measured time-domain results indicate that better than 18-dB return loss is achieved for the antenna. TDR Response

Measured Simulation

Design and Performance of the CMOS Transmitter-Antenna As shown in Fig. 5, the CMOS tunable monocycle pulse generator chip is mounted directly onto the edge of the UWB antenna without a feed line. A quasimicrostrip antenna operating from 0.2 to more than 20 GHz is used as the receiving antenna for pulse transmission measurement of the UWB transmit

114

module. Figs. 8 and 9 show the signals received from the tunable impulse and monocycle pulse signals, shown in Figs. 3 and 4, transmitted by the UWB transmit module. The pulse-duration tunability is clearly visible in the received pulses. All the received signals have shape similar to the first derivative of the transmitted pulses, as expected from the designed antenna. Both the measured impulse and monocycle-pulse transmission results clearly demonstrate the workability of the developed CMOS-based tunable UWB transmit module. 20

10

1.5 V 2.0 V 2.3 V

15

5 Voltage (mV)

Voltage (mV)

10

1.5 V 2.0 V 2.3 V

5 0 -5 -10

0

-5

-15 -20 0

500

1000

1500 Time (ps)

2000

2500

3000

Fig. 8. Received signals of the impulses transmitted by the UWB transmit module.

-10 0

500

1000

1500 2000 Time (ps)

2500

3000

Fig. 9. Received signals of the monocycle pulses transmitted by the UWB transmit module.

Conclusion A low-cost, low-power fully-integrated CMOS-based transmitter-antenna module with tunable pulse duration has been developed for UWB impulse systems. Performance was verified experimentally for both impulse and monocycle pulses, demonstrating its possibility for use in various UWB applications including UWB communication systems, sensors, and radars. Acknowledgement This work was supported by the National Science Foundation. References [1] J. Han and C. Nguyen, “Ultra-Wideband Electronically Tunable Pulse Generators,” IEEE Microwave and Wireless Components Letters, vol. 14, No. 3, pp. 112-114, March 2004. [2] J. W. Han and C. Nguyen, “On the Development of a Compact SubNanosecond Tunable Monocycle Pulse Transmitter for UWB Applications,” IEEE Trans. on Microwave Theory and Techniques, Vol. 54, No. 1, pp. 285293, January 2006. [3] J. R. Andrews, “UWB signal sources & antennas,” Picosecond Pulse Labs, Boulder, CO, Application Note AN-14, February 2003. [4] M. Miao and Cam Nguyen, “On the Development of an Integrated CMOSBased UWB Tunable–Pulse Transmit Module,” IEEE Transactions on Microwave Theory and Techniques, vol. 54, pp. 3681-3687, Oct. 2006.

115

9/29/2010

Frequency Stability Measurement: Technologies, Trends, and Tricks

Presented at Microwave Update 2010 John Miles, KE5FX

The importance of time 

Time is a wide wide-range range parameter – scales of interest range from femtoseconds to years! – Time is also the most precise physical quantity we know how to

measure. Almost every measurement made by engineers and physicists ultimately relies on a timebase



When we talk about “stability”, we must specify the timescale of interest – Long-term stability (“Drift”) – what timescale(s)? – Short-term stability (“Phase noise”) – what offset(s)? – These look like different phenomena, but are really two aspects of

the same problem: unwanted changes in phase over time.

116 1

9/29/2010

Why measure longlong-term stability? 

  

Debugging gg g a new pproject? j Mysterious y pproblems are sometimes obvious in the long-term time domain – Loop stability-margin problems – “Phase hits” – Unwanted vulnerabilities to temperature, power, vibration…anything periodic Comparing and tweaking clocks: OCXOs, GPS/Rb/Cs standards, and more Understanding and optimizing your station’s behavior under different environmental conditions Precision timing opens new research areas to amateurs: bistatic radar, long-baseline interferometry, GPS enhancement…

Long--term stability measurement Long 

Frequency counter – Like spectrum analysis for PN – ‘measurement floor’ is not great – Best frequency counters resolve about 11 digits/second



Time Interval Counter (TIC) – Better resolution through interpolation and other techniques – Best TICs have single-shot resolution in the 10-ps range – Various kludges (DMTD, etc) used to improve resolution

117 2

9/29/2010

Long--term stability measurement Long 

Direct digital test sets – Measures phase like a TIC, but with SDR-like “process gain”  Can often measure phase noise as well – State-of-the-art resolution is in the 1E-15/second range  1000x better than the best counters!

118 3

9/29/2010

Why measure phase noise? 

Phase noise is a common topic p of discussion when serious homebrewers get together, from HF to microwave – PN tells you more about the health of your signal source than perhaps any

other measurement

– Historically one of the more difficult/awkward measurements to make  

Weak-signal work demands precise, repeatable tuning Weak signals may also be vulnerable to MDS degradation – WA1ZMS put it best: stability determines what signals sound like.

 

Diagnosing link problems is easier when you know what to expect Instrumentation design – the analyzer has to be cleaner than the DUT! (…or does it?)

119 4

9/29/2010

Phase noise is everywhere…  

No source or device above 0 Kelvin can avoid contributing jitter Multiplied references common in UHF-microwave work suffer 20*log(N) effect – 20*log(N) = simple consequence of jitter  Lag/lead time of any given edge remains constant through multiplication, but the carrier period shrinks – +60 dBc/Hz from 10 MHz to 10 GHz  Sometimes much worse – many PLLs divide before they multiply!  Even clean references can be degraded by process noise  Throwing money at the problem does not guarantee improvement

120 5

9/29/2010

… so how do we measure it? 

Direct spectrum analysis – Simply tune a spectrum analyzer to the USB half of the carrier



Indirect (baseband) spectrum analysis – Phase detector method – Frequency discriminator method – Two-port device measurements



Direct digital analysis – Recover and measure phase variations with DSP techniques

Direct spectrum analysis   

Measures composite p (AM+PM) ( ) noise Limited by instrument’s LO noise floor Calibration process involves a few factors… – – – –

Subtract carrier level if not 0 dBm Subtract 10*log(RBW) to normalize to 1 Hz BW Add 2.5 dB to account for averaging power in “dB space” Subtract ENBW of the RBW filter 



Usually about 0.5 dB for xtal/LC filters or 0.25 dB for FFT

S t measurements Spot t are often ft supported t d by b dBm/Hz dB /H markers k

– Note difference between dBm/Hz and dBc/Hz – use reference-level offset

to avoid confusion

– Better to use software!  PN from www.ke5fx.com/gpib/pn.htm  OEM phase-noise personality software (HP 85671A, R&S FS-K4…)

121 6

9/29/2010

Direct spectrum analysis

Direct spectrum analysis: some typical instrument floors

122

7

9/29/2010

Indirect PN analysis: Phase Detector method 

Downconvert signal from DUT to 0 Hz (“baseband”) ( baseband ) – Simple PLL with mixer as phase detector – Commonly-cited references  HP Product Note 11729B-1  www.wenzel.com/documents/measuringphasenoise.htm

Indirect PN analysis: Phase Detector method    

Requires q a reference at the same frequency q y as the DUT Injection locking can be a problem – need isolation amps Lots of options, with manuals the size of phone books Calibration process is much more detailed…

– All factors in direct spectrum analysis apply here as well – Plus the need to account for the test set’s response  Mixer’s sensitivity when used as phase detector (volts per radian)  Post-mixer LNA gain, if any  6 dB to t convertt ffolded ld d DSB bbaseband b d tto L(f)  Effect of PLL, if its bandwidth overlaps desired measurement range



Only a masochist would attempt indirect PN measurements without software support! 

KE5FX PN, HP 3047A, HP 3048A, Agilent E5500…

123 8

9/29/2010

Indirect PN analysis: Phase Detector method

Phase Detector method: some typical measurements

124 9

9/29/2010

Indirect PN analysis: Frequency Discriminator method 

Instead of a separate reference reference…. – Delay line converts df to dphi, then mixer converts dphi to dv – See HP 3048A manuals, HP Product Note 11729C-2

Indirect PN analysis: Frequency Discriminator method

125 10

9/29/2010

Indirect PN analysis: Two--port residual measurements Two   

Useful variation on discriminator measurement Replaces delay line with DUT Must drive splitter with a clean signal source or its broadband noise will decorrelate and fold…

Indirect PN analysis: Two--port residual measurements Two

126

11

9/29/2010

Typical indirect PN analysis gear: HP 11729B/C, HP 3048A

See www.hpmemory.org/news/3048/hp3048_01.htm Great collection of HP app notes on indirect PN measurement!

Homebrewing a quadrature PLL – Simple p type-2 yp PLL with DBM and opamp p p  http://www.wenzel.com/documents/measuringphasenoise.htm  Several other references at end of this slide deck – Can measure two sources with a microwave mixer – Can also use a downconverter for a single microwave source, with

a stable HF reference on the other port 

HP 11729B/C block diagram is a good example of this technique

– As with the commercial 3048A and E5500 packages, almost any

spectrum t analyzer l can bbe used d 

Quadrature-PLL measurements with RF analyzers are supported by PN.EXE – See last FAQ entry at http://www.ke5fx.com/gpib/faq.htm



Baseband analyzers offer some advantages, though…

127

12

9/29/2010

Baseband analysis for indirect measurements 

Advantages g of popular p p surplus p FFT analyzers y

– Faster ‘sweeps’ with FFT versus conventional RF analyzers – Resolves noise at offsets down to 1 Hz or better  Not too important at RF/microwave but can be good for HF work



Disadvantages versus RF spectrum analyzers

– Less dynamic range  Common to overdrive the front-end mixer in an RF analyzer for improved range, but ADCs don’t tolerate this  High-amplitude content near DC offsets has to be HPF’ed to see the broadband response  HP 3048A hardware+software switches filters for you, but it complicates homebrew solutions – Less third-party software support  PN doesn’t work with popular baseband analyzers like HP 3561A, 3562A

Baseband analysis: alternatives to older surplus gear 

SDR hardware S a dwa e with w t good LF response espo se

– RFSPACE SDR-IQ supported by TimeLab  www.ke5fx.com/timelab/readme.htm  $500 retail, 14-bit ADC, can ‘see’ from ~100 Hz-30 MHz



PC sound cards

– Planned support in TimeLab – Range similar to SDR-IQ, but with widely varying performance



Homebrew data-acquisition data acquisition hardware

– Analog Devices EVAL-AD7760 boards are about $150 each  100+ dB dynamic range at 2.5 MSPS  Overall highest performance LF-to-MF ADC I’m aware of  Need a very fast PC to perform realtime analysis at full rate!

128 13

9/29/2010

EVAL--AD7760 as baseband analyzer EVAL

EVAL--AD7760 as baseband analyzer EVAL

129 14

9/29/2010

EVAL-AD7760 as baseband analyzer: EVALHP 8642B measured via 11729C

Build a direct digital analyzer instead? 

State of the art pperformance of commercial gear is better than most users will need –

– –



Symmetricom TSC 5120A: about $25,000  ADEV floor near 3E-15/s  PN floor near –175 dBc/Hz Agilent E5052B: about $90,000  PN floor near –180 dBc/Hz A phase-noise analyzer with 10-15 dB worse performance would still be extremely useful

ADC eval boards to the rescue again… – –

–

2x AD9446100LVDS/PCBZ-ND ($220 each)  100 MSPS x16 bit, jitter = 60 fs RMS Nexys2 FPGA trainer ($129)  Spartan3E FPGA with 1.2M equiv gates  USB 2.0 high-speed interface, 30+ MB/sec Surplus Wenzel 38.025 MHz OCXO from eBay used for initial experiments ($25)

130 15

9/29/2010

Prototype direct digital phase noise/timing analyzer

Prototype direct digital phase noise/timing analyzer

131 16

9/29/2010

Timing performance shootout

Phase noise performance shootout

132 17

9/29/2010

http://www.ke5fx.com/stability.htm Collection of useful links for pphase noise and timingg metrology, gy, updated frequently Special thanks to Marc Mislanghe of http://www.hpmemory.org for contributing photos and artwork for this presentation, and to Agilent Technologies for their support in making the material available.

133 18

Compass Basics And Some Representative Types In much of our microwave field work, we use a compass to set the direction of our dishes. With smaller dishes it seems that a basic compass is all that is needed. As bigger dishes or higher frequencies are used the demands placed on a compass’s accuracy may not be matched by the design. There are five basic types of compasses that are in current usage 

A Boy Scout or really a Lensatic compass



A sighting compass



A Solar compass



A Fluxgate compass.

Knowing the advantages and disadvantages of each of these types of compasses, your needs for pointing accuracy, the local isogonic variation and other sources of error in measurements are all elements in getting your dish pointed where it is supposed to be. Sources of Errors Books have been written about the subject, but here we will center on a few of the most obvious sources of error. The first source is the construction of the compass. The liquid in the compass should provide enough dampening to allow for stable readings. Magnifiers and sighting mechanisms should be sturdy and easy to use. A compensator for local isogonic variation makes making true bearing readings much easier. A second source of very strong errors is local magnetic variations such as nearby vehicles and buried pipes and iron in rocks. Take a sample reading of North at several spots near where you are set up and see if the bearing changes. Try to isolate a place where you can get reading that has a minimum of interference. Know your local isogonic variation. The difference between true North and Magnetic North at any location is called its magnetic variation. The line along which that reading is the same is called the isogonic line. For short the whole thing is called the isogonic variation. That variation changes as a predictable but consistent amount. Our local yearly variation is about .1 degree in Long Beach, Ca. If you are using old data and are using an antenna with a 2degree beam width, you will have difficulty lining up correctly with another station if you do not have current data. Here is a map of the worldwide variation. Notice that it is not smooth or even.

134 1 

Data for the local isogonic variation can most easily be found on a pilot’s IFR chart for a nearby airport. They are also available on the internet and called an “Airport Diagram”. Here is an example of our local airport LGB. Notice in the upper left the isogonic variation of 13.2 deg E, the date of the measurement and the yearly variation. If you are planning a long hop over 100miles or more, it would be good to consult a nearby airport diagram in the area for their local variation. There may be local anomalies, like mineral deposits, that can give a unexpected difference in the variation.

135 2 

136 3 

With the above information a microwaver should be able to accurately find a bearing to within one degree with some measure of certainty. Compass Types Here are samples of each of the main types of compasses The first is a cheap lensatic type. It looks like it has the right parts. It has a sighting lid, a thumb holder, a gun sight and a way to match up the dial to True North. However it is so badly built that it is useless. A cheap Lensatic type

Military Compass The second is a Chinese Military compass available on Ebay for about $20. It has all the same features as the model above, but it works and works extremely well. It has a magnified sight that allows you to match up the direction the compass is pointed with a view of the compass direction. It does not have isogonic variation built in. the green triangle is matched up with the triangle on the top and then readings can be taken.

137 4 

Hiking Compass A more complex model is the Brunton 8099. It even comes with waterproof directions. It is more expensive, but it is built to last and to survive in adverse circumstances. It is also a sighting type. It includes a compensation for isogonic variation, so that readings are automatically in True bearings. It also has an inclinometer. Matching the dial to the card is very simple and very accurate. The circle on the top is adjusted so that it is over the N on the bottom. I believe this compass will easily read below one degree. Excellent sighting type A Brunton 8099

Boating Compass A top of the line boating compass, that floats, is a Morin compass. It has no isogonic compensation and is deceptively easy to use. Simply hold the compass up to the eye and read the bearing. Sighting is automatic. It is one of the most stable and easy to read compasses I have ever used. Morin sighting type

138 5 

Solar Compass This is a solar compass. It takes a bit of setting up and has to be recalibrated every half hour or so, but it can be used on a moving vehicle and is immune to local magnetic disturbances. Unfortunately it is useless at night, even with a flashlight. Well, unless you have the flashlight pointing exactly south. You also must know the time accurately and a few other measurements. It was used with great success in WWII in Africa. Solar Compass with Gnomon

Fluxgate Compass Many of the newer GPS units, like the Etrex Vista, come with a compass that combines readings from the GPS and a fluxgate sensor to give the user a bearing. All of the sources of error mentioned above apply here along with a few others. The GPS operates on batteries and generally ends up being left on for long periods of time. When you are trying to make the most of a location and contacts, a GPS compass might be much more fiddling and problem than a good analogue compass. The GPS compass is more expensive than an analogue compass with no increase in accuracy. I have an Etrex Vista and I like it very much for hiking and navigating. However for microwave use, I don’t use it. Conclusion Here are my conclusions for what they are worth. First, get accurate isogonic information. The Chinese military compass is probably the best bang for the buck. Next would be the Brunton. Although it is more expensive, it gives excellent information. I have found a couple of Morins that were inexpensive and enjoy them very much. I will have examples of everything but the solar compass available to look at. Contact information: [email protected]

139 6 

Signal Level Meter Throw Down* By Doug Millar K6JEY *With apologies to Bobby Flay. Signal Level Meters When I started radio back in the 50’s most of what we used for tuned signal level meters were a tuned circuit and a meter or lamp. Of course any receiver is technically a signal level meter and I had a variety of those. Most receivers though only cover a narrow frequency range and are not normally very portable. One group has specialized in highly portable receivers that are meant to be used in the field. The needs of the TV and cable industries have led to the development of highly portable units that usually tune from 30 MHz to near 1000 MHz, give respectable signal level readings and demodulated audio. Recently with the change in the TV allocations and switch to digital modulation some of the commercial units have become available pretty cheaply. With a modern level meter, one is able to take the instrument to the antenna, operate it on batteries, quickly tune in a frequency and read the level in DBm or DBv quickly and easily. There are many circumstances in which a box like this can be useful. Since these units tune both 144 and 432MHz, they can listen to the output of converter systems. With an accurate level meter one can observe small changes in sky to ground noise, and beacon levels. Altogether it can be a handy item to have in the lab or in the field. What follows is a review of three of the less expensive models that have been available. Each has its own charm, as it were, and price. All tune in the VHF/UHF band. They all use batteries or 12v for power sources. They are made to be relatively simple to use and strong enough to be used in the field. The nice part about these three choices is that you won’t be too upset if you drop it off a tower or dish, as the replacement cost is small and they may be built well enough to take quite a tumble. Example One An Old Solid State Sadelco The first example is from the older crowd and is based on a Sadelco 600. I say based because there are a variety of meters from different manufacturers that all seem to share the same layout. They go by lots of names

140

As you can see, the dial and meter are prominent. The db level can probably be read to a little less than a db. There is an adequate attenuator, a speaker and volume control, a charger input, and F connector for input. The power switch is often configured so that if the lid is put on, the power is turned off. The IF has about 230kc selectivity and makes for apparently great stability. Since the signal you are going to look for is probably strong, the selectivity is adequate. However, I was able to tune in WWV on 10 and 15 MHz with a wire antenna. That says something for the receiver qualities. These are generally available pretty cheaply. One reason is that they are pretty boxy. If you want to run it on batteries, I suggest you replace what is probably a proprietary battery pack with one from perhaps Ebay that will be much cheaper. Be sure to use the original voltage and at least the same ampere hour rating. These should be available cheap, but because of their strength, may only need a fresh set of batteries to get them going again. Example Two- The Next Generation Sadelco MinMax-M The next generation is represented by the Sadelco MinMax M. It is still supported but replaced by a newer meter. This one is available used for less than $100. It is extremely rugged and contains a full receiver and PLL. It is outwardly simple but inwardly very complex. In operation one just needs to turn it on, enter the frequency and read the dial. It shows level to 0.1db and tunes from 5 MHz to 875MHz. It has an 18v battery that lasts for 16hours. It also auto calibrates itself for each measurement. The bandwidth is 125KHz. It’s MDS signal level is at least 75uv. It is considerably lighter than the first meter and actually seems stronger. It is very easy to use in the field. There are a wide variety of features in the unit. For that reason a good look through the manual will make you a happier user. Of the three units this is the one that seems the easiest to use and the best option for signal level measurements. Some of the uses are cable loss measurement, antenna pattern measurement, comparison of antenna gain, and as a microwave IF receiver. In the field it would be easy to use one of these as and IF receiver to see if the rest of the system is functioning properly. Sadelco MinMax0M

Newer and Smaller

141

Example Three More Complex More Expensive The last receiver for consideration goes by lots of names but is principally known as a Protek 3201. The giveaway is the 3201 model number. It is principally an inexpensive spectrum analyzer with a very good level meter. It has much more advanced features. The receiver has both a 125 KHz filter and a 2.3 KHz filter. It can demodulate AM, FM and SSB. It has an internal speaker and volume control. It tunes from 100 KHz to about 2000MHz depending on model. It is very light and the strap helps keep it attached to the person. Like the MinMax the manual bears a good look through, but once you know the interface the meter is easy to use. Almost any parameter can be set and traces and setups can be memorized. There is also an RS232 port on the side. The display is also backlit. You can also make comparative level measurements and have the difference displayed. It runs on AAA cells and has excellent battery life. Sensitivity is excellent. The meter indicates frequency very well and can be used as a counter. The units come on EBay for from $300 to $800 dollars. Below are some pictures of the unit and its displays.

Number 3‐ the Protek 3201

142

Display

Conclusion I have looked at these units for years both wondering what to do with them and how well they worked. I hope this talk has been helpful in that regard. Depending on your needs and budget, one of the three may work for you. I have found a signal level meter to be a handy and useful tool. Contact information is dmillar at moonlink.net

143

C-Band LNB to LNA conversion by Christian Shoaff N9RIN The following paper describes the conversion of a C-Band LNB to an LNA that can be used for the amateur radio 6cm band. This provides an inexpensive LNA for the 6cm band. I was able to purchase from ebay the LNB for about $15US and the transition was about $10US. I did this conversion instead of designing one mainly out of cost and time. I do not think I could design and make an amp for this cost unless I did a large production run. This also took a lot less time to do than starting an amp from scratch. Tools and parts needed: Exacto knife Small blade screwdriver Small Phillips screwdriver Soldering iron and solder Tweezers Drill and drill bits. Size of drill bits is up to the user. A tap may be needed as well. Screws, size is up to the user depending on the holes for the transition. Hookup wire. Washers .141 semirigid with an SMA connector or another connection method to get the RF signal out. Multimeter Power supply: +12VDC with current limiting is ideal. Set current limit to no more than 100mA. This LNB was left in the original housing since I do not have a mill to make a box. I found a WR229 to SMA transition for coupling the 3.4GHz signal in. For those that want to make a milled box, keep in mind that there is an inductor on the input of the first FET that is part of the matching circuit. It plays a big part in setting the NF of the amplifier. It will most likely require tweaking this into the correct position in order to get the lowest NF again. Here is the LNB as it looks out of the box.

144

To open the lid, use the excato knife to cut through the silicone seal.

Use a flat blade screwdriver to pry the lid up.

This is what you should see with the cover off. Unscrew and remove the metal cover that is under the foam. It will not be needed again but save the screws. This cover is to shield the DRO. The DRO will be disabled since we do not need it.

145

The items in the picture above will disable the DRO, mixer and the final amp. This will help reduce the amount of power the amp uses.

Take care in removing the F connector as it is easy to damage the pc board. The hole for the pin can be used for the +12VDC in wire.

As noted in the picture, be careful drilling the holes for the WR229 transition if you go that route. It is very easy to drill too far and damage the pc board. I have a LNB sitting on the shelf as proof of that.

146

Here I used the existing hole from the F connector that was removed to route the power wires in. How the connections are made are up to you. Soldering the wires directly to the voltage regulator will work fine also. For the RF out, I drilled a hole large enough to pass the .141 semirigid through and soldered the center conductor as shown in the picture. The shield was soldered to the ground along the trace that was for the DRO shield cover. Make sure the shield does not touch the power line. I installed one of the screws from the DRO cover back into the open hole. It will need a washer or two added to it depending on the thickness of the washers. Check for any shorts. If none are found, put the lid back on and attach the WR229 to SMA transition.

A top view with the lid back on and the transition attached. Front view.

147

Test the amp as you would for any other LNA. My converted amp has about 30dB of gain and around a 1.2dB NF. I am happy with that and I am going to leave it as is. Your conversion may vary due to how you couple the signal out and if the wire on the input FET was moved. The noise figure on the LNB says 13K, which comes out to about 0.2dB. This noise figure is a rating at a particular frequency and not across the entire band. While there is definitely room for improvement on my amp, I do not see spending the time it would take to try fro the 0.2dB level. I will do another amp later and see what I get and do some tweaking on it then. If you have any questions, I can be reached at: [email protected] Chris n9rin

148

A Personal Beacon for 10 GHz (Which can’t possibly work)

Paul Wade W1GHZ ©2010 [email protected]

A beacon signal is extremely useful for microwave operation. It can provide you with at least three important things: 1. Assurance that your receiver is working 2. A frequency reference 3. A heading for aligning your antenna Even if the beacon frequency is not accurately known, any two stations that hear the beacon can use it to meet on a common frequency, for example, 50 KHz below the beacon. If there is no beacon within normal range on a particular band, we may have to resort to a local weak signal source. Hearing this source can at least provide the first two items on our list. The source can come from a frequency reference, for instance a GPS-locked oscillator, to provide an accurate microwave frequency. The N5AC A32 Synthesizer available from Down East Microwave is an easy way to do this, providing a number of selectable output frequencies for various microwave bands. A popular choice is 1152 MHz – calling frequencies on almost all the microwave bands are multiples of this frequency. The A32 source alone has enough harmonic output to be heard at close range up to at least 10368 MHz, so it can be used as a weak signal source. But what about a beacon? The LO chain in my 10 GHz transverter gets from 1134 MHz to 10 GHz with a few MMICs and pipe-cap filters. Certainly this could be retuned to make a 10 GHz beacon, but I’d hate to cut up an expensive transverter Teflon PC board to do it. What about a cheap PC board from ExpressPCB? The ordinary fiberglass-epoxy boards in my multiband transverter don’t seem to know that they can’t work work at 3456 MHz, so I decided to try one for 10 GHz. The experts say this can’t possibly work, but we aren’t going to tell the boards. Our design philosophy is still: GAIN IS CHEAP. Keep line short and use enough MMICs to overcome losses, and there might be something left at the output. The thick board will definitely radiate, but so what? We can put it in a box, as long as it doesn’t oscillate. I sketched out a rough schematic, prettied up in Figure 1, that I thought might work. It starts with a MMIC tripler and a ¾” pipe-cap filter tuned to 3456 MHz, then an amplifier before another tripler to 10368 MHz. Here it is cleaned up and amplified by two ½” pipe-cap filters and two MMIC amplifier stages. All the capacitors are ordinary chip caps – they are probably lossy, but gain is cheap.

149

I drew up the artwork using ExpressPCB software and made it part of a Miniboard, to take advantage of space left over from another project on the Miniboard. When cut the boards up and started to put one together, I found that I had left out some of the platedthru holes that provide good grounding. I had to drill some holes and add ground wires, which never work as well for RF.

Figure 1 – Schematic Diagram of the 10 GHz multiplier I assembled the first multiplier and amplifer stage, so that I could tune the first pipe cap. Then I found that I had also forgotten to put in the intermediate test point – that’s what happens when you try to rush something. Tacking on a scrap of semi-rigid cable gave enough output to tune the pipe cap, but it wasn’t as much as I’d hoped for. Still, why not finish assembly and see what happens. I added the remaining stages and tuned it up – it actually had output at 10 GHz. After some fiddling, and adjusting the voltage on each multiplier for maximum output, it really had output – roughly +13 dBm. Since this was far more than expected, I added a waveguide filter at the output to make sure it was really at 10368 MHz. Without the filter, the output is clean enough for a temporary low-power beacon, but a filter is needed before adding an amplifier or putting one in a high place.

Figure 2 – Prototype of the 10 GHz multiplier board My prototype unit is shown in Figure 2, and the pipe-cap side in Figure 3. I’ll have to clean up the artwork and build a couple of clean units to determine final component values and performance. Since I did everything wrong in the prototype, a nice clean unit might work almost as well.

150

Figure 3 – Pipe-cap side of the 10 GHz multiplier Phase noise from the A32 Synthesizer is definitely audible in the receiver, and shows up on the spectrum analyzer – Figure 4 is a poor photo of the spectrum.

Figure 4 – Spectrum of the 10 GHz output (20 KHz/division)

151

High-Power Directional Couplers with Excellent Performance That You Can Build Paul Wade W1GHZ©2010 [email protected] A directional coupler is used to sample the RF energy travelling in a transmission line – useful for measuring power, frequency, and VSWR or impedance. If it is truly directional, then it can separate the power flowing in opposite directions, for instance, forward power transmitted toward an antenna and reflected power returning from the antenna. High-power couplers that are truly directional are rare, but it is quite possible to build one using hand tools that outperforms commercial units.

Figure 1 – High Power Directional Coupler with 7/16 DIN Connectors

152

Directional Couplers A directional coupler, shown conceptually in Figure 2, is characterized by coupling and directivity between a main transmission line, Port 1 to Port 2, and a second transmission line, Port 3 to Port 4, coupled to the main line. The coupling is the ratio of transmitted, or forward, power, going in to Port 1 and coming out Port 2, to coupled power at Port 3, measured in dB. A coupling of -30 dB would couple one milliwatt out for each watt travelling through. Reflected power, going in to Port 2 and coming out Port 1, would be coupled to Port 4. Coupling usually varies with frequency.

Figure 2 – Directional Coupler Directivity, a measure of how well the coupler separates the two directions, is the ratio of coupled power out at Port 3 to power out at Port 4 when power on the main transmission line is only flowing in one direction, into a perfect termination at Port 2. This leakage from poor directivity limits the return loss or VSWR that we can measure – for instance, a directional coupler with only 20 dB of directivity would indicate a return loss of 20 dB, or VSWR = 1.22, for a perfect load. If it were used to measure an antenna with an actual VSWR of 1.22, the unwanted coupled power due to low directivity would add to the coupled reflected power. Depending on the phases of the reflected and leakage power, the total could be twice as much as the reflected power, for an indicated VSWR of 1.5, or zero if the phases cancelled, for an indicated VSWR of 1.0, or anything in between. Higher directivity is needed to measure low VSWR. Low directivity can also affect power measurement – leakage from reflected power adds to the coupled forward power, again at unknown phase, so that measured power varies with VSWR. Commercial directional couplers, like those we find in surplus, are often designed for relatively constant coupling over a frequency range. Typically, directivity is not high over the whole range – often as low as 15 to 20 dB. Couplers found in instrumentation such as network analyzers usually have higher directivity, but are not intended for high power. The coupling in these couplers is typically -20 to -25 dB, so that measurements may be made with relatively lower power levels.

153

Homebrew Directional Couplers I’ve been trying to make a good directional coupler for a long time. My goal is one that has high directivity, good power handling, and is robust and easily reproduced with simple hand tools. Some new projects with solid-state amplifiers gave me impetus to make this a reality. One obvious approach is a printed-circuit board – easily reproduced in any quantity. I built one years ago, and have tried other designs in software, but never with good results. A microstrip coupler, with transmission lines on top of the board and a ground plane on the bottom, has part of the energy in the dielectric and part in air, travelling at different speeds. The two parts arrive at different times, creating a phase difference, so the result is poor directivity. One alternative, a stripline coupler with ground planes on both sides, could be better, but requires multilayer PC boards, which are significantly more expensive, particularly in low-loss dielectric materials. The preferred dielectric is clearly air. One advantage for hams is that we are looking for weak coupling, so that hundreds of watts couples only milliwatts to the coupled port. This makes the design considerably easier. Textbooks1 on coupler design describe odd-mode and even-mode impedances where the two lines are coupled, and the necessary impedances and spacing for the desired coupling ratio. For weak coupling, less than -30 dB, the numbers reduce to two lines with impedance very close to 50 ohms, spaced relatively far apart. For high powers, we want wide lines, to carry high currents, with large air gaps, for high voltages. However, the high power is only on the main line – the coupled line only sees low power, and is spaced a good distance away, so it can be much smaller. One of my directional couplers is shown in Figure 3. A diecast aluminum box provides a robust enclosure that will not flex, so spacings are consistent and performance is constant. A wide, flat stripline centered between top and bottom makes a nice main transmission line. The width is great enough that a small error has little effect, and any slight offset up or down from the center is inconsequential. The smaller coupled line is along one edge of the box, using the wall as a ground plane. The main line has type-N connectors, while SMA connectors are adequate for the coupled ports.

154

Figure 3 – Homebrew Directional Coupler (Larger version) To achieve accurate 50-ohm lines and an estimate of coupling, I simulated the coupler using Ansoft HFSS software2. This 3D electromagnetic simulator is much more accurate than simple programs like AppCAD3 and graphs in books, particularly for odd shapes that I tried. For the coupled line, I started with a simple round rod over ground but found the spacing above ground to be much less than AppCAD calculated (probably because the equation is not accurate for low impedances). A flat strip would be good, but not rigid enough to maintain constant spacing and impedance. Rectangular hobby brass tubing (or WR-22 waveguide, for the extravagant) makes a rigid line and makes the space large enough to allow for reasonable tolerance. These simple shapes, wide stripline for the main line and microstrip (with air dielectric and a very thick line) for the coupled line, calculate to between 50 and 51 ohms in both HFSS and AppCAD. The frequency range is set by the length of the coupling line – maximum coupling should occur when this length is an electrical quarter-wave. Above and below this center frequency, the coupling decreases, but directivity typically does not decrease at lower frequencies; sometimes it improves. Knowing this also allows us to use surplus directional couplers at frequencies well below the rated frequency. The maximum coupling is set by the separation between the two lines, and falls off predictably at lower frequencies. Thus, we can achieve a desired coupling at a particular frequency by adjusting the spacing or the coupled length, or both.

155

Results I built two couplers with Type-N connectors using this basic design – a large one for 1296 MHz down to VHF, Figure 3, and a smaller one to get up to 2400 MHz, shown in Figure 4. Maximum frequency is limited by resonances in the boxes at roughly 1.7 GHz for the larger box and 3.0 GHz for the smaller. Higher frequencies would require smaller boxes and better transitions from coax to stripline.

Figure 4 – Homebrew Directional Coupler (Smaller version) Measured S-parameters are shown in Figure 5 for the larger directional coupler, and in Figure 6 for the smaller one. For those who aren’t fluent in S-parameters, relevant quantities are shown individually: Coupling in Figure 7, Directivity in Figure 8, Loss in Figure 9, Return Loss in Figure 10, and VSWR in Figure 11. Finally, numerical values for amateur bands are listed in Table 1. Copies will not have these exact values – calibration at the desired frequency is required for exact results.

156

Figure 5 – Measured S-parameters for Larger Directional Coupler The larger coupler has excellent directivity up to 432 MHz, and is still good at 1296 MHz. The smaller one has excellent directivity up to 1296 MHz, and is still good at 2400 MHz. Measured directivity in Figure 7 appears noisy because the reflected power is more than 90 dB down, near the noise floor of the VNA. At the VHF frequencies, the smaller version has very weak coupling, so the larger one would be preferable.

Figure 6 – Measured S-parameters for Smaller Directional Coupler The wide lines have very low loss and should be adequate for high powers. I was only able to test them at 500+ watts at 144 MHz and 100 watts at 903 MHz, but they work fine at these power levels.

157

Figure 7a – Coupling for Large Directional Coupler

Figure 7b – Coupling for Small Directional Coupler

Figure 8a – Large Directional Coupler Directivity

Figure 8b – Small Directional Coupler Directivity

Figure 9a - Large Directional Coupler Insertion Loss

Figure 9b – Small Directional Coupler Insertion Loss

Figure 10a - Large Directional Coupler Return Loss

Figure 10b – Small Directional Coupler Return Loss

Figure 11a - Large Directional Coupler VSWR

Figure 11b – Small Directional Coupler VSWR

158

A Higher Power Directional Coupler The two directional couplers described above recently appeared in DUBUS4. When I saw them described as QRO, I knew some of the EME community would take issue – some of them run enough power that type N connectors are inadequate. The popular coax connectors for real QRO seem to be the 7/16 DIN connectors, so I decided to make another directional coupler using these connectors, shown in Figure 1 above. The 7/16 DIN connectors are relatively expensive, but I located a few connectors at a reasonable price on the internet and bought them. The mounting flange is larger than a type-N connector, so a larger enclosure is required. This suggested a lower frequency design, so I chose to aim for -40 dB coupling and a center frequency of 432 MHz. A stock aluminum chassis, 7x9x2 inches, seemed just about right, and I had one on hand. The construction is very similar to the other couplers – a wide copper stripline for the main line, and an air microstrip on the side wall for the coupled line. I chose to leave the full length of Teflon insulator on the connectors to maximize power capability, at the cost of a small discontinuity that increases VSWR slightly at 432 MHz. Minimum spacing is about 6.5 mm for the main line, so it should handle any power level that amateurs can generate.

Figure 12 – S-parameters for Higher Power Directional Coupler Performance is excellent, and very close to the goal – coupling is -40.7 dB at 432 MHz, increasing to -46.6 dB at 144 Mhz and -55.3 dB at 50 MHz. Directivity is better than 30 dB through 432 MHz. The measured S-parameters are shown in Figure 12, and separated into individual graphs in Figures 13 through17. If higher coupling is desired, increase the separation between the main line and coupled line. Simulations show an increase of 0.5 dB for each additional millimeter of spacing – an additional 20 mm of spacing will yield a -50 dB coupler, and a further 20 mm results in -60 dB coupling.

159

Figure 13 – Coupling for Directional Coupler with 7/16 DIN Connectors

Figure 14 – 7/16 DIN Coupler Directivity

Figure 16 – 7/16 DIN Coupler Return Loss

Figure 15 – 7/16 DIN Coupler Loss

Figure 17 – 7/16 DIN Coupler VSWR

160

161

-51.3 34.9 0.03 45.7 1.01 136 7.37 -71.3 19.7 0.03 44.0 1.01 13423 0.07

-55.3 33.0 0.04 36.6 1.03 338 2.96

Small Coupling Directivity Insertion Loss Return Loss VSWR Power per mw Coupled @ 1 KW

7/16 DIN Coupling Directivity Insertion Loss Return Loss VSWR Power per mw Coupled @ 1 KW

50

Large Coupling Directivity Insertion Loss Return Loss VSWR Power per mw Coupled @ 1 KW

Band

-46.6 33.3 0.06 31.6 1.05 45 22.1

-62.7 29.4 0.04 36.7 1.03 1847 0.54

-42.3 32.3 0.04 38.9 1.02 17.0 58.7

144

Table 1

-43.4 34.8 0.07 34.2 1.04 22 45.5

-58.6 43.2 0.05 33.9 1.04 722 1.39

-38.7 29.7 0.04 37.1 1.03 7.4 135.5

222

-40.7 30.6 0.11 23.3 1.15 12 84.5

-53.1 31.7 0.06 31.3 1.06 206 4.86

-33.6 27.0 0.06 39.4 1.02 2.3 441.4

432

-50.9 21.2 0.16 28.3 1.08 124 8.1

-46.8 31.5 0.13 29.6 1.07 47.7 21.0

-30.3 23.6 0.10 27.2 1.09 1.1 933.9

902

W1GHZ 2010

-39.0 15.3 0.45 16.3 1.36 8 126.5

-43.8 32.1 0.10 30.3 1.06 24.2 41.4

-32.9 22.5 0.15 23.0 1.15 2.0 511.7

1296

-39.6 22.2 0.17 21.2 1.19 9.1 109.7

2304

Directional Coupler Performance at Amateur Bands

-39.4 21.4 0.18 20.9 1.20 8.6 115.9

2400

Watts/mW mW

dB dB dB dB

Watts/mW mW

dB dB dB dB

Watts/mW mW

dB dB dB dB

MHz

Construction A sketch with important dimensions indicated is shown in Figure 18, and a cross-sectional view in Figure 19, and a summary of dimensions I used in Table 2. More important than the exact dimensions I used are the factors needed to realize a different coupling ratio, to utilize metric materials, or to make a higher power version using 7/16 DIN connectors, so that a coupler can be customized for individual requirements.

Figure 18 – Homebrew Coupler Dimensions, Top View

Figure 19 – Homebrew Coupler, Cross-section View

162

Design

Large inches

Hammond 1590BB

Enclosure Coupled Length Height Main Line Width Taper Separation Coupled Line " above wall

Small mm

 1/4  or less enclosure 1.35 * Height best VSWR for coupling 50 ohms 50 ohms

inches

7/16 DIN mm

inches

mm

Hammond 1590-B Bud AC-406

3.4 1.18

86.4 30.0

1.36 1.05

34.5 26.7

7.5 2.0

190.5 50

1.59 0.1 1.75

40.4 2.54 44.5

1.42 0.16 2.0 0.25 x 0.125 0.065

36.1 4.1 50.8

2.7 0 0

68 0 0

0.25 x 0.125 0.065 1.65

1.65

0.25 x 0.125 0.065 1.65

Table 2: Directional Coupler Dimensions

The line dimensions for 50 ohms are most important – the main line, 0.020 inches (0.5mm) thick, has a width of 1.35 times the inside height of the box. The coupled line impedance is set by spacing above ground. For the coupled line, I used rectangular hobby brass, 0.250 x 0.125 inches. For this aspect ratio, the spacing above ground is about 0.27 times the width of the coupled line. Taper at the end of the main line affects VSWR and directivity – 100 mils (2.5mm) for the large version and 160 mils (4.5mm) for the small one seemed to give good results. I also rounded off sharp corners of all the lines with a file. The main line connectors may be moved side-to-side to adjust coupling, and need not be centered in the box. In the smaller coupler, I put them slightly off-center to increase the coupling slightly. Calculated curves of coupling vs frequency are shown for several different separation distances, measured from the end wall (under the coupled line) to the main line connectors, are shown in Figure 20 for the large coupler and Figure 21 for the small coupler. Measurements of coupling on my units are within one dB of the calculated values. The center frequency, where maximum coupling occurs, is slightly higher in Figures 20 and 21 than expected from quarter-wavelength calculation, and the center frequency increases with weaker coupling. By scaling these curves, it should be possible to estimate the dimensions for a desired coupling at any frequency.

163

Figure 20 – Calculated Coupling vs Separation and Frequency for Large Coupler Directivity and VSWR vary a few dB calculated values, since small variations in construction affect these values. Generally, looser coupling tends to have higher directivity, and careful construction improves both VSWR and directivity.

Figure 21 – Calculated Coupling vs Separation and Frequency for Large Coupler Construction is simply marking and drilling holes, and tapping the threads for the screws. The main line is cut to size, fit in place, and soldered. The coupled line has two holes drilled at the same distance as the connectors, then slid over the connector pins. To control the spacing above ground, I use a pad of Post-It notes. Remove sheets until the pad is the desired thickness (0.065 to 0.070 inches, or 1.65 to 1.77 mm) as measured with a caliper or micrometer, slide the pad under the line as shown in Figure 22, push down, and solder the connector pins. Remove the pad, file the pins flush, and clean up. For the longer lines of the coupler with 7/16 DIN connectors, I used a pad at each end to make the spacing uniform.

164

Figure 22 – Setting spacing above ground for Coupled line

Uses for Directional Coupler Power Measurement – a directional coupler, accurately calibrated, can be used to measure high powers accurately. A chart like Table 1 listing Watts per milliwatt of coupled power at each frequency assists in quick, accurate calculation. Test sessions at many VHF and up meets make a Vector Network Analyzer available for quick calibration; otherwise, comparison to a known power meter may be used. VSWR and impedance – measurement of both forward and reflected power yields Return Loss, which is easily converted to VSWR if preferred. Complex impedance requires a network analyzer connected to the two ports, either a surplus unit or homebrew like my Handheld Network Analyzer4. Sampling – a tiny bit of the signal is available to measure frequency, or spectrum, or to use for ALC or other feedback. Spectrum measurements will require correction for the variation in coupling with frequency. Amplifier protection – many solid-state amplifiers are less tolerant of high VSWR than are large triode tubes. Sensing both forward and reflected power with a directional coupler can monitor Return Loss continually, and it can be used to automatically shut down when VSWR is excessive. One simple way to do this would be to put an attenuator on the forward port with value equal to the desired maximum return loss. Then a simple comparator IC can sense when reflected power is greater than attenuated forward power and operate a relay or solid-state switch. For instance, a 10 dB attenuator at the forward port would make the forward and reflected outputs equal for a Return Loss of 10 dB, or a VSWR of 1.92.

165

Note that any power reflected from the forward port travels along the coupled line and appears at the reflected port as erroneous reflected power, just like poor directivity. So, if the power detector does not have good VSWR, an attenuator will improve matching and allow more accurate measurements. Accuracy in power measurement also depends on the accuracy of the power meter used to measure coupled power. Surplus power meters like the HP-432 are fairly common, and can be easily calibrated. A simple IC power detector5 is probably less accurate, but can be adequate for many uses, such as amplifier protection or remote monitoring of power. The amateur standard for high-power measurements has been the Bird 43 wattmeter. At VHF and UHF frequencies, these are quite accurate when new and have good directivity, but they are rather expensive to leave in the line for constant monitoring. Used units of questionable provenance also have questionable accuracy and are hard to repair.

Summary These easy-to-build directional couplers demonstrate that it is possible to homebrew a quality coupler suitable for high power measurements. Curves are included which should make it possible to tailor one to any desired frequency and power level.

References 1. G. Matthaei, L. Young, and E.M.T. Jones, Microwave Filters, Impedance-Matching Networks, and Coupling Structures, McGraw-Hill, 1964. 2. www.ansoft.com 3. http://www.hp.woodshot.com/ 4. Paul Wade, W1GHZ, “High-Power Directional Couplers with Excellent Performance – That You Can Build,” DUBUS, 2/2010, pp.10-23. 5. Paul Wade, W1GHZ, “Antenna Ratiometry Measurements for the 21st Century Using a Homebrew Ratiometer (also a Handheld Network Analyzer),” Proceedings of Microwave Update 2005, ARRL, 2005, pp. 40-63. also available at http://www.w1ghz.org/small_proj/hna.zip 6. Paul Wade, W1GHZ, “Microwave Integrated Power Detectors,” DUBUS, 4/2008, pp. 5873.

166

Analysis of the WA1MBA 78 GHz Low Noise Amplifiers By Al Ward W5LUA

Introduction The workshops at the Microwave Update conferences generally provide a means of testing components and devices that the attendee may not necessarily have access to at their home. Capabilities may include equipment for measuring noise figure, gain, return loss of amplifiers and phase noise of oscillators. At MUD 2009 held in Dallas, Texas, we attempted to up our frequency capability to 78 GHz to help with the evaluation of WA1MBA’s 78 GHz amplifiers. We had noise figure measurement capability to 78 GHz and scalar network analysis through 80 GHz. Part 1 of this 2 part article will cover the analysis of the WA1MBA 78 GHz LNAs since MUD 2009 and Part 2 will be presented by Tom WA1MBA on moving forward with the 78 GHz LNA project. The WA1MBA 78 GHz Amplifier The WA1MBA 78 GHz amplifier [1] incorporates 2 stages of the CHA-1077 MMIC. Input and output for the amplifiers was WR-12 which is optimum for the 60 to 90 GHz frequency range. In years past we attempted to measure noise figures of various mixers for 78 GHz but this was the first year we got serious about measuring Tom’s 78 GHz amplifiers. The measurements we made at MUD 2009 were DSB measurements which means that the noise figure is really the average of the noise figure at LO +/IF. We generally chose a 2M IF so it was not expected that the noise figure of the amplifier would vary much over 2X the IF bandwidth. Noise Figure Measurements at 78 GHz Over the years we have experimented with several different noise sources. The dominant two were supplied by Will W0EOM and Tom WA1MBA. The W0EOM noise source was a Noise Com NC5115 noise source with WR-15 waveguide flange. Its calibration chart ranged from 50 to 75 GHz. We attempted to extrapolate the ENR table to 78 GHz with limited success. Tom’s noise source was a similar Noise Com unit with a WR-10 flange which was calibrated from 75 to 110 GHz. Neither noise source included any sort of isolator on the output. As it turns out both noise sources may have been initially calibrated with an isolator attached. Although we gave it our best attempt at MUD2009 to measure noise figure, I was not totally convinced that we really had accurate numbers. So the project came home and was the focus of a major investigation for months to come after MUD2009 was over. Over the winter Barry VE4MA was able to purchase a WR-10 Noise Com noise source with an isolator that also included a calibration chart. Obviously this third noise source should be the tie breaker and everything should fall in place. Well the results certainly did give us a third number but it was way high. This is more of a psychological thing because we always want to believe that lower is more real than high but in this case the real eye opener was that we were measuring optimistically low number with our first two sources…My conclusion was

167

that the ENR tables on both Will’s and Tom’s noise sources probably did include an isolator that WAS attached to the source when calibrated and subsequently removed. Most of the WR-15 and WR-10 isolators tested had several dB of loss which was about the difference in noise figure measured. However as it turns out when dealing with active devices, even MMICs, their gain is sensitive to mismatch at the input and as the noise source is turned on and off as part of the noise figure measurement , the impedance presented to the DUT changes. This change in impedance can affect the gain of the device and subsequently can affect the indicated noise figure. So why is the isolator important? The explanation is very similar to explaining why a low ENR noise source like the Agilent 346A provides more accurate noise figure measurements than the higher ENR Agilent 346B noise source when measuring low noise amplifiers. The 346A noise source is a nominal 5 dB ENR noise source while the 346B is a nominal 15 dB ENR noise source. The 346A is calibrated with an internal 10 dB attenuator to lower the ENR from 15 dB to 5 dB. The 10 dB attenuator provides 10 dB greater isolation between the DUT and the impedance presented by the noise diode. Therefore as the noise diode in the noise source is cycled on and off to make the noise measurement, the change in impedance is less visible to the DUT. Generally higher ENR noise sources are used to measure higher noise figure passive mixers with high accuracy. However, active devices are more sensitive to impedance changes and if the gain of the DUT changes as the noise source is cycled on and off then this gain change will alter the apparent noise figure measurement. The isolator used with most waveguide noise sources provides similar if not greater reverse isolation than the attenuator with less loss in the forward path. Once it was determined that each of the 78 GHz noise sources should have an integral isolator permanently in place, they were individually calibrated against Barry’s “gold standard” noise source. Scalar Network Analyzer Measurements With some encouragement from Will, W0EOM and Mark, N0IO I resurrected my old HP 8757 scalar analyzer and with the help of a multiplier and a WR-15 detector from W0EOM was able to measure swept gain from 50 to 80 GHz. I used my existing HP 8340A synthesized sweeper locked to a GPS reference as my source. The system had greater than 40 dB of dynamic range. Armed with our simple setup we were able to make numerous broad band gain measurements at MUD 2009 including WA1MBA’s 78 GHz amplifiers. A typical plot of gain vs frequency for one of Tom’s amplifier is shown in Figure 1. The amplifier has about 18 dB gain at 78 GHz. Note that below 75 GHz, the gain rises significantly. The gain at 72 GHz has risen to about 35 dB. A look at the S Parameters from the CHA-1077 only reveals data from 75 to 90 GHz. Quite often the design of a very high frequency MMIC has more gain below the lowest intended frequency of operation. We must however make sure we don’t compress the second stage such as our mixer in our noise figure set up. A simple S parameter analysis of

168

a cascade of CHA-1077 MMICs reveals the response shown in Figure 2.

Figure 1. Swept gain response of the WA1MBA 78 GHz LNA #5 from 65 to 80 GHz. Each horizontal division is 1.5 GHz

Figure 2. ADS Analysis of 2 stages of CHA-1077. This is solely based on simple cascade of manufacturer supplied probed S parameters. Note dip in gain and poor return loss at 78 GHz.

Since there were no manufacturer supplied S parameters below 75 GHz, the expected lower frequency performance is an unknown. The ADS analysis does predict 30 plus dB of gain for the two die in series but it does not take into account ground inductance, cavity effects and the transition to waveguide. All these parameters will have a major effect on performance at 78 GHz. The S parameter analysis also suggests that some additional tuning may required for optimum performance at 78 GHz. Tom has already incorporated some manufacturer suggested tuning.

169

The substantial low frequency gain of the amplifier prompted the concern about hitting the mixer with too much gain over too wide a bandwidth. This was verified one day by accidentally removing the LO from the mixer and still be able to read noise figure and gain on my 8970! I was now convinced that the noise power at the lower frequencies was actually acting like an LO for my broad band 50 to 75 GHz WR-15 mixer. This is another consequence of trying to use components above their normal intended frequency range. 78 GHz Filters It was quickly decided that maybe it was time for incorporating a filter in front of the mixer to remove the image and out of band noise from hitting my WR-15 mixer. Now we could generate SSB noise figure numbers at a single frequency and hopefully increase the accuracy of the noise figure measurements. Mark, N0IO has been doing some work on some 78 GHz filters but I decided to take a look at my junk box and see what I could come up with. I came across some 47 GHz cavity filters designed by OE9PMJ so I decided to analyze these on my scalar analyzer setup and see if they could be pushed to 78 GHz. Much to my surprise there seemed to be a secondary mode that could be tuned to 78 GHz which had fairly low loss. The filter now optimized for 78 GHz has the response shown in Figure 3. The filter has several hundred MHz of bandwidth and is suitable for a 432 MHz IF with 20 dB of image rejection. One of the downsides of the filter when tuned to 47 GHz was an extraneous response at 39 GHz. When tuned to 78 GHz , the same filter also has multiple responses below 75 GHz. The dominant secondary responses are at 73.5 GHz and below 73 GHz. The solution was to incorporate a small section of WR-8 waveguide in series with the filter. WR-8 waveguide has a cut-off frequency of approximately 73 GHz. The resultant plot of the 78 GHz bandpass filter in series with the short section of WR-8 is shown in Figure 4.

Figure 3. Measured Response of the OE9PMJ 47 GHz Filter tuned to 78 GHz.

170

Figure 4. Measured Response of the retuned 78 GHz filter in series with section of WR-8 waveguide. SSB Noise Figure Measurements The SSB noise figure of the bandpass filter / highpass filter / WR-15 mixer measured approximately 22 dB of which 4 dB was filter loss. The remaining is due to using a WR-15 fundamental mixer that is somewhat starved for LO. The X2 39 to 78 GHz multiplier which was driven by one of W0EOM’s X3 multipliers was only putting out 0dBm. Although higher than desired, the 8970 noise figure meter was able to calibrate out this second stage noise figure. Testing of the WA1MBA amplifiers with our improved setup showed higher than expected noise figures based on the VE4MA “gold standard” noise source. Plots of input and output return loss suggested that the optimum performance was somewhat lower in frequency. I then decided to make a multiple stub tuner in WR-15 waveguide. With Craig, KA5BOU’s help, we made several tuners. The results shown in Table 1 show a substantial improvement both in noise figure and gain with the additional of an input tuner. Compared to the scalar analyzer results the gain numbers achieved with the noise figure setup appear to be several dB low. This still needs resolution. Although an output tuner was only quickly looked at, it is felt that with an output tuner, gains upwards of 25 dB are possible. Figure 5 shows a WR-15 tuner added to LNA #5. Figure 6 shows the setup for measuring noise figure and gain. LNA # 4 5 6

Noise Figure w/o Tuner 10.6 dB 11.8 dB 11.5 dB

Noise Figure w/ Input Tuner 8.5 dB 7.7 dB 6.9 dB

Gain w/o Tuner 14.6 dB 14.9 dB 9.7 dB

Gain w/ Input Tuner 17.3 dB 19.1 dB 14.5 dB

Table 1. Noise Figure and Gain with and without input tuner

171

.443” WR-15

.070” Only 2 0-80 screws needed to lower NF from 11.8 dB to 7.7dB ~ 4 dB improvement

.140” 2pl on top 3 0-80 tapped holes on top and 2 on bottom, ones on bottom are offset by .070” from top

Figure 5. WR-15 Tuner added to input of LNA #5

Calibrated 8970 at this point Figure 6. Setup used for measuring gain and noise figure of the 78 GHz LNA What’s Next? The WA1MBA 78 GHz LNAs offer a substantial improvement in noise figure of a bare mixer. At W5LUA, I have incorporated one of Tom’s LNAs into a 78 GHz system coupled to a 2.4 m offset fed dish and am presently looking at sun noise. Results will hopefully be presented at MUD2010. Part II of this article will present work by Tom WA1MBA on the plan for the incorporation of an integral waveguide tuner and other improvements. Thanks Al Ward W5LUA August 31, 2010 References. 1. 78 GHz LNA Status Report, Microwave Update 2009 Proceedings, Dallas, Texas , pp 32-36

172

Moving Ahead with the 78 GHz Low Noise Amplifier Project By Tom Williams WA1MBA

Introduction The 78 GHz LNA project moved from essentially a one-person project at MUD 2009 into an analysis phase. Al Ward W5LUA spent significant effort at the MUD conference trying to measure the first few prototypes for gain and noise figure. The results were inconclusive, and we decided that we had to investigate and create better and consistent measurement techniques. As presented in a related paper, subsequent investigations by Al with help from Will W1EOM, Craig KA5BOU and Barry VE4MA yielded results that show us the need to improve the amplifier and attempt to achieve the results from an earlier prototype. The primary problems seem to be poor gain and noise figure due to ineffective tuning. The early development was focused on effective bias supply, good bypassing, elimination of sources of oscillation, and finally, tuning. An initial concept was to incorporate an input tuner to allow each unit to be optimized for performance. Unfortunately, the existing circuits seem to provide only 10 to 15 dB of gain, and an output tuner assists in increasing overall gain. In late breaking news, over 25 dB of gain has been achieved repeatably, and very low noise figures achieved at the same time. This was accomplished by removing broken MMICs, replacing some microstrip boards which were fabricated using a prototype milling machine with boards fabricated through precision photo-etching, and using 2 x 0.5 mil ribbon with very short low inductance bonds. Furthermore, with some external tuning we were able to get noise figures below 4 dB. This is excellent. We will present the latest results at the Microwave Update conference. It is possible that the unit without a tuner will be very good for terrestrial work, and that with the addition of a tuner, EME class performance could be obtained. Therefore, two possible next steps are worth considering. In one, we just build the amplifiers as designed, and offer an external tuner for those serious about getting the last dB of noise figure. The other possible next step is to incorporate a tuner into the design. Another (less important) improvement will set the input to output dimension to exactly 2 inches so that interfacing with identical waveguide components is easy when connecting to a waveguide switch where the standard dimension is also 2 inches. As of this writing, revision 5 has been designed, which consists of an input tuner version of the latest design and measures 2 inches boss-to-boss. Of course, the latest status update will be presented at the MUD 2010 conference.

173

Rev 5 preliminary design. The outline (left) shows the input tuner section, on right the split block bottom.

Typical gain plot of an improved Rev 4 LNA. This one has 28 dB gain at 78.19 GHz

174

An Improved 2 x MRF286 Power Amplifier for 1296 MHz By Darrell Ward VE1ALQ In recent years there has been a lot of amateur construction activity surrounding the Motorola MRF286 transistors on 1296 MHz. All of the known designs have been empirically derived and have produced some good results with approximately 150 W out and a gain of about 12-13 dB for a pair of transistors. Computer modeling has shown that a big improvement of output power is possible with a considerable improvement in input return loss. The modeling has shown that at these power levels great care has to be taken in the choice of materials in order to achieve the predicted results. The improvements have been confirmed experimentally Amplifier Design I have been fortunate to have available an up- to- date copy of the Microwave Office software to analyze the MRF 286 designs. The software allows the complete analysis of the RF circuitry including the power, gain, frequency response, return losses, gain compression and harmonic content. Of course the circuit board layout is fundamental to the design of the circuit and can be exported directly. The design work was stimulated by problems encountered while trying to build the W6PQL amplifier boards. There have been several versions of the boards released and I received the Version 7.2 which gave me problems. Similar problems have been experienced by Dominique HB9BBD, who has spent an extensive amount of time in modifying the Version 7.2 W6PQL boards to get them to work properly. Please refer to the write-up by HB9BBD elsewhere in the 2010 EME conference Proceedings, which describes the modifications he has made. The problems with the Version 7.2 boards appear in several areas: 1. Input match to each device 2. Lack of balance in the hybrid couplers Input Match The input section on the version 7.2 W6PQL boards is too short to complete the matching. Some improvement was achieved by replacement of the single turn trimmer with a 4.5 pF high Q multi-turn piston trimmer. The adjustment is very sharp. Hybrid Coupler There is a problem with the design of the hybrid couplers in that all of the input and output ports should share a half each of the 35 and 50 Ohm legs of the hybrid. When they do not the hybrid balance of the output ports is upset. The design of these hybrids is not obviously different between different versions of the board, yet on an earlier version V7.11 used by VE4MA the hybrid balance was apparently not a problem and did meet Jim’s specifications. Choice of Substrates One of the fundamental choices to be made in the design is the choice of substrates. 1296 MHz is at a frequency where some of the lower frequency substrate choices are possible as well as the high frequency ones. Of course the amount of surface area is relatively large for almost any 1296 design which will have an impact on the production cost if the high frequency substrate is chosen. The substrate material chosen is Rogers 4003C (the same as W6PQL used) however others considered were Rogers RT-5889LZ and Taconic TLX-8-200.

175

The lower dielectric constant materials result in wider traces and subsequently larger circuit boards. With the high current density on these circuit boards, there is some RF efficiency gained by using larger traces, but at the cost of the larger board and cost of the material. The modeling showed that there is also a significant disadvantage in the use of thicker 0.062 inch board material vs. the more common 0.020 or 0.032 inch material. The thicker substrate material, i.e. 32mil results in wider copper traces and radiation loss from the board is not so much of an issue. Please note however that with the high powers being used here the radiation hazard does exist for the eyes! The final consideration is the thickness of the copper plating on the boards. The skin depth for copper at 1296 MHz is about 0.0006 inches (or 0.6mil) and the copper thickness for 1oz copper is 1.3779527559055mil and for 2oz copper is 2.75590551mil Therefore there is no advantage to the use of 2oz copper board material when using the RO4003C 32mil. The Hybrids and matching Pads are of sufficient width to handle the power capabilities of the active devices being used in this design. Sources of Transistors & Choice of mounting The MRF286 / XRF286 transistors are no longer in production by Motorola or its descendent company Freescale. These transistors are available by salvaging from surplus “PyroJoe” amplifier boards available on EBay but they are also being supplied from stock piles in China. It is a sad fact that copies of semiconductors are being produced in China that are cosmetically excellent but the RF (and other) characteristics may not match the original device performance. There is considerable variation of the input and output capacitance of some devices but this has not been correlated to RF performance. The effect of the capacitance changes should be negligible at 1296 MHz since the impedances are so low. There has been a concern with purchases made by several ham operators but it is not clear if the MRF286 devices being supplied are originals or copies and if there has been a problem with the devices supplied but as the old saying goes “Buyer Beware”. There are two (2) mounting choices for the MRF286, with the standard being the flanged package carrying the MRF286 designator and the flangeless package that is designated as the MRF 286S. The spacing of the Gate and Drain leads above the bottom of the package is the same for either version. The flangeless MRF286S must be soldered down to something that ultimately is intimately attached to the heat sink. In both cases some form of heat spreading plate is desirable and great care taken to mount the devices with the lowest thermal resistance possible, and greatest RF return path to the device Source. I do recommend the MRF286S over whose you see on eBay which are the MRF286F with flange mounting because the F version can only be secured with 4-40 bolts, unless you drilled the slot for 6-32. Then you only have something like Wakefield Thermal compound to conduct the device heat to the heat sink, and I found the device was hotter around each bolt than the spreader was. Whereas the S version is soldered to the heat spreader allowing heat to be more evenly dispersed. I suggest if using the Flange mounting version of the MRF286 is to simply cut the bolting tabs off and solder the device directly to the copper heat spreader as you would with the MRF286S version, this should eliminate any localized heating around the device and improve greatly the RF return path to the Device Source. Design Results The designs were completed for several board materials but this report will concentrate on the .020 and .032 inch RO4003c substrates. The results with the 0.032 inch substrate were very interesting in that the best power output is 320W with 13.5 dB gain at the 1dB gain compression point! This result is shown in Figure 1, while Figure 2 shows a DC power efficiency of 58%. The gain and return loss vs. frequency are shown in Figure 3.

176

The results for the 0.020 inch substrate are similar except that the maximum output power at 1 dB gain compression is reduced to about 280 W. Nothing could be done to the design to improve the output power. Another big reason for using RO4003C 32mil material over the thinner 20mil substrate was that at 300+ Watts output the output Hybrid was heating and starting to shine like a Mirror and very certainly would have lifted from substrate over a period of time. The modelling program also predicts the harmonic power levels and this is shown in Figure 4.

Figure 1 DC and Power Added Efficiency vs. Drive for .032 inch RO4003C Substrate Amplifier

Figure 2 Gain & Return Loss vs. Frequency

177

Final Circuit Board Design Layout & Schematic I have no intentions of putting this board into production, but all Drawing and Gerber files will be made available to those who ask. I can also provide a source connection for the MRF268S Devices if needed. Similarly if anyone is interested in the board information for the 20 mil substrate design can be made available by asking me by email to [email protected] or check the Web Site http://www.ve1alq.com for more information and a complete set of ZIP files. Four Port Power Combiner Design In my investigations it became apparent that 4 of these 330W modules could not be combined using a hybrid coupler made of even the 0.032 inch RO4003 substrate material. I did follow through with a design using 0.062 inch Taconite substrate (see Figures 6 & 7 below) which includes a directional coupler for forward and reflected power monitoring purposes. Once again those interested can contact me for Drawing and Gerber files. A sub-set of this 4 Stage combiner was extracted in order to combine 2 amplifier modules. It is essencially the top 1/3rd of the 4 Stage Combiner (figure 6) and is shown in Figure 8. The Coupler coupling and isolation performance is shown in Figure 9. The excellent Port to Port Balance and very low insertion loss should be noted. Ports 1 & 2 would be the driven Ports, Port 3 is the Isolation Port and Port 4 the combined output port. Once again those interested can contact me for Drawing and Gerber files.

Figure 3 Output Spectrum vs. Frequency @+41dBm Drive Level  

Actual Results The 2 stage amplifier design has been tested using the 0.020 inch Taconic’s TLX substrate as shown in Figure 10at a saturated power of 330 W when drive into compression. Those Coils or chokes were not

178

needed and later removed with same results. This version was still with the flanges on the devices and as a result generated higher heating around the #4-40 machine screws, which was the largest that the flange mount would allow. Soldering the devices in rather than bolting in would allow much better heat distribution to the heat spreader, or heat sink. As further confirmation of the design, it has been checked and confirmed by Mr. Dane Collins, the CEO of AWR/ Microwave Office and his Support staff who provided excellent support and corrective pointers as the project developed. This brings up a very IMPORTANT point: This board is not, nor will it be produce for profit by any one and I mean ANY ONE…….PERIOD, including myself or my temporary licence will be revoked. I am sure there will be someone who perhaps would like to see that happen, but in the spirit of Amateur Radio let’s hope no one attempts to do it. I am attempting to locate a PCB Manufacture who will be able to produce the board at a reasonable price for those who do not wish to produce their own boards using either the Positive Sense, or Negative Sense approach, and will advise all who that will be.

Figure 4 Circuit Board Layout for 0.032 inch RO4003C Substrate

179

Figure 5 MRF286 Amplifier Schematic

180

Figure 6 4 Port Power Combiner Layout    

Figure 7 Isolation & Coupling Response of 4 Port Power Combiner

181

Figure 8 High Power Output Combiner for 2 Amplifiers  

Figure 9 Isolation & Coupling Response of High Power Output Combiner for 2 Amplifiers

182

Figure 100 Early Prototype of Amplifier with Taconic 20 mil Board

183

Modifying a DMC Dielectric Resonator Oscillator for Amateur 10 GHz Use Brian Yee, August 2010 Introduction Some time back I had the thought of replacing the “brick” local oscillator in my 10 GHz transverter with something a little more modern. The “brick” oscillator is a well-proven design with excellent performance, but often requires a 20 volt (sometimes negative) supply and consumes 4 to 12 watts, and requires a locked 100 MHz reference frequency. I was looking for a small package that would operate from 8 to 15 volts DC and take a 10 MHz reference frequency directly from an OCXO. There are some synthesizer products in the 1 GHz range on the market but when multiplied to 10 GHz had less than desirable phase noise. Some YIG synthesizers are available in surplus but they consume a lot of power. Then I remembered I had some DMC (Digital Microwave Corp.) dielectric resonator oscillators in my junkbox and thought that these might be good candidates for experimentation. They have excellent phase noise and were designed for 8.4 volt operation. The drawback was that they used a simple ovenized 100 MHz crystal similar to the “bricks” and I would have to build a PLL to lock them. Since I had to build a PLL anyway, why not lock the DRO directly? A dielectric resonator oscillator (DRO) is a small ceramic cylinder with a large dielectric constant and low dissipation factor, and acts like a high-Q resonant cavity. When used in an oscillator, it is used as a series or shunt feedback element. Since the puck is sensitive to external coupling, circuit board traces can be used to couple energy into and out of the puck. Also, for coarse tuning, a tuning screw supported above the puck can be used to tune the oscillator over a few hundred MHz. A varicap on one of the coupling lines can be used to provide voltage tuning. The advantage of dielectric resonator oscillators is in addition to their high Q and excellent phase noise, they consume a lot less power than YIG oscillators.

DMC Oscillator Internal Architecture A little reverse engineering revealed the internal architecture of the DMC design. The crystal oscillator output is divided into two paths. One path is multiplied to produce a 1253 MHz signal, and the other path divides the crystal oscillator frequency by 4 to provide a signal in the 25 MHz range to be used as the reference frequency for a phase comparator. The dielectric resonator oscillator (VCO) is sampled and divided by 8 to produce a signal 25 MHz higher than the multiplied crystal oscillator frequency. These two signals are mixed to produce a 25 MHz difference frequency. This is the other input to the phase comparator. A loop filter integrates the phase comparator output and provides the tuning voltage to the 10 GHz VCO.

Oscillator Modifications I decided to prescale the 10 GHz VCO and lock it directly with a synthesizer IC, the Peregrine PE3341, which has an internal flash memory to enable automatic frequency programming on power up. The prescaler is a Hittite HMC383-S8, which has an upper frequency limit of 12 GHz.

184 1

I wanted a DRO VCO output frequency of 10224 MHz, so a divide by 8 prescaler produces an output frequency of 1278 MHz. The PE3341 synthesizer chip is programmed to lock 1278 MHz with a 100 KHz reference frequency, derived from the 10 MHz input from an external OCXO. The output of the PLL is integrated and is used to control the tuning voltage for the 10 GHz VCO. Two PC boards were designed and etched on 0.031” FR4. The layout is designed so that the prescaler input is close to the original sampling point for the 10 GHz VCO, therefore the loss from the FR4 is minimal. A second board in the bottom of the case provides voltage regulation and signal conditioning for the 10 MHz OCXO input. Figure1 shows the completed oscillator shown with the original unmodified DMC unit. Figures 2 and 3 show the top and bottom of the modified oscillator.

Figure 1. Unmodified (left) and modified DMC oscillator

Figure 2. Upper PC board, showing prescaler and PLL IC. Dielectric resonator oscillator is at the top center.

185 2

Figure 3. Lower PC board: regulator and 10 MHz signal conditioning board. The board at the top is the original 5 volt regulator and phase comparator.

Performance A phase noise measurement was made and measured approximately –77 dBc/Hz at 10 KHz away from the carrier. Figures 5 and 6 show the spectrum at spans of 10 MHz and 100 MHz. Except for a couple of 2 MHz spurs 60 dB down from the carrier, the spectrum is clean. This oscillator has been used in several contests and has performed well over temperature and voltage. Further improvements are being investigated to improve the phase noise.

Figure 4. Phase noise plot at 10224 MHz

186 3

Figure 5. Spectrum plot at 10 MHz span

Figure 6. Spectrum plot at 100 MHz span

187

4

188 5

References 1. Peregrine Semiconductor PE3341 data sheet, www.peregrine-semi.com 2. Hittite Microwave Corporation HMC363G8 data sheet, www.hittite.com 3. Analog Devices ADIsimPLL, http://www.analog.com/en/tools-software-simulationmodels/resources/rfif-components/pll-synthesizersvcos/index.html

189 6

Connectorize Your IF Radio! Wayne Yoshida, KH6WZ San Bernardino Microwave Society (SBMS) [email protected] One of the most time-consuming things needed for an SHF station is the inter-connection to and from the IF radio to the transverter. Yaesu, Icom, MFJ, or Kenwood - make your choice, but all of them have one thing in common: You must have access to some of the control circuits inside the rig in order to interface a transverter to the unit. Many times strange, hard to use, or hard to find connectors are used on the rear panels of these rigs. And if you must change the IF radio due to a malfunction or rig change, you most likely need a new interface cable. Over the last several years, I have been collecting various all-mode VHF rigs, both new and used, to become IF radios for my transverters. This includes classic radios such as the IC-202, TR-751A, TR-9130, TM-255A, FT-817 and HTX-10. At first I thought I could just make up some sort of adapter to convert the connectors for one radio to another, but this became very cumbersome after a very short time. So, I solved this issue by connectorizing my IF radios, and adding some simple but useful features. See Figures 1 and 2. In this example, the FT-817’s mini-DIN connectors have been converted to standard, ordinary and available-everywhere RCA connectors. Although only two connections can be used to enable the FT-817 as an IF radio, I decided to bring out all of the connections to the outside world via the RCA connectors. In addition, locking toggle switches are used for a “Key Lock” function (closes and holds the CW key contacts for a continuous tone) and “PTT Lock” (enables transmit) for sending a continuous carrier signal. LEDs indicate status. The red push button is a backup CW key that will never get lost. Now, no matter what IF rig I choose to use (or, in case of a field failure, a change by force), any IF rig can be used with any of my transverter systems since RCA connectors are used to interface the IF radio functions.

Figure 1. The connectorized FT-817. The front panel has locking toggles for keyand PTT-lock. The red button is a CW key.

190

Figure 2. The rear panel, showing the RCA jacks and the wires going into the interface box.

9/15/2010

WORKING ON THE MICROWAVES Seeing is Believing

W3ZZ & WA3OFF MUD 2010

NOW FOR SOMETHING DIFFERENT • Most MUD presentations focus on microwave tx, rx, antennas and ancillary equipment that makes them work as a system • This presentation will assume you have all of that and emphasize the techniques needed to actually work one another

191 1

9/15/2010

STEPS TO A CONTACT • You must find and detect the other station • They must find and detect you. Sounds simple. But the microwaves have specific problems. We will examine 902 MH tto 10 GH MHz GHz b butt much h off th the di discussion i will be applicable to 24 GHz

MICROWAVE ADVANTAGES • Little or no QRM • Sensitivity is paramount. Strong signal handling capability is less important

192 2

9/15/2010

MICROWAVE DISADVANTAGES • Signals are often extremely weak • High gain antenna have very narrow beamwidths • Frequency accuracy more difficult to determine q y stabilityy more difficult to maintain • Frequency • High power txs not easy to acquire or build Many MUD presentations deal specifically with these problems.

WEAK SIGNALS • Low noise front ends – ≤ 0.5 dB at 902 MHz to ~±1 dB at 10 GHz

• Mast mounted transverters and preamps to limit transmission line losses

193 3

9/15/2010

NARROW BEAMWIDTH ANTENNAS • Typically ≥20 dBi gain, often ≤15° b beamwidth idth • Use 6 digit grids to know accurately where you are pointed – Bore sighting – Fluxgate compasses – Beacon directions – 1° readout digital rotators • Both readout and rotator must have 1° resolution

STABLE ACCURATE FREQUENCY • GPS locking • Precision reference oscillators – Nominally @ 10 MHz

Many papers at MUD and elsewhere cover this subject in great detail

194 4

9/15/2010

TRANSMITTER POWER I • Base transverters often very low power ( (mw) ) • Amplifiers are expensive – Buy commercially --DEM, DB6NT, etc. – Convert commercial amplifiers • 902 cell amps; 2304 Spectrians; 2345 Toshibas; 5760 Avanteks

– Design and build from scratch or kit • W6PQL 23 cm

TRANSMITTER POWER II • To make a contact one of the two stations must initially hear the other • Once you have a decent antenna [~≥20 dBi] increasing tx power may be more effective ff ti th than iincreasing i antenna t gain i • You can never have enough tx power

195 5

9/15/2010

LIAISON FREQUENCIES A SLIPPERY SLOPE

• Most µW operations maintain a liaison f frequency on 2M or 70cm 70 to t help h l fifind d th the other station on the µW s • Once you have found the other station, all communication on the liaison frequency should cease while the µW contact is being made – but often it doesn’t • Without the liaison frequency many µW Qs would never be made. Is there a better way??

A DIFFERENT APPROACH I • Even if you do all these things the µW station you are trying to work has likely done few or none of them • What if we could look for weak signals simultaneously i lt l in i a chunk h k off µW W spectrum t 50 KHz wide or more?

196 6

9/15/2010

A DIFFERENT APPROACH II • In this case – The other station’s lack of frequency accuracy/stability does not matter – You can see them

• If you run enough transmitter power he mayy well hear yyou even if he is not quite q pointed at you and hopefully he will peak his antenna

ENTER THE SDR • Soft Software are that controls p pure re SDRs can look at relatively wide bandwidths and display the results on a spectrum scope and a waterfall • But the spectrum p scope p must be sensitive enough to see anything you can hear – or this approach will not work

197 7

9/15/2010

TYPES OF SDRS • Pure P re SDRs – Transceivers Flex Radio, Soft Rock – Receivers RF Space, Perseus, Quicksilver

• Hybrid SDRs Elecraft K3 Hybrids require a pure SDR to tune their IF frequencies: e.g., LP-PAN or Soft Rock and suitable software like a modified Power SDR Console. Or a P3 panadaptor.

W3ZZ SPECTRUM SCOPE

198 8

9/15/2010

W3ZZ µW STATION I • Built around a K3 hybrid SDR • Transverters are selected by the Mother of All Switching Units MOASU [March/April 2008 World Above 50 MHz] • 902/1296 are in the shack; 2.3 GHz and above are tower mounted at the antennas • 8.2 MHz Soft Rock tunes the K3 IF • Modified WU2X PowerSDR/IF stage looks at a 96 KHz chunk of µW spectrum

W3ZZ µW STATION II • The LP-PAN is a more elegant solution to tune the K3 8215 KHz IF • The LP-PAN or the Soft Rock can be built to interface with several other tcvr IFs • The W3ZZ system sees 96 KHz with a Delta 44 soundcard. The E-MU 0202 soundcard d d will ill see 192 KHz KH • The Elecraft P3 panadaptor is probably an even better interface and has better integration

199 9

9/15/2010

SYSTEM SENSITIVITY • When I started this project I found that the stock t k WU2X PowerSDR/IF P SDR/IF Stage St fails f il to t see many signals that the ear can hear • WHY??? • Roger Rehr W3SZ had the answer. – There are too few bins for adequate sensitivity over the 96 KHz coverage of the sound card

• WHAT TO DO??

RATIONALE I • As downloaded the WU2X software is hard coded for 4096 ((212) bins in the spectrum display buffer, or 23.4 Hz/bin with a sound card that has a 96 KHz sample rate • That is insufficient; you can hear things you cannot see • Changing to 65536 (216) bins reduces the bandwidth to 1.46 Hz/bin; now you can see everything you can hear

200 10

9/15/2010

RATIONALE II • The sensitivity [essentially the SNR] of a 1 46 Hz 1.46 H bin bi iis 12 dB b better tt th than 23 23.4 4H Hz bin • By inspection I have determined that as downloaded the IF Stage software fails to see the weakest signals by <12 12 dB. • Thus the narrower bin allows the spectrum scope and its waterfall to see all the signals audible in the receiver

MODIFYING THE WU2X POWERSDR/IF STAGE • Change the console software so that the spectrum t di display l b buffer ff size i iis user configurable from 4096 to 262144 • Change the DttSP [DSP processing DLL] to accept display buffer size parameter from the console • Change the Wisdom file generator to create information for the fast Fourier transforms for buffer sizes up to 262144

201 11

9/15/2010

K3UO/B 2304 FM09rc

K3UO/B 3456 FM09rc

202 12

9/15/2010

SUMMARY • The spectrum scope is a powerful tool to fi d stations find t ti whose h ffrequency accuracy and stability is poor • You can find stations that you could easily miss if you were tuning by ear • It obviates the need for a liaison frequency • SEEING IS BELIEVING!

ACKNOWLEDGEMENTS • The authors wish to thank Roger Rehr W3SZ who originally brought the basis for the sensitivity problem of the spectrum scope to our attention and who provided an example to implement the change in the size si e of the spectrum spectr m display displa b buffer ffer and supplied his modified code for the Wisdom file generator

203 13

Physical Optics Demonstrations with Microwave Or Fun with Microwave, Other than Talking and Cooking Of all the fields of light the most mathematical is “physical optics.” Light of course, is an electromagnetic wave of a wavelength different from that of radio, but physical optics applies to all bands of the electromagnetic spectrum. It is the study of polarization, diffraction and interference; terms familiar to radio engineers as well as the optical engineer. There is also “geometric optics,” having to do with lenses and there is the “met-trees” of light; photometry, and colorimetry. The word optics in “physical optics” is because the earliest work in this field was done using the eyes; going back to the early 19th Century. When math is applied to physical optics the more general term electrodynamics is used. In this lecture you will see examples of famous physical optics experiments using microwaves. Without the math. Instead of mathematical evidence, I will create the conditions for nature to show you. It is my hope you will not only find physical optics interesting but will be inspired to enjoy microwaves in ways other than modulation. You can take as much understanding of these demonstrations as you can by seeing them happen. This sort of “witness-understanding” allows you to remain in awe and yet still be able to use that understanding to invent useful things. Polarization, diffraction and interference occur in all bands of the electromagnetic spectrum. The microwave band however, is best for demonstrations because the waves are of a size that is very convenient for the lecture table. For example, the slats of a diffraction grating for the microwave can be seen from across the room, yet the wavelength is small enough that a beam can be formed and passed through the grating to emerge at a noticeably different angle. That choice of wavelength is why I joined SBMS. I’ve learned a great deal from these guys about the plumbing and detection of microwaves.

Interference When two frequencies play on the same detector inside a cavity the difference between those two emerge as a signal on the wires that support the detector. That difference in frequency means the world to you guys. All hams are familiar with heterodyning . Microwave hams usually heterodyne twice; a down-converter beats it down to some frequency a VHF receiver accepts through its antenna and in there, another local oscillator beats it down yet again to a more familiar IF. Heterodyning is an interference phenomenon that happens in time. The interference of physical optics is heterodyning in space. With multiple sources and a detector that is free to roam about, interference happens in places as well as in time. You are familiar with that as “multipath.” The reason multipath is of little concern in broadcast radio is that with a strong signal, multipath interference is rarely perfectly canceled. A null has to drop below the noise level to be revealed as a dead spot. In physical optics it is the “places” of interference that are of most interest. And these nulls don’t have to drop to zero to be seen. The following is a list of demonstrations that will show this form of “spatial” heterodyning.

204

  





Classic Young’s Double Slit Experiment o The line down the middle is always a place of reinforcement Double Source (use the tee structure for a stronger signal) Introduce a second source that is not coherent with the first o Block one source; at the null, at the peak and somewhere in between o The lesson here is that coherent waves add in a way that is not intuitive o Place sheets of plastic of various thicknesses to show ½ wave of phase change Turn to the Michelson interferometer and show the fringes in the longitudinal direction o Bring out the spark source to show the fall-off in the fringe contrast o Use a different detector to show the filtering effect of the cavity. Rearrange the Michelson to make it inline o Show how sensitive the DC amplifier is to changes in position. o Make the speaker sound as the motion is faster o Connect it to Spectran to show velocity o Show Eddie’s video and how much fun this aspect microwaves can be o Show that vibration and spinning antenna are not Doppler o Show Bill’s phase sensitive motion sensor

This paper examines just a few of the things to be shown in the lecture. The photos are just a sample of what the lecture will show. A very important lesson with the equipment to the right is that demonstrated by the covering and uncovering of one source when both sources are coherent to each other. In one configuration, it could be less than the sum of each measured alone. And in another, it could be more than the sum of each measured alone. Adding Two Incoherent Sources When they are not coherent, the covering and uncovering makes sense; the detected power is the sum of the two sources when measured by themselves. Since most of you are radio engineers, you will recognize that the two distinctly different sources will result in a beating that may be in the MHz, but the power resulting from that mixing is simply one plus the other. If the two sources are almost the same, say different by only one cycle per second, the power meter needle will deflect from zero to some maximum at 1.0 Hz. If the difference in frequency is faster than the needle will respond, the needle will ride at some RMS value that is less than the peak but obviously more than zero. It is when it is at its peak that the incoherent sources are behaving like coherent sources, but normally, detection of two mingling incoherent signals is the average.

205

The power meter is merely a detector and a voltmeter and it is good to remind ourselves that the output of a detector is always on the same side of ground. When modulation is to be amplified, it is passed through a capacitor where ground is moved to the middle and the modulated signal now spends half its time negative. Do not confuse these negatives (the negatives of the modulation) with the RF negatives that are playing on the detector which we do not measure. Adding Two Coherent Sources But when the sources are coherent, that is, when they have their beat frequency adjusted to zero, that’s when things seem not to add up. When one source is covered and uncovered, the result is sometimes more than the sum of the two when measured alone, some places complete cancelling and more often than not, some intermediate result including the curious condition of no change when either one of the two sources is covered and uncovered. Interference happens because in certain places, the waves from one of the sources has gone a bit further than the other and if that difference in distance is just right the waves from the two sources always arrive out of phase and no signal is measured. We know this because we are taught that radio waves, light waves, even sound waves cancel when the phases are 180° out. There are other places where they are “in” phase and add. We were taught that too. Not taught is why the in-phase place results in more signal than the sum of the sources when measured alone. Waves in sound and on water were well understood in the early 19th century when light and later radio waves were being studied for the first time. The important part of Huygens’ wave theory is that waves spend half their time in an opposite state. These opposite states can be made to cancel another source of waves if the positive of one is running in the same direction and comingling with the negative of the other. The important thing to remember is that a detector can only measure plus/minus fields arriving together one after the other. If the cancelation is incomplete the remainder is still detected with the same sign as when they are perfectly in phase. A detector cannot detect only the negatives or only the positives. The minus and plus arrive alternately billions of times per second, and the detected signal is a DC value. The DC value may change but it is never less than zero. Those places where there is a maximum it is because the plus peaks from both sources and minus peaks from both sources are arriving together. It is a place where peak to peak for one source always arrives with the peak to peak for the other source. Where the minuses arrive with the pluses there is a continuous canceling. Everywhere else it is somewhere between peak + peak and zero.

206

Let me apologize for this long and not very convincing explanation. Perhaps seeing it happen in the demonstration will bring sufficient closure to the observation. It is the observation that counts. It is that in some places in the interference field the two signals playing on the detector measure more than the sum of the sources measured alone and in other places (where canceling is going on) the signals measured together are less than the sum of the sources measured alone. It is as if there’s an aspect of electromagnetic radiation that is “raw” and capable of adding with sign and an aspect of electromagnetic radiation we can measure which doesn’t show a sign. The raw aspect is called amplitude and the part we can measure is called intensity or power.

Diffraction It is common knowledge that diffraction is the bending of light due to an edge. In astronomy it is known for the spikes on star images caused by the struts that hold the secondary optics. In the microwave it is a little less easily observed. Most of you know it as the bending over a mountain where knowing the diffraction angle for your particular frequency can tell you where to put your antennas. What is less well known is that diffraction is not caused by the presence of an edge but the absence of an aperture. It’s the area as well as the direction of the obstruction that counts, not the length of the edge. By making the struts that hold the secondary of the telescope thinner, the diffraction spikes will be less noticeable. In microwave, diffraction happens over a sheet metal fence, whereas a wire, even a chain link fence has practically no effect on the signal. The following demonstrations will show diffraction as an interference phenomenon.  



 



Start with a grating of metal strips; the most familiar example of diffraction Demonstrate the Young’s Double Slit experiment. o Go to the slides to show the introduction of more radiators between the two o More slides to tie the double slit to beam divergence and aperture Return to the beam configuration and show the effect of o a single edge (hard to see) o a surrounding edge (a ring of wood) return to the strips then show the dielectric grating Show the narrowing of the beam with: o The flat flange vs the choked flange o Poly rod radiator Return to the slides to show what diffraction is not

207

Diffraction is the Cause of Beam Divergence The reason diffraction is taught in a separate chapter in physics is historical. It was more noticeable, easier to quantify than interference. In the demonstration you saw that Young’s Double Slit produces an interference field out in front of the sources. When additional in-phase sources are made between the separated two sources, the diffraction pattern becomes more beam-like with less pronounced nulls within the beam. When the sources are continuous between the original separation, there are no nulls within the beam and the width of the beam is easily predicted. The most useful aspect of diffraction is the formula for the narrowness of the beam given the diameter of the aperture. This formula is the famous Rayleigh Criterion:

It was at first derived by merely plotting the size of a star with diameter of aperture, but by the late 19th Century it was derived from first principles. The following Rule of Thumb is from http://www.microwaves101.com/ Rule of 70

Angular beam width (degrees) = 70 degrees / (D/�)

for example, a X-band (3 cm) and 30 cm dish:

70° / 30/3  7°

By definition an isotropic radiator (radiates equal intensity in all directions), is 0 dB. Directional antennas are measured as a ratio of power per square degree compared to the power per square degree of an isotropic radiator. As you might guess, the gain is merely the smaller solid angle divided into the larger. The directivity or “gain” is…

Directivity ≈ 10 D2 / �2 Gain ≈ � * 10 D2 / �2 208

where � is antenna efficiency

Using the same X-band example and an � of .5 .5 * 10 * (30 cm)2 / 32  .5 * 10 * 100  500 and ten times the log of that is 27 dB Here's Microwave 101’s rule of thumb for the above formula in terms of beam widths:

Gain = 27,000 / (Θ1* Θ 2)

where Θ1 and Θ2 are the 3 dB (half-power) beam widths, measured in degrees (not radians). Using the above 7° example: 27,000 / (7*7)  551 and ten times the log of that is 27.4 dB

Non Dish Beam Diameters What about the gain of a 12 element Yagi? I’m sure there’s a formula for that, but it would be much more complicated than any of these rules of thumb. The important point is that the more elements, the higher the gain. A receiving antenna with lots of directors is not just more directional; it actually sucks power out of a larger area. A longer Yagi has a wider shadow behind the antenna than a short one. In the case of a quad array of Yagis, the longer the antenna, the more widely separated they have to be on the masts so as not to rob from each other. As transmitting antennas, they actually radiate into a larger beam if there are more directors. You won’t hear a professor of physics teach that because the explanation is probably very difficult. (I can’t explain why an antenna can reach out beyond its actual physical area, but notice that, that doesn’t keep me from telling you about it.) Polyrod Radiators Even more interesting than long Yagis are the polyrod radiators. These are tapering rods of plastic that emerge from a waveguide. This method of creating a beam was preferred (over dishes) by the Germans in WW2. As you saw in the lecture demonstrations, the polyrod radiator is as directional as a 3” wide horn yet is only a fraction of an inch in diameter. However, when metal with various diameter holes are put over the polyrod radiator, it can be seen to reduce the beam directivity. This shows that the actual radiating area extends beyond the dielectric just as in the case of the Yagi. It is as if the radiation emerges away from the rod (away from the line of directors in the Yagi) but then propagates in the direction of the rod. It is this virtual aperture that confines the beam. And again, the bigger it is, the less diffracted is the beam.

209

Polarization The third aspect of physical optics is polarization. This is the orientation of the antennas in the direction that most antenna rotators don’t move in (rotation about the direction of propagation). The following demonstrations will show polarization phenomenon as distinct from interference and diffraction. 

projected beam of white light and cross polarizers and plastic under stress.



show polarization also at 72 MHz, and then 10 GHz.



show Malus’ Law demonstrator using my orthomode detector and the rotatable source. o Show slides that illustrate how Malus’ Law can be used to show the relationship of power and amplitude.

Linear Polarization The main lesson in linear polarization is that the matching of orientation is not like the tuning a station in frequency. There’s no way to make it any sharper than the cosine squared of the angle. One can use the null side to more accurately define a direction. (This is like a direction finding null; to be accurate you have to have a strong signal and a direct path to the source.) But all you will have done was determine the exact orientation of his antenna; not nearly as useful as direction finding. The reason polarization alignment in broadcast radio is of so little concern is because matching the orientation has such a broad optimum. And of course there’s so much power available. In addition, there’s so many multipath reflections, there’s bound to be a path that has sufficiently correct polarization to give good reception. (Reflections cause a tilt in the polarization.) When proper polarization counts is when the signal is very weak. Hams of course arrange their antennas by convention; 6 meters and longer horizontal, 2 meters and shorter vertical. Just to experiment, hams will have orientations different from the standard, but you always communicate that a head of time with the other ham. Some of you may have played with diversity reception where you join two receivers tuned to the same frequency but being fed by two different antennas. Diversity in polarization means one is horizontal and one is vertical. It is better than a single antenna when trying to get the weakest of signals by way of bounces. But it is not as effective as diversity in azimuth or diversity in location; that is if you have the room for them. (Vertical and Horizontal antennas can be within each other.) In optics Polaroid dark glasses are the most common example of polarization. But it’s not a very useful one. It reduces only some kinds of glare. In the radio world, you don’t even have that. Polarization is just something to keep in mind when setting up your antenna. Its most important feature is insight into the nature of electromagnetic waves. That’s what follows.

210

Nineteenth Century Physics Historically this was the earliest physical optics phenomenon to be explored; the 1830s. There was no industry for research in those days. Physics was even new to the universities. Scientists were almost always inheritors of wealth. They had the time for research and could pay for the equipment. But nonscientists who were otherwise well educated liked to show their awareness of the sciences. It was very common in the 19th century homes to have terrariums, or aquariums. In physics, the most common object was a polariscope. It consisted of cross polarizing filters and in between a specimen that was transparent. Mineral, and glass under stress were the most common objects. The appearance was quite colorful. This was at a time when dies weren’t very saturated and photography hadn’t been invented yet. It was a kaleidoscope with a science lesson associated with each display object.

Circular Polarization The colors of the polariscope involve circular polarization. Color is difficult to show in microwaves, but circular polarization can be produced and it can be sensed. The helical antenna is not a common site above the roofs of hams, but all hams seem to be familiar with it. That it works is much more difficult to explain than it looks. So I am comfortable with simply showing you how to make one and how to see that it is working. The following list of demonstrations show how circular polarization in X-band is made from linear, how it is used and to some extent what it looks like.     

Explain the spinning detector Install the various circular polarizing devices Show the random polarizer Show slides of TEM modes and finish lecture with the Goubau line and the radiated radial polarization

Helicals Those hams who are familiar with helical antennas know that there’s right hand twist and left hand twist. What is less well known is which way the twist has to be to receive it. If you think too deeply about it, you will never be sure. The best way is to make some and play with them. Helicals are surprisingly easy to make, particularly in the microwave since the helix can be made self supporting. (Use 17 gage wire

211

and make many of them. The 17 gage is lovely to work with and it is a slip fit into the center conductor of a Type N female connector.) As to which way to wrap it, it turns out the simplest understanding is the correct one. A right helical on the transmit side needs a right helical on the receive side. Left helical needs a left helical. It helps to know that it doesn’t matter how you look at a helix. If you’ve ever cut the head off a bolt and then deburred the cut, you know the same nut fits on either end. Indeed a nut works no matter which side starts. There are so few hams using helicals, there really is no standard of handedness. But if you do experiment make sure you and your pal agree. There’s an interesting and somewhat rare exception to the left-left, right-right. Although there doesn’t seem to be any advantage to circular polarization for EME, when you do use it, the ham catching the reflection off the moon has to have the opposite handedness. That is also the case when bouncing off a mountain. (Fortunately, you don’t need to switch between LH and RH when you switch from talk to listen.) There is no such shifting from vertical to horizontal when using reflections with linear polarization. It has been observed that noise is rarely circularly polarized, but that can’t be taken advantage of because at the microwave end of the ham spectrum, amplifier noise is larger than natural radiation noise. Oh but there is one advantage with helicals. They look so cool; especially tilted up. Squeeze Tube There are other ways to make circular polarization and these were shown in the lecture. Squeeze tube is the easiest. Whatever method you use, it is critical that you have a linear detector that you can rotate to see if the radiated power is the same in all orientations. If it is not, you will have a mix of linear and circular; elliptical. The amount of squeeze is very critical. Helicals don’t have that problem. They are always perfectly circular. Circular Polarization Feeds Squeeze tube, the diagonal plastic, and the row of screws method are interesting but even when it comes to feeding a dish you are better off with a helical. The longer the helical, the narrower the beam. As to how long, just make some and use a field strength meter to see how much fill you will get for your dish. You can find formulas for the diameter and pitch of the helix on line (Paul Wade’s Helical Feed: http://www.w1ghz.org/antbook/conf/Helical_feed_antennas.pdf) but it is far more fun to just wind your own around different diameter dowels. (Start with 3/8” for X-band) The pitch can be made adjustable by merely stretching. Another neat thing about a helical is that where the first turn emerges from the ground plane can be used to match the impedance. If you don’t want to bother with a circulator and

212

detector to use as a SWI, just put some sort of power meter in the far field and squeeze that first turn closer or farther from the ground plane for maximum good. The optimum is very broad so you can’t miss it. Waveguide Polarization Waveguides propagate microwaves in modes. The simplest is with the E-field running between broad walls. (10 GHz through a WR-90) A larger waveguide can support two side by side out of phase waves between broad walls such that when they emerge from the end, there are two beams with a null in between. The most bizarre modes have either the E-field or the B-field in the direction of travel. This mode seems contrary to common sense in that there will be oscillation in the direction of travel and you would think that on each half cycle the field would have to expand faster than the speed of light. The fact that nature has no dilemma with that, illustrates the weakness of our model for oscillation. We think of it and draw it as a sine wave that goes out and comes back in. But that’s just how we draw it. In reality, fields get stronger and weaker. They don’t go out and come back in. It is our drawing of that strengthening and weakening that makes it look like the fields move back and forth. Even stranger than the longitudinal wave modes, are the radial modes where either the E-field or the Bfield is radial inward away from the walls of a round waveguide. If the magnetic field is radial, the Efield is a whirl. Don’t confuse this whirl with circular polarization. Both need round waveguide but one is a rotating linear, and the other is a fixed E-field orientation that involves all angles at the same time. Radial Polarization in Free Space The easiest waveguide-mode-radial-polarization to launch into free space is the radial E-field. Believe it or not they do emerge from the waveguide in a legitimate polarization state; an unusual one that very few people know about. There is even an optical analog for it also called radial polarization. In the lecture when the funnel was centered over the core of the beam, it was able to capture the radial polarization. Away from that radial core, the detector does very poorly. It is a very sensitive to centering. The polarization away from the core is linear but a different orientation depending on where the detector is within the beam. Such a curious polarization state, you would think would be difficult to generate. But it turns out the launching of an E-field radial mode is the easiest of all forms of transition from coax to waveguide and from waveguide to free space. You simply expand the shield of the coax away from the center conductor until it is the diameter is that of the waveguide. Here’s an interesting fact about radial modes. The round waveguide for radial mode has to be larger by a factor of two than normal mode. For X-band it has to be more than

213

an inch and quarter in diameter, much larger than the normal X-band waveguide. It is not as efficient in terms of tube diameter, but it is just as efficient and stable as the ordinary WR-series of rectangular waveguides. And of course it transitions back to coax just as easily and efficiently. The freely radiating radial polarization seems almost too good to be true. Well, there is a problem. Although you can focus it with a lens, horn or dish into a beam, the beam still diverges and right at the center is the weakest part of the beam, instead of the strongest. You can’t even use this strange polarization for stealthy communications. All around the beam it’s linear; anybody can detect it. G-Line Let me hand you yet another level of strange. The E-field radial can be constrained to follow a wire. It is a phenomenon Professor Goubau spent nearly a lifetime studying, starting in the early 50s. It is a fascinating phenomenon; a sort of inside out waveguide. If you have to send a lot of power a long way efficiently, a Goubau line or G-line is the cheapest way to go; far cheaper than waveguide and with much less loss than coax. Why isn’t it more popular? Because there’s rarely any need to send a lot of microwave power a long way. It’s one of those really neat inventions without much of a need to fill.

214

Microwave Update 2009 Dallas, Texas October 24, 2009 Testing by: W5LUA, WD5AGO and WA8RJF Equipment supplied by W5LUA and WD5AGO Compiled by WA8RJF HP 8970B and Eaton 2075 Measurement Accuracy +/- 0.2 dB at 10 GHz and below Band 902/903 MHz

1296 MHz

Call

5760 MHz

Noise Figure (dB)

Gain (dB)

0.70 0.99 1.18

32.8 16.2 14.3

1.20 2.67

25.0 14.5

NE32584 / ATF28184 ATF 36077 TAMPS 242 ? NEC 325/ ATF54143 ATF 54143 ATF ATF

0.17 0.29 0.50 0.53 0.58 0.58 0.81 0.82 2.30

33.0 20.1 32.3 16.5 35.6 18.2 19.0 17.8 15.5

FHC04/ATF36163 NEC 3210/ATF101 NEC 325

W7GBI

0.34 0.39 0.40 0.45 0.94 1.95 2.45

29.0 31.9 26.5 15.4 21.3 8.2 15.4

Device

K5GNA K0MHC

HB HB DEMI XVTR Comm SSB Elec DEMI

WD5AGO WA6ZKY K4QF W1AIM KM5PO K5RUS KM5PO KM5PO K0MHC

HB - Reference DEMI HB DEMI G4DDK W6PQL kit DEMI DEMI DEMI

WD5AGO WD5AGO KM5PO K5RUS K4QF K7ICW NU8I

HB @ 0F degrees HB G4DDK DEMI HB HB @ 1994 HB

WD5AGO WA9FWD WD5AGO

AGO-LUA AGO AGO-LUA

NE3210/MGA61563 NE3210/MGA61563 NE3211/MGA 61563

0.48 0.50 0.57

28.5 27.8 27.1

WD5AGO WD5AGO K5GW W4ZST

HB HB HB Comm

NE3211/FHC40 ATF 36077/FSCM57 NEC 325 ?

0.76 0.99 1.15 4.52

26.2 30.8 10.4 16.9

K4QF W1AIM N0EDV

2304 MHz

3456 MHz

Design

TAMPS 242 ?

ATF 10136

TAMPS 242

215

10368 MHz

24192 MHz

K6MGM W5RLG WA5YWC WA6ZKY WA6ZKY W5RLG AD6IW KM5PO WA9FWD WA9FWD WA5YWC

HB Kuhne HB LUA DEMI kit DEMI kit Kuhne HB DEMI-LUA Comm #5188 Comm #2797 Comm #1391

W5LUA W5LUA W5LUA W5LUA NU8I

DB6NT #1 HB HB XVTR #2 HB Comm

47 GHz

W5LUA W0EOM W5LUA

SMT-1240 Test Xvtr

78 GHz

WA1MBA VE4MA W5LUA

LNA#4 Converter Converter

216

AD6IW design NE3210 HEMT 2X ATF 36077

0.78 0.90 1.03 1.10 1.22 1.23 1.27 1.90 1.95 2.12 2.23

11.2 24.8 24.1 8.5 10.7 20.5 24.2 20.4 29.5 27.7 30.3

1.83 1.92 4.00 4.03 5.23

26.3 13.8 25.0 19.4 17.3

AMMC-6241 40-45 GHz LNA

4.93 9.45 19.8

14.2 15.7 15.3

2 Stage CHA1077 Mixer/IF Amplifier Isolator/BPF/Mixer/IF Amplifier

7.75 14.00 20.20

19.8 12.2 15

XVTR NE3210