Jump down to your favorite software company of yore:
Much of the material below is a product of the combined experience of the Microwaves101 Professor (an ex-EEsof guy) and the Unknown Editor (a recovering MMIC designer with Tourettes Syndrome), with a ton of edits by the father of microwave CAD himself, Les Besser, and even a conversation or two with Randy Rhea, James Rautio and now Ulrich Rohde (how's that for name dropping!) If anyone can contribute to this history lesson, please send in your stuff, even if it's just a correction! We'd like names and dates, and for really important CAD visionaries, we'd like to put them into the Microwave Hall of Fame!
This was Les Besser's precursor to COMPACT. In his own words:
"I wrote the SPEEDY on the General Electric Timeshare system while I worked at Fairchild's Microwave Division in 1970. The main goal was to promote Fairchild's new line of microwave devices, including the first commercially available GaAs FETs. The program had two-port analysis only, but it stored all the Fairchild transistor S-parameters. Users accessed the program worldwide through local telephone calls with terminals (i.e., teletype) and modems. First-generation terminals operated at 110 baud (bits/s), but later the "high-speed" Texas Instrument portable terminal had the blazing speed of 300 baud. Note that I am not using "K" in front of the bauds! Even though by today's standards that sounds awfully slow, commercial timesharing brought the design software to engineers through telephone lines. A few companies, like Texas Instruments, had internal software on their mainframe computers, but most of the microwave engineers did not have such capability and they were happy to use timesharing.
SPEEDY is not an acronym, we just wanted to advertise its speed, compared to the early day time-domain simulators that operated at low frequencies.
The second-generation program, COMPACT, had circuit-optimization and it was commercially introduced through six different timeshare companies. Its transistor databank, in addition to the Fairchild products, also included other manufacturers' devices. In-house installations began in 1973, with Communication Research Center (CRC) of Canada being the first one. See more on COMPACT below."
According to Robert P. (Bob) Coats, one of the first CAD tool for microwave circuits, CAIN-01, was developed about 1971 for use by designers at Texas Instruments. CAIN-01 was available on an IBM 360 computer and accessed using punched cards. It was never marketed by the corporation. Here we quote Bob's historical perspective on this achievement:
"CAIN-01 stands for "Computer Aided Integrated Networks version 1". In July 1973 I published a paper in the July issue of the MTT transactions of the IEEE. I used the software on the work described, and referenced an article by its designers. The authors were T. W. Houston, L. W. Dyer, and G. J. Policky. The paper was entitled "Computer-aided design of Microwave Integrated Circuits" and it was published in the August 1969 IEEE Wescon Convention Record. The software was developed initially in support of the first phased array development program at Texas Instruments (the MERA program). The software was widely used by microwave circuit designers at TI for many years."
This information came from Albert, thanks!
For a while in the 80's I had a program called "CADEC" that ran on a HP 9845 desktop. It was stored on a HP-200 tape cassette and did OK with passive circuits. It relied mostly on ABCD parameters and Z/Y matrix analysis (like in low frequency AC circuit analysis with matrices). The cool part was it came in Rocky Mountain basic and you could go in and change anything, add new circuit elements, and mess around with display options. At the time Martin Marietta was strongly opposed to PC's due to a very entrenched computer department, and we would order HP computers as "instrument controllers" so we could get some free computing power. I may still have a manual around.
Albert, we'd like to get some scans of that manual! - UE
For most of us, in the beginning, there was COMPACT. The acronym stands for "Computer Optimization of Microwave Passive and Active CircuiTs". Introduced in 1973, it was the only commercially-available microwave circuit optimization software product available during the disco years (in-house programs at General Electric and Texas Instruments were capable of analysis without optimization, see above). During the hippy epoch and for all times before, microwave design was done using a Smith chart and a pencil and even the simplest designs took many months to develop. Les Besser was the originator of the COMPACT program, and appears in our Microwave Hall of Fame for laying the groundwork for microwave CAD industry. Les is a Hungarian-Canadian-American. Born and raised in Hungary, he escaped to the West in 1956, became a Canadian citizen in 1961, later moved to the U.S.A. and was naturalized to USA citizenship in 2000. Welcome aboard!
Here's some references for Les's early work in this area:
Besser, L., Newcomb, R., "A Scattering Matrix Program for High Frequency Circuit Analysis" IEEE Conference on Systems, Networks, and Computers, Mexico, January 1971
Besser, L. "Computer Aided Design of High Frequency Circuits" Electromechanical Design, August 1971
Besser, L. "A Fast Computer Routine to Design High Frequency Circuits" IEEE ICC Conference, San Francisco, California, June 1970
Les Besser submitted a description of his COMPACT computer program to the IEEE Circuit Theory Transaction for publication in 1972, but they rejected it, perhaps because it didn't contain enough fancy equations... below Les Besser fills us on this chapter of microwave history in an email to Microwaves101 dated September 1, 2004:
"The article you mentioned was initially rejected by the IEEE Circuit Theory
(now Circuits and Systems) Group. I felt quite bad at first, until I found out that Phil Smith's original article on the Smith Chart was also rejected with "No practical contribution" comment. After rewriting the COMPACT article (removing most of the explanations and adding lots of complex matrix equations) it was published."
No practical contribution indeed! There can be no doubt that the IEEE will never publish an article on Microwaves101 either! Here is a scan of as close to the original COMPACT user's manual as you will ever get:
Because it predated the personal computer, COMPACT ran on mainframe "timeshared" computers such as the monsters from Control Data Corporation and IBM. Most of us thought that COMPACT was so wonderful, no one could ever compete with it or improve upon it. The first versions only analyzed and optimized circuits made from ideal transmission lines (and could only handle elements with cascaded or series/parallel connected two-ports, ergo no multiloop feedback circuits!) Soon microstrip elements and discontinuity junctions were added. Later versions could do crude matching network synthesis and even drew Smith charts using alphanumeric characters. What could be better than using a line editor on a remote printer terminal? This was so much better than punch cards! Recall the simple pleasures of using remote terminals hooked up with 300 baud acoustical modems, while paying $3000-$5000 per CPU hour. Didn't every engineer take pride in knowing the proper connect settings, such as seven or eight bits, odd or even parity? Come to think of it, the situation was not that good. But by 1980, if you were lucky enough to own a Digital Equipment VAX, you could run COMPACT on it, thereby owning your own design workstation, which probably cost only $250K! Below is some COMPACT marketing material.
COMPACT was bought out by COMSAT in 1980, and Les Besser was associated with it for three more years only. During that time a third-generation version, SuperCOMPACT was released that became the industry standard. While the COMPACT SOFTWARE group was fat, dumb and happy, a revolution started in the early 1980s, due to the IBM personal computer. Even with its incredibly limited memory and low speed, some visionaries recognized that microwave design would some day be done on this type of platform. Two specific groups of visionaries included some California dudes from "the valley", who chose the name EESof, and some guys located at Stone Mountain Georgia (near Dogpatch), who chose the really cool name Circuit Busters (who ya gonna call?)
COMSAT could read the writing on the wall and eventually offered SUPER-COMPACT-PC, a version of which could run on an IBM PC. They added their own layout tool, AUTO-ART. However, SuperCOMPACT PC was a "chopped-up" version of the large mainframe program, while Touchstone was written for the PC architecture, taking advantage of the new operating system, and the decade-long leadership of Compact Software was successfully challenged by EESof. After Comsat lost interest in CAD, COMPACT Software again became a private company, owned by Ulrich Rohde. It was located across the street from his other microwave enterprise, Synergy Microwave, in beautiful Patterson NJ, birthplace of funnyman Lou Costello.
"The 201 McLean Boulevard Patterson location on Route 20 near the river is and was a nice location and Compact was in a fancy building. No need to shoot at Patterson
When I bought the company from Comsat, both the main frame version and the PC version where unstable and models such as T-junctions, crosses and others fairly inaccurate at higher frequencies. Becoming a partner in the DARPA MIMIC program, and mostly financed from earnings rather the DARPA money, these things were changed by developing EM-based models, and verified by Raytheon and Texas Instruments.
The most significant contributions were:
N-dimensional nodal noise analysis for all types of both linear and nonlinear circuits, including oscillators, mixers and amplifiers under large signal conditions.
EM-based models with extremely short computational time replacing lookup tables and curve fittings. The multi-coupled line model by itself is a full EM-based simulator with extreme accuracy .
The first complete nodal-based harmonic balance simulator with the most stable convergence for multi-tone analysis, and still the leading approach in the industry. This is important for mixers (3-tones) and n-stage designs where more harmonics are generated.
Complete nonlinear microwave nonlinear models , much improved over the SPICE-type models, and new important models such as GaAs, HBT (even physics based) and others became available.
A new yield analysis capability was added and other improvements in optimization (nonlinear optimization) were made. As an important example, oscillators can be optimized for best phase.
There were a large number of other enhancements made and utility programs added. After the merger with Ansoft, the Harmonica Program became Ansoft Designer."
Ulrich L. Rohde, PhD, ex-President Compact Software, Chairman of Synergy Microwave Company.
Did we mention that Ulrich was on the Ansoft board of directors in 2004? Ulrich had a few other comments but they they sounded too much like free advertising for Ansoft so we cropped them out. One part of the original COMPACT, Smith Tool, is rumored to exist in an Ansoft program even today.
At one time, there was a version of COMPACT that could be run as a module on the HP41 programmable calculator, so you could take your designs to the "reading room". Thanks to Dave, who still has a copy!
We need to add some history here, but for now we'll just copy some stuff we grabbed from the Ansoft web site. Ansoft was founded in Pittsburgh Pennsylvania in 1984 by Dr. Zoltan Cendes. At some point they merged with SuperCOMPACT.
OK, that wasn't much, maybe next time Zol Googles his name he will come across this reference and send us some stuff!
EESof was a well financed enterprise from the beginning, the brainchild of Chuck Abronson and a former Compact employee, Bill Childs. Their first product was called TOUCHSTONE, introduced in 1983. We don't think the name was an acronym. To the user, it seemed like they had ripped off the entire COMPACT netlist interface, but behind that they had created an entirely new engine (another hardware term unfortunately ripped off by geeks). It must have been very efficient, because it ran fast enough on an IBM-AT PC so that it wasn't too annoying (no doubt you had to spring for the math coprocessor on the PC). It also could run on H-P minicomputers. This quickly got a lot of attention from engineers who were sick of mainframes, where they generally had to play second fiddles to accounting and data processing departments.
New for June 2010: Here's the perspective of someone who was there (thanks!)
I can add a little background to the Touchstone story. I worked at Amplica during the period of time when EEsof was formed. Our president Chuck A. had sold the company to Comsat the year before I joined; Comsat had also bought Compact Software the year before that. One of the Comsat engineers that had worked with the Compact part of the business (Bill Childs) was transferred to Amplica to help develop our amplifier designs. It was Bill who authored Touchstone. He would carry a Compaq portable PC back and forth to work every day (it looked like a sewing machine in its case), and thatâ€™s the platform he developed much of the code on. Chuck was the money and business man, he made no technical contributions to the program.
To analyze circuits using TOUCHSTONE, you had to have two display screens for your PC, one for handling the netlist and the other for displaying results. One huge innovation they are credited with is the ability to "tune" the circuit and watch the response move. TOUCHSTONE required a hardware key on the printer port to keep track of the license, making it portable as well. By 1987 EESof's TOUCHSTONE was linked with a crude layout tool (MiCAD), and a version of SPICE. Another cool program they offered was ANACAT, which allowed you to control a Hewlett Packard 8510 network analyzer or the equivalent Wiltron ANA from a PC, and be able to sort data in an acceptable format for reading into TOUCHSTONE. By this time the guys at COMPACT must have soiled themselves with fear for their jobs.
EESof often ran ads that always showed food on top of a computer in a covert but successful effort to promote obesity and chair stains in the industry.
EEsof's TOUCHSTONE product eventually became "Libra" when the harmonic balance analysis was added, which was a truly great piece of software (some MMIC geeks get all misty when you mention it) but a step backward to PC users since it typically ran on a UNIX platform. Most of the physical circuit elements first became perfected on this package. Hewlett Packard had their own CAD software called MDS microwave design microwave (thanks for the correction, John D!). Eventually they bought out EEsof, pink-slipped a bunch of their competitors and over the years abandoned the Libra code (which started as TOUCHSTONE) and came up with an entirely new code stream which is now ADS. ADS is not an improvement in user-friendliness over any of its predecessors, according to a lot of microwave engineers. However, there is hope for this product, we can report that a test copy of the next 2004 release being used by a "Microwaves101 unnamed source" runs much faster, which could indicate that Agilent has finally gotten around to cleaning out all the dead code in the bastard grandson of TOUCHSTONE.
Attention Intel and other politically correct companies... the word "bastard" is not a "swear word", it means illegitimate. Did you know that bastards are a fast growing demographic in the good 'ol diverse USA? Just ask your kids or grandkids how many of their little friends have parents that never married. Maybe someday the IEEE will form an interest group called "Microwave Transactions from Bastards", or MTB for short!
In case you are looking for Mr. Abronson, he's now Chairman of CAP Wireless.
CIRCUIT BUSTERS (what's with all the capital letters that were used in the 1980s?), was founded in 1985, by a filter designer Randy Rhea. While his team was at the task of developing a competing product for linear analysis, they put together some filter synthesis software. The first product was called STAR and you could buy it from a little ad in the back pages of trade mags for $99 (and they took VISA and MasterCard, another brilliant move not anticipated by COMPACT), delivered on a 5.5 inch floppy disc. This was truly an unheard of price, probably less than the square-root of what COMPACT was selling for. We asked Randy about this, and he recalled that he had to rethink the original price, considering the limited market. CIRCUIT-BUSTERS "second generation" product was called SUPER-STAR and was marketed at $595 less than two years later.
Since CIRCUIT BUSTERS was a garage shop, they didn't get around to putting a key on the software until future versions. Either they had some financial help from their in-laws, or they didn't quit their day jobs for a few years. Below is the earliest advertisement we could turn up for CIRCUIT BUSTERS (MW&RF January 1987), featuring an earnest-looking and young Randy Rhea (nice tie!), who personally guarantees his software. The Unknown Editor liked the price, and still has a copy of SUPER STAR 1.0 somewhere under a pile of other priceless artifacts!
What happened to CIRCUIT BUSTERS? They renamed the company Eagleware in 1991, and they are a huge success (ask any of their well-treated employees!) Eagleware is an inspiration to microwave garage operations all over the world, such as a certain microwave web site. In 2005, Eagleware sold out to Agilent. Time will tell if the Eagleware linear simulator survives, it certainly undermines the sales of ADS. As Paul Harvey might say, "now you know the rrreeest of the story!
A relative newcomer to Microwave CAD is Applied Wave Research. This company was founded by a couple of ex-Hughes MMIC designers, well before Raytheon bought Hughes and deep-sixed their Gallium Arsenide Operations foundry. In any case, AWR's Microwave Office is a slick package that combines a linear simulator with E-M analysis. They even allow you to plug in EM software from their competitors, such as Sonnet. Interoperability is the name of the game going forward, these dudes know what they are doing.
Meanwhile during the 1980s, some other folks were working on applying Maxwell's equations to solve arbitrary two and three dimensional geometries. Sonnet was among the first of these software products. James Rautio started Sonnet in his home in 1983, while designing MMICs for General Electric in Syracuse, jumping ship in 1988 and making his first sale the following year. Sonnet is still privately held, unlike Super COMPACT, EEsof, or Eagleware. Learn more about the early days of Sonnet here on their web site.
Here's your opportunity to help us, there are other software products out there that we'd like to know the history of.
In 1992, Zeland Software started offering IE3D, in competition with Ansoft's HFSS.
The first nonlinear simulators (besides SPICE products) also arrived in the late 1980s, and used harmonic balance techniques.
Today, there is huge competition from software vendors for your microwave analysis budget. Enjoy it! We'll add more history of microwave software later, especially if some more of you "founding fathers" contribute!