The Costas Loop - Dr. John P. Costas
The Costas Loop and the Best Engineer.
By Bill Jones.
This paper is to note some of the work of Dr. John P. Costas and to tell of my work and thoughts of my time with John. He was the best engineer that I have ever known. His work was in a relatively narrow field not well known by most engineers. His work was done starting at General Electric in the early 1950’S and continued over his life time. He contributed greatly to several areas of communications, Sonar and mathematics, for example, the Costas Array. He wrote many papers with some being classic. He continued as an inventor even after his retirement.
The term Engineer goes back some years, probably to the 1300’s AD. While I do not mean to compare Dr. Costas to these early genius engineers it should be clear that these early engineers did provide the great foundations we have in our present day curriculums. I am thinking of John as a man that used the foundations to invent an outstanding system - the Costas Loop. If you were to ask the present day communications engineers to rate “their best” I believe that you would find Dr. Costas to be at the top of the list. Of course, in my book John was the best. I knew him well since I worked for and with him for about ten years. I left John in 1962, but I know that he continued to produce important papers and work even after his retirement. I can comment only on the time that I spent with John.
I do confess that this note about John will probably be as much about me -William H. (Bill) Jones as John Costas. I have been thinking about John and his work for some time and wanted to record my thoughts about him and some of the systems that we worked on. While he is famous in his area of engineering it is now difficult to find out much about John as an engineer, or who he really was. While John died in 2008 there are many who assumed him dead long before that.
After I finished Oklahoma University with a degree in electrical engineering and a specialty in electronics I went to work on the General Electric “test program”. The test program required that you spend three months (one test) with various organizations within G.E. to see just how much interest you might have in the groups that you worked for, and also to see the interest they might have in you. You could take four tests. My first two tests were at the G.E. Electronics Park facility. The first test was with a radar group and the second test was with Dr. Costas at the Electronics Laboratory. My first test on the GE program was with a group that had the responsibility to design shipboard radar. It was complicated radar and had an output of 10 megawatts – L band. I was impressed with the size of the project, but it was in its final phase and not of great interest to me.
I asked that my second test be at the G. E. Electronics Laboratory. It was on the cutting edge of electronic systems technology. I wanted to do more advanced work. I was apparently impressed with my engineering expertise. Most of the engineers at the Electronics Laboratory had their Doctoral degrees.
I got my request for a test at the Electronics Lab and went to work for Dr. Costas. John needed help since he was working alone and building an unusual radio. He had no help, and was building what he would often call a “double sideband suppressed carrier receiver” which is precisely what it was. John was fairly new out of MIT where he earned his Doctorate. He had done all the design on this radio, and built a breadboard, but it would not work. We worked together on the receiver, but it had been built in a fairly deep chassis, and was difficult to work on. John made some circuit design changes, and I built the second breadboard which still would not work. John spent a great deal of time with me explaining how the receiver was supposed to work along with the mathematics of the system. He was very kind, and he could always answer a stupid question with a straight face. The receiver was known to us by several names but I will refer to it as a Synchronous Detector, because I believe that was the name we used most often. It was here at the Electronics Laboratory that my eyes were opened and I began to understand that I had a lot to learn, and with John I could learn a lot. I knew that I wanted to work on this receiver with John and he said that he would offer me a job. But I had two more tests to take.
My last two tests were at the G.E. Plant in Schenectady N. Y. I had my third test with the Gas Turbine group actually testing gas turbines that were used to run large alternators for electric utility company use. I had a Professor at Oklahoma University that consulted for G. E. in the summer and he would mention the G.E. test program occasionally during class. Professor Bruce Wiley actually was responsible for arranging an interview for me with the G. E. office in Oklahoma City. During the interview I learned about the Electronics Park facility and decided to join G. E. because of the work that was done there.
For my third test, I decided to ask for a mechanical engineering test –unusual for an electrical engineer. It was Professor Wiley that suggested that all Engineers should take a test in the mechanical engineering area. It was another world, and while it was interesting and I enjoyed it – it was not for me, but it was free, and I got paid to do it. My fourth test was with the well-known G. E. General Engineering Laboratory, and was in the guided missile area, interesting, but again it was not for me.
After I finished my fourth test I wanted to go back to Electronics Park and work for Dr. Costas and John was aware of my desire. I let him know when I about to finish the fourth test. John sent me a letter in his own hand telling me to come for an interview. I still have the letter. I was of course thrilled, and it was clear that John would offer me a position. Arriving at the Electronics Laboratory it was a surprise to me when I learned that I had talk to a number of managers before I would get to talk to John.
The Electronics Laboratory Managers that I talked with seemed to believe that I was mainly interested in working at the Electronics Laboratory. Of course, I was only interested in working for John, at the Laboratory, but it seemed best not express that desire. The first interview was with Dr. L. C. Maier. Dr. Maier was an outstanding manager and a noted engineer. He was quite outgoing with a great personality. He apparently wanted to get some help for John. He told me that I would be asked certain questions by one of the interviewers. Dr. Maier knew the questions and the correct answers; Dr. Maier gave me the answers. It was apparently necessary to have the approval of this particular manager. I think that Maier reported to him, and I believe that John reported to Maier. I was surprised at some of the questions, and the correct answers. It bothered me that I had to go through all these interviews. The answers to some of the questions were not my desired goals. I was only interested in system design, and development. The thought of having a goal of being a manager was frightful. In my later years I did have to be a manager for some time, but I was not excited doing it and always found a way to get back to system design.
I finally got to talk to John. The Laboratory hired few people that only had a Bachelor’s degree, but they offered me a job. I was John’s first employee.
John was still trying to get the Synchronous Detector to work. When I was on my last two tests he was working alone. John was clearly happy to have a helper. He did not spend all of his time on the receiver during the six months that I was in Schenectady because he had other tasks such as getting a contract to build his receiver. He got a contract with the Air Force. He had to write a monthly report for the Air Force, and these monthly reports were well written usually with mathematical considerations of all the various areas of the Synchronous Detector. He spent a lot of time on the monthly reports and I still have a number of them. Electronics Park had a lot of government contracts and I am quite sure that John had help in getting his Air Force contract.
Dr. Costas and I were the only ones that were working on the project. Most of the managers and engineers were working on other G.E. projects, with G. E. money. The Electronics Laboratory had numerous interesting research projects that were looking far into the future. One of their most favorite Laboratory projects was a thin glass television that would hang on the wall and was called ”A Picture on the Wall”, but it did not get very far – this was 1953. We finally have them, but not from G.E.
Dr. Costas apparently only wanted to work on the Synchronous Detector. I assume that John had some agreement with G.E. before he was hired that would allow him to work on his receiver. When I reported to work for John we began work on the last receiver that I had built and spent several months trying to determine the critical areas of the receiver. The oscillator-reactance tube was critical. The oscillator stability was most important but a reactance tube can also provide undesired oscillator drift. We spent some time studying and designing various oscillator circuits. The feed-back signal from the phase detector could cause the reactance tube to modulate the oscillator. The quadrature signals from the oscillator were critical but that was easy to fix. We had great difficulty with the phase detector.
The demodulators were only a minor problem and it was only important to keep the input to them small, because it was necessary to prevent them from becoming non-linear. The demodulators were not a serious problem, but we did spend some time on deciding on the tube that we should use. The output of the demodulators was of course audio. The bandwidth of the receiver was determined by the audio filters used. We would finally use an outside manufacturer for our audio filters, but initially we were making our own filters. John enjoyed filter design and he did teach courses in the Electronics Park for interested engineers. I believe he taught four or five courses, and I did attend them. The courses covered R.F. and audio filter design, and tube circuits. John would hand out problems for the class to do and he graded them. I thought that I had to get every problem correct, and I did. Some of the classes had around forty to fifty engineers attending.
The Costas Loop was totally new. There had been early thoughts of locking an oscillator to the carrier of an amplitude modulated signal, and then using the carrier to demodulate the signal and this would be relatively easy to do. There were some that had investigated an injection locked oscillator. The injection lock is actually a phase lock but mathematically different than a phase lock loop. Locking an oscillator to the incoming carrier has the advantage of using a clean or noise free oscillator for demodulation. This provides an improvement in signal to noise ratios and frees the carrier from notch fading.
The Synchronous receiver did away with the carrier and allowed the carrier power to be put into the sidebands. John noticed in his Doctoral Thesis (Interference Filtering) that it was possible to obtain the carrier frequency and phase information from the sidebands. The Synchronous Detector was actually described in his thesis. He noted that if you did two quadrature demodulations of the amplitude modulated signal (with or without the carrier) and compared their outputs in a phase detector the detector would provide the proper information to lock the receiver oscillator to what would have been the carrier. It worked fine on paper, but we could not get the actual system to work.
I suggested that we separate the various parts of the receiver, and use a chassis for each part. We would then hook the chassis’ together with coax cables, and that way when we were satisfied with one circuit we could start on another circuit. This then gave us three chassis. The oscillator-reactance tube circuit with its quadrature outputs was one chassis. The two demodulator circuits with their filter and audio amplifiers was a single chassis and the phase detector was the final chassis. The demodulator chassis was no problem. The oscillator-reactance tube was slightly more difficult because of oscillator modulation from the reactance tube, and of course the stability of both the oscillator and the reactance tube circuit was a problem.
We spent some time on the oscillator circuit. John did a complete analysis on each circuit we considered. The Colpitts may have been what he considered the best oscillator. The most difficult circuit was the phase detector. John had originally treated the phase detector as a simple nonlinear device, and as such the input audio signals from each channel appeared at the output of the phase detector along with the phase information. Separating the audio and phase information was done using a low pass filter. It appeared to me that we got too much audio information with the phase information, speaking of course as a matter of signal to noise. It required a heavy low pass filtering to remove the audio from the phase information.
I designed a single balanced phase detector using the in-phase audio for the balanced input and this gave us a much nicer phase detector output. The system would still not lock. It did however make it clear that we needed a double balanced phase detector and another new chassis build. The double balanced phase detector gave a nice looking phase signal and required much less filtering on the phase detector output. This of course gave us a wider bandwidth for the phase lock loop, but again the system would not work. John pretty much left me alone with the phase detector work – I believe that he was busy writing his usual monthly report. Some of his reports were quite long, and could have been papers for publishing had it not been for security concerns. His Air Force contract was classified.
We had some very nice test equipment – a great Marconi signal generator, and the newest Tektronix scopes. The scope had an X and Y input and I used these inputs for the in-phase and quadrature audio signals. This gave a scope presentation of a rotating straight line. As you adjusted the frequency of the incoming signal it was possible to slow the rotation of the straight line and almost make the line stop if the stability of the oscillators were good. The system would still not lock.
For some reason I put a potentiometer on the phase detector output and used the center post as the input to the reactance tube. This was actually a system gain control. I tried locking the system with the control at various positions starting with full gain and gradually turning down the gain. I got the gain down close to zero and the system locked! The Q audio was zero and the In-phase audio at maximum. A straight line with no rotation, but because of oscillator stability it would not last very long with such a low loop gain.
The loop gain was the problem. We were treating the system as a first order servo system which is a system with only a six dB per octave drop in the loop gain. We were assuming that the system was like an automatic gain control on ordinary radios. I was quite excited to see the system lock, and ran in to get John!
I showed John what I was doing to get the system to lock and showed him that the system would lock if the loop gain was very low. John was thrilled and he knew what the problem was of course; an unstable servo. He did not immediately understand why the servo was unstable, but I think that he had strong suspicions of what might be wrong. John had a friend in TV engineering that had done his thesis on phase lock loops. He called his friend and we went over to the TV department to talk with him, and read his thesis. We came back to the Laboratory and John sat down with me, and took the initial assumptions of any phase lock loop, and derived the conditions for lock. This took him an hour or so because he explained to me what he was doing. It turns out that an oscillator that has a frequency control is actually an integrator, and thus has a six dB per octave gain drop. The concept of an oscillator with reactance control having such a response seemed to cause John a problem, and he discussed the integrator for a day or so before he became comfortable with the concept. It took me a lot longer to become comfortable with the concept. It is quite simple mathematically, and obvious.
It is a bit more difficult to explain it physically. Of course, it only means that when you to apply a sinusoidal input to a frequency controlled oscillator that the total frequency (phase) deviation will become larger and larger as you lower the frequency. With a low frequency it has more time to change. With a high frequency it has less time to change. As you increase the frequency the total phase deviation drops, and it drops six dB per octave.
The way we were treating the loop gave us a twelve dB per octave gain drop, and this would guarantee an unstable loop. The only way it would lock was near unity gain. John had calculated the loop gain and was set such that the phase error would be only a few degrees. John was somewhat familiar with servos, and used the Bode means of loop stability. It was only necessary to put in a lead-lag network to obtain a stable loop. We stayed with the double balanced phase detector which allowed a greater loop bandwidth.
I was surprised when it occurred to me that I was the first person to see a Synchronous Receiver system lock. I did not realize it until the writing of this paper. While Dr. Costas invented the system he was the second person to see it lock. I do not believe that it ever occurred to either of us. It was only years after my retirement that I realized that I was the first person to see such a system work.
We normally think that an unstable system will oscillate in some sinusoidal or nonlinear repetitive manner but it was not the case with the Synchronous Detector. As I recall the Synchronous Detector unstable output appeared quite random. We might have obtained some hint of the stability problem had we actually listened to the I or Q output. Strange, but I do not recall that we ever listened to the audio signals. I would suggest that the system has an unusual instability and exhibits a chaotic audio waveform. As the system begins to become unstable the I signal stays close to the maximum with some noise modulation and the Q signal begins to increase in amplitude. I would guess that the system became unusable when the Q signal became about one-half of the I signal. We never investigated the type of waveforms obtained with the unstable Synchronous Detector. It would be an interesting study.
Getting the system to work was most important, and allowed John to hire engineers so that we could build a usable Synchronous receiver for the Air Force. John wanted a good looking and a workable receiver for the final product. The final receiver was all of that, but it was over designed with too many tubes, too many adjustments and too much wasted space. We probably had at least twenty tubes and a separate power supply. I would love to see a circuit diagram of the Synchronous Detector that we designed. I usually kept diagrams of systems that I designed, but this was classified because of the Air Force and I never kept any of that information.
For the final design I was given the RF section with the oscillator/reactance tube, the demodulators, and the phase detector. My first attempt at a band switched RF amplifier was a disaster. I quickly learned that it was necessary to short all the unused coils because unshorted coils will resonate with a coil in use and cause a drastic drop in gain. A picture of all the chassis’ is shown at the end of this paper.
Our frequency range was 2 to 32 megahertz. My second attempt for the R.F. amplifier was successful. We had a design goal for our total gain to be one million with a noise figure of three dB which was considered optimum at the time. The R.F. amplifier had three stages. There was a careful physical separation between the R.F. amplifier variable capacito, and the oscillator variable capacitor, because the oscillator was at the same frequency as the incoming signal. As I recall I used an insulating universal joint between the two variable capacitors. It also required some good shielding. I designed the sine wave oscillators for the receiver, and the receiver required three oscillators. The oscillator had an unusual automatic gain control that kept the oscillator output both sinusoidal and constant to within a dB or so over the entire frequency range of the Synchronous Detector. John insisted on a sine wave oscillator so that there would be no harmonic problems. He also insisted on constant amplitude so that the demodulator gain would be constant. I should have put in a patent docket on the circuit, but I was too busy designing, and then winding the coils for the receiver. The total number of coils was twenty four. As I recall we built three receivers in the Laboratory. Another group built another ten or so and it seems that some other military service wanted a few.
John needed a frequency synthesizer for the receiver and he spent a great amount of time trying to come up with a synthesizer system. It was paramount to have a stable oscillator. He considered a number of systems. He hired Roger Swanson to build the system that he and Roger agreed on. It was a clever system that used a one hundred kilohertz oven controlled crystal oscillator that would with circuitry produce an sharp voltage impulse. The impulse gave a harmonic output every one hundred kilohertz beyond thirty two megahertz. It was then necessary to lock an oscillator to each one hundred kilohertz harmonic from 2 to 32 megahertz. It was then necessary to devise a means to scan one hundred kilohertz between each harmonic. A second oscillator was then offset by a low frequency oscillator that covered a one hundred kilohertz range and the sum of the two locked to the first oscillator. The actual tuning of the receiver was controlled by the low frequency oscillator and the tuning range of the receiver was then one hundred kilohertz. It actually required two adjustments to tune the receiver. You had to lock the first oscillator to one of the impulse harmonics, and then you could cover one hundred kilohertz above that frequency. The dial spread of the low frequency oscillator was about two hundred and seventy degrees with the numbers starting at zero, and going to 100 Kilohertz. This gave a bandspread of about ten inches for 100 kilohertz.
The low frequency oscillator generated a lot of discussion trying to decide on the circuit that should be used for best stability and I believe that John finally decided on a bridge oscillator, and the frequency was controlled by a potentiometer. I do not remember the frequency of the oscillator but it was probably around two hundred kilohertz. It was of course sinusoidal.
John hired Jack Cuniff to build the power supply and audio circuits for the receiver, and I believe Jack did the low frequency oscillator and the alpha –beta networks, but I believe that Roger did help him. The networks were not well known and it was John that knew of them. The alpha-beta networks allowed the Synchronous receiver to function as a single side band receiver. Jack was fond of power supplies, and audio circuits. We did have an outstanding product designer that was loaned to us from the TV department. He did the front panel of the receiver and the necessary gearing behind the front panel. There was a large amount of design here. He did a beautiful job, and he was great to work with, but I do not remember his name. He did everything on the front panel including the beautiful aluminum knobs.
The receiver was a bit difficult to operate at first, but if you knew the frequency that was desired it was easy. It was difficult to search across the dial because you had to first choose one of the harmonics and then you could only tune one hundred kilohertz above that harmonic. This was a great band- spread feature. It was perhaps the first receiver to use a frequency synthesizer.
While John’s Synchronous Detector scheme was to eventually contribute to data rates only dreamed of by him; it was years before it happened. The system was poorly understood by others and at the time the receiver market belonged to the Single-Sideband system. The Air Force was sold on Single Sideband. The Collins Company built a fine Single Sideband receiver and it is still admired and still used by hams today. The Air Force loved the receiver and would not give it up. It was a good system and worked well, but it was not optimum as John continued to point out in his papers.
The Synchronous Receiver has many advantages over single sideband, but John was never able to convince the Air Force of these advantages regardless of what he did. We performed an over the air test using the Synchronous Detector verses the Single Sideband receiver from the Rome Air Force base to Bolder Colorado using a word recognition test. The Synchronous Receiver beat the Sideband system badly. We taped the results and had a group of people listen and write down the words that they thought they heard. Roger Swanson took the results to Washington, and the results were not well received. Roger felt that they had accused us of cheating. We used differentiation and clipping in the transmitter exciter, and this was a means of increasing our average power. and improved word recognition. One of the problems with sideband is that you cannot use fast rise time signals as a transmitter input, because it will cause large transmitter output peaks that the transmitter cannot handle. This is because of the missing sideband. We took advantage of that. Single sideband will also not provide a good data system because of this problem. The Synchronous receiver that we designed had Single Sideband capability. This capability is easy to implement into the receiver.
G.E. did not pursue the business. Our Synchronous Receiver was big and heavy and would have required at least a couple of design iterations to condense it into a really marketable receiver.
When we finished the Synchronous Detector Dr. Costas had no idea that his contribution would allow communication systems to approach their optimum capabilities – no one did. It was years before his system was recognized for its capabilities and really put to work.
Sometime after we had finished the receiver and completed the project John hired an excellent engineer who was from Bell Labs, and had several years of experience. We were all impressed with Joe Reddeck because of his experience at Bell Labs. Joe took me under his wing and taught me to use some design tools that I had not been aware of. We worked together on several projects, and it was a joy to work with Joe. I believe that Dr. Costas was busy with other projects, and perhaps writing papers, I have always suspected that John expected Joe to be my mentor. We worked well together and Joe seemed to enjoy the roll as much as I did. He was an excellent circuit designer.
Dr. Costas was approached by C.R. Wayne who had an Advanced Engineering group in the G.E. Heavy Military Department. Wayne got John to move our group from the Electronics Laboratory to Heavy Military. I believe that we had six engineers not counting John. After we moved to Heavy Military John hired two or three engineers. Wayne had a large group of engineers working in diverse advanced Engineering areas. Wayne had acquired a fine organization. I have no idea how he got John to move. Wayne was a great manager, outgoing, and quick to smile, I liked him a lot.
After we moved to Wayne’s group John decided to build a Spread-Spectrum system. He was always concerned with jamming of communication systems. John conceived a wide band data system that was jam proof. I did not work on the system, but followed it closely. He finished the system, and actually put it on the air in the 20 meter ham-band. As I recall it covered the band completely and could not be detected by a normal receiver because the power per cycle was so small. John named the system “Phantom” and I do not recall him writing much about the system. The system used a tapped delay line with the data being sent depending on how the taps were set. He demonstrated the system, and then seemed to abandon it. I assume that it was covered by G.E. money rather than military contract.
While John was working on Phantom, I started working with an excellent engineer whose name was Jim Kovaric. I was helping Jim get an unusual system to work. The system was called a “Side Lobe Canceller”. The system was to be used in the Berlin supply crisis. Our Berlin supply planes often had difficulty landing because the Berlin radars were being jammed by the Russians. The Side Lobe Canceller was to help circumvent the jamming. I believe that Wayne also had others working on the problem.
Because of my work with Jim I believe that I was given the job of building an “Over the Horizon Radar System”, but I still reported to John. The system was called a Velocity Indicating Coherent Integrator. The system used a long delay line, and while I do not remember the delay time it consisted of a large six foot high cabinet with three hundred feet of wire inside. It was built by the Electronics Laboratory by a group headed by Steve Tehon. The delay line was delivered to me just as I finished the electronics for the system. The delay line had a sharp drop in output at an important frequency, and I had to build a chassis that had a matching sharp peak at this same frequency. This then made the delay line frequency response flat and this was imperative for the system. Setting beside the delay line was a full rack of electronics that made up the Velocity Indicating Coherent Integrator electronics. The system was to measure the movement of the ionosphere as a missile (Russian) passed through the ionosphere. We tested the system at the Rome Air Force base and the backscatter did show the movement of the ionosphere. I was not given the results of any missile tests or other effects. The system was turned over to the Air Force and I heard no more of the system. My work with the system was in the late nineteen fifties as I recall. I think that the Syracuse Electronics Park was still in the “Over the Horizon” business at least until the nineteen sixties.
John was given a Navy contract to build a fifty megahertz version of the Synchronous Detector for a data link between Boston and Cape Cod. The Navy was interested in the error rate at this particular frequency and we were to use a one kilowatt transmitter. The Navy required that the receiver antenna had to be floating on the water (submarine). There were several groups involved. One group built the data generator and error checking equipment and another built the transmitter, and got the sites for the transmitter and receiver. John gave me the receiver to build and the time frame was around three months. The tests lasted for several months and I got to visit both sites. We were happy with the results, but the Navy of course did not give us any information as to how they felt about the system.
I put a one inch oscilloscope on the front panel of the receiver to show the I and Q channels. This was nice because you could see the noise from the system because the line on the oscilloscope is very thin when the noise is low and will widen when the noise gets large. This led me to an automatic Signal plus Noise to Noise ratio indicator on which I received a patent. Johns Synchronous Detector actually led me to a number of patents, and his system suggested applications for patents in other fields. My first patent was with John as W.H. Jones ET AL.
G. E. decided to open a Military Communications Department in Oklahoma City. A meeting was called, and our group was called to attend. We were told that we had to go to Oklahoma City, and had no other choice. I was delighted, but the rest of John’s group was not happy with the announcement.
John called a meeting of our group plus a few others as I recall, and the meeting was held at his home. He wanted to discuss the possibility of starting a company. John had already picked the name for the company as Cogent Systems, and he called the meeting to discuss the various markets and the financing for the company. While I understood that he was interested in starting a company I assumed that the meeting was only a first casual discussion of the idea. John put together a talk regarding his ideas, and lasted perhaps less than thirty minutes.
As soon as he finished talking he said that he was going to poll the group as to whether each person would join the company or not. I was surprised because I felt that the invitation was rather sudden, and I did not have an answer. To my horror, I was the first to be asked - John said- “Bill, will you join?”. Going into business with Dr. Costas was a dream come true. I knew however, that I wanted to be in Oklahoma City and the Military Electronics Department was a great match for me, and also John. It had not occurred to me that John had decided that he would definitely not go to the Military Communications Department, but he had. I had a huge dislike for the weather in Syracuse and wanted to go back “home” in the worst way – as did my wife – she was at the meeting along with the other wives. To my dismay I had to come up with an answer. John knew of my admiration of him as an engineer, and as a person – even a friend. He certainly expected that I would be a safe “Yes”. I had to say something and it was a simple “no”. It surely required a long pause for the “no” to come out because of the many things that went through my mind. I have always thought that John asked me first because he expected a resounding “Yes!”. I also felt at the time that he would come to his senses and actually go to Oklahoma City. It was a perfect match for John but he was adamant and would have no part in the transfer.
An excellent manager by the name of Ben Walker was picked to manage the department. I have always wondered what would have happened if Dr. Costas had been asked to be the department manager. I think that it would have been difficult for him to turn down. (It is possible that John knew of the new department, and he may have been asked to manage it. It was a large department with a number of advanced projects, and two large buildings with a large factory area.
John wrote a letter for me to Ben Walker telling him that I wanted to go to Oklahoma and asked Ben to find a position for me in the Department. I wound up working for Dick Ellis in Oklahoma, and enjoyed working for him. Dick was a great manager. Dick and I got along very well. I was the first of his group to move to Oklahoma and Dick arrived a few months later. He found a house across the street from our house. We were both unaware of this for some time.
I was of help to Dr. Costas because I had been interested in radio since I was a child. I was shown how to build a crystal receiver when I was ten years old. I even learned how to make my own crystals. I built one tube receivers and would often visit furniture stores looking for old used radios that would not work and buy them at very reasonable prices. I learned how to solder and how to take radios apart. I learned about all the various parts of the radio. I could only repair a few of the radios that I purchased.
I was extremely fortunate to get a job at Al Robertson’s radio shop while attending high school. Here, I really learned to repair radios. Al had an expert radio repairman that would help me with those radios that I could not repair. Ray Binghamton could repair any radio that was brought into the shop. He seemed pleased in showing me what was wrong with any radio that I was having trouble with. After about one year I could repair any radio that came into our shop. This background helped me greatly during the G.E. test program, and it was most useful in working with John on the Synchronous Detector My work with John was an important area of my life and it was an important area for radio and data communication because of Dr. Costas. I consider myself to be an excellent engineer because John was my mentor. John was the best Engineer that I ever met. John could do it all. He could design circuits and I would consider him a mathematician as well as an engineer. Some years after the Synchronous receiver were finished John retired from G.E. and moved to Boston where he continued to make important contributions in other areas.
John was a family man, and had a beautiful wife –Helen. My wife and I would go to their house for an occasional dinner and they came to our house for dinner. They lived in an upscale neighborhood while we lived in a downscale neighborhood. Helen was from Boston and when John retired from G.E. they moved to Boston. John did some consulting work in Boston for a company by the name of Cogent Systems. Strangely enough, Cogent was the name for the company that John had wanted to start when he was working for General Electric.
I really missed John after the Oklahoma move because it was difficult for me to find the optimum capabilities for any unusual system. I always missed John’s mathematical skills and John as a mentor. He was an outstanding teacher.
John was always interested in the optimum that was possible for any system. He had worked under or for some of the best known names at MIT. These Professors were an important part of John’s career. He spoke of them often. Shannon, Weiner and Lee were part of his vocabulary. Dr. Lee was his thesis advisor. Generally, they were ones that had mathematical theories for various systems with the optimum values that could be achieved for such systems. That is to say, the best performance that could be obtained with that particular system, but the way to reach such performance was usually unknown.
In the early fifties the data rates possible for communication systems were far from the theoretical maximum reachable and John was always curious as to why this was so. He had a wide ranging interest in many areas and of course did not have the time to contribute to all the areas of interest. He wrote many papers for the Institute of Electrical and Electronic Engineers. John wrote a paper concerning Synchronous Detection for the I.E.E.E. in 1956 and it was such a classic paper the I.E.E.E. reprinted it years later. He was a prolific writer and had a wide ranging interest in many areas. John was an active Ham operator (K2EN) and built his own kilowatt transmitter. This was typical at the time. It was well built in a nice six foot rack and he had a rotating beam to go with it. He had a top of the line AM receiver.
After moving to Oklahoma and separated from John I worked for a number of companies. I worked for the G.E. Military Communication’s department but it stayed in Oklahoma City for only a few years and then went back to Syracuse. It was shortly thereafter disbanded. I was able to stay in Oklahoma because the G.E. Computer Department moved into the City just when the Military Communications moved out.
After G.E. I worked for Honeywell, and Magnetic Peripherals. The work was the same for the three companies because G.E. sold our group to Honeywell, and Honeywell sold us to Magnetic Peripherals. I quit Magnetic Peripherals, and started to work for Silicon Valley start-ups. Worked in California for ten years.
We were working on a high density disk drive at Konica Technology- a Silicon Valley start-up. I was the Read/Write engineer. While at Honeywell I became what was called a Read/Write engineer. After Honeywell, I was a Read/Write engineer for each company were I worked. As it turns out being a Read/Write engineer is quite the same as being a Communications engineer. The Read/Write engineer needs to know phase lock loops and the Synchronous receiver was one big phase lock loop. I had the perfect background to design read/write circuits for disk drives.
One day while at Konica, a group of us were discussing a problem, and someone brought Dr. Costas’s name into the conversation. The Japanese engineers were well acquainted with the Costas Loop, and Johns work. I said –“ I understand the problem because I worked for Dr. Costas for years”. One of the Japanese engineers said “no – Dr. Costas is dead ”. I assured the gentleman that John was not dead. It seemed clear to him that I was incorrect and that I was probably thinking about some other Costas rather than Dr. John P. Costas. It so happened that shortly thereafter I had a business trip to Boston, and I called John and we had dinner. Over dinner John discussed some of his areas of business, and in the discussion he said “you know Bill – there are a lot of people that think that I am dead.” I recalled the Japanese gentleman, (I do not remember his name). I told John about the discussion with the Japanese. John was amused. I asked John to autograph his business card (Cogent Systems) so that I could carry it back to the Japanese engineer who was certain of Dr. Costas’s demise. John addressed the card to the Japanese gentleman--by his name. I also asked John to autograph his card for me--I still have it.
Had John not been such an accomplished engineer and mathematician he would still would have been highly regarded for his intelligence and integrity. I always thought that his IQ was in the genius category. He was a great theoretician, and loved to teach. He was a Mathematician. He could analyze systems, optimize systems, design circuits, and optimize circuits. He knew well how to use a soldering iron. Dr. John Costas was an outstanding Engineer.
Dr. Costas was a down to earth gentleman in the best sense of the word, and the best Engineer that I ever knew.
By Bill Jones.
This paper is to note some of the work of Dr. John P. Costas and to tell of my work and thoughts of my time with John. He was the best engineer that I have ever known. His work was in a relatively narrow field not well known by most engineers. His work was done starting at General Electric in the early 1950’S and continued over his life time. He contributed greatly to several areas of communications, Sonar and mathematics, for example, the Costas Array. He wrote many papers with some being classic. He continued as an inventor even after his retirement.
The term Engineer goes back some years, probably to the 1300’s AD. While I do not mean to compare Dr. Costas to these early genius engineers it should be clear that these early engineers did provide the great foundations we have in our present day curriculums. I am thinking of John as a man that used the foundations to invent an outstanding system - the Costas Loop. If you were to ask the present day communications engineers to rate “their best” I believe that you would find Dr. Costas to be at the top of the list. Of course, in my book John was the best. I knew him well since I worked for and with him for about ten years. I left John in 1962, but I know that he continued to produce important papers and work even after his retirement. I can comment only on the time that I spent with John.
I do confess that this note about John will probably be as much about me -William H. (Bill) Jones as John Costas. I have been thinking about John and his work for some time and wanted to record my thoughts about him and some of the systems that we worked on. While he is famous in his area of engineering it is now difficult to find out much about John as an engineer, or who he really was. While John died in 2008 there are many who assumed him dead long before that.
After I finished Oklahoma University with a degree in electrical engineering and a specialty in electronics I went to work on the General Electric “test program”. The test program required that you spend three months (one test) with various organizations within G.E. to see just how much interest you might have in the groups that you worked for, and also to see the interest they might have in you. You could take four tests. My first two tests were at the G.E. Electronics Park facility. The first test was with a radar group and the second test was with Dr. Costas at the Electronics Laboratory. My first test on the GE program was with a group that had the responsibility to design shipboard radar. It was complicated radar and had an output of 10 megawatts – L band. I was impressed with the size of the project, but it was in its final phase and not of great interest to me.
I asked that my second test be at the G. E. Electronics Laboratory. It was on the cutting edge of electronic systems technology. I wanted to do more advanced work. I was apparently impressed with my engineering expertise. Most of the engineers at the Electronics Laboratory had their Doctoral degrees.
I got my request for a test at the Electronics Lab and went to work for Dr. Costas. John needed help since he was working alone and building an unusual radio. He had no help, and was building what he would often call a “double sideband suppressed carrier receiver” which is precisely what it was. John was fairly new out of MIT where he earned his Doctorate. He had done all the design on this radio, and built a breadboard, but it would not work. We worked together on the receiver, but it had been built in a fairly deep chassis, and was difficult to work on. John made some circuit design changes, and I built the second breadboard which still would not work. John spent a great deal of time with me explaining how the receiver was supposed to work along with the mathematics of the system. He was very kind, and he could always answer a stupid question with a straight face. The receiver was known to us by several names but I will refer to it as a Synchronous Detector, because I believe that was the name we used most often. It was here at the Electronics Laboratory that my eyes were opened and I began to understand that I had a lot to learn, and with John I could learn a lot. I knew that I wanted to work on this receiver with John and he said that he would offer me a job. But I had two more tests to take.
My last two tests were at the G.E. Plant in Schenectady N. Y. I had my third test with the Gas Turbine group actually testing gas turbines that were used to run large alternators for electric utility company use. I had a Professor at Oklahoma University that consulted for G. E. in the summer and he would mention the G.E. test program occasionally during class. Professor Bruce Wiley actually was responsible for arranging an interview for me with the G. E. office in Oklahoma City. During the interview I learned about the Electronics Park facility and decided to join G. E. because of the work that was done there.
For my third test, I decided to ask for a mechanical engineering test –unusual for an electrical engineer. It was Professor Wiley that suggested that all Engineers should take a test in the mechanical engineering area. It was another world, and while it was interesting and I enjoyed it – it was not for me, but it was free, and I got paid to do it. My fourth test was with the well-known G. E. General Engineering Laboratory, and was in the guided missile area, interesting, but again it was not for me.
After I finished my fourth test I wanted to go back to Electronics Park and work for Dr. Costas and John was aware of my desire. I let him know when I about to finish the fourth test. John sent me a letter in his own hand telling me to come for an interview. I still have the letter. I was of course thrilled, and it was clear that John would offer me a position. Arriving at the Electronics Laboratory it was a surprise to me when I learned that I had talk to a number of managers before I would get to talk to John.
The Electronics Laboratory Managers that I talked with seemed to believe that I was mainly interested in working at the Electronics Laboratory. Of course, I was only interested in working for John, at the Laboratory, but it seemed best not express that desire. The first interview was with Dr. L. C. Maier. Dr. Maier was an outstanding manager and a noted engineer. He was quite outgoing with a great personality. He apparently wanted to get some help for John. He told me that I would be asked certain questions by one of the interviewers. Dr. Maier knew the questions and the correct answers; Dr. Maier gave me the answers. It was apparently necessary to have the approval of this particular manager. I think that Maier reported to him, and I believe that John reported to Maier. I was surprised at some of the questions, and the correct answers. It bothered me that I had to go through all these interviews. The answers to some of the questions were not my desired goals. I was only interested in system design, and development. The thought of having a goal of being a manager was frightful. In my later years I did have to be a manager for some time, but I was not excited doing it and always found a way to get back to system design.
I finally got to talk to John. The Laboratory hired few people that only had a Bachelor’s degree, but they offered me a job. I was John’s first employee.
John was still trying to get the Synchronous Detector to work. When I was on my last two tests he was working alone. John was clearly happy to have a helper. He did not spend all of his time on the receiver during the six months that I was in Schenectady because he had other tasks such as getting a contract to build his receiver. He got a contract with the Air Force. He had to write a monthly report for the Air Force, and these monthly reports were well written usually with mathematical considerations of all the various areas of the Synchronous Detector. He spent a lot of time on the monthly reports and I still have a number of them. Electronics Park had a lot of government contracts and I am quite sure that John had help in getting his Air Force contract.
Dr. Costas and I were the only ones that were working on the project. Most of the managers and engineers were working on other G.E. projects, with G. E. money. The Electronics Laboratory had numerous interesting research projects that were looking far into the future. One of their most favorite Laboratory projects was a thin glass television that would hang on the wall and was called ”A Picture on the Wall”, but it did not get very far – this was 1953. We finally have them, but not from G.E.
Dr. Costas apparently only wanted to work on the Synchronous Detector. I assume that John had some agreement with G.E. before he was hired that would allow him to work on his receiver. When I reported to work for John we began work on the last receiver that I had built and spent several months trying to determine the critical areas of the receiver. The oscillator-reactance tube was critical. The oscillator stability was most important but a reactance tube can also provide undesired oscillator drift. We spent some time studying and designing various oscillator circuits. The feed-back signal from the phase detector could cause the reactance tube to modulate the oscillator. The quadrature signals from the oscillator were critical but that was easy to fix. We had great difficulty with the phase detector.
The demodulators were only a minor problem and it was only important to keep the input to them small, because it was necessary to prevent them from becoming non-linear. The demodulators were not a serious problem, but we did spend some time on deciding on the tube that we should use. The output of the demodulators was of course audio. The bandwidth of the receiver was determined by the audio filters used. We would finally use an outside manufacturer for our audio filters, but initially we were making our own filters. John enjoyed filter design and he did teach courses in the Electronics Park for interested engineers. I believe he taught four or five courses, and I did attend them. The courses covered R.F. and audio filter design, and tube circuits. John would hand out problems for the class to do and he graded them. I thought that I had to get every problem correct, and I did. Some of the classes had around forty to fifty engineers attending.
The Costas Loop was totally new. There had been early thoughts of locking an oscillator to the carrier of an amplitude modulated signal, and then using the carrier to demodulate the signal and this would be relatively easy to do. There were some that had investigated an injection locked oscillator. The injection lock is actually a phase lock but mathematically different than a phase lock loop. Locking an oscillator to the incoming carrier has the advantage of using a clean or noise free oscillator for demodulation. This provides an improvement in signal to noise ratios and frees the carrier from notch fading.
The Synchronous receiver did away with the carrier and allowed the carrier power to be put into the sidebands. John noticed in his Doctoral Thesis (Interference Filtering) that it was possible to obtain the carrier frequency and phase information from the sidebands. The Synchronous Detector was actually described in his thesis. He noted that if you did two quadrature demodulations of the amplitude modulated signal (with or without the carrier) and compared their outputs in a phase detector the detector would provide the proper information to lock the receiver oscillator to what would have been the carrier. It worked fine on paper, but we could not get the actual system to work.
I suggested that we separate the various parts of the receiver, and use a chassis for each part. We would then hook the chassis’ together with coax cables, and that way when we were satisfied with one circuit we could start on another circuit. This then gave us three chassis. The oscillator-reactance tube circuit with its quadrature outputs was one chassis. The two demodulator circuits with their filter and audio amplifiers was a single chassis and the phase detector was the final chassis. The demodulator chassis was no problem. The oscillator-reactance tube was slightly more difficult because of oscillator modulation from the reactance tube, and of course the stability of both the oscillator and the reactance tube circuit was a problem.
We spent some time on the oscillator circuit. John did a complete analysis on each circuit we considered. The Colpitts may have been what he considered the best oscillator. The most difficult circuit was the phase detector. John had originally treated the phase detector as a simple nonlinear device, and as such the input audio signals from each channel appeared at the output of the phase detector along with the phase information. Separating the audio and phase information was done using a low pass filter. It appeared to me that we got too much audio information with the phase information, speaking of course as a matter of signal to noise. It required a heavy low pass filtering to remove the audio from the phase information.
I designed a single balanced phase detector using the in-phase audio for the balanced input and this gave us a much nicer phase detector output. The system would still not lock. It did however make it clear that we needed a double balanced phase detector and another new chassis build. The double balanced phase detector gave a nice looking phase signal and required much less filtering on the phase detector output. This of course gave us a wider bandwidth for the phase lock loop, but again the system would not work. John pretty much left me alone with the phase detector work – I believe that he was busy writing his usual monthly report. Some of his reports were quite long, and could have been papers for publishing had it not been for security concerns. His Air Force contract was classified.
We had some very nice test equipment – a great Marconi signal generator, and the newest Tektronix scopes. The scope had an X and Y input and I used these inputs for the in-phase and quadrature audio signals. This gave a scope presentation of a rotating straight line. As you adjusted the frequency of the incoming signal it was possible to slow the rotation of the straight line and almost make the line stop if the stability of the oscillators were good. The system would still not lock.
For some reason I put a potentiometer on the phase detector output and used the center post as the input to the reactance tube. This was actually a system gain control. I tried locking the system with the control at various positions starting with full gain and gradually turning down the gain. I got the gain down close to zero and the system locked! The Q audio was zero and the In-phase audio at maximum. A straight line with no rotation, but because of oscillator stability it would not last very long with such a low loop gain.
The loop gain was the problem. We were treating the system as a first order servo system which is a system with only a six dB per octave drop in the loop gain. We were assuming that the system was like an automatic gain control on ordinary radios. I was quite excited to see the system lock, and ran in to get John!
I showed John what I was doing to get the system to lock and showed him that the system would lock if the loop gain was very low. John was thrilled and he knew what the problem was of course; an unstable servo. He did not immediately understand why the servo was unstable, but I think that he had strong suspicions of what might be wrong. John had a friend in TV engineering that had done his thesis on phase lock loops. He called his friend and we went over to the TV department to talk with him, and read his thesis. We came back to the Laboratory and John sat down with me, and took the initial assumptions of any phase lock loop, and derived the conditions for lock. This took him an hour or so because he explained to me what he was doing. It turns out that an oscillator that has a frequency control is actually an integrator, and thus has a six dB per octave gain drop. The concept of an oscillator with reactance control having such a response seemed to cause John a problem, and he discussed the integrator for a day or so before he became comfortable with the concept. It took me a lot longer to become comfortable with the concept. It is quite simple mathematically, and obvious.
It is a bit more difficult to explain it physically. Of course, it only means that when you to apply a sinusoidal input to a frequency controlled oscillator that the total frequency (phase) deviation will become larger and larger as you lower the frequency. With a low frequency it has more time to change. With a high frequency it has less time to change. As you increase the frequency the total phase deviation drops, and it drops six dB per octave.
The way we were treating the loop gave us a twelve dB per octave gain drop, and this would guarantee an unstable loop. The only way it would lock was near unity gain. John had calculated the loop gain and was set such that the phase error would be only a few degrees. John was somewhat familiar with servos, and used the Bode means of loop stability. It was only necessary to put in a lead-lag network to obtain a stable loop. We stayed with the double balanced phase detector which allowed a greater loop bandwidth.
I was surprised when it occurred to me that I was the first person to see a Synchronous Receiver system lock. I did not realize it until the writing of this paper. While Dr. Costas invented the system he was the second person to see it lock. I do not believe that it ever occurred to either of us. It was only years after my retirement that I realized that I was the first person to see such a system work.
We normally think that an unstable system will oscillate in some sinusoidal or nonlinear repetitive manner but it was not the case with the Synchronous Detector. As I recall the Synchronous Detector unstable output appeared quite random. We might have obtained some hint of the stability problem had we actually listened to the I or Q output. Strange, but I do not recall that we ever listened to the audio signals. I would suggest that the system has an unusual instability and exhibits a chaotic audio waveform. As the system begins to become unstable the I signal stays close to the maximum with some noise modulation and the Q signal begins to increase in amplitude. I would guess that the system became unusable when the Q signal became about one-half of the I signal. We never investigated the type of waveforms obtained with the unstable Synchronous Detector. It would be an interesting study.
Getting the system to work was most important, and allowed John to hire engineers so that we could build a usable Synchronous receiver for the Air Force. John wanted a good looking and a workable receiver for the final product. The final receiver was all of that, but it was over designed with too many tubes, too many adjustments and too much wasted space. We probably had at least twenty tubes and a separate power supply. I would love to see a circuit diagram of the Synchronous Detector that we designed. I usually kept diagrams of systems that I designed, but this was classified because of the Air Force and I never kept any of that information.
For the final design I was given the RF section with the oscillator/reactance tube, the demodulators, and the phase detector. My first attempt at a band switched RF amplifier was a disaster. I quickly learned that it was necessary to short all the unused coils because unshorted coils will resonate with a coil in use and cause a drastic drop in gain. A picture of all the chassis’ is shown at the end of this paper.
Our frequency range was 2 to 32 megahertz. My second attempt for the R.F. amplifier was successful. We had a design goal for our total gain to be one million with a noise figure of three dB which was considered optimum at the time. The R.F. amplifier had three stages. There was a careful physical separation between the R.F. amplifier variable capacito, and the oscillator variable capacitor, because the oscillator was at the same frequency as the incoming signal. As I recall I used an insulating universal joint between the two variable capacitors. It also required some good shielding. I designed the sine wave oscillators for the receiver, and the receiver required three oscillators. The oscillator had an unusual automatic gain control that kept the oscillator output both sinusoidal and constant to within a dB or so over the entire frequency range of the Synchronous Detector. John insisted on a sine wave oscillator so that there would be no harmonic problems. He also insisted on constant amplitude so that the demodulator gain would be constant. I should have put in a patent docket on the circuit, but I was too busy designing, and then winding the coils for the receiver. The total number of coils was twenty four. As I recall we built three receivers in the Laboratory. Another group built another ten or so and it seems that some other military service wanted a few.
John needed a frequency synthesizer for the receiver and he spent a great amount of time trying to come up with a synthesizer system. It was paramount to have a stable oscillator. He considered a number of systems. He hired Roger Swanson to build the system that he and Roger agreed on. It was a clever system that used a one hundred kilohertz oven controlled crystal oscillator that would with circuitry produce an sharp voltage impulse. The impulse gave a harmonic output every one hundred kilohertz beyond thirty two megahertz. It was then necessary to lock an oscillator to each one hundred kilohertz harmonic from 2 to 32 megahertz. It was then necessary to devise a means to scan one hundred kilohertz between each harmonic. A second oscillator was then offset by a low frequency oscillator that covered a one hundred kilohertz range and the sum of the two locked to the first oscillator. The actual tuning of the receiver was controlled by the low frequency oscillator and the tuning range of the receiver was then one hundred kilohertz. It actually required two adjustments to tune the receiver. You had to lock the first oscillator to one of the impulse harmonics, and then you could cover one hundred kilohertz above that frequency. The dial spread of the low frequency oscillator was about two hundred and seventy degrees with the numbers starting at zero, and going to 100 Kilohertz. This gave a bandspread of about ten inches for 100 kilohertz.
The low frequency oscillator generated a lot of discussion trying to decide on the circuit that should be used for best stability and I believe that John finally decided on a bridge oscillator, and the frequency was controlled by a potentiometer. I do not remember the frequency of the oscillator but it was probably around two hundred kilohertz. It was of course sinusoidal.
John hired Jack Cuniff to build the power supply and audio circuits for the receiver, and I believe Jack did the low frequency oscillator and the alpha –beta networks, but I believe that Roger did help him. The networks were not well known and it was John that knew of them. The alpha-beta networks allowed the Synchronous receiver to function as a single side band receiver. Jack was fond of power supplies, and audio circuits. We did have an outstanding product designer that was loaned to us from the TV department. He did the front panel of the receiver and the necessary gearing behind the front panel. There was a large amount of design here. He did a beautiful job, and he was great to work with, but I do not remember his name. He did everything on the front panel including the beautiful aluminum knobs.
The receiver was a bit difficult to operate at first, but if you knew the frequency that was desired it was easy. It was difficult to search across the dial because you had to first choose one of the harmonics and then you could only tune one hundred kilohertz above that harmonic. This was a great band- spread feature. It was perhaps the first receiver to use a frequency synthesizer.
While John’s Synchronous Detector scheme was to eventually contribute to data rates only dreamed of by him; it was years before it happened. The system was poorly understood by others and at the time the receiver market belonged to the Single-Sideband system. The Air Force was sold on Single Sideband. The Collins Company built a fine Single Sideband receiver and it is still admired and still used by hams today. The Air Force loved the receiver and would not give it up. It was a good system and worked well, but it was not optimum as John continued to point out in his papers.
The Synchronous Receiver has many advantages over single sideband, but John was never able to convince the Air Force of these advantages regardless of what he did. We performed an over the air test using the Synchronous Detector verses the Single Sideband receiver from the Rome Air Force base to Bolder Colorado using a word recognition test. The Synchronous Receiver beat the Sideband system badly. We taped the results and had a group of people listen and write down the words that they thought they heard. Roger Swanson took the results to Washington, and the results were not well received. Roger felt that they had accused us of cheating. We used differentiation and clipping in the transmitter exciter, and this was a means of increasing our average power. and improved word recognition. One of the problems with sideband is that you cannot use fast rise time signals as a transmitter input, because it will cause large transmitter output peaks that the transmitter cannot handle. This is because of the missing sideband. We took advantage of that. Single sideband will also not provide a good data system because of this problem. The Synchronous receiver that we designed had Single Sideband capability. This capability is easy to implement into the receiver.
G.E. did not pursue the business. Our Synchronous Receiver was big and heavy and would have required at least a couple of design iterations to condense it into a really marketable receiver.
When we finished the Synchronous Detector Dr. Costas had no idea that his contribution would allow communication systems to approach their optimum capabilities – no one did. It was years before his system was recognized for its capabilities and really put to work.
Sometime after we had finished the receiver and completed the project John hired an excellent engineer who was from Bell Labs, and had several years of experience. We were all impressed with Joe Reddeck because of his experience at Bell Labs. Joe took me under his wing and taught me to use some design tools that I had not been aware of. We worked together on several projects, and it was a joy to work with Joe. I believe that Dr. Costas was busy with other projects, and perhaps writing papers, I have always suspected that John expected Joe to be my mentor. We worked well together and Joe seemed to enjoy the roll as much as I did. He was an excellent circuit designer.
Dr. Costas was approached by C.R. Wayne who had an Advanced Engineering group in the G.E. Heavy Military Department. Wayne got John to move our group from the Electronics Laboratory to Heavy Military. I believe that we had six engineers not counting John. After we moved to Heavy Military John hired two or three engineers. Wayne had a large group of engineers working in diverse advanced Engineering areas. Wayne had acquired a fine organization. I have no idea how he got John to move. Wayne was a great manager, outgoing, and quick to smile, I liked him a lot.
After we moved to Wayne’s group John decided to build a Spread-Spectrum system. He was always concerned with jamming of communication systems. John conceived a wide band data system that was jam proof. I did not work on the system, but followed it closely. He finished the system, and actually put it on the air in the 20 meter ham-band. As I recall it covered the band completely and could not be detected by a normal receiver because the power per cycle was so small. John named the system “Phantom” and I do not recall him writing much about the system. The system used a tapped delay line with the data being sent depending on how the taps were set. He demonstrated the system, and then seemed to abandon it. I assume that it was covered by G.E. money rather than military contract.
While John was working on Phantom, I started working with an excellent engineer whose name was Jim Kovaric. I was helping Jim get an unusual system to work. The system was called a “Side Lobe Canceller”. The system was to be used in the Berlin supply crisis. Our Berlin supply planes often had difficulty landing because the Berlin radars were being jammed by the Russians. The Side Lobe Canceller was to help circumvent the jamming. I believe that Wayne also had others working on the problem.
Because of my work with Jim I believe that I was given the job of building an “Over the Horizon Radar System”, but I still reported to John. The system was called a Velocity Indicating Coherent Integrator. The system used a long delay line, and while I do not remember the delay time it consisted of a large six foot high cabinet with three hundred feet of wire inside. It was built by the Electronics Laboratory by a group headed by Steve Tehon. The delay line was delivered to me just as I finished the electronics for the system. The delay line had a sharp drop in output at an important frequency, and I had to build a chassis that had a matching sharp peak at this same frequency. This then made the delay line frequency response flat and this was imperative for the system. Setting beside the delay line was a full rack of electronics that made up the Velocity Indicating Coherent Integrator electronics. The system was to measure the movement of the ionosphere as a missile (Russian) passed through the ionosphere. We tested the system at the Rome Air Force base and the backscatter did show the movement of the ionosphere. I was not given the results of any missile tests or other effects. The system was turned over to the Air Force and I heard no more of the system. My work with the system was in the late nineteen fifties as I recall. I think that the Syracuse Electronics Park was still in the “Over the Horizon” business at least until the nineteen sixties.
John was given a Navy contract to build a fifty megahertz version of the Synchronous Detector for a data link between Boston and Cape Cod. The Navy was interested in the error rate at this particular frequency and we were to use a one kilowatt transmitter. The Navy required that the receiver antenna had to be floating on the water (submarine). There were several groups involved. One group built the data generator and error checking equipment and another built the transmitter, and got the sites for the transmitter and receiver. John gave me the receiver to build and the time frame was around three months. The tests lasted for several months and I got to visit both sites. We were happy with the results, but the Navy of course did not give us any information as to how they felt about the system.
I put a one inch oscilloscope on the front panel of the receiver to show the I and Q channels. This was nice because you could see the noise from the system because the line on the oscilloscope is very thin when the noise is low and will widen when the noise gets large. This led me to an automatic Signal plus Noise to Noise ratio indicator on which I received a patent. Johns Synchronous Detector actually led me to a number of patents, and his system suggested applications for patents in other fields. My first patent was with John as W.H. Jones ET AL.
G. E. decided to open a Military Communications Department in Oklahoma City. A meeting was called, and our group was called to attend. We were told that we had to go to Oklahoma City, and had no other choice. I was delighted, but the rest of John’s group was not happy with the announcement.
John called a meeting of our group plus a few others as I recall, and the meeting was held at his home. He wanted to discuss the possibility of starting a company. John had already picked the name for the company as Cogent Systems, and he called the meeting to discuss the various markets and the financing for the company. While I understood that he was interested in starting a company I assumed that the meeting was only a first casual discussion of the idea. John put together a talk regarding his ideas, and lasted perhaps less than thirty minutes.
As soon as he finished talking he said that he was going to poll the group as to whether each person would join the company or not. I was surprised because I felt that the invitation was rather sudden, and I did not have an answer. To my horror, I was the first to be asked - John said- “Bill, will you join?”. Going into business with Dr. Costas was a dream come true. I knew however, that I wanted to be in Oklahoma City and the Military Electronics Department was a great match for me, and also John. It had not occurred to me that John had decided that he would definitely not go to the Military Communications Department, but he had. I had a huge dislike for the weather in Syracuse and wanted to go back “home” in the worst way – as did my wife – she was at the meeting along with the other wives. To my dismay I had to come up with an answer. John knew of my admiration of him as an engineer, and as a person – even a friend. He certainly expected that I would be a safe “Yes”. I had to say something and it was a simple “no”. It surely required a long pause for the “no” to come out because of the many things that went through my mind. I have always thought that John asked me first because he expected a resounding “Yes!”. I also felt at the time that he would come to his senses and actually go to Oklahoma City. It was a perfect match for John but he was adamant and would have no part in the transfer.
An excellent manager by the name of Ben Walker was picked to manage the department. I have always wondered what would have happened if Dr. Costas had been asked to be the department manager. I think that it would have been difficult for him to turn down. (It is possible that John knew of the new department, and he may have been asked to manage it. It was a large department with a number of advanced projects, and two large buildings with a large factory area.
John wrote a letter for me to Ben Walker telling him that I wanted to go to Oklahoma and asked Ben to find a position for me in the Department. I wound up working for Dick Ellis in Oklahoma, and enjoyed working for him. Dick was a great manager. Dick and I got along very well. I was the first of his group to move to Oklahoma and Dick arrived a few months later. He found a house across the street from our house. We were both unaware of this for some time.
I was of help to Dr. Costas because I had been interested in radio since I was a child. I was shown how to build a crystal receiver when I was ten years old. I even learned how to make my own crystals. I built one tube receivers and would often visit furniture stores looking for old used radios that would not work and buy them at very reasonable prices. I learned how to solder and how to take radios apart. I learned about all the various parts of the radio. I could only repair a few of the radios that I purchased.
I was extremely fortunate to get a job at Al Robertson’s radio shop while attending high school. Here, I really learned to repair radios. Al had an expert radio repairman that would help me with those radios that I could not repair. Ray Binghamton could repair any radio that was brought into the shop. He seemed pleased in showing me what was wrong with any radio that I was having trouble with. After about one year I could repair any radio that came into our shop. This background helped me greatly during the G.E. test program, and it was most useful in working with John on the Synchronous Detector My work with John was an important area of my life and it was an important area for radio and data communication because of Dr. Costas. I consider myself to be an excellent engineer because John was my mentor. John was the best Engineer that I ever met. John could do it all. He could design circuits and I would consider him a mathematician as well as an engineer. Some years after the Synchronous receiver were finished John retired from G.E. and moved to Boston where he continued to make important contributions in other areas.
John was a family man, and had a beautiful wife –Helen. My wife and I would go to their house for an occasional dinner and they came to our house for dinner. They lived in an upscale neighborhood while we lived in a downscale neighborhood. Helen was from Boston and when John retired from G.E. they moved to Boston. John did some consulting work in Boston for a company by the name of Cogent Systems. Strangely enough, Cogent was the name for the company that John had wanted to start when he was working for General Electric.
I really missed John after the Oklahoma move because it was difficult for me to find the optimum capabilities for any unusual system. I always missed John’s mathematical skills and John as a mentor. He was an outstanding teacher.
John was always interested in the optimum that was possible for any system. He had worked under or for some of the best known names at MIT. These Professors were an important part of John’s career. He spoke of them often. Shannon, Weiner and Lee were part of his vocabulary. Dr. Lee was his thesis advisor. Generally, they were ones that had mathematical theories for various systems with the optimum values that could be achieved for such systems. That is to say, the best performance that could be obtained with that particular system, but the way to reach such performance was usually unknown.
In the early fifties the data rates possible for communication systems were far from the theoretical maximum reachable and John was always curious as to why this was so. He had a wide ranging interest in many areas and of course did not have the time to contribute to all the areas of interest. He wrote many papers for the Institute of Electrical and Electronic Engineers. John wrote a paper concerning Synchronous Detection for the I.E.E.E. in 1956 and it was such a classic paper the I.E.E.E. reprinted it years later. He was a prolific writer and had a wide ranging interest in many areas. John was an active Ham operator (K2EN) and built his own kilowatt transmitter. This was typical at the time. It was well built in a nice six foot rack and he had a rotating beam to go with it. He had a top of the line AM receiver.
After moving to Oklahoma and separated from John I worked for a number of companies. I worked for the G.E. Military Communication’s department but it stayed in Oklahoma City for only a few years and then went back to Syracuse. It was shortly thereafter disbanded. I was able to stay in Oklahoma because the G.E. Computer Department moved into the City just when the Military Communications moved out.
After G.E. I worked for Honeywell, and Magnetic Peripherals. The work was the same for the three companies because G.E. sold our group to Honeywell, and Honeywell sold us to Magnetic Peripherals. I quit Magnetic Peripherals, and started to work for Silicon Valley start-ups. Worked in California for ten years.
We were working on a high density disk drive at Konica Technology- a Silicon Valley start-up. I was the Read/Write engineer. While at Honeywell I became what was called a Read/Write engineer. After Honeywell, I was a Read/Write engineer for each company were I worked. As it turns out being a Read/Write engineer is quite the same as being a Communications engineer. The Read/Write engineer needs to know phase lock loops and the Synchronous receiver was one big phase lock loop. I had the perfect background to design read/write circuits for disk drives.
One day while at Konica, a group of us were discussing a problem, and someone brought Dr. Costas’s name into the conversation. The Japanese engineers were well acquainted with the Costas Loop, and Johns work. I said –“ I understand the problem because I worked for Dr. Costas for years”. One of the Japanese engineers said “no – Dr. Costas is dead ”. I assured the gentleman that John was not dead. It seemed clear to him that I was incorrect and that I was probably thinking about some other Costas rather than Dr. John P. Costas. It so happened that shortly thereafter I had a business trip to Boston, and I called John and we had dinner. Over dinner John discussed some of his areas of business, and in the discussion he said “you know Bill – there are a lot of people that think that I am dead.” I recalled the Japanese gentleman, (I do not remember his name). I told John about the discussion with the Japanese. John was amused. I asked John to autograph his business card (Cogent Systems) so that I could carry it back to the Japanese engineer who was certain of Dr. Costas’s demise. John addressed the card to the Japanese gentleman--by his name. I also asked John to autograph his card for me--I still have it.
Had John not been such an accomplished engineer and mathematician he would still would have been highly regarded for his intelligence and integrity. I always thought that his IQ was in the genius category. He was a great theoretician, and loved to teach. He was a Mathematician. He could analyze systems, optimize systems, design circuits, and optimize circuits. He knew well how to use a soldering iron. Dr. John Costas was an outstanding Engineer.
Dr. Costas was a down to earth gentleman in the best sense of the word, and the best Engineer that I ever knew.
Pictures/Letters
Left to Right- Bill Jones, Dr. Costas, Roger Swanson, Jack Cuniff. This Picture subject is the Synchronous Receiver A/N FRR-48 XW-1
Bottom of RF chassis by Bill Jones
Top of RF chassis by Bill Jones
Top of Synthesizer by Roger Swanson
Top of RF chassis by Bill Jones
Fifty millisecond Delay Line from Right B. Jones, Steve Tehon. Other unknown
Patentees. Standing, Right-Dr. Costas, Sitting, left-Dr. Maier. Sitting Number 2 Left-Townsend, Number 5 W.H. Jones, Others Unknown.
W.H. Jones at Al Robertson Appliance Store. Age 18
Letter from Dr. Costas to Bill Jones, Subject; Employment
John’s letter to Captain King of Australia (four pages)
Questions or Comments? Please e-mail me at whnj
Amazing account. Thank you. Also, "Costas" is a Greek name. Was Dr. Costas Greek-American?
ReplyDelete