James Long, Ph.D., P.E. Retired Analog and RF Consulting Engineer
History of Mixer Technology
(And Other Stories of Misplaced Credit)
Things are seldom as they seem. Skimmed milk masquerades as cream.
If you arrived here from a search engine, you will probably be interested in other similar pages.
Brother Cysa Dime has authored a book on the seven deadly sins. I recommend it and his how to study the Bible book. Both are available on Amazon
All of mixer technology in use today was mathematically known several hundred years ago and reduced to practice by 1935 when the DPDT vibrator modulator was used in low frequency amplifiers. By 1949 this function was performed by the 7360 vacuum tube in radio circuits and six vacuum tubes in rocket telemetry circuits. In 1963 Jones of Burroughs Corp. patented a common mode feedback addition to the transistor version of the circuit to mitigate the poor performance. This circuit in integrated circuit form was market by Sylvania in 1964 as an ECL exclusive or gate. People today who get their history from journalists instead of reality call this topology the "Gilbert Cell Mixer."
In 1924 and 28, Nyquist and Hartley published the limits to communication over a noisy channel. In 1949, Shannon and Weaver published a book on the same subject. Shannon got the credit for Nyquists' and Hartley's work. He also claimed the 34 year old sampling theorem as his own work.
H. Nyquist, "Certain Factors Affecting Telegraph Speed," Bell Systems Tech. Jour., vol. 3, April 1924, p. 324
H. Nyquist, "Certain Topics in Telegraph Transmission Theory," A.I.E.E. Trans., vol. 47, April 1828, p. 617
R. V. L. Hartley, "The Transmission of Information"Bell Systems Tech. Jour., vol. 7, July 1928, pp. 535-564
Helmholtz devised a theory of equivalent sources. Thevenin and Norton got the credit for reinventing them.
The Smith Chart was invented and published in Japan by Mizuhashi 13 months prior to Smith publishing it. Campbell of filter fame published it 25 years earlier.
Frequency hopping spread spectrum was a public domain idea by 1917. The Germans used it in WWII. Hitler wanted to win by bluff and before the war started, invited public figures from England and the US to see how invincible his military was. Hedy Lamarr, who has similar scientific-mathematical skills to Cher and Edith Bunker, was among one of these groups who was shown the "invincible" communication system the German's had. When the group got back to the US, they applied for a patent and possibly as a joke put only Hedy's name on it. The patent office examiners then, as they are now, are not practicing engineers and are spread over a wide range of technology, they are jacks of all trades and masters of none. They are also short of time to keep up with what is going on in the engineering world or to study engineering history. They only see if a similar patent was issued in the past. Finding none, they issued the patent. This is why so many patents are being issued these days on public domain prior art.
Udi invented a radically new antenna type. Yagi could speak and write in English. He got the credit for the antenna. It wasn't until several decades later that Udi's work was examined and found to have the optimal explanation of how the antenna worked which had eluded western "antenna experts."
Heaviside recast Maxwell's equations in their present lucid form. He also predicted the ionosphere and its effects on radio waves. He further theoretically derived all aspects of transmission lines in the form we use today. Since his results were different than those of the Cambridge professors, he was doomed to obscurity by the in crowd who controlled the technical press. No one remembers the work of the Cambridge professors because it was all wrong.
Watson-Watt was in charge of a government laboratory where a rank and file engineer devised a way to use radio to locate airplanes. Watson-Watt arranged to get the entire credit for the invention. Even the employee was 15 years too late. The RADAR principle had been used since 1926 to measure the height of the various layers in the ionosphere. A Race to the Edge of Time by David E. Fischer describes several European public demonstrations of working RADAR systems in the several decades before the war. Secret Weapon by Kathleen Broome Williams describes a microwave RADAR developed by the French division of ITT. This RADAR was undergoing sea trials on a French passenger ship when the war started. US Electronic Warfare vol. iipage 335 by Alfred Price describes how Soviet RADAR was much advanced than the British, including high power (many kW) microwave magnetrons, until the purges. Each government department head that wanted RADAR done in his/her department made false charges on the talented people in the other departments. After all of the talented people were killed there was no one to continue the development with any degree of success.
Harvey is given credit for the discovery of circulation of blood. Almost 100 years earlier Servetus, a physician, published this theory as a sidelight in a book on his unorthodox religious theories.
Millikan had a graduate student who devised an experiment to measure the charge on an electron. Millikan had the paper published with just his name as the author. This same experiment and calculate results were done nine years earlier in England by Thompson.
"Intellectuals" in the 1930s believed that compound interest was the cause of all social evils, such as economic depression and war. This was one of the basic pillars of socialism. No wonder that the countries that used socialism had such economic disasters. England was once the leading industrial nation in the world. Now it is so far behind that it may well be classed as a third world power in the future.
The 1932 Pulitzer prize went to Walter Duranty a New York Times journalist in Moscow who reported that the defendants in the show trials were genuinely guilty and that the farm collectivization program was a success. (25% of the population died from the famine.) This journalist was being blackmailed into making the favorable reports. What is inexcusable is the refusal of the prize committee to withdraw the prize now that it is known that the reports were false. I find that most prizes are politically motivated and have more deserving recipients. The most recent example is the Noble prize being given to improvers of the MRI device and not to the inventor of it.
President Kennedy's book was ghost written. He got the Pulitzer prize. In contrast to this, in the music industry in around 1995, the award was withdrawn from the group that lip synched to a back up group and the award was given to the back up group. President Kennedy quoted a Greek writer around the time of Plato in his "ask not what your country can do" statement. The journalists were so culturally challenged that they did not recognize the source and President Kennedy was so morally challenged that he did not correct their error.
Charles Darwin was not the first to publish ideas about "evolution", he copied from Herbert Spencer ["survival of the fittest" was one item copied.] and others without giving them credit. The Greek and Roman philosophers predated him by about 2200 years. I have seen descriptions of selective breeding and cross breeding from works of about 3500 years ago. Unfortunately, his writings are similar to the person who did not have a 40 year career, but instead had a one month career 480 times. He presents the same flawed argument over and over. His works were made before the discovery of genetics and the extreme chemical and electrical complexity of single cell animals. With the discovery of the relative young age of the universe (18 billion years, the earth being much younger) and the stability of genes [The genes of the Neanderthal "man" are so different from modern humans that it is not an ancestor of us.], professional biologists are looking for some other explanations of how we got here. It is the second rate amateur biologists and hobby biologists that still cling to the "Darwin theory." Currently there is a lot of work going on in the area of adaption, which by a slight of hand trick is called "Darwin." This gives the false impression that the shifts between vastly different types of organisms is being proven.
The credit for inventing the klystron usually goes to Varian. Several years earlier (1935) A. A. Heil and O. Heil published the concept in Z. Physica vol. 95 p. 752. A few months earlier, W. C. Hanh and G. F. Metcalf published the general theory of velocity modulated tubes with a complete mathematical analysis in Proc. IRE vol. 27 ,p. 106-116, February 1939.
Major Armstrong is given credit for inventing the superhetrodyne radio topology. Prior to his "invention" there were issued patents on the same topology in England, France, and Germany.
The Baker Clamp was common knowledge as early as 1953 and was described in passing in an introductory text on transistors written by Shea.
The Wheatstone bridge was invented by S. H. Christie in 1833.
Wheatstone did invent a hand cypher. It was shown to the Foreign Office by Playfair. Playfair got the credit even over his objections.
The US Patent Office has been issuing patents on very old public domain technology.
The Gibbs Phenomena was discovered and mathematically analyzed 7 years earlier by Wilbraham of Trinity College Cambridge.
The Laplace transform was invented earlier by Poisson.
Hertz and Marconi are given the credit for inventing radio. Twenty years earlier in West Virginia a person signalled 14 miles with radio. He applied to the US Army for funding and they turned him down. Since he was not an academic the work never got published.
President Kennedy is given credit for avoiding global war over Cuba. What really happened is that the CIA military operations against Cuba were too successful. The soviets did a ploy where they suckered Kennedy into agreeing to not do what he would have done (CIA operations) in exchange for the Soviets not doing what they were not going to do (put nuclear weapons in Cuba).
McCarthyism was invented by President Truman three years before Senator McCarthy's committee, which included Senator Kennedy, became prominent. Now that the Soviet, FBI, and code breaking records have become declassified, it turns out that there were 450+ Soviet spies in the government including several of President FDR's top advisers. There were also several influential academics and journalists who were Soviet agents and used their public platform to further the Soviet cause.
Mixers operate by performing the trigonometric function of multiplying two sines. Specifically, 2(CosA)(CosB) = Cos(A+B) plus Cos(A-B). Where A and B are changing angles (frequencies) with possibly different time origins (phases). If A and B are different frequencies, two new frequencies are produced. If they are the same frequency, a DC term related to their phase difference and a double frequency term is generated. The first situation is used mostly in radios, and the second is used mostly in instrumentation and servo systems. People in each of these fields know about both of these cases and must design their system to avoid having the unwanted outputs produce system aberrations. If a square wave is used for one signal, the mathematical operation of bipolar sampling is implemented.
This mathematical process is linear. That is, there is a 1:1 correspondence between multiple inputs and multiple outputs. This is why blocks of signals can be frequency translated. In radio work, one of the inputs is a pure spectral tone and the other input is the signal. A defect occurs when the local oscillator spectral tone has phase noise which is then modulated onto the output signals. (If A has a phase jitter, so will A+B or A-B.) This produces random phase rotations in QAM modulations which make their detection more error prone. The phase noise sidebands of strong signals overlap the frequencies of smaller signals and reduce their signal to noise ratio.
There are several topology choices in making mixers and phase detectors. They involve several mutually independent areas. Each area has several selections.
1. Should the mixer/phase detector use some semblance of multiplication or should it use a polynomial nonlinearity which has a product term? The polynomial method is such a disaster that it is only used when cost is more important than performance or when technology has no alternatives. One example of the latter was the diode mixer used in WWII RADAR receivers.
2. Should the multiplication be linear with a sine wave local oscillator or should the local oscillator produce a multiplication by a square wave (hard switching action also called chopping)? The square wave is composed of harmonics which produce harmonic mixing terms. This can usually be mitigated by the system frequency plan. Pure analog multipliers have limited frequency response and higher intermodulation distortion terms. Hard switching circuits are less prone to intermodulation distortion. High performance systems always use hard switching mixers.
3. Should the multiplication be done on the voltage signal or should the signal be converted to a current which is then multiplied? The weak link in the current method is the conversion operation which has poor intermodulation distortion. The output of the mixer is a different frequency that the input and so overall feedback cannot be used. The best of a poor selection is to use source/emitter degeneration resistors. High performance mixers usually use the voltage method.
4. How many levels of pairing and phasing should be used? An unbalanced mixer has no port to port isolation. The local oscillator and input signal appear at the mixer/phase detector output along with the mixing products. Phasing two unbalanced mixers allows the cancellation of one input at the output (singly balanced mixer). The RF is usually chosen as the cancelled output as the local oscillator is just one signal and at a known frequency. Two singly balanced mixers can be phased together to cancel the other input at the output. This way, both the local oscillator and RF input are kept from the output. Two doubly balanced mixers can be phased together to remove one of the mixing outputs (single sideband mixer). The choice is mainly economic. The mixer that minimizes the total system cost is usually chosen. This includes the cost of mitigating the effect of the unwanted outputs.
5. What technology should be used? This changes with time as manufacturing efficiencies reduce the cost and improve the performance of various electron devices.
These choices are very old and well known. It has been changing economics and needs that have made different combinations of the above choices the optimum and most popular method at any given time in history.
Surprising Information for Members of the X Generation
People usually assume that present conditions are the same as past conditions or better. This is rarely true. In days gone by very few people went to college. Because of this, engineers of the past were significantly more talented than today's average engineer. I would estimate that engineers in the past were equivalent to today's Ph.D. students. This resulted in many ideas that would result in an IEEE journal article today being commonplace knowledge in the past derivable by any engineer in an afternoon with paper and pencil. Because of this, these ideas never went into journals and therefore, people doing literature searches today will not find them. Every circuit topology commonly used today was thought of in the past. The high cost of vacuum tubes and discrete transistors kept them from being implemented in commercial products although products used by the military and intelligence agencies did use them.
In the early days of radio, there were not many stations on the air and they had narrow signals. The signals were digitally modulated using a sequence of five symbols that were distinguished by their duration of logic 1 and logic 0. The symbols were grouped in sets of variable length to optimize English language text transmission speed a la Huffman. The symbol for E was the shortest and the symbol for Q the longest. Punctuation marks and numbers were even longer. The spacer between letters was logic 0 for three time units, or bauds, the spacer between words was logic 0 for 5 bauds. The Baud was officially defined by the CCITT in 1936 as the smallest time unit that the members of a signal set were composed from.
Before the invention of vacuum tubes, mechanical contacts in the form of a rotating disk with multiple conductors on the edge were used to sample and down convert radiotelegraph signals to the audio range. When triodes were invented, the radiotelegraph signals were converted to the audio range by injecting a continuous sine wave at the antenna using a frequency slightly different from the signal to be received. The nonlinearities in the triode provided the mixing action.
By the 1914-1918 War
By 1914, vacuum tubes became economical and more radio stations were on the air. Major Armstrong of the US Army Signal Corps invented the superhetrodyne principle in 1917 and reduced it to practice. (It had been invented a few months before in England and Germany, but those inventors stopped at the paper design stage.) This topology allowed easier tuning between stations and a constant bandwidth. Major Armstrong also invented frequency modulation.
At this time, the telephone company devised a way to put more calls over the long distance metallic lines by changing the frequency of the audio signals. This was called frequency division multiplexing.
This produced a divergence in mixer technologies between the two systems. The telephone lines were broad band systems that had the potential to suffer from spurious mixer products. Therefore from 1918 on they used the diode ring double balanced mixer initially with copper oxide rectifiers. The radio receivers only received a single signal, and therefore, it was more economical to use the polynomial nonlinearity of vacuum tubes for the mixing process and then to use a tunable input filter to reduce the IMD products.
The major use of radio receivers after the 1914-1918 war was for AM broadcast reception where cost was the most important criteria. The mixer stages were quite crude with the local oscillator and signal fed into two, or even the same, electrode element. (polynomial nonlinearity)
As the number of stations increased and the power levels also increased; receiver performance had to improve. The pentagrid converter, such as the 6BE6 tube (British EK90 valve), was introduced which was optimized for mixer performance. This included the isolation of local oscillator and input signal as well as allowing the first grid and cathode to work in an oscillator configuration. These tubes allowed a closer approximation to the multiplication of the two sine waves which produced fewer spurious mixing products.The British used a 7 grid tube with similar multiplication properties in FM broadcast receiver demodulators. They picked signals off the primary and secondary of the last IF double tuned transformer (which were 90 degrees out of phase) and multiplied them.
Throughout this time, the technical journals, such as the Bell System Technical Journal, RCA Review, and the Proceedings of the Institute of Radio Engineers, were filled with theoretical analysis and practical description papers. All textbooks had sections on mixers from 25 - 75 pages in length. There were also entire books on mixers.
The stations on the AM broadcast band were limited in number and power by government regulations. The short wave bands were another matter. Numbers of stations and their powers were monotonically increasing. This made it economically feasible to use more complex tubes if they had increased performance.
By 1935 the vacuum tube differential amplifier and operational amplifier functions were in commercial use. A common shortcoming of these was the DC offset drift that occurred as the temperature of the tubes changed with airflow. To work around this, the chopper stabilized amplifier was invented. This function was also used for low frequency signals to eliminate the cost of high value capacitors and inductors that were needed to extend the amplifier frequency response to low values.
The chopper function shifted the input signal up in frequency by a SPDT or a DPDT switching function performed by a set of mechanical switches (relay contacts) that were thrown rapidly between the two states by forming an oscillator using one set of contacts in series with the coil and a DC source. This produced the +/-1 multiplication (or chopping) of the diode ring double balanced mixer. From the sound of the mechanical vibrations, this device was know as a vibrator. One further advantage was the DC frequency response of the mechanical switching contacts.
This function was officially termed the "modulator." The signal was AC amplified and then converted back to the lower frequency by synchronous detection using another set of switch contacts that were mechanically coupled to the first set. This produced a multiplication of +/-1 and this function was officially called the "demodulator."
The choice of SPDT or DPDT switching was an economic consideration. DPDT was used when there was a common mode signal that needed to be eliminated.
Advances were being made in servo systems that relied on phase. Phase mixer/phase detectors were used to control the drive on electric motors which rotated a physical member. The aeronautical radio compass was one example.
By 1949 it became economically feasible to include the +/-1 switching function (chopping) from the 1918 diode ring double balanced mixer (or alternately, the DPDT switching function from chopper stabilized amplifiers) with a gain function. This was first done in England and later RCA in the USA commercially introduced the 7360 vacuum tube. This has significantly fewer mixing products and the oscillator was totally isolated from the signal.
The tube was constructed as a cross between a standard pentode and a CRT. The cathode emitted electrodes which were accelerated and controlled by the grids, one of which was the signal input. The deflection plated directed the electron beam to either of two plates. Since the beam had a large diameter, the transfer function between the deflection electrodes and the two plates was linear for small signals and +/-1 for large signals. The deflection plates and two collecting plates were used in a tube type prior to 1940 in an attempt to increase the frequency response, but the manufacturing process of conventional tubes improved enough for them to have the same performance at a lower cost. A similar idea was proposed by Vreeland using magnetic deflection prior to 1910.
This mixer tube was widely used in high performance receivers, such as those made by Collins and those in amateur home construction projects.
In the cost is no object government missile program, a vacuum tube mixer used in the FDM stack up of channels in the telemetry system used a voltage to current (0 degree and 180 degree outputs) converter with current steering to a load by a square wave local oscillator. Since this was a broad band system, a double balanced circuit was used. The transistor equivalent of this circuit was used in the Jones phase detector patent. (See Jones below.)
As transistor costs came down, all of the vacuum tube circuits were implemented with transistors, except for this circuit. There were two basic reasons for this exception. The first was the cost of using many transistors and transformers. The second was performance. The poor matching of discrete transistors made for poor balance and nulling. The exponential nature of the PN junction made for poor IMD performance. The poor frequency response was another limiting factor. This explains the lack of journal and magazine articles on this topology using transistors. This does not mean that engineers did not think of using transistors in this topology. Because of declining costs and rising performance requirements, the diode ring double balanced mixer was used with Schottky diodes which could be well matched, and the 7360 circuit was used less frequently. Low performance consumer and hobbyists products used cheaper 6BA6 mixer circuits.
One interesting servo system got double duty out of the phase detector. Short wave single sideband systems came into prominence. This required precise tuning to hetrodyne the audio to baseband. The carrier was reinserted at a 10 dB reduced level and used in phase locked receivers. The phase detector served double duty. It mixed the RF to baseband as well as produced a DC output to drive the phase correction circuit which was an electric motor which rotated a variable capacitor in the local oscillator.
The following is from a visitor to this site: The circuit for the six-transistor analogue multiplier appears in it's
basic form in the Feb 1966 issue of "The Review of Scientific Instruments",
and is therefore public domain. In the text "Precision Electronics"
(Philips, 1967), the circuit appears in both vacuum tube and transistor
The 7360 circuit, FDM telemetry circuit, and choppers were not totally forgotten. In 1963, Howard E. Jones of Honeywell, Inc. filed for a patent using the transistor emulation (using what the British called a long tail pair which was in turn a transistor equivalent of one stage of a vacuum tube operational amplifier) of the chopper/7360/FDM telemetry circuit as a phase detector. It also included common mode feed back to correct for the poor matching of discrete transistors. This was issued March 15, 1966 as US Patent 3 241 078. Since the US patent office disallows claims involving prior, public domain art as well as those ideas that are easily devised by practitioners in the art, there were no claims for the circuit use in radio receivers or transmitters as a mixer. I suspect that if the common mode feedback was not claimed, the patent would not have been issued. The usual limitations of this circuit did not cause problems in a low frequency phase detector. Sylvania marketed this circuit in integrated form in 1964 as an ECL exclusive or gate. I have seen ECL parts used in commercial products as mixers and amplifiers.
Over the years several IC companies have tried to market this mixer with little success. This circuit only came into its own when the cellular systems were put in place and the handsets had severe restrictions on size and battery performance. The limitations of this circuit are mitigated by using FET transistors which have less IMD that bipolar transistors. Also, the cellular system is carefully crafted so that the handsets work in small signal environments and the frequency plan keeps the IM products out of the receiver IF. Also, there is power feedback so that handsets that do experience increased Bit Error Rate have their received signals increased.
When this mixer circuit is used in ham radio VHF handsets in a slightly less ideal environment of less controlled frequency plans and power levels, they produce poor system performance. It is necessary to use tone squelching to prevent the IMD products from reaching the audio speaker.
In 1968, Barrie Gilbert published a paper on a four quadrant analog multiplier which combined the circuit of the Jones patent along with predistortion circuitry to make a large signal four quadrant multiplier. The entire content of the paper was about accuracy of multiplier circuits used in analog computers. In 1974 a further paper was published showing how to reduce the multiplication errors even further. It was the integrated nature of the circuitry that made the transistors match which made the circuit accurate. No reference was made to radio mixers in either paper. The circuit The Jones patent was referenced.
Gilbert cites the Jones patent in his first paper. He, like tens of thousands of engineers in the previous 15 years, had independently conceived of using bipolar transistors in the topology that had previously used vacuum tubes. Also cited were two papers that used the topology in radio circuits and a master's thesis. There was a discussion on commercially available versions using discrete transistors needing temperature control ovens to reduce errors.
Not referenced is the following paper:
A simple d.c. to 10 Mc/s analogue multiplier
,R R A Morton 1966 J. Sci. Instrum. 43 165-168
with the following abstract:
R R A Morton
Electrical Engineering Department, Monash University, Clayton, Victoria, Australia
Abstract. A transistor analogue multiplier having a frequency response of 0 to 10 Mc/s with a maximum attenuation of 1 dB on either input and an accuracy to 1%, and which operates over four quadrants, has been developed. The multiplier inputs have a range of ±3 V and the outputs, ±2 V; residual noise (which determines the minimum useful output) is 3 mV r.m.s. Temperature stabilization over 20 to 50°C is achieved by using two matched transistor pairs and adding two thermistors to the circuit. The multiplier and a detailed analysis of the errors arising in it are described.
Print publication: Issue 3 (March 1966)
Received 14 September 1965
The Sad State of Ethics in Modern Science and Technology
Several decades ago, digital signal processing was a new research field. Two workers devised a method of reducing the number of computations in performing the Discrete Fourier Transform. This method was named after them. Once this method became widely advertised, workers in mathematical fields came forward with evidence that this method had been commonly used in their area for a long time. It goes back at least to 1805 and Gauss. Thereafter the method was no longer described by the names of the two DSP workers and was instead called the FFT. It is a poor engineer who does not reinvent the wheel on a frequent basis and has to copy designs from handbooks without understanding how the work.
In the mixer area we have a case of vacuum tube versions being followed by discrete versions and yet the mixer is called by the name of the person who thought of it later than these versions.
In biology, things are even worse. Darwin is still given credit for the work plagiarized from others and his mistakes are quietly corrected and assigned to him and presented as his work as done in his lifetime. (This reminds me of several home grown US religion where the prophet's writings have been carefully revised to remove errors both theological and scientific. These revised versions are presented as the prophet's original writings.)
In medicine things are pretty bad. Drug companies do research on new drugs and write up scholarly reports that are given to famous medical school professors to submit to research journals as their own original work.