While amplifiers using both feedback and non-feedback topologies were being refined in the late 1930's and throughout the 1940's, there were some very interesting developments within the realm of differential amplifiers.

Today's op amps utilize differential topologies to a high degree, but the reader should understand that this wasn't universally so, back in the days of vacuum tube amplifiers. In fact, vacuum tube op amp topologies that utilized differential techniques fully never really became well established before the breed began dying off. Nevertheless, it is still a useful thing to examine some key differential amplifier publications up through about 1950, at which point it represented a maturing of the art. The fully differential, defined gain precision DC amplifier was of course, the forerunner of what we know today as the instrumentation amplifier.

The earliest vacuum tube differential amplifiers were reported well back in the 1930s, and evolved steadily over the next 15-20 years. Many of these authors addressed the problems of low-level instrumentation amplifier circuitry used in obtaining signals from living tissue, thus the apparatus involved was often called a "biological amplifier".

One of the early authors in this field was B. H. C. Matthews, writing on a special differential input amplifier in 1934 (Reference 1: B. H. C. Matthews, "A Special Purpose Amplifier," Proceedings of the Physiological Society, Vol. 81, March 1934, pp. 28, 29. (A dual-triode differential amplifier, with dual triode common-cathodes biased to ground through a high resistance, AC-coupled at the output) ). Matthews' amplifier did indeed have differential inputs, but since the common cathodes were tied directly to the power supply common, it wasn't optimized towards minimizing response to common-mode (CM) inputs. Note that in those days CM signals were often referred to as push-push signals, to denote signals in-phase at both inputs.

Alan Blumlein's UK patent 482,470 of 1936 went a step further in this regard, by biasing the common-cathodes of a differential pair through a common resistance to ground (see dual triodes within Fig. 2 of Reference 2: A. D. Blumlein, "Improvements in or Relating to Thermionic Valve Amplifying Circuit Arrangements," UK Patent 482,740, filed July 4, 1936, issued April 4, 1938. (A basic common-cathode/resistance-biased-to-common, dual-triode differential amplifier, AC-coupled at input and output) ). Blumlein's patent was concerned with wideband signals, not biological ones, using AC-coupling. Nevertheless, it was a distinct improvement over the Matthews amplifier, since it provided bias conditions more amenable to CM signal rejection.

In 1937 Franklin Offner discussed a variety of differential amplifiers, and among them is found one similar to Blumlein's configuration (see Fig. 3 of Reference 3). Like those of Blumlein, Offner's circuits also used AC-coupling. A useful technique that appears in this paper is the use of common-mode feedback to increase CM rejection (Fig. 4 of Reference 3: Franklin Offner, "Push-Pull Resistance Coupled Amplifiers," Review of Scientific Instruments, Vol. 8, January, 1937, pp. 20, 21. (AC-coupled, cascade differential amplifier circuits, including the use of the dual-triode common-cathodes biased to ground through a high resistance) ). To enable this, a CM sample from a downstream stage is fed back to an earlier stage. This feedback decreases the CM gain, and thus it improves CM rejection.

Otto Schmitt discussed a common-cathode, cathodes-to-common dual pentode circuit in 1937 (Reference 4: O. H. Schmitt, "A Simple Differential Amplifier," Review of Scientific Instruments, Vol. 8, No. 4, April, 1937, pp. 126, 127. (A basic common-cathode/cathode-to-common, dual-pentode differential amplifier, with plate-screen coupling, single-ended DC-coupling at the output) ). This circuit, while novel in the operation of the pentode screens, didn't minimize response to CM input signals (similar to the Matthews circuit, above).

In 1938 J. F. Toennies discussed what might be the first form of what has subsequently come to be known as the long-tailed pair (see Reference 5). In this form of differential input amplifier, the push-pull input signals are applied to the dual grids of the stage, and the common-cathodes are returned to a high negative voltage, through a high value common resistance. Toennies' fundamental circuit (Fig. 1 of Reference 5: J. F. Toennies, "Differential Amplifier," Review of Scientific Instruments, Vol. 9, March, 1938, pp. 95-97. (A differential amplifier circuit that uses a triode and pentode in a common-cathode configuration, biased to a negative voltage through a high resistance) ) used dual triodes with a plate supply of 135V, and a cathode bias supply of –90V.

The action of the large value cathode resistance biased to a high negative voltage acts to optimize the differential coupling of the stage, while at the same time minimizing the CM response, as noted in Figure H-2 below. This may intuitively be appreciated by considering the effect of the large cathode resistance to a high negative voltage -VS, as in B, versus the simple cathode-coupled pair as in A. In A the cathode resistance RK is returned to ground, the same point common to the grids (the return for the +VS supply).
A comparison of simple differential pair biasing in A (left) and longtailed biasing as in B (right)
Figure 2: A comparison of simple differential pair biasing in A (left) and longtailed biasing as in B (right)

The constant current action of the long-tailed biasing shown in H-2B tends to minimize response to CM inputs (while not impairing differential response). Later on, some more advanced designs were even to go as far as using a pentode tube for the "long tail" common-cathode bias, capitalizing on a pentode's high incremental resistance.

Otto Schmitt also discussed a long-tailed pair form of amplifier in 1938 (Reference 6: O. H. Schmitt, "Cathode Phase Inversion," Journal of Scientific Instruments, Vol. 15, No. 3, 1938, pp. 100, 101. (A circuit is presented as a single-input phase inverter, but is also a true DC-coupled, differential-input/differential-output amplifier topology. The dual triodes have their common cathodes biased to a negative voltage through a high resistance) ). The context of his discussion was not so much aimed towards optimizing CM rejection, but rather using such a stage as a phase inverter. With one input grid of such a stage grounded, and the opposite grid driven in single-ended fashion, out-of-phase signals result with equal plate load resistors. Schmitt was a prolific inventor, and was to return later on (below).

Lionel Jofeh, within UK patent 529,044 in 1939, offered a complete catalog of eight forms of cathode-coupled amplifiers (Reference 7: Lionel Jofeh, "Improvements in Electric Circuits Comprising Electronic Discharge devices," UK Patent 529,044, filed May 9, 1939, issued Nov. 13, 1940. (A catalog of the various forms of dual triode, dual pentode differential amplifier circuits, with variations in methods of cathode coupling) ).

Harold Goldberg presented a complete multi-stage, direct-coupled differential amplifier in 1940 (Reference 8: Harold Goldberg, "A High-Gain DC Amplifier for Bio-Electric Recording," Transactions AIEE, Vol. 59, January 1940, pp. 60-64. (A completely differential, multi-stage, direct-coupled, high gain amplifier for biomedical work. Illustrates the use of common-cathode bias resistance for reducing sensitivity to in-phase inputs) ). Using power pentodes within a unique low voltage differential input stage, Goldberg reported an equivalent input noise of 2μV for the circuit. This work parallels some of the earlier work mentioned above, apparently developed independently.

In 1941, Otto Schmitt published another work on the differential amplifier topic, going into some detail of analysis (Reference 9: O. H. Schmitt, "Cathode Phase Inversion," Review of Scientific Instruments, Vol. 12, No. 11, November 1941, pp. 548. (An expansion of previous work of the same name; includes a true DCcoupled, differential-input/differential-output amplifier topology, with the dual-triode common-cathodes biased to a negative voltage through a high resistance) ). In this work he clearly outlines the advantages of the long-tailed pair, in terms of the stage's phase-inversion properties. He also covers the case of a degenerated long-tailed pair, where a common cathode-cathode resistance is used for gain adjustment, and the individual cathodes are biased to a negative voltage with resistors of values twice that of a single cathode-coupled stage.

Walther Richter wrote on cathode follower and differential circuits in 1943 (Reference 10: Walther Richter, "Cathode Follower Circuits," Electronics, November 1943, pp. 112-117, 312. (Analysis of various cathode follower circuits, including the long-tailed differential pair) ). While primarily focused on single-ended cathode followers, this article also does an analysis of the long-tailed pair.
Harold Goldberg wrote again on his multi-stage differential amplifier, in 1944 (Reference 11: Harold Goldberg, "Bioelectric-Research Apparatus," Proceedings of IRE, Vol. 32, June 1944, pp. 330-336. (Another differential, multi-stage, direct-coupled, high gain amplifier for biomedical work. Illustrates the use of pentode bias resistance for long-tailed pair; uses battery operated front end) ). The 1944 version still used batteries for most of the power, but did add a pentode to supply the bias current of the first stage long-tailed pair.

Writing in 1944, G. Robert Mezger offered a differential amplifier design with a new method of interstage level-shift coupling (Reference 12: G. Robert Mezger, "A Stable Direct-Coupled Amplifier," Electronics, July, 1944, pp. 106-110, 352, 353. (A stable, high gain, differential-input/differential-output, direct-coupled amplifier. Uses novel active device inter-stage coupling technique) ). Previous designs had used either a resistive level shift like Nyquist, or the more recent glow-tube technique of Miller. Mezger's design used a 12J5 triode as the bottom level shift element, which acts as constant current source. Working against a fixed resistance at the top, this allows a wideband level shift. Good overall stability was reported in a design that used both differential and CM feedback. Regulation was used for plate and critical heater circuits.

Franklin Offner wrote a letter to the editor in 1945, expressing dissatisfaction with other differential amplifier authors (Reference 13: Franklin Offner, "Balanced Amplifiers," Proceedings of IRE, Vol. 33, March 1945, pp. 202. (A criticism of other differential amplifier circuit papers) ). In this work he comments on the work of Toennies (Reference 5, again), …"merely an application of in-phase degeneration by the use of a large cathode resistor,"… Offner also overlooked Blumlein's patent.

D. H. Parnum published a two-part survey of differential amplifier techniques, in 1945 (Reference 14: D. H. Parnum, "Biological Amplifiers, Parts 1 and 2," Wireless World, Nov. 1945, pp. 337-340, and Dec. 1945, pp. 373-376. (A survey of various differential amplifier techniques, with a multi-stage AC-coupled design example) ). This work analyzed some previously published designs, and presented two differential-throughout amplifier examples, both DC and AC-coupled.

In a comprehensive study of differential amplifier designs from 1947, Denis L. Johnston presented a three-part article on design techniques, with a finished design example (Reference 15: Denis L. Johnston, "Electro-Encephalograph Amplifier, Parts 1-3," Wireless Engineer, August 1947, pp. 231-242, Sept. 1947, pp. 271-277, and October 1947, pp. 292-297. (A comprehensive overview of various biological differential amplifier techniques with many example circuits, both DC and AC-coupled. Includes 61 references) ). This article is notable not only for the wealth of detailed information, but it also contains a bibliography of 61 references to related works.

The input stage of Johnston's design example amplifier used an input long-tailed pair based on the 6CS7 dual triode, with the cathode current supplied by a 6J7G pentode (Fig. 10a of Reference 15). The second stage was also a long-tailed pair, directly coupled to the first stage, with CM feedback. Multiple stages of supply regulation are used.

D. H. Parnum also published another work on differential amplifiers, in 1950 (Reference 16: D. H. Parnum, "Transmission Factor of Differential Amplifiers," Wireless Engineer, April 1950, pp. 125-129. (An analysis of the common-mode and differential signal response of differential amplifiers) ). In this paper he presented a critique of the input stage design of the Johnston design (Reference 15, again), pointing out necessary conditions for optimizing CM rejection for multiple stage amplifiers.

The P. O. Bishop and E. J. Harris design paper of 1950 is similar in overall scope to the Johnston work noted above (Reference 17: P. O. Bishop, E. J. Harris, "A DC Amplifier for Biological Application," Review of Scientific Instruments, Vol. 21, No. 4, April 1950, pp. 366-377. (A very sophisticated differential amplifier system, employing common-mode feedback, active differential pair current sources, input shield bootstrapping, power supply stabilization, etc. Many references cited.) ). It reviews the work of many other designers in the biological area, and presents a sophisticated example design. In this circuit (Fig. 3 of Reference 17) a 954 pentode pair is used for input athode followers, driving a 6J6 dual triode long-tailed pair. Both the input stages as well as the next two stages used 12SH7 pentodes for the tail current sources. Highly stabilized power supplies are used for the plate supplies, with critical heaters also stabilized.

Richard McFee published some modifications useful to improve the CM rejection of a single dual triode stage, in 1950 (Reference 18: Richard McFee, "Improving Differential Amplifier Rejection Ratio," Review of Scientific Instruments, Vol. 21, No. 8, August 1950, pp. 770-771. (A feedback modification to a standard dual triode differential stage to improve common-mode rejection) ).

One of the better overview papers for this body of work appeared in 1950, authored by Harry Grundfest (Reference 19: Harry Grundfest, "Biological Requirements for the Design of Amplifiers," Proceedings of the IRE, Vol. 38, September 1950, pp. 1018-1028. (An overview of amplifier requirements for biological measurements; includes commentary on the history of differential amplifier development) ). This paper also gives greater insight into how the biological amplifiers were being used at this time, and offers many references to other differential amplifier work.

It is notable that Grundfest credits Offner (Reference 3, again) with the invention of the long-tailed pair. However, it can be argued that it isn't apparent from Offner's schematics that a true long-tailed pair is actually being used (there being no negative supply for the cathode resistance). The type of biasing that Offner (and Blumlein) use is a simple resistor from the common cathodes to circuit common, which would typically have just a few volts of bias across it, and, more importantly, would have a value roughly comparable in magnitude to the cathode impedance.

Unfortunately, Grundfest also overlooks Blumlein (Reference 2, again), who preceded Offner with a similar circuit. This similarity is apparent if one compares Blumlein's Fig. 2 against Offner's Fig. 3, in terms of how the biasing is established.

One of the deepest technical discussions on the topic of DC differential amplifiers can be found in C. M.Verhagen's paper of 1953 (Reference 20: C. M. Verhagen, "A Survey of the Limits in D.C. Amplification," Proceedings of the IRE, Vol. 41, May 1953, pp. 615-630. (A technical discussion of both tube and circuit parameters which impact DC stability) ). Verhagen goes into the electron physics of the vacuum tube itself, as well as the detailed circuitry around it, as to how they both effect stability of operation. This paper includes detailed mathematical expressions and critiques of prior work. Many other topical papers are referenced, including some of those above.

The above discussion is meant as a prefacing overview of DC differential amplifiers, as this technology may impact op amp designs. It isn't totally comprehensive, so there are likely other useful papers on the topic. Nevertheless, this discussion should serve to orient readers on many of the general design practices for stable DC differential amplifiers.
Share this article :


Post a Comment

Please wait for approval of your comment .......