Insufficient minimum potentials are indicated at highly negative grid voltages by the presence of secondary-emission effects but they have no harmful consequences inasmuch as this section of the plate characteristic is not utilized in practical operating conditions.—O. H. Shade, "Beam Power Tubes," Publication ST-59 (RCA: Harrison, New Jersey, March 1938), pages 168–169.

Incidentally, the author writes in connection with the July article that 6L6's have been found to be unsuitable for the push-pull-push circuit. Although satisfactory as a straight push-pull oscillator, when connected p-p-p the principle result is a vicious parasitic oscillation causing a high circulating current in the crystal and holder and giving practically no second-harmonic output. This is in reply to the numerous questions about using the tubes in the circuit.—Editor's footnote to J. Stanley Brown, W3EHE, "Some Trick Crystal Circuits," QST, September 1936, pages 19–21.

A condition is frequently encountered in practice when an 807 will appear perfectly stable when running loaded with uninterrupted excitation and without modulation and yet will develop widespread clicks when the transmitter is keyed, or splatter when the amplifier is modulated. We have seen more than one commercial installation with heavy loading resistors across the tanks and they weren't put there to increase tank-circuit efficiency!—Don Mix, W1TS, "Amplifier Instability in Transmitters," QST. June 1948, pages 19–24 and 110.

Although parasitic oscillations can and do occur in triode, true-tetrode and true-pentode RF power amplifiers, parasitic-oscillation troubles and their nostrums became commonplace as amateur radio experimenters went to work with the 807, the first among many RF-specialized beam power tubes—screen-grid tetrodes designed to exhibit pentodelike characteristics. According to multiple anecdotal accounts I've encountered, parasitic-oscillation problems were even more prevalent with the circa-1950 5763/6417, a miniature beam-power design for which, unusually and I think significantly, audio/linear amplifier ratings were never promulgated.

This page will outline my theory that significant negative-resistance effects endemic to first-generation beam-power-tube topology can set up conditions of one-port oscillation readiness that are neither explained nor predicted by such two-port characteristics as transconductance and feedback capacitance; that RCA's strangely constrained 5763 specs were yet another exercise in positively spinning a flawed design; and that recognizing that design's tendencies for harboring one-port, negative-resistance oscillations allows us to intelligently apply stabilization measures that give acceptable practical stability without resorting to the ill-understood, hit-and-miss procedures described in decades of amateur and professional literature.

Many 6CL6s I have used, all the way from audio to six meters, and nary a whimper or any taking-off-unbidden. Not so with the 5763. Physicist friend who long ago forgot more than I'll ever know about thermionic valve design said to me once, "Look at the RCA data on a 5763. It'll tell you the tube was designed as an oscillator-multiplier. And if you give it half a chance, it'll oscillate or multiply."—John McAulay, Glowbugs mail reflector, May 20, 1997

Talking Points

Negative resistance in tetrodes: A look at the eye-opening characteristic curves of the 24A tetrode.

The dynatron oscillator vogue: For a short year or three, radio amateurs of the early 1930s turned the tetrode's negative-resistance bug into a feature.

Pentode characteristics: How characteristic curves show that a true suppressor grid really does suppress secondary-emission effects.

807 characteristics: As Shade indicated was true of the 6L6's second-emission-suppression design, negative resistance at low plate voltages and high negative control-grid biases is evident in the 807's published plate curves. And that's the problem with the 807: The 6L6 was designed for Class A and AB1 operation in which the grid never goes highly negative and the plate voltage does not drop far below that of the screen, and the 807, obviously just the 6L6 beam-suppression design with better insulation, stem shielding and a top-cap plate connection, would, when used for Class C amplification with its screen voltage supplied via a voltage-dropping resistor from the plate supply, routinely pass through or be ongoingly operated under exactly the conditions Shade reported would result in secondary emission in the 6L6!

We decided on the 2E26 because it is small, can be run at the 25 or 30 watts input that was wanted, and seems to be less prone to parasitic oscillations than its larger cousin. If some kind tube manufacturer ever brings out an 807 with short leads, it will be logical tube for this unit, but we played it safe for the present and used the 2E26.—Byron Goodman, W1DX, "A 1950 VFO Exciter," QST. September 1949, pages 29–33.

A survey of parasitic-oscillation lore and remedies: A survey of the literature of parasitic-oscillation suppression reveals increasing confusion over possible oscillation modes once beam power tubes—aligned-grid tetrodes—made the scene.

Bugs. Bacause an 807 has such a high transconductance and because it is almost impossible to get perfectly isolated return circuits to the cathode, a tendency toward instability may be encountered if the mechanical layout illustrated is not followed in every detail. But unless the design is particularly bad, the instability will be apparent only when both the load and the excitation are removed from the 807 stage. Because the instability will disappear when either excitation or load is applied to the 807, it should not be bothersome. However, if you so desire, the 807 stage can be made stable as the rock of Gibraltar at a small sacrifice in the 10- and 20-meter output by placing 50-ohm 1/2-watt carbon resistors in both the control grid and screen grid leads right at the tube socket (before connection to anything else). This method of taming 807's was suggested by W6BHO and works in all but the most stubborn cases.—W. W. Smith, W6BCX, "The '4-25' Exciter," Radio, December 1938, pages 11–16

Parasitics with Oscillator Keying. When keying in the crystal stage, or, for that matter, any stage ahead of the final amplifier, the stages following the keyed stage must be absolutely stable so that parasitics or output frequency oscillation will not occur when the excitation is rising on the beginning of each keying impulse. This type of oscillation gives rise to extremely offensive clicks which cannot be eliminated by any type of filter; in fact, a filter designed to slow up the rate at which signal comes to full strength may only make them worse.—W. W. Smith, W6BCX, and others, editors, The Radio Handbook, 10th edition (1946), Chapter 7,"Transmitter Theory," page 186

A remedy which works in most cases is to insert a small 50-ohm resistor between the screen terminal and the by-pass capacitor.—Donald Mix, W1TS, "Unstable Signals," QST, August 1946, pages 23–26, 126

The most common way of suppressing v.h.f. parasitics with tetrodes has been the use of a small choke in the grid lead in conjuction with a small un-bypassed noninductive resistor at the screen [between the screen terminal and the screen bypass capacitor]. While this combination rarely fails to suppress the parasitic, it has been found that even a small amount of resistance—as low as 10 or 12 ohms—used in this manner has a very serious effect upon the plate-grid isolation at the operating frequency....Don't use a suppressor resistor at the screen.—Donald Mix, W1TS, "Amplifier Instability in Transmitters," QST, June 1948, pages 19–24 and 110.

In this passage, the manufacturer-promulgated "no neutralization required" myth drives a fallacy of the excluded middle. Aren't we after a stage that is unconditionally stable? So after hitting upon a form of parasitic suppression that "rarely fails to suppress the parasitic" only to encounter "a very serious effect upon the grid-plate isolation at the operating frequency," why isn't our next natural step to go on to neutralize our now-parasitic-free amplifier to achieve a parasiticless stage that also doesn't self-oscillate or regenerate at the signal frequency? (To be fair, hamdom was at least two years away learning from learning via several venues about an all-capacitive neutralization approach—nowadays standard—that dealt more easily and effectively than "plate neutralization" with the very small [relative to its equivalent in triodes] grid-to-plate capacitance of beam power tubes and pentodes.)

Due to their relatively high slope these valves are prone to parasitic oscillation and it is advised that a small resistance on the order of 47 ohms, or less if essential, should be wired in series with the plate, directly connected to the top cap.—"Typical Operation at Radio Frequencies: General Recommendations," 807 Beam Tetrode Application Report, Standard Telephone and Cables Pty. Ltd., Sydney, Australia, 1954.

One cure for [low-frequency parasitic oscillations] is to change the type of choke in either the plate or the grid circuit....If...the stage under investigation uses a beam-tetrode tube, negative resistance can exist in the screen circuit of such tubes. Try larger and smaller screen by-pass capacitors to determine whether or not they have any effect. If the condition is coming from the screen circuit an audio choke with a resistor across it in series with the screen feed lead will often eliminate the trouble....It may be said in general that the presence of low-frequency parasitics indicates that somewhere in the oscillating circuit there is an impedance which is high at a frequency in the upper audio or low-r-f range. This impedance may include one or more r-f chokes of the conventional variety, power-supply chokes, modulation components, or the high impedance may be presented simply by an RC circuit such as might be found in the screen-feed circuit of a beam-tetrode amplifier stage.
      Beam-tetrode stages, particularly those using 807-type tubes, will almost invariably have one or more v-h-f parasitic oscillations unless adequate precautions are taken in advance....These considerations involve the facts that a beam-tetrode amplifier stage has a greater power sensitivity than an equivalent triode amplifier, such a stage has a certain amount of screen inductance which may give rise to trouble, and such stages have a small amount of feedback capacitance....Basically, parasitic oscillations in beam-tetrode amplifier stages fall into two classes: cathode-grid-screen oscillations, and cathode-screen-plate. Both of these types of oscillation can be eliminated through the use of a [parallel RL] parasitic suppressor between the screen terminal of the tube and the screen by-pass capacitor. Such a suppressor has a negligible effect on the screen at the operating frequency....The grid-screen oscillations may occasionally be eliminated through the use of a [parallel RL] parasitic supressor in series with the grid lead of the tube. The screen-plate oscillations may also be eliminated by inclusion of a [parallel RL] parasitic suppressor in series with the plate lead of the tube. A suitable grid suppressor may be made of a 22-ohm 2-watt Ohmite or Allen-Bradley resistor wound with 8 turns of no. 18 enamelled wire....The parasitic suppressor of the plate circuit of a small tube such as the 5763, 2E26, 807, 6146 or similar type may consist of a 47-ohm carbon resistor of a 2-watt size with 6 turns of no. 18 enamelled wire wound around the resistor.
      ....A series of rapid dots should be sent, and the frequency spectrum for several megacycles each side of the carrier frequency carefully searched. If any vestige of parasitic is left, it will show up as an occasional "pop" on a keyed dot. This "pop" may be enhanced by a slight detuning of either the grid or plate circuit....If the parasitic shows up, it means that the stage is still not stable, and further measures must be applied to the circuit. Parasitic suppressors may be needed in both screen and grid leads of a tetrode, or perhaps in both grid and neutralizing leads of a triode stage. As a last resort, a 10,000-ohm 25-watt wirewound resistor may be shunted across the grid coil, or grid-tuning condenser of a high-powered stage. This strategy removed a keying pop that showed up in a commercial transmitter operating at a plate voltage of 5000.—William I. Orr, W6SAI, editor, The Radio Handbook, 14th edition, 1956, pages 295–300.

"An occasional 'pop' on a keyed dot" is almost certainly not a VHF-parasitic phenomenon, but a low-frequency or audio effect if it's heard along with the wanted signal at the frequency of the wanted signal.

It is recommended that a 100-ohm resistor be connected in series with grid No. 2, as close as possible to the socket, to prevent the generation of parasitic oscillations.—"Operational Considerations," 6939 twin pentode, RCA Transmitting Tubes, TT-5 edition.

Simple methods of suppressing v.h.f. parasitics are available and one or other of them should always be used as a safeguard. The most commonly adopted method is the inclusion of "stopper" resistances in the leads to the valve electrodes. These are connected directly at the valve socket or at the cap of the valve. These resistances should be of carbon and should not have a higher value than is necessary to stop the parasitic oscillations; suitable values are 10–100 ohms for [the grid stopper] and 10–22 ohms for [the plate stopper]. The stopper resistance in the screen circuit is usually of about 50 ohms and is included because there is often quite a large mutual conductance between the control grid and screen, with the consequent possibility of v.h.f. parasitics being generated in the screen circuit. The presence of this resistance, while preventing the screen from being properly earthed at the carrier frequency, is unlikely to make neutralization necessary. If difficulty is in fact encountered, the use of two or more bypass capacitors of different values in parallel (e.g. 0·01 μF and 100 pF) to give a low impedance over both h.f. and v.h.f ranges may be tried as an alternative to the use of the screen stopper.—The Radio Society of Great Britain Amateur Radio Handbook, Third Edition (1961), pages 186–187.

A reality-check data-point on the losses involved with plate and screen stoppers: In a stable, well-neutralized, parasitic-free 1631 (12.6-V 6L6) class C vacuum-tube amplifier operating at 3.55 MHz, adding a 47-ohm plate stopper reduced the output power from 7.98 W to 7.69 W—that is, by 0.16 dB. (I had to retune the stage slightly to remaximize its output after adding the resistor.) Relative to the original figure of 7.98 W, operating the stage with a 100-ohm screen-grid stopper and the 47-ohm plate stopper reduced the output power to 7.76 W, a reduction of only 0.12 dB. Recognizing that adding the screen stopper might affect the stage's neutralization, I determined that no instability could be detected, even when keying the entire transmitter without drive and varying the output stage's TUNING capacitor through its full range. At 7 MHz, the stage's output with plate and screen stoppers installed was 7.12 W, fully in keeping with my earlier characterization of the rig's output as "8 W on 80 and 7 W on 40."

What the physicist saw: Zeroing in on the 5763 plate curves—and the plate curves of a miniature two-beam-tubes-in-one-envelope RCA twin-"pentode" design so prone to screen-negative-resistance instability problems that even its one-paragraph description in the fifth edition of the RCA Transmitting Tubes Manual recommends that a 100-ohm resistor be used between the tube's screen terminal and its bypass capacitor.

Snivets: How parasitic oscillations in beam-power-tube electron-beam-deflection amplifiers caused a singular picture defect in early post-World-War-2 television receivers; an RCA publication zeroes in on load-line hysteresis associated with snivets; and how the latter-day built-in-beam-plates-to-cathode-diode design of the 6JQ6 sought to tame both.

Another theory holds that snivets interference is caused by a form of Barkhausen oscillation. This theory is logical because the plate voltage swings appreciably below the screen-grid (grid no. 2) voltage in many receivers. This condition is especially severe in modern flyback transformer designs which drive the plate voltage as far into the knee region as possible. An example of the load line of a horizontal-deflection tube illustrates this phenomenon quite well. The most familiar load line to most engineers is that drawn for resistance-coupled amplifiers, which is simply a straight line. Figure 1 shows the load line of a typical deflection tube in a television receiver which exhibited strong snivets. Is it any wonder that interference resulted?—from W. E. Babcock, "Circuit Troubles Caused by Unusual Tube Effects," Electron Tube Design (RCA: Harrison, New Jersey, 1962), pages 736–742.

Figure 1—Load-line graph for a 25CD6 beam power tube operating as a horizontal deflection amplifier. Wrote the RCA's W. E. Babcock of this presentation, "Is it any wonder that interference resulted?"

Current Division in Region A—Special precautions must be taken to ensure current stability in region A (Figure 2) to permit tube operation where a portion of the plate-voltage swing is below screen-grid potential. Regeneration resulting from a negative slope in the plate current can cause extraneous oscillations. [The author footnotes Schade's "Beam Power Tubes," Proceedings of the Institute of Radio Engineers, Volume 26, Number 2, pages 137"181—the source of the quote at the beginning of this page—here.] Current division in Region B—All tetrode structures exhibit stable current characteristics in this region.—from J. W. Gaylord, "Some Factors Affecting the Design and Application of Small Power Tubes," pages 743–762.

Figure 2—Graph of typical tetrode current characteristics from J. W. Gaylord, "Some Factors Affecting the Design and Application of Small Power Tubes." Hysteresis in the plate and screen curves indicates that both elements exhibit negative resistance in region A. When such intrinsic negative resistance is present, connecting either element to a means of reactive energy storage—lead inductances, internal and external, in conjunction with external reactances that provide a path to common, are sufficient—can result in oscillation.

Why neutralization can't stop parasitics caused by intrinsic negative resistance: Intrinsic negative resistance is a one-port effect that requires only a two-terminal connection to the negative resistance of some form of sufficiently low-loss reactive energy storage. An inductor, a capacitor, a series or parallel combination of inductance and capacitance, or a sufficiently low-loss distributed (transmission-line) equivalent of any of these, will do.

If a pair of phones is connected in series with the plate of the dynatron, it will oscillate at a high audio frequency. This frequency may be lowered to three cycles per second simply by increasing the inductance of the circuit and inserting capacitance in parallel with the phones.—O. P. Susmeyan, W1BLH, "Experiments with Dynatron Oscillators," QST, September 1930, pages 33–35.

My own crazy-oscillations-in-a-final-amplifier story: An encounter with a 10JA5 Class C stage that oscillated so strongly at AF and RF that I could hear the parasitics acoustically—likely as a result of bypass and/or blocking capacitors acting as piezoelectric speakers.

Eye-opening research: A college student's electrical engineering honors thesis provides eye-opening graphs depicting negative resistance in the 5763, 6146, and other types. Writes she of the 5763, "Interestingly, in this tube I was not able to turn the secondary emission effect off. Grounding the third grid only succeeds in reducing the effect, not eliminating it....Apparently, for the typical applications of the 5763 tube, secondary emission does not adversely affect the operation."

How users of modern power tetrodes do it right: Modern power tetrodes do not include beam-forming plates and are applied with much more attention given to supply and management of screen voltage and negative resistance. From the datasheet for the Svetlana 4CX7500A tetrode:

The screen current may reverse under certain conditions and produce negative indications on the screen current meter. This is a normal characteristic of most tetrodes. The screen power supply should be designed with this characteristic in mind, so that the correct operating voltage will be maintained on the screen under all conditions. A current path from the screen to cathode must be provided by a bleeder resistor or a shunt regulator connected between screen and cathode and arranged to pass approximately 10% of the average screen current per connected tube. A series regulated power supply can be used only when an adequate bleeder resistor is provided.

—a passage very similar to wording in Ian White, G3SEK, "Power and Protection for Modern Tetrodes," QEX, October 1997, page 15. Might we be able to tame negative-resistance-related parasitics in first-generation beam-power tetrodes by applying today's shunt screen-voltage-regulation methods?

Beam-power-tube creep: Some tubes promoted as, or which originally appeared as, true pentodes always were, or have morphed into, or are or were sometimes produced as, beam-power tubes, including such suspects as the 6CA7/EL34, 6GK6, 6GW8/ECL82 pentode section, and more. The answer, my friend, is blowin' in their characteristic curves—short of examining their internals through destructive inspection.

Beam-power-tube avoidance as performance art: If you're hobby-experimenting with screen-grid tubes for Class C amplifiers that operate at relatively low powers, you too can escape the dubious tyranny of the beam power tube. Here are some suggestions for true pentodes capable of output powers from 5 to 50 watts. Think 1613, 38, 41, 42, 47, 59, 6AG7, 6AK6, 6AR5, 6F6, 6G6, 6K6, 802, 804, 837, and 89.

References:

807 Beam Tetrode Application Report, Standard Telephone and Cables Pty. Ltd., Sydney, Australia, 1954.

The Amateur Radio Handbook, Third Edition (1961), Radio Society of Great Britain.

W. E. Babcock, "Circuit Troubles Caused by Unusual Tube Effects," Electron Tube Design (RCA: Harrison, New Jersey, 1962), pages 736–742.

Byron Goodman, W1DX, "A 1950 VFO Exciter," QST. September 1949, pages 29–33.

J. W. Gaylord, "Some Factors Affecting the Design and Application of Small Power Tubes," pages 743–762.

David D. Meacham, W6EMD, "Understanding Tetrode Screen Current," QST, July 1961, pages 26–29.

Donald Mix, W1TS, "Unstable Signals," QST, August 1946, pages 23–26 and 126.

Don Mix, W1TS, "Amplifier Instability in Transmitters," QST. June 1948, pages 19–24 and 110.

William I. Orr, W6SAI, editor, The Radio Handbook, 14th edition, 1956, pages 295–300.

O. H. Shade, "Beam Power Tubes," Publication ST-59 (RCA: Harrison, New Jersey, March 1938).

W. W. Smith, W6BCX, "The '4-25' Exciter," Radio, December 1938, pages 11–16.

W. W. Smith, W6BCX, and others, editors, The Radio Handbook, 10th edition (1946), Chapter 7,"Tansmitter Theory," page 186.

O. P. Susmeyan, W1BLH, "Experiments with Dynatron Oscillators," QST, September 1930, pages 33–35.

RCA Transmitting Tubes, Technical Manual TT-5, Antique Electronic Supply reprint.

Ian White, G3SEK, "Power and Protection for Modern Tetrodes," QEX, October 1997, page 15.