This page presents notes on some hobby-radiocommunication explorations at amateur radio W9BRD.

20251012: The 1936 ARRL Radio Amateur's Handbook included in its Receiver Construction chapter an odd regenerative-receiver design (Figure 1) based on the use of a 6F7 "dual-purpose" vacuum tube — a low-mu triode and a remote-cutoff, small-signal-RF-amplifier pentode in one envelope, sharing a common cathode — and a 41 power pentode (which in its octal version would later be known as the 6K6GT and in its 7-pin-miniature version would even later be known as the 6AR5) as an AF voltage amplifier. I recently constructed a good-performing varation on that design, using Hartley feedback instead of tickler-coil ("Armstrong") feedbck in its detector, and coupling antenna-system energy in at the detector cathode instead of via a link on the detector-tuned-circuit coil.

Figure 1—This "two-tube" receiver design from the ARRL's 1936 Radio Amateur's Handbook actually uses three stages: a 6F7 pentode as a regenerative detector with Armstrong (ticker coil) feedback; a 6F7 triode as first AF amplifier; and a 41 power pentode as an AF voltage amplifier. (Yes, this schematic of what is actually a simple circuit is badly laid out such that determining signal flow and stage function is a chore. That the original used component designators—cryptic, gotta-keep-referring-to-the-parts-list IDs like C2 and R8 instead of the actual capacitance and inductance values shown—made the jumble even harder to parse. My version, below, is easier to understand.)

Figure 2—The "BRDized" 6F7–41 receiver reconfigures the detector as a Hartley; couples antenna-system energy to the detector cathode; and decouples the 6F7 and 41 cathodes for dc. Oh, and I use a resistive plate load for the detector—150 kilohms—rather than the unobtainium-in-2025-and-never-actually-necessary 1080-H choke specified in the Handbook writeup of the circuit, and a resistive plate load for the 41 instead of the 30-H choke.

I used the Hartley configuration instead of the original's tickler-coil feedback so I could use my existing BRD-80 regenerative receiver, with its compact, already-tunes-80-meters-as-I-want-it-to tuning head, as a basis for trying the 6F7–41 circuit rapidly. Because the BRD-80 receiver has space for only two tube sockets, I had to omit the grounded-grid-triode RF buffer amplifier I usually use between antenna system and detector. Instead, the Figure 2) circuit uses mismatch loss across a 20-pF capacitance to provide some passive buffering.

The circuit worked well from first power-up but was initially strongly microphonic, easily going into speaker-driven audio oscillation if I turned up the station amplifier/filter-box gain too high with the station speaker in use. The much-higher-than-necessary AF gain of the 6F7-triode/41-pentode pair was at least part of the basis of this, of course. The tacked-on 220-kilohm R from the 41 control grid to common in Figure 2 was my initial response to the problem, raising the onset of feedback to a higher gain-setting level by lowering net gain of the set's two-tube AF strip. But then I tried a different 6F7—one made by RCA instead of the Sylvania part I used first—and the set's tendency for microphonics-driven AF feedback vanished.

20251015: Note in Figure 2 above the use of a 41 power pentode as a voltage amplifier rather than a power amplifier—an approach taken, I assume, because a screen-grid tube intended only for small-signal amplification was found to be incapable of (sufficiently) driving headphones directly. The 41 works so well in that role—in my case, as a voltage amplifier with more gain than that achievable with one of the relatively low-gain pre-octal triodes—that I tried the same idea in the last stage of my 6A7 converter|37 RF buffer|36 Hartley detector|37 AF amplifier superhet, substituting a 38 power pentode for the design's second 37 triode will similarly good results. The only change relative to the Figure 2 implementation was to cathode-bias the 38 with 1100 ohms instead of the 680-ohm R used with the 41.

20251017: But back to evaluating the behavior of the Figure 1 circuit and my Figure 2 variation thereon: I'm nearly convinced that there's something wrong with the design or manufacture of the 6F7 tube itself. Our first clue to this is the strange participation of the 41's cathode current in biasing the 6F7 (Figure 3):

Figure 3—From our first look at the 6F7–41 circuit, we notice the strange participation of the 41's cathode current in biasing the 6F7. The Handbook explanation of this arrangement superficially satisfies—until we experience how the circuit actually behaves in practice.

From the circuit's Handbook writeup:

It will be noticed that the d.c. grid return of the 6F7 pentode section is to the cathode terminal of the socket. This is necessary to provide bias for the triode (audio) section of the tube, derived from R2 [the 50-ohm resistor in Figure 3]. The bias for the 41 is taken through this resistor, as well, in order to hold the 6F7 bias fairly constant despite the cathode current variation in that tube due to the plate current changes with detection.

Neglecting for now the fact that the above does not fully describe for those not well-versed in vacuum-tube biasing how the Figure 3 arrangement operates to bias the 6F7 triode and pentode sections differently, might some practical, behavioral issue have driven the inclusion of that cross-biasing and the hand-waving-class description of its necessity and operation? What possible defect, perhaps an unwelcome discovery during initial operation of the circuit, might have suggested to its originator that such elaboration was necessary? ("Plate current changes with detection" are, in effect, merely a form of noise—so why not just add across the 0.01-uF bypass cap on the cathode line an AF-appropriate electrolytic bypass cap to shunt that noise to common, stabilize the 6F7-triode bias, and keep the tubes' cathode circuitry separated?)

As soon as my initial "Cool—it works" impression wore off—after I first solved the set's microphony problem reducing the AF-strip gain and then finding that only my Sylvania-made 6F7 was microphonic and my RCA 6F7s weren't such that I switched over to using an RCA 6F7"as I kept fiddling with my operating Figure 2 implementation, I encountered a problem: The 6F7 triode was noisy—continuous-frying/fizzing/popping-class noisy, noise that was all-too-noticeable even beneath and within my antenna system's high noise floor at 80 meters. All three of my RCA-made 6F7s exhibited the problem; the overly-microphonic Sylvania 6F7 stayed in its box.

20251019: Multiple experiments proved that the noise was confined to the 6F7 triode section. With AF drive from the pentode detector disconnected from the triode, the triode was still noisy—even with the set's REGENERATION control turned all the way down (pretty much factoring out detector-cathode-current variations as the noise source). The noise continued unchanged with the triode's control grid shorted to common. Shorting the 6F7 cathode to common; biasing the triode with a 1.5-V dry cell; powering the 6F7 cathode from a floating (unconnected to common) 6.3-V ac supply—no change in the noise. I briefly rebuilt the circuit with a 6AZ8 subbed for the 6F7—noise gone. I briefly bolted one ear of an octal socket to the receiver tin and flying-wires-connected in a 6C5, its cathode connected to the 6F7 cathode and its heater in series with the 6F7 heater for heating from the set's 13.5-ish-V dc heater supply, as a substitute for the 6F7 triode. Noise gone. And feeding the 41 AF amplifier directly, the 6F7's remote-cutoff pentode operating as regenerative detector proved itself to be frying/fizzing/popping-noise-free.

(At about this point, having proven both that the 41 as voltage amplifier worked well and then some, and that the detector was usable even with direct antenna-system feed via that 20-pF capacitance to its cathode, I removed the 41, replaced its 6-pin socket with an octal socket, and reinstalled my standard dual-triode-based one-triode-as-RF-buffer-and-the-other-as-first-AF-amp circuitry—this time with a 12AH7GT instead of a 6C8G or 12SC7.)

20251020: With the circuit's more-gain-than-I-really-need 41 pentode replaced by one of the 12AH7GT's mu = 19 triodes, I decided to re-try my one addtional 6F7: the overly-microphonic Sylvania part that I used when first trying the circuit. Frying/fizzing/popping gone!

20251022: So why not also try going back to the 6F7–41 lineup, this time with a few tweaks based on lessons learned with the Figure 2 lineup?

Figure 4—A few revisions (red values) to the Figure 2 6F7–41 circuit make it work even better. (Blue changes were incidental to experiments I did in the meantime; I left them in place. The 72-V line was regulated from the beginning, but I call out its regulatedness here to emphasize the importance of powering the detector from a regulated source for better stability.) Reducing the antenna-coupling C from 20 pF to 10 pF still allows my antenna system's noise floor to override the detector's noise floor at 80 meters while further isolating the detector from the antenna system as load. Redistributing the reduction of the design's AF gain (red Rs associated with the AF-amplifier stages) practically banishes speaker-driven AF oscillation that easily occurred with the Figure 2 circuit. (And maybe happening to use a different, possibly less microphonic 41, which this time around is now enclosed in a large shield can, helped.)

So now some detail on how the 6F7's pentode and triode are actually biased. Yes, from the ARRL Handbook writeup on the Figure 1 circuit, we do have

It will be noticed that the d.c. grid return of the 6F7 pentode section is to the cathode terminal of the socket. This is necessary to provide bias for the triode (audio) section of the tube, derived from R2 [the 50-ohm resistor in Figure 3]. The bias for the 41 is taken through this resistor, as well, in order to hold the 6F7 bias fairly constant despite the cathode current variation in that tube due to the plate current changes with detection.

but the second sentence of that passage is actually a non sequitur, as the point of connecting the 6F7 pentode's grid return to the tube's common cathode is to make the pentode's grid-leak-based bias independent of the triode's cathode-resistor-based bias—not to make the triode's bias independent from the pentode!

Control-grid bias is a relative voltages proposition: To bias a grid negatively, we can either connect it to a source of negative voltage such that it's more negative than the cathode, or we can make the cathode more positive than the grid, thereby making the grid relatively more negative than the cathode. The most common means of achieving such cathode bias is to insert a resistance between the cathode and common such that current flow through the tube causes a voltage drop across the R, thereby raising the cathode above common and making it more positive than the common-referenced grid.

The Figure 3 biasing circuit looked strange to me from the first time I saw it, and it looked even stranger when I read the Handbook's unsatisfactory explanation of the reason for its inclusion and its supposed operation. Really: If cathode-current pulses from the 6F7 pentode—which will not be strong, considering the low plate and screen voltages at which the pentode operates—somehow did cause noise or distortion traceable to same in the output of the Figure 1 circuit, the fix would be simple: Just connect several tens of microfarads in parallel with the bottom-of-the-tuned-circuit 0.01-uF RF-bypass cap and the 50-ohm bias R to shunt that noise to common! And while we're at it, modify both circuits such that the 6F7 has its own triode-appropriate, suitably-bypassed cathode-bias R (2.2 kilohms will work fine, also, in effect, making a tens-of-microfarads AF bypass across it work even better), and then connect the common end of the 41's cathode-bias R to common as it should be. The voltage drop across that 6F7-stage triode-appropriate R will be higher than "triode-caused only" because the pentode's combined plate and screen currents contribute to the drop, but, most critically, the voltage drop across any 6F7-stage cathode-bias R present in Figure 1 and Figure 3 circuits won't affect the pentode's bias because its grid-bias R is returned directly to the cathode. Because the pentode's grid R is returned directly to the cathode, "above" our triode-appropriate cathode-bias R, as it were, any voltage drop across that R merely subtracts from the pentode's plate and screen voltages by raising their reference above common. The pentode's bias is affected by neither the 50-ohm R in Figure 1 nor the 2.2-kilohm R in Figure 2.

In practice, the Figure 2 circuit (as further evolved to the Figure 4 circuit) works fine. Whatever the real or imagined resultants of "cathode current variation in [the 6F7] due to the plate current changes with detection" may have been, I haven't heard any across many days of using the Figure 4 circuit for listening and for two-way Morse communication. (Along the way I did further modify the set by replacing the 41 with a 38 for more-straightforward seriesing of the tubes' heaters across 13.5 V regulated dc, and to reserve the 41 for use with my RAL-5 Navy regen.)

It occurs to me to question why the 6F7–41 circuit was included in the ARRL Handbook at all. From experience I and many other hams with regenerative-receiver experience know that a screen-grid detector followed by just a medium-mu triode's worth of AF amplification provides more than enough gain for basic, practical reception even for two-way work, on lower-frequency ham bands. Here the 6F7 unfortunately cannot quite serve sufficiently well on its own because of the miserably low mu (8) and transconductance (0.5 mS) of its triode section—hence the need for one more stage of audio amplification. So why use one? I suspect that free ARRL Lab tube stock provided by a manufacturer—likely RCA, at least—drove "see what you can do with it"-class exploration. Whatever, the 6F7–41 "advanced" receiver appeared only in the 1936 ARRL Handbook and did not return in the 1937 edition.