4x 6AG7 Amplifier

Discussion in '"Boat Anchor" & Classic Equipment' started by NW2K, Feb 11, 2019.

ad: L-HROutlet
ad: l-rl
ad: l-BCInc
ad: Left-3
ad: L-Geochron
ad: L-MFJ
ad: Left-2
ad: abrind-2
  1. AA5CT

    AA5CT Ham Member QRZ Page

    Show the need for other than a Class AB "linear" for amateur service. Isn't that what we're talking about here? All these 'amplifiers' we run are after-market add-ons to rigs that run (do I have to put 'predominantly' here) linear modes. Even CW modes benefits from a bit of 'envelope shaping' to reduce keying sidebands, thereby requiring some linearity in an amplifier.

    Apparently, there isn't a need for Class C cathode-drive amplifiers; cathode-driven also has inherently have lower gain than grid-driven designs, thereby requiring perhaps another stage in the 'chain'.. The older tube Class C operation amplifiers I've seen for commercial 2-way radio service run push-pull neutralized PA stages at that. One thing- it is easy peasy to incorporate neutralization with P-P grid-driven designs.

    I don't know what direction you want to take this; I've got a more pragmatic view, one driven from an engineering point of view.
  2. K9STH

    K9STH Ham Member QRZ Page


    There is still the cathode as well as filament / heater parameters.

    Improper operation can "strip" a cathode resulting in reduced emissions which will greatly affect operation.

    Glen, K9STH
  3. W9BRD

    W9BRD Ham Member QRZ Page

    Very good, and yes. Forgetting plate, screen, and grid issues--including possible leakage or flashover between pins or internal elements as a result of overvoltage--the primary limiting factor in a design as stupid as this CQ mistake is the cathode emission of the tubes. The cathode's emission capacity is what we "use up" across the lifetime of a tube, even if we don't overload it.

    Operating the tube internals far too hot can also poison the cathode, also reducing its emission. But issues of element dissipation aside, trying to push far too much current through a tube destroys its cathode emission early.

    Although the 6AG7 ratings--at least from RCA, who developed the tube--don't give a maximum plate current spec for the design, "30 mA" is the figure we see for max-signal plate current in the specs. Add that to the screen-current spec, and add a few more mA for grid current, and you have a working normal-operation cathode-current spec. So "pinning a 500 mA meter" with four 6AG7s is idiotic.
    N2EY likes this.
  4. W9BRD

    W9BRD Ham Member QRZ Page

    The main attractiveness of grounded-grid power amplifiers at MF and HF is that they don't need neutralization if built well. This is so because a tube's control grid--and any other grids between the cathode and plate--when grounded, serves as an electrostatic shield between cathode and plate. If one is willing to neutralize, grounded-grid offers no advantage over grounded-cathode and even afford a key disadvantage over g-c if you're after linear amplification of a complex signal.

    The reason grounding a tube's "control grid" and driving its cathode instead works is because a tube amplifies by means of variation in its grid-cathode differential, not by variations in its control-grid voltage per se. That is, we can either hold the cathode voltage steady and vary the grid, or we can hold the grid steady and vary the cathode--the tube doesn't "know" the difference. Assuming equal operating conditions in g-c and g-g (and assuming that any filament/heater and/or dc biasing components aren't in ac parallel with the cathode RF), a tube will have the same gain when operated class A in both modes. It doesn't "know" that the grid is varying relative to the cathode in g-c and the cathode is varying relative to the grid in g-g; all it "sees" is a grid-cathode wiggle that changes its plate current, and off we go.

    What is very different about the two modes is the impedance of the driven tube element. Until we drive it with a high enough voltage to make it conduct, a tube's control grid at MF and HF "looks like" high-impedance capacitive load. If we bias the grid with a sufficiently high negative voltage (and don't need linear amplification), we can enjoy the high efficiency of class C operation, in which the driving signal overcomes that high negative bias for only about 1/3 of each cycle, causing the tube's plate current to flow in pulses.

    A parallel tuned circuit driven through a low-impedance link is the usual way we develop sufficiently high driving voltage to overcome that high negative bias for class C. We have this option because a tube's grid is a relatively high impedance until the grid conducts. As a result of the voltage step-up from the link to the tuned circuit, only a fraction of a watt to a few watts is enough to drive a grounded-cathode class C amplifier to full output.

    Because as a first-order approximation the impedance of a tube's cathode (or a BJT's emitter or a JFET's source) is 1 divided by the device's transconductance, grounded-grid operation doesn't afford the equivalent option for voltage step-up because the impedance of the cathode as control element is relatively low. Per its specs, a 6L6 operating class A at 250 V plate and screen exhibits a transconductance of 6 mS, which equates to a cathode Z of 166 ohms. The class C bias for a 1614 (RF-selected 6L6; an 807 or 1625 is dc-identical) is 45ish volts. Assuming that impedance level, just to equal that bias we'd need to apply 45 V, peak, to the cathode--12.2 W in 166 ohms! (45 squared divided by 166).

    The situation for cathode drive will be even more problematic, because a tube's transconductance, and thereore its cathode impedance, is set by its operating conditions, which vary from instant to instant as the driving signal varies. This is why we read for linear amplification of SSB in g-g that the driving stage must be able to handle without distortion wide variations in load impedance. Therefore the "166 ohms" I mentioned above as the cathode Z of a g-g class A 6L6 doesn't even begin to describe what would happen in practice if the tube's operating point was varying from plate cutoff to saturation across the driving-signal cycle.

    OTOH, were we to ac-ground the cathode, connect a 166-ohm resistor from grid to common, and drive the grid instead of the cathode, we'd achieve the same grid-cathode voltage variation without crazy variation of control-element Z across the driving-signal cycle. (The term passive grid was coined for this approach; the technique has a following. Although we no longer enjoy the operation of the control grid as a shield, and the Miller effect comes into play as a result, driving the grid across a low-value resistance can be sufficiently stable without neutralization. The gain will be pretty much the same.)

    Bottom line is that we don't see class C g-g amps because (among other reasons) we'd need a crazy-high driving-signal power level at the cathode's low Z to overcome the stage's cutoff bias--compared to g-c operation, in which we can use the voltage step-up from a low-Z link to a tuned circuit at the grid to let us overcome the tube's cutoff bias with a driving signal of only a fraction of a watt to a few watts, "depending."

    Building that really bad CQ circuit is a way of passing one's time. Especially because it's a stupidly destructive application of devices that are no longer made, I propose that a more fun challenge would be to build a well-engineered grounded-cathode amplifier, neutralization and all, keeping all devices within their ratings.

    That done, our correspondent's main challenge will be to reduce the output of his AT-1 far enough not to overdrive the amplifier. (Pssst: Screen-grid control on the rig's 6L6.)
    Last edited: Mar 25, 2019
  5. AA5CT

    AA5CT Ham Member QRZ Page

    @W9BRD Okay then ...

    One of my stints was with what was called the Ti GaAs Facility, which fabricated MMICs (microwave monolithic integrated circuits) out into the mm wave range ... and all types of circuit topologies, from WBDAs to cascaded narrow band power amps, so, been there , seen that as far as 'tech' ...

    It is my hope that others can benefit from your treatise above, as my focus is on to power mosfets these days. Not much call for tube designs (or designers) today, exc in the case of special requirements or re-work to accommodate increased capabilities, like adding 160 to an SB220.

    Anymore design work calls for 'modelling' work, as that works out to be a much quicker to try a variety of 'what ifs' that our forefathers would have found difficult to try in a reasonable amount of time to try for the most beneficial balance of trade-offs in any given design.
  6. NW2K

    NW2K Ham Member QRZ Page

    Perhaps, but I'll make no apology for risking the lives or well being of six pre-owned 6AG7's. :)
    WB4IUY likes this.
  7. W9BRD

    W9BRD Ham Member QRZ Page

    :) What's interesting to me is that the authors were on the right track in that they had experimentally figured out that what efficient g-g operation needed was a triode or triodes with a much higher amplication factor--mu--than what triode-connected beam power tubes (like the 6L6, and even the TV sweep tubes) could provide. They just adopted a stupid implementation in proving their point. (In this they were aided by the fact that the editor of CQ at that point was Wayne Green, W2NSD, who it can be argued had little interest in the quality of the magazine's technical content and who was also within a year or so of being fired as editor by CQ's publisher, for allegedly stealing circulation information [that, as it later played out, could well have been useful in starting up his own magazine, 73, in 1960].)

    High-mu triodes--triodes with high amplification factors--are exactly what's needed for grounded-grid power amplfication to work well. Such tubes, by now well-known and well-applied, include the 3-500Z (amplification factor, 130) and the more modern 3CX800A7 (amplification factor, 200). The 6AG7 abusers in that CQ article were figuring this out; from specs we don't know exactly what the triode-connected mu of a 6AG7 is (if a "grid 1 to grid 2" mu was given, that would pretty much tell us), but by rooting around we can discover that the grid-1-to-grid-2 mu of the miniature equivalent (6AH6) of the 6AG7's octal little brother (6AC7) is 40. A triode-connected beam power tube will generally have a mu of less than 10; the TV sweeps, maybe even less than 5, making them pretty bad choices for g-g linear service.

    Another class of tubes might interest you as an experimenter in grounded-grid: beam-triode high-voltage regulators. One you can find pretty readily on Ebay, the 6JD5, has a 35-watt plate, a peak plate current rating of 325 mA (dc rating is probably about a third of that, say 110 mA), an amplification factor of 300, and a rather stupendous transconductance (for a tube) of 55 mS! (Which value, assuming the tube was biased to that point for linear amplification, would equate to a cathode impedance of 18 ohms!) A combination of high plate voltage and some positive control-grid bias will be necessary to make tubes of that type perk as an amplifier; Google around and you can find tube-audio folk making them work in Class A2 at a few hundred volts plate.
    N2EY likes this.
  8. N2EY

    N2EY Ham Member QRZ Page

    I suspect a bit of....exaggeration....in such stories. How would they ever tune up such a transmitter?

    What I know for a fact was that amateurs would indeed run tubes far beyond published ratings (see the thread "Thirty Three Watts Per Dollar" in the Amplifiers forum) and get away with it....for a while.

    One reason they did this was because tube cost went up exponentially with power...if they could get high-power tubes at all; the radio store in a small or medium size town probably wouldn't carry such an expensive item. Another reason was that new designs were coming out rather rapidly, and a tube that was the bee's knees one year would be considered obsolete in just a few years because newer, lower-cost ones were developed.

    Finally, the published ratings were usually for long life in continuous-duty commercial-type service, where failures were a big problem, and tubemakers were trying to build reputations for reliability. 24/7 operation means 8760 hours per year, (8784 on a leap year), but the amateur who actually transmits about 1 hour per day will need 24 years to put that much time on a tube!

    After WW2, the trend continued somewhat because of odd surplus tubes hitting the market.

    The problem got to the point that "CCS" and "ICAS" ratings were developed.

    All different today. There's NO reason to push tubes beyond ratings!

    73 de Jim, N2EY
  9. N2EY

    N2EY Ham Member QRZ Page

    There's FM....which is mostly used by amateurs on VHF/UHF.

    A couple of points:

    - There were some cathode-driven Class C transmitter designs in QST and other publications in the early 1950s, mainly to eliminate the need for neutralization in driver stages. The tradeoff was lower gain, and more stages. Never really caught on.

    - There is a myth that "grounded grid doesn't require neutralization". There's a bit of truth to that, when stated correctly: "grounded grid doesn't require neutralization when certain operating parameters are met". For example, a quad of 811As in grounded grid will often show signs of instability on 10 meters because the filament-to-plate capacitance is high enough to provide significant feedback at that frequency (and sometimes even 15 meters). The length of the grid lead in an 811A doesn't help the situation.

    - As you state, the typical amateur HF GG amplifier is usually operated as an "add on" to a 100-watt-class transmitter/transceiver. As such, there is no need for high gain, and the drive will be provided at about 50 ohms.

    And now the big one:

    - In the AM days, operating high-power 'phone was an expensive proposition. A typical legal-limit plate-modulated 'phone transmitter filled a six foot rack (or two) with power supplies, modulator, amplifier, etc. High power 'phone was so expensive that to my knowledge only two manufactured transmitters were made for the amateur market (the Johnson Desk Kilowatt, which cost $1600 without the desk or driver transmitter - or shipping) and the Collins KW-1, which was in the if-you-have-to-ask-you-can't-afford-it price range. Only about 400 Desk KWs and about 150 KW-1s were ever made.

    Then came SSB. At first amateur HF SSB was complex and expensive, but then came The Perfect Storm.

    First was the 100 watt SSB transceiver, such as the Collins KWM-2, which set the standard for decades to come. One small box that was easy to operate, eliminated zero-beating, cost about as much as separate tx and rx of equal quality, and could be used mobile and portable with just a changed of power supply. When the KWM-2 concept was followed by less-expensive models from other manufacturers, there was no stopping the trend.

    Second was the development of the "tabletop" linear amplifier, with solid-state voltage-doubling power supply and tubes either specifically made for GG service (3-400Z family) or well adapted to it (811A, 572B/T-160L). Such amplifiers needed lots of drive, but with a 100 watt exciter that was no problem. They could be small, light - and cheap. No modulator, no screen or complicated screen supply, no filter chokes, all in one box.

    Put them together and it's clear why SSB took over in the early 1960s.

    Consider the Heathkit SB-100 ($360) and SB-200 ($200) when first introduced about 1964. With AC power supply for the SB-100, you could have 1200 W PEP input on SSB, all HF bands, for less than $600....and set up the station on a sturdy card table. (An SB-200 weighs 35 pounds - less than what many popular receivers of the 1950s weighed).

    Last edited: Mar 29, 2019
  10. N2EY

    N2EY Ham Member QRZ Page


    There are so many better options. And the 6AG7 has lots of uses other than as a short-lived final amplifier.

Share This Page