Plate-fed allows more output than Cathode Follower?

[GlideOn]

I had a rather curious and interesting chat with a very reputable amp tech who took in my amp recently while fixing some various things I was a bit beyond my depth to fix (oscillations, DC transformer saturation, oscilloscope readings, etc).

Anyways, I have a lot of admiration for this person as they have a great wealth of experience and have been fixing amplifiers for decades and I've taken a few pages from them in my journey thus far. I don't know if they frequent this forum so I will respect their identity and privacy. Very few people like them left and they should absolutely be treasured.

My amp in question is a Single Ended Parallel EL34, cathode bias that is currently making around 15-16w RMS when measured before clipping. Preamp is pretty much Marshall JMP. Even though it is biased slightly cold around 42ma per tube, decreasing cathode resistance probably wouldn't raise the power much. I was hoping for around 20w-25w RMS clean as that seems to be the sweet spot for getting some decent edge-of-breakup sound in a room with an aggressive drummer. My 2.5k/25w output transformer and 600v/200ma transformer should handle any increase in power no problem.

It currently sits outside its' intended 1x12 combo chassis originally loaded with a Mesa MK4 amp (since damaged and discharged by original owner). Still has original C90 speaker which sounds great.

With that context now established, I learned something interesting while chatting with this person: if one is aiming for maximum output or RMS watts, they should consider a Plate Fed tonestack vs a Cathode Follower. Even though a Cathode Follower is designed to provide relatively lossless EQ control, it actually starts to clip earlier as a side effect and can limit the clean headroom and consequently RMS rating.

A Plate-Fed stack however has more loss, but it actually helps to increase headroom and RMS as a result.

I have already have Marshall and 60s Bassman style amps; one is Cathode Follower, the other is Plate Fed. They sound more similar than not though the Cathode is maybe brighter and the Plate Fed control work a bit better - that's about it. No huge difference in gain I can detect.

Is there some truth to this?

Would converting to a Plate-Fed tonestack help to increase power? And realistically: how much so?

[R.G.]

Interesting thought.

In my mind, the issue would be whether the signal loss in a plate-fed tone/volume stock was less than the signal limitations caused by a cathode follower AND whether the distortions in a cathode follower were actually desirable.

Maybe you could try it both ways - disconnect the cathode follower, hook up the T/V stack to the plate, measure the output power. Then reinsert the cathode follower and measure. That's probably as quick and easy as you can do without an oscilloscope or AC voltmeter.

Another possibility is to use a MOSFET follower; see http://geofex.com/Article_Folders/mosfe ... tfolly.htm for how this might work. A proper MOSFET follower will directly sub for a cathode follower, but have a bigger output before distortion. Many people have reported good results with them. It also frees up one triode section that could possibly be used to add in a little gain after the T/V stack, juicing up the signal a bit before it hits the PI. The gain could be messed with so it's a little more signal, not a metal gain-monster addition, if you like.

All this ignores what you have already assumed, such as the power supply being OK with the added power, side-effect distortions, and so on.

[GlideOn]

Thanks for the feedback.

I certainly don't have to worry about layout or giving up room to run an Anode fed tonestack, even more so I could also free up one side of the triode and add another gain stage or change the insertion point via DPDT switch to have one dedicated clean channel with new spare triode. It would only take an extra wire, extra cap and possibly extra resistor to make it work.

I don't have any interest in adding additional solid state complexity at this time nor really have anywhere practical to mount it.

I don't have any strong feeling towards the compressing nature of a Cathode Follower and it is probably a moot point when the amp featuring 50w or more can spare to lose a bit of headroom for said effect. But only having around 20w or less to work with, I am seeing the wisdom in trying to extract as much headroom and bite as possible or delaying the distortion a bit, which is what an Anode stack would help do best. The added benefit of have greater tone control is potentially more useful for a combo amp trying to stand out in a room with 50w push pull amps.

[R.G.]

I certainly don't have to worry about layout or giving up room to run an Anode fed tonestack, even more so I could also free up one side of the triode and add another gain stage or change the insertion point via DPDT switch to have one dedicated clean channel with new spare triode. It would only take an extra wire, extra cap and possibly extra resistor to make it work.

That's good. Options are always nice.

I don't have any interest in adding additional solid state complexity at this time nor really have anywhere practical to mount it.

I understand. I get that a lot when I suggest MOSFETs. On the other hand, consider: it's literally one MOSFET and one resistor. That is, it's the same complexity as a cathode follower. Just sayin'

I don't have any strong feeling towards the compressing nature of a Cathode Follower and it is probably a moot point when the amp featuring 50w or more can spare to lose a bit of headroom for said effect. But only having around 20w or less to work with, I am seeing the wisdom in trying to extract as much headroom and bite as possible or delaying the distortion a bit, which is what an Anode stack would help do best. The added benefit of have greater tone control is potentially more useful for a combo amp trying to stand out in a room with 50w push pull amps.

It's worth checking out. I'm still not sure whether the higher source impedance of a plate feed to the T/V loading would necessarily result in a bigger, less distorted signal at the volume control wiper when it's full up, versus the buffering of a cathode follower. Probably, like most things, the devil is firmly hidden in the details.

Try it. Let us know.

[martin manning]

...I get that a lot when I suggest MOSFETs. On the other hand, consider: it's literally one MOSFET and one resistor. That is, it's the same complexity as a cathode follower. Just sayin' :D

Less complexity, no heater!

Here are some plots from a sim... Marshal tone stack, plate driven by a standard 12AX7 stage (100k Ra, 1k5 Rk, 22u Ck), and the same with a 12AX7 CF inserted (100k Rk). Transient is at the stack input, the Bode shows the stack input and output with all controls at noon. More gain from the CF version, different distortion characteristics.

Screenshot 2024-10-26 at 11.35.00.png

Screenshot 2024-10-26 at 11.33.14.png

And with LND150 SF (100k Rs):

Screenshot 2024-10-26 at 11.53.17.png

Screenshot 2024-10-26 at 11.54.28.png

[R.G.]

Good illustration Martin!

@ others...
The plate of a 12AX7 acts like the plate has a resistance of 60k to 80k in series with it, depending on the plate voltage and such. This plate resistance means that the plate can only pull up and down that strongly. Any external plate resistor is then tacked on to this. A 12AX7 plate will then look like a signal voltage source feeding the 60k to 80k plate resistance and the real plate resistor in series, and the output signal on the plate pin as the voltage divider voltage fed through the parallel combination of the plate resistance and the real plate resistor. Any external loading appears in parallel with the real plate resistor, and lowers the signal voltage further.

Adding a cathode follower to a 12AX7 plate removes the external loading to an extent, at least as far as the plate signal voltage is concerned. But the 12AX7 follower itself has that 60k-80k plate resistance internally too. It can only pull up on its cathode resistor by that amount. The small signal impedance is smaller by the amount that feedback reduces the effect of the plate resistance, but this gets diluted when feedback can't cover up the internal impedance by expending mu (transconductance) to do so. The effective small signal output impedance of a well done CF is 400-500 ohms. In the case of wanting every last volt of signal out of a plate, it's no longer "small signal" so you run into the limits of pulling up through a 60k-80k plate resistance and down through the resistor to ground when the tube turns fully off.
A 12AX7 will have a transconductance of 1200 to 1600 uMho - that is, 1.2 to 1.6 milliampere of plate current change per volt of change on the grid.

A sub-$1 MOSFET can have Vds of 400-1000V, a transconductance of > 1000 milliamperes per volt, an internal drain resistance of under 1 to a few ohms, and an insulated-tab TO220 package. That is, it has under a thousandth of the internal resistance of a 12AX7, and more than a thousand times the tranconductance. It can effectively follow that plate it's attached to into signal extremes where a CF just can't go. The gate-source capacitance which might otherwise limit frequency response is reduced by the source feedback into insignificance.

And there's no heater... sorry, I'm rambling.

[nuke]

Good discussion of hi-z vs. lo-z input to the tone stack. It's a good discussion, I think some of the online tonestack simulators also allow selection of the prior stage impedance, so the effects of plate vs cathode-follower (or source-follower) are accounted for.

The OP's conjecture remains:

With that context now established, I learned something interesting while chatting with this person: if one is aiming for maximum output or RMS watts, they should consider a Plate Fed tonestack vs a Cathode Follower. Even though a Cathode Follower is designed to provide relatively lossless EQ control, it actually starts to clip earlier as a side effect and can limit the clean headroom and consequently RMS rating.

A Plate-Fed stack however has more loss, but it actually helps to increase headroom and RMS as a result.

If we break the OP's question up into the various stages, all that the power amp "sees" beyond the tonestack is the signal present at the input of the phase-inverter. The PI in most amps (long tail) is a relatively hi-z input, so has little effect on the prior stages.

Thus, one can simply inject a signal there. The PI/power stage will amplify to the limits of its power and then clip. It doesn't really "know about" the prior stages implementation details.

The effect of plate input tonestack is that the signal loss is higher across the stage, due to the source impedance of the feeding stage. This means the loss across the stage is higher and the effect of the tone controls somewhat different than a cathode-follower feeding it.

"Clean headroom" is a perception.
"RMS output" is a measurement.

Let's tackle RMS:

As long as either the plate-loaded or cathode follower stages do not clip before the power amp does, then the RMS output at the speaker terminal is the same at the point the power amp begins to clip. Root-mean-square power does not change. True whether the tone stack is plate-fed, or cathode-follower fed, or simply driven directly at the PI input, provided the preamp stages do not clip first.

"Clean headroom" being a perception, is subjective. It isn't a measurable quantity. One person may feel one is better than the other, and someone else may feel the opposite.

I can say that each of the tone-preamp topologies "sound different" having built plenty of them, perhaps not as different as people may opine. One can also alter the sequence of gain, tone-stack, volume control, into gain, volume, gain, tone or whatever else. (see Blues Jr. vs AB763 vs Tweed Fenders for example) and get different "feel".

Just sort of depends on which flavor you prefer actually.