Hammond AO-44-1 Reverb Amplifier

[patlaw]

I'm refurbing a Hammond AO-44 reverb amp. It's working fine, but I have a design question. What is the purpose of C9 in the cathode of the output tubes? The reason I ask is that adding or removing the cap with any tone being fed through the amp at any amplitude does not cause a change in level. Whether I use 20 Hz, 100Hz, 1kHz, 5kHz, or 10kHz, the capacitor makes no difference at all. The cap is not bad. The resistor measures 130Ω. The audio output is clean and symmetrical, so the tubes are good. Now, having asked what will be perceived as a dumb question (particularly for someone who has been repairing electronics for more than 50 years), I understand that bypass the cathode resistor is supposed to increase the AC gain in the stage. But it doesn't. What am I missing?

[Chuck H]

I'm not the first person to trust regarding operating class, but according to the schematic those are 9 watt tubes idling at 20 watts each. That's enough past AB that any conduction is class A. I've read here and elsewhere that there is no advantage to a cathode bypass cap on power tubes for class A (someone should have told VOX ***) So while I can't say why the designer put the cap there I hope this can answer to why it's presence makes no difference. And on that note...

Why were you compelled to compare the amps output with and without that cap? Did you expect what you found?

*** FWIW I've also read here that the AC30 circuit can sometimes shift into AB under clipping conditions. I was being glib on the matter above

[Helmholtz]

The purpose of the cathode cap is to short unwanted signal to ground.

A pure Class A PP stage develops no cathode signal as the sum of the cathode signal currents is zero. So a bypass cap is not needed/makes no difference.

But most (if not all) cathode biased guitar amps leave Class A operation at some medium output when tubes alternately are driven into cutoff (Class B).
So these (including the AC 30) are all Class AB amps.
In class B a cathode signal develops and that's where the bypass cap is needed.

The transition from Class A to Class B shows by an increase of cathode DC voltage.
The cathode voltage in an AC30 is 10V at idle and 12.5V at full unclipped power. That's a clear indicator of Class AB operation.

I can't believe the 20V across the 130R cathode resistor. No serious designer would bias the tubes at 240% plate dissipation.
Tubes should be redplating.

[patlaw]

Why were you compelled to compare the amps output with and without that cap? Did you expect what you found?

It was a purely pedagogical exercise. And no, I didn't find what I expected. I thought the cap would do something. I like to try to get into the designer's head and understand why each component was added.

[patlaw]

The purpose of the cathode cap is to short unwanted signal to ground.

A pure Class A PP stage develops no cathode signal as the sum of the cathode signal currents is zero. So a bypass cap is not needed/makes no difference.

But most (if not all) cathode biased guitar amps leave Class A operation at some medium output when tubes alternately are driven into cutoff (Class B).
So these (including the AC 30) are all Class AB amps.
In class B a cathode signal develops and that's where the bypass cap is needed.

The transition from Class A to Class B shows by an increase of cathode DC voltage.
The cathode voltage in an AC30 is 10V at idle and 12.5V at full unclipped power. That's a clear indicator of Class AB operation.

I can't believe the 20V across the 130R cathode resistor. No serious designer would bias the tubes at 240% plate dissipation.
Tubes should be redplating.

As usual, thank you for the awesome explanation.

Regarding the 20V across the 130Ω resistor, that could be incorrect. I'm redrawing schematics from terrible original drawings. I may be reading the voltage wrong. I haven't measured it yet. I'll update when I do.

[patlaw]

An unrelated question also regarding this amp: Notice that C4 is a 1µF cap. A friend ran a simulation on the circuit, and he discovered that there is a bump below 10 Hz. When he changed the value of the cap to 10µF, the bump disappeared. Why? Since it's a reverb recovery amp, the input is a spring. The bump probably makes no difference, but I think I'm going to start changing them to 10µF. (The 1µF cap is good.)

The customer is complaining that when he hits a bass pedal on his B3 organ, the bass saturates the reverb spring, and the reverb on the higher notes disappears momentarily. He wants to create a high pass filter to the spring since the pedals don't need reverb. That idea seems to have some merit. Adding reverb to frequencies below 100Hz seems unnecessary. Thoughts?

[Helmholtz]

Q1 voltage gain depends on its collector load.
If C4 is large (e.g. 10µ) it acts as a power supply filter cap and creates a filter node (=AC ground) so collector load essentially is R3.

A 1µ C4 gets high impedance at very low frequencies, so Q1 gain increases.

The rolloff at even lower frequencies should be mainly due to coupling caps.

Nothing wrong with using 10µ for C4 if it sounds better.

[patlaw]

Q1 voltage gain depends on its collector load.
If C4 is large (e.g. 10µ) it acts as a power supply filter cap and creates a filter node (=AC ground) so collector load essentially is R3.

A 1µ C4 gets high impedance at very low frequencies, so Q1 gain increases.

The rolloff at even lower frequencies should be mainly due to coupling caps.

Nothing wrong with using 10µ for C4 if it sounds better.

Could the choice of 1µF have been intentional, or do you think they thought it was "good enough" to act as a power supply filter cap? The frequency is so low, I'd bet they didn't pay any attention to it.

The voltage across the 130Ω is actually 9 volts. I blew up some bad schematics and then measured it. Thanks for catching it.

[Helmholtz]

Could the choice of 1µF have been intentional, or do you think they thought it was "good enough" to act as a power supply filter cap?

As an experimental phycicist I don't like to speculate without data. Relevant data would be the frequency response of the reverb tank including driver.
Otherwise I'd let my ears decide.

The voltage across the 130Ω is actually 9 volts.

Now that means around 110% plate dissipation and makes sense.

[Chuck H]

Now that means around 110% plate dissipation and makes sense.

And still biased class A. That is, biased at idle. With conduction levels pushing the output to clipping the voltage on the cathode resistor will rise due to current across it and the bias condition will become 'cooler' (as has already been discussed). Perhaps re do the experiment with and without the cathode circuit bypass cap with the amp under clipping conditions and see if you don't measure/hear a more significant difference between the two.

[Helmholtz]

And still biased class A.

How do you know?
As there's a bypass cap I assume it runs class AB.
The cap might not noticeably increase gain but will lower distortion caused by discontinous cathode currents when driven into class B.
.
You can't tell operation class from thermal dissipation at idle.
With a PP class A amp tubes are never pushed into cutoff before clipping.
A typical AC 30 is biased at around 115% dissipation but turns class B at some medium output.
OTOH, you could design a pure class A amp with only 50% idle dissipation. Just requires tubes having a large enough dissipation limit.

[Chuck H]

Great point. I was definitely generalizing. The 100% (ish) dissipation rule for class A doesn't have to apply and it IS more dependent on the tubes operational conditions and dissipation limit. For most circuits we discuss here (including this one I think) it's usually a useful generalization.

I did rebuild a Vibro Champ once with a bigger 4k/8ohm OT, EL34 power tube and an efficient 10" speaker and I found that I had to limit screen current quite a lot. Though I'm still not sure exactly why. The typical 1k rail resistor after the OT was bumped to 3.3k and I had to limit any drive signal 2V greater than the bias voltage with a zener circuit to keep the screen grid resistor from incinerating under overdrive. So max dissipation was limited to a little over 7W. If I remember correctly under this condition the EL34 was biased at about 22W at idle (whole tube plate and screen) to get the most symmetrical performance both clean and clipped. Though some technical aspects go over my head I suppose that I HAVE seen circumstances like you mention above.

[Helmholtz]

For most circuits we discuss here (including this one I think) it's usually a useful generalization.

Fact is that I haven't seen a pure class A push-pull guitar amp yet.
Reason probably is that max. power output would be disappointing (or non-competitive).

You could built a textbook class A amp with 2xEL84 biased at 100% dissipation with a max. B+ of 250V.
As soon as you increase B+ (for more power), the amp will inevitably turn class AB even though this increases dissipation.
A countermeasure could be increasing OT primary impedance but is not typically done because it would lower output power again.

[Chuck H]

Fact is that I haven't seen a pure class A push-pull guitar amp yet.
Reason probably is that max. power output would be disappointing.

You could built a textbook class A amp with 2xEL84 biased at 100% dissipation with a max. B+ of 250V.
As soon as you increase B+ (for more power), the amp will inevitably turn class AB even though this increases dissipation.
A countermeasure could be increasing OT primary impedance but is not typically done because it would lower output power again.

I really want to get this. Doesn't the Vox AC30 circuit (for example) operate class A to full, unclipped output power and then only moves into AB when pushed beyond that? This has been my understanding from rote, read material for years. I never considered the notion of any push pull amp maintaining class A operation when clipping.

[Helmholtz]

Doesn't the Vox AC30 circuit (for example) operate class A to full, unclipped output power ...

No, increasing cathode DCV clearly shows that operation shifts to class B before clipping.
It would run pure class A if B+ was only 250V.