Class AB tube overdrive pedal - anyone ever done this?

[poodad]

I've seen a gazillion variations on tube overdrive pedals, but they all have one thing in common: class A operation. I'm thinking that an AB circuit would provide for a far more symmetric clipping of the tubes verses a class A.

I'd use 12V transformer to supply heater voltage and also to drive a one or two stage voltage doubler to 24V or 48V for plate voltage.

One section of an op-amp (powered from the 12V) would provide variable gain. This would feed the grid of half a 12ax7. Another op-amp section would invert that signal (unity gain) and feed it into the grid of other half of the 12ax7.

The cathodes of the 12ax7 would have a variable resistor to ground to provide variable bias.

The plates would be fed via the 48V through plate resistors.

The outputs of the tubes would be voltage divided down to a reasonable level to feed into the + and - inputs of a third unity gain op-amp to "combine" the out of phase signals from the tube plates.

Before I go to the trouble of breadboarding this up, has anyone ever tried it? Ever heard of a pedal that does this? I can't imagine that no one has thought of this before. But the fact that I've never seen this implemented leads me to believe that it isn't worthwhile.

[Gus]

https://www.tapatalk.com/groups/unofficialacousticcontrolcorpmessageboard/swr-history-and-vintage-models-t3956-s10.html

It has a transistor based PI and an output transformer
Look for a schematic of the opamp PI circuit in a Musicman amp.

[diydave]

Why not go for the 'power-amp' section of the firefly, and drive this with one or more preceding opamp stages?
The OT-transformer is really, really small.

[teemuk]

Vox/Korg Valvereactor. 

[Steben]

You could use low voltage tubes, such as the 12u7 (triode) or 12k5 (pentode). they were designed to work on 12V.
A pentode has more gentle onset of clipping (which gets harder with feedback loop).

[mozz]

Never heard anything good about triode push-pull or SE for output stages, too clean,  for guitar anyway. I'd go pentode 12k5 as was stated or something else more common such as 6au6(if it would work with lower plate voltages).

Here's a good read on space charge tubes.
http://www.junkbox.com/electronics/lowvoltagetubes.shtml

[merlinb]

I'm sure there was an amp that had an unusual distortion control where one side of a sort-of-LTP was driven, and you could gradually feed more signal to the other side, to blend bewteen SE and PP-ish distortion. But for the life of me I can't remember what it was. I want to say a Peavey or Laney?

[vigilante397]

But the fact that I've never seen this implemented leads me to believe that it isn't worthwhile.

A pretty good assumption In my limited understanding (especially standing next to merlinb) the big reason class AB is popular compared to class A is that you can get more power more efficiently. If there was an efficient way to built a 50W class-A amplifier I'm sure it would be wildly popular as people tend to like class A for tonal reasons. For overdrives efficiency of power isn't a huge issue as your overdrive isn't acting as a power amplifier, so I see no reason to go class AB in a pedal.

[teemuk]

...But for the life of me I can't remember what it was. I want to say a Peavey or Laney?

Peavey.
Behringer also uses that idea in their "Tubegain" signal processors. Those also employ "starved" 48V B+.

[iainpunk]

i have build and sold a push pull tube drive, it was based on grandpa's notes (he was an electronics hobbyist just like me), the circuit had 2 "single row push pull" (srpp) tube stages. sounds alright, but nothing special, the guy wanted something with tubes so badly, claimed to be able to hear the difference between transistors and tubes. anyway heres the schem:

[mozz]

Isn't that cascode?

[iainpunk]

Isn't that cascode?

no, a cascode has an external bias, this relies on itself to bias well, also, a cascode's output is at the anode/plate of the tube, not the kathode.

the srpp is basically a self contained push pull pair, while a cascode is based on simulating a pentode with 2 triodes

the output impedance is also way lower in the srpp, but it runs a bit hotter than a cascode (more quiescent current)

ow yeah, by the way, the srpp isn't really class AB but it behaves about the same, linearity wise

[amptramp]

One thing about SRPP is it needs a load to remain in the linear region.  The current through the lower tube biases the upper tube and as it increases, the bias gets more negative on the upper stage causing its current to drop.  But the triodes are in series - without a load, the voltage slams up against the rails.  You could build a nice fuzz with a variable resistor on the output so at maximum resistance, you get fuzz and below a critical resistance, you get Class A amplification.

[iainpunk]

One thing about SRPP is it needs a load to remain in the linear region.  The current through the lower tube biases the upper tube and as it increases, the bias gets more negative on the upper stage causing its current to drop.  But the triodes are in series - without a load, the voltage slams up against the rails.  You could build a nice fuzz with a variable resistor on the output so at maximum resistance, you get fuzz and below a critical resistance, you get Class A amplification.

owyeah, sorry, i forgot to mention that because i had to much to drink already, but it really depends on the tubes you use, because some tubes load themselves enough to stay linear (e80cc) while others deliver a more "unstable" comparator like output (12ax7)(a bit like a fuzz).

if you try this topology with Jfets, they load themselves enough to stay stable, and you can use diodes instead of resistors

[merlinb]

...But for the life of me I can't remember what it was. I want to say a Peavey or Laney?

Peavey.

Can you remember which model?

[poodad]

But the fact that I've never seen this implemented leads me to believe that it isn't worthwhile.

A pretty good assumption In my limited understanding (especially standing next to merlinb) the big reason class AB is popular compared to class A is that you can get more power more efficiently. If there was an efficient way to built a 50W class-A amplifier I'm sure it would be wildly popular as people tend to like class A for tonal reasons. For overdrives efficiency of power isn't a huge issue as your overdrive isn't acting as a power amplifier, so I see no reason to go class AB in a pedal.

The reason for going class AB has nothing to do with power. The reason is that tubes have a very different clipping characteristic when being driven into saturation versus going into cutoff. So with an overdriven  class A amp, the "top" of the waveform is clipped by the saturation curve of the tube, and the "bottom" of the waveform is clipped by the cutoff curve. Very different animals resulting in a very non-symmetric clipping.

In other words, if you feed a sine wave into an overdriven class A amp and look at the output on an o-scope, the resulting waveform will be squished at the top in a very different manner than the bottom.

With a class B amplifier, the "top" half of the waveform is clipped by the saturation curve of tube A, while the "bottom" half of the waveform is clipped by the saturation curve of tube B. Since the halves of a twin triode should be almost identical, this should result in a symmetric clipping of the final waveform.

In other words, the top and bottom half of the waveform would be squished almost exactly alike. That should make for less odd harmonics, and thus a better tone.

Class B operation results in crossover distortion at around the zero voltage as the one tube goes into cutoff and the other tube starts to conduct. So by keeping a bit of bias on each triode, you have them both conducting a bit in the waveform region near zero volts. This is class AB, and it eliminates the zero point crossover distortion that you'd normally get with pure class B.

[Gus]

Did anyone look at the link I posted

I am surprised it has not been cloned

[poodad]

That looks like a normal hybrid tube power amp with the unusual novelty of using a transistor as the phase inverter to drive the power tubes.

Oh, and there may be a huge mistake in that schematic - it shows the EL84 screen grids connected directly to the 250V supply. I'd expect to see resistors in there.

[teemuk]

Can you remember which model?

At least some of the Valveking series amps. I have a faint memory Peavey also has a patent for such feature.

[teemuk]

That looks like a normal hybrid tube power amp with the unusual novelty of using a transistor as the phase inverter to drive the power tubes.

So? The only difference to earlier scheme (with lower B+ and preamp tube as active device) is that this one just produces more output power. The unit then dissipates that power in a dummy load and then 'reamps' that signal from the power amp. Simultaneously it introduces post power amp fx loop and tone controls.

Symmetric clipping produces mainly odd order harmonics, by the way. It is not any better tone than that of asymmetric clipping. Just different and more similar to tone of an overdriven push-pull power amp. For obvious reasons.