• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Wavebourn Alligator amplifier

Questions/suggestions/objections?

alligator10.gif
 

EC8010

diyAudio Moderator Emeritus
2003-01-18 7:57 am
Near London. UK
OK, I'll bite the alligator...

I saw a circuit similar to this recently on AA that described itself as PP. You seem to have a conventional input stage with an output valve buffered by a source follower. The op-amp and MOSFET are a bias servo(?) to balance the currents in the output transformer and allow use of a PP transformer? It's hard to tell exactly what's going on without values.

Oh, and I forgot the time machine. May 2008?
 
Re: OK, I'll bite the alligator...

EC8010 said:
I saw a circuit similar to this recently on AA that described itself as PP. You seem to have a conventional input stage with an output valve buffered by a source follower. The op-amp and MOSFET are a bias servo(?) to balance the currents in the output transformer and allow use of a PP transformer? It's hard to tell exactly what's going on without values.

It's a PP amp with triode SE signature where one shoulder is a triode, another shoulder is a controlled current source. Yes, it allows to use better transformer (no need for massive iron with gap so no need for an extra copper) and twice more of power from one triode.

As you know a dynamic resistance of a current source is very high so transfer function is determined by a triode only that works in a privileged more, i.e. less plate current variations.

V102 -- power triode,
A1 и Q103 -- controlled current source,
V101 -- driver for a power triode, also a phase splitter. Triode control signal goes from anode, VCCS control signal goes from cathode, to the same cathode goes positive feedback by current and negative feedback by voltage.

Q101 -- source follower to drive a grid positively,

Q102 -- bias regulator for the output triode.

Resistors in cathode of V102 and in source of Q103 are equal so DC currents through them are equal. Because of time constant of R113 and C103 currents are balanced by average value. However, it introduces some transients, but anyway SE amp is prone to them.
Plus some positive feedback by current will go through C103 to cathode of the 1/st tube, but it may be made very small if to use an opamp with small input currents and increase R113 value in respect to R103.


This idea is result of my previous success with "Tower" amplifiers where I used SS output stage (a voltage follower loaded on a variable current source). They sound very pristine. Alligator is the way to implement similar trick in order to get a SE triode signature.
 
EC8010 said:
Ah, so you have a concertina phase splitter. I wondered about that earlier, but it could just have been a feedback path, depending on values. Have you considered the balance of the capacitive loading on each output?

Yes. But disbalance is not a big deal since CCS has huge output dynamic resistance. Also, the Concertina is already very disbalanced (VCCS needs much less voltage swing than a triode).

Meanwhile, I'm working on a 2'nd version that will be much easier to tune. I'm going to avoid a feedback path to cathode of a phase splitter that would add a dynamic disbalance.
 
Actually, the amp of a version 2 consists of 2 parallel parts: tube SE amp and a SS VCCS, input signal goes to both of them simultaneously.

The tube amp has self-balanced by a P-MOSFET bias.
DC bias for a VCCS is taken from output tube's cathode so idle currents of both shoulders are equal and cancel bias of output transformer.

Tube amp is inverting, VCCS is not inverting, so AC currents they supply equal and counter-phased (a tube part has a negative feedback by voltage, a SS part has a negative feedback by current). Since dynamic resistance of a CCS is very high a transfer function is determined by a tube part that is single ended with triode output.
 
gain matching?

The FET in the lower "shoulder" is not running as anti-current
mirror slaved (in reverse) to the tube "shoulder". It is servo'd
to balance DC well enough, but the AC gate drive drive through
the op-amp is a unity gain (because of your NFB capacitor.)
Gate tracks the cathode of your Concertina splitter, but not the
anti-current of the other shoulder.

Nothing here forces your FET to have a similar voltage gain to
the tube, nor any sort of anti-tube SEPP curvature... The way
you got it driven is sort of Concertina driven Pentode-ish FET.
Which would probably cancel a lot of SE in the real Pentode of
the other shoulder...

I do like your A2 superdiven grid for the output tube...
I am uncertain about automatic baising. I don't think
there is anything wrong with it, just don't understand
your circuit yet. I probably need to stare at it for awhile...

This might be a fun one to tie grid #1 to the cathode,
and drive screen grid #2, just to see what happens...
No sense leaving any good rules unbroken!
 
Looks like C207 in Version 2.0 is just 10 pF for HF stability. So the R218 current pickoff is allowing the Op Amp to program the Mosfet current linear to its + input.

I believe Anatoliy is not trying to program accurate anti-triode current here (in the Mosfet), but just lighten the plate loading on the triode side. So resistors are used to set the VCCS gain here, which could even be set to greater than -1*triode AC current. This would allow the triode to operate at more nearly constant current with more constant Mu as a result.

I think the design goal here is more along "why not do SET better than SET can" instead of just equaling SET. Clearly the triode characteristic will be modified some, even with nominal -1 current gain, since the non-linearities of the triode will be altered. Whether this will still sound like a SET, or some sort of triode/pentode(Mosfet) P-P remains to be seen.

Don
 
Re: gain matching?

kenpeter said:
The FET in the lower "shoulder" is not running as anti-current
mirror slaved (in reverse) to the tube "shoulder". It is servo'd
to balance DC well enough, but the AC gate drive drive through
the op-amp is a unity gain (because of your NFB capacitor.)
Gate tracks the cathode of your Concertina splitter, but not the
anti-current of the other shoulder.

Nothing here forces your FET to have a similar voltage gain to
the tube, nor any sort of anti-tube SEPP curvature... The way
you got it driven is sort of Concertina driven Pentode-ish FET.
Which would probably cancel a lot of SE in the real Pentode of
the other shoulder...

I do like your A2 superdiven grid for the output tube...
I am uncertain about automatic baising. I don't think
there is anything wrong with it, just don't understand
your circuit yet. I probably need to stare at it for awhile...

This might be a fun one to tie grid #1 to the cathode,
and drive screen grid #2, just to see what happens...
No sense leaving any good rules unbroken!

Peter;
a first, a cap os very small, just for stability on high end. Second, concertina is very asymmetrical, with gain needed for proper balance. Anyway, I've abandoned this first version because NFB to the same cathode causes dynamic asymmetry.

smoking-amp said:
Looks like C207 in Version 2.0 is just 10 pF for HF stability. So the R218 current pickoff is allowing the Op Amp to program the Mosfet current linear to its + input.

I believe Anatoliy is not trying to program accurate anti-triode current here (in the Mosfet), but just lighten the plate loading on the triode side. So resistors are used to set the VCCS gain here, which could even be set to greater than -1*triode AC current. This would allow the triode to operate at more nearly constant current with more constant Mu as a result.

I think the design goal here is more along "why not do SET better than SET can" instead of just equaling SET. Clearly the triode characteristic will be modified some, even with nominal -1 current gain, since the non-linearities of the triode will be altered. Whether this will still sound like a SET, or some sort of triode/pentode(Mosfet) P-P remains to be seen.

Don

Do you mean a triode loaded on a CCS sounds less triode? I expect just more output power with equal distortions, or less distortions on equal output power, that is the same. If I get 50 W from a single trioded GU-50 it will be a big success!

But the main advantage is in more linear and less frequency dependent output transformer...

Probably 6F12P is overkill so older tubes with less transconductance may be enough, I do not know yet, need to experiment.
 
Re: Wavebourn

I'm not doubting that distortion can be reduced by lightening the load on the triode. And "AMEN" to getting rid of those awful air-gapped xfmrs.

But the question remains as to whether SET buyers will like the sound of a "SET" with less distortion. I couldn't tell ya at this point. But as to more power, that should be attractive indeed, one can use "SET" with ordinary speakers then.

Don
 
What! Not putting the big tubes in front with a see thru acrylic panel?

I have decided on a name for my emulated "SET" amplifier now:

Hodja Nasreddin's Magic Tubeless SET

Since the SET sound will be so convincing, no one will believe that there are really no tubes in it. So the emulation tubes will be partially hidden, but not so well that someone cannot discover their "hidden" presence, thus "confirming" there suspicion that it is really a tube design in disguise. Only fitting for an amplifier with many levels of revelation. (Pentode emulation of triode, and Mosfet P-P, not SE, breaks all the rules!)

Don :D
 
smoking-amp said:
What! Not putting the big tubes in front with a see thru acrylic panel?


They will be visible through vent holes on top.

I have decided on a name for my emulated "SET" amplifier now:

Hodja Nasreddin's Magic Tubeless SET

Since the SET sound will be so convincing, no one will believe that there are really no tubes in it. So the emulation tubes will be partially hidden, but not so well that someone cannot discover their "hidden" presence, thus "confirming" there suspicion that it is really a tube design in disguise. Only fitting for an amplifier with many levels of revelation. (Pentode emulation of triode, and Mosfet P-P, not SE, breaks all the rules!)
:D

One more version: Ali Baba and Forty Transistors. :D