Thoughts on this 807 design

[thevoice]

Hey people, I'm looking at building something along the lines of this 807 PP amp. It's basically a re-worked Williamson design (UL), which has taken into account a few of the critiques that have been written of that design over the years. More info on it here

Obviously it's not my design and I am much more highly skilled in the construction than in the real expert understanding of such things, though I view this as a way to learn and experiment.

Does anyone have any comments, critique of the design etc?

I have the chance to get custom made trannies for it, made by a retired transformer winder who remembers when valve technology was all there was, he still does jobs for friends. My impression is that the amp has been designed around available iron, rather than the other way, is there a way to work out exactly what OT specs would suit the amp? I'm thinking impedance here and any other specifiable characteristics, I'll leave the choice of how to actually achieve those things to him.

Thoughts?

 

[Tubes4e4]

Hi TheVoice,

Hey people, I'm looking at building something along the lines of this 807 PP amp. It's basically a re-worked Williamson design (UL), which has taken into account a few of the critiques that have been written of that design over the years. More info on it here

Thank you for mentioning my homepage - I am Tom Schlangen, the author

Does anyone have any comments, critique of the design etc?

Yes. On later research (the amp was finished already) I found out that the chosen UL operation point of the 807 output tubes is quite suboptimal, but I never changed it anyway, this amp will stay as is.

I suggest to consult the 807 UL OP curves available somewhere on the net. Don´t remember the URL right now, but it may have been at the Sofia / Automatica curve tracer homepage.

Tom

 

[Tubes4e4]

Ah, I found the curves again:

807 tetrode, UL and triode plate plate curves

Lots of other interesting plate curves at that, too.

Tom

 

[thevoice]

Cheers Tom, I spoke to you briefly on email a little while ago about the PS for this.

Thanks for the link, now I have to work out how to use the curves properly, like I said I deal much better with the physical construction than understanding the electronics.

I've got some reading to do.

 

[Miles Prower]

"Does anyone have any comments, critique of the design etc?"

Yes. I'm not a real big fan of the Williamson topology. It seems absurd to use a cathodyne to split the phase, and then use a LTP to further amplify the two signals. The LTP could do both jobs at once. But that's just me.

Preferences aside, there are only three problems that I see. First off (see attached) the tail on the differential amp is too short. You might get away with that with a solid state LTP, however, 6SN7's don't have the gain that transistors do. That thing needs a CCS instead. Secondly, I don't see any provisions for adjusting either AC or DC balance. Finally, the 807s are missing the plate suppress resistors. The 807 Application Report calls for 100 ohm plate stoppers if operating class AB1 (like here) or 47 ohm stoppers for Class AB2. (Adding a 10 turn coil across the resistors wouldn't hurt either.) After all, the 807 is an RF power amp tube.

 

[cerrem]

To give you an idea of whats happening in the tail of the diff. amp... The Tail R is 1.2K ohms.....Now looking into the cathode of each 6SN7 you will have roughly 3.1K ohms.... at that operating point...
Idealy you want this tail to be infinite in a perfect world.... Since you get a current divider affect ...you want all the AC current to couple....
SO you can do the math and see whats acceptable in terms of coupling co-efficent and then determine what value you need in the tail to accomplish this.... On the other hand, the coupling was known to be poor...that why they used a balance pot to adjust the AC signal balance....
Also, some amps use feedback technique in the tail...this method known as "bootstrapping" is used to make a small Tail R look effectively like a larger impedance to AC signal.... this feedback method also self balances itself and the output stage when properly implemented.....

Chris

 

[gingertube]

The 807 Application Report

All you ever wanted (almost) to know about using the 807
The definitive 807 Users Guide.
http://www.retrovox.com.au/STC807.pdf
Cheers,
Ian

 

[audiousername]

[B First off (see attached) the tail on the differential amp is too short. You might get away with that with a solid state LTP, however, 6SN7's don't have the gain that transistors do. That thing needs a CCS instead.[/B]

It is not necessary for that stage to be truly differential. The phase has already been split by the cathodyne anyway. I can't speak for Tom, but I would say that the idea of the stage is to buffer the outputs of the cathodyne and to increase the gain of the amp - the common cathode resistor is probably there as a convenience rather than anything else.

 

[Tubes4e4]

Hi Miles,

First off (see attached) the tail on the differential amp is too short. You might get away with that with a solid state LTP, however, 6SN7's don't have the gain that transistors do. That thing needs a CCS instead.

Why do you think so? This is not a LTP splitter stage after all, but just a driver stage with highly balanced input signal already. So, what real advantage should a "long" long tail or CCS give there? Maybe a bit more CMRR? Not worth the effort, I think.

Secondly, I don't see any provisions for adjusting either AC or DC balance.

Yes, that would be an improvement in most cases. In my implementation I did go for well-matched (3 point and gm) 807 tubes simply because I have a lot of them and the output stage works almost completely (up to 10 watts) in class A1, were balancing is much less important as with class AB or even B, as you surely now. And the output tubes run nowhere near their max. Pd (less than 80%, IIRC).

Finally, the 807s are missing the plate suppress resistors. The 807 Application Report calls for 100 ohm plate stoppers if operating class AB1 (like here) or 47 ohm stoppers for Class AB2.

Again, this amp runs up to 10 watts purely class A, with maybe 2 additional class AB watts on top. No parasitics whatsoever, no plate stoppers needed in this case.

If one goes for class AB instead (which conceptually would be a completely other amp, able to crank out at least 30-40 watts and fixed bias should be used), plate stoppers probably would be helpful, indeed.

After all, the 807 is an RF power amp tube. [/B]

Nah, it is just a 6L6 adapted to RF class C high voltage duty by adding a top cap plate connector

Tom

 

[Tubes4e4]

Hi Jason,

It is not necessary for that stage to be truly differential. The phase has already been split by the cathodyne anyway. I can't speak for Tom, but I would say that the idea of the stage is to buffer the outputs of the cathodyne and to increase the gain of the amp - the common cathode resistor is probably there as a convenience rather than anything else. [/B]

Exactly.

Tom

 

[Miles Prower]

In my implementation I did go for well-matched (3 point and gm) 807 tubes simply because I have a lot of them and the output stage works almost completely (up to 10 watts) in class A1, were balancing is much less important as with class AB or even B, as you surely now.

Stop calling me Shirley.

No, I didn't know that. Up till a few months ago, all my experience was with solid state. I'm still working on it, however I have some ideas borrowed from SS practice that's worked well for me.

My design philosophy is different.

 

[thevoice]

Cheers guys, some good food for thought.

Here's something of a noob question, to use the plate curves properly I should pick the flattest line, so for ultralinear operation this would be somewhere around the -15V or -10V bias point and around 130V on the plate, giving a plate current not too far under the 125mA max? (Going simply by eye.) This assumes a screen voltage of 400V and I think it is less with this design.

Taken from the earlier posted http://www.mclink.it/com/audiomatica/tubes/807.htm

I'm reading through the STC app report right now.

Peter.