Tube Amp Compensation Q

[atiq19]

Dear Friends,

I have been wondering why compensation mechanisims are not frequently used in tube voltage amplification stages. Contrastly, I have seen the following methods being applied in solid state amps:

1. Miller compensation (known as Lag comp??)

2. Phase lead compensation ( used by JLH and others)

3. Combination of the above to tame over the top reactive load.

4. 2-pole compensation (well documented by D. Self)

5. PILL compensation scheme..

Now the question is why don't we see these in tube amps? Is it the case that tube amps don't suffer from instability issue?

Could someone shed light on this matter?

Happy Holiday!

-Atiq

[SY]

I've moved this over to "Tubes."

In any case, tube amps have been built with all sorts of feedback schemes. Certainly because of the presence of the output transformer, the feedback tends to be much lower (20dB is a lot in this world!), so not as many tricks need be played to keep things stable. But compensation and the equivalent of Miller compensation are quite standard. There are also quite a few schemes (such as cathode feedback around the output stage) peculiar to tubes that are not seen in transistor amps. Different devices, different topologies, so unsurprisingly, different means of lowering distortion and output impedance while keeping things stable.

[atiq19]

Thanks SY, I was thinking of moving the post but u did it.

Correct me if I am wrong but the compensation techniques native to ss amps are also used in tube amps. However, out put x-former somehow limits nfb so does typical compensation scheme. Also triodes got its own inherent feedback properties..

I sarted this question with a hybrid amplifier in my mind where a buffer satge like the following one, developed by Steve Dunlop, driven by a tube front end:

Krill - The little amp that might...

What sort of compensation do you recommend in this hybrid? Lets assume it is driven by Broskie Aikido front end with 5751 tubes for gain.


Going back to original question, how compensation is realised in this modified ultrapath amp where output is connected with a transformer:


http://www.tubecad.com/2008/08/blog0147.htm

Cheers

Atiq

p.s: how do u attach image?

[atiq19]

Sorry, just realised, links in the above post does not work. You have to cut & paste to get there, I'm afraid.

[SY]

I fixed your links.

Re JB's Ultrapath, no compensation is needed because there's no feedback loop. This is usually the case with SETs, since the idea of the amp is not to have a linear path from input to output, but rather to "beautify" the sound. The distortion figures and source impedance bear testimony to my admittedly cynical observation.

[atiq19]

Ok.

I got it in simple terms,

'no gnfb calls for no mandatory frequency compensation'

but without gnfb, I am open to higher distortion (desirable SET sound by many listener) and uncontolled ouput impedence (???) or damping factor ...


Therefore, If I wish to drive steve dunlop's non-switching class-B solid state buffer with an Aikido tube amp, I wouldn't worry about lead/lag or 2-pole compensation.

Now what happens when a gnfb loop or error correction loop (as in Pax amp by Jan Didden, utilise AD844 CCII for MJ Hawksford principle) is put on across steve dunlop's non-switching class-B buffer, while driven by a non feedback tube gain stage ? Do we care about compensation for VAS stage?

Many thanks for your patience.

[SY]

I'm unfamiliar with that buffer- can you provide a link?

[atiq19]

here is PAX H.Ec buffer:

http://www.linearaudio.nl/paXdocs/pa...20apr%2008.png

Cheers.

[SY]

I know the Pax circuit, I meant the other buffer you mentioned. In either case, the tube block will not have a lot of gain. It will generally be run as an independent unit, with the loop running around the solid state block. IOW, you run the tube circuit open loop (it's pretty linear that way and has appropriate gain), then set the ec to give unity gain for the solid state block that the tube stage is driving. Compensation goes inside the ss block, it's not needed in the tube stage.

[atiq19]

here is a strip down version of CCII HEc loop:

http://www.diyaudio.com/forums/attac...amp=1207509381

Note that pin X, Y, Z of AD844 form basic HEc loop which could be applied across instead of any traditional gnfb loop. Now the question is: is there any benefit of this loop over traditional gnfb loop. I can understand the availability of ready-made Current conveyer ad844..

Later..