Going on from simons 45 thread

Subjects that don't have their own home
steve s
Shed dweller
Posts: 2829
Joined: Wed May 30, 2007 6:19 pm
Location: east yorks

#16 Re: Going on from simons 45 thread

Post by steve s »

andrew Ivimey wrote: Mon Aug 15, 2022 6:47 pm Yeah but doesn't push pull throw out a different kind of distortion or..... it's the way I make'em.
I wonder Andrew, is there something going on yet to be uncovered ..
The tube manual is quite like a telephone book. The number of it perfect. It is useful to make it possible to speak with a girl. But we can't see her beautiful face from the telephone number
User avatar
Paul Barker
Social Sevices have been notified
Posts: 8863
Joined: Mon May 21, 2007 9:42 pm

#17 Re: Going on from simons 45 thread

Post by Paul Barker »

The effect on 2nd should be calcable from the curves and measurable.

I have some questions Ive not got a view on but might need an answer.

Take two identical valves in series, ignore loading and coupling for simplicity. I mean the “effect” isnt limited to transformer coupling.

Perhaps Morgan Jones would say to me if he were in my room now, “measure the effect RC coupled” to limit other distortions. He said to me in Chris’s flat about my then 212 amplifire. “Use a high value load resistor for the input valve to use up all that surplus voltage.”

Perhaps another grounding is to find a valve with curves that demonstrate equivalence in 2nd distortion not amplitude but shape. Hold that thought.

Next consider valve one drives valve 2 so valve one’s distortion is multiplied, valve 2’s distortion isnt amplified.

Question 1. Is this the end of the story. Case for coupling identical valves disproved?

Question 2. Not enough gain from two in series. So what happens if we make this 3 series identical valves? Surely we have now obliterated the dual stage theory. To fix this must we make 4 stages?
"Two things are infinite, the universe and human stupidity, and I am not yet completely sure about the universe." – Albert Einstein
Max N
Old Hand
Posts: 1453
Joined: Thu Aug 16, 2007 6:10 pm

#18 Re: Going on from simons 45 thread

Post by Max N »

On distortion cancellation - an example where this is successfully used is the Aikido with the same valves used in the gain stage and the follower stage, with the same loading on each stage. Because the follower has (approximately) zero gain, it is operating at the same signal level as the gain stage.
So Steve, if you used an Aikido gain stage and follower in front of the output valve, you would have close to a distortion free front end and driver and would be left with just the distortion of the output stage.
Steve, exploring a bit further your original idea of two identical output valves in series to cancel distortion. The second valve (the one driving the speaker) would need to be configured as a cathode follower. The first valve would somehow have to see a similar load to the second valve. I can’t see a way to achieve that.
But you might want to try an Aikido gain/driver - that would let you hear what the output valve sounds like when driven with a very low distortion signal. It also has the advantage of a low impedance driving the output valve.

Sorry Paul, this crossed with your post, so I wrote this before reading yours.
User avatar
Nick
Site Admin
Posts: 15706
Joined: Sun May 06, 2007 10:20 am
Location: West Yorkshire

#19 Re: Going on from simons 45 thread

Post by Nick »

ignore loading and coupling for simplicity.
But that's the problem, you can't ignore those without getting into an entirely made up world of spherical cows.

The transfer curve of a valve, any valve, is a set of curves that define a complex surface in 3D. Or to put it another way, the transconductance of a valve varies with current, so unless you have the two valves seeing the same load and traversing the same voltage, the two curves will not be equal so you can not expect any repeatable cancellation.

There is a way that the two valves can see the same load and the same voltage excursion, but I have already mentioned that.

Yes, it may be that you like the sound of a valve, so using two of those may make a sound you like more. The same may also be true of two different valves.
On distortion cancellation - an example where this is successfully used is the Aikido with the same valves used in the gain stage and the follower stage
Is that true? I don't think it is because the two points the load is attached (the anode and the cathode) are not equivilant, so I cant see any reason to assume the distortion characteristics would be the same. What you will see is the total current draw of the two stages would be close to constant as equal loads would be swept over equal but opposite voltages ranges so the total current is constant, so you would get improved PSRR which is good. But I don't think either theory or practice means a anode follower followed by a cathode follower with equal load resistances would have zero distortion. AFAIK, the Aikido bit (though I may not be right, not been following that closely) is the injection of power supply noise into the cathode follower to further improve PSRR.
Whenever an honest man discovers that he's mistaken, he will either cease to be mistaken or he will cease to be honest.
steve s
Shed dweller
Posts: 2829
Joined: Wed May 30, 2007 6:19 pm
Location: east yorks

#20 Re: Going on from simons 45 thread

Post by steve s »

Nick wrote: Tue Aug 16, 2022 10:36 am
ignore loading and coupling for simplicity.
But that's the problem, you can't ignore those without getting into an entirely made up world of spherical cows.

The transfer curve of a valve, any valve, is a set of curves that define a complex surface in 3D. Or to put it another way, the transconductance of a valve varies with current, so unless you have the two valves seeing the same load and traversing the same voltage, the two curves will not be equal so you can not expect any repeatable cancellation.

Thats the bit that me thinking and inspired my opening post.

My thinking was valves have a very fast rise time, with Miller capacitance the 'slugger' so theres every chance they should give a reasonable accurate amplification ( excluding a variable 2nd h in the main based on the voltage being amplified)?

My own thinking (scarey as it may seem!) had me thinking if the coupling could be phase stable, the 2nd harmonic from the 1st stage as long as it is present, would be canceled by the 2nd stage, but it would be subtraction leaving the difference, if that makes sense ?

Its the same music voltage going through both( the signal and 2nd H would be amplified by the 2nd stage ) so the distortions should be in the same places, but still with varying amplitude between the stages.

So partial cancellation is a possibility

If that is happening to some varying level with amplifiers, is it worth including for that in our design if it actually has some substance, which it of course may not !

I my personal view is 2 stages sounds better than 3, until now I've always blamed the extra stage for just being there ?


My understanding is none of what I'm saying would change the distortion at maximum output, but if true would change the slope of the distortion
The tube manual is quite like a telephone book. The number of it perfect. It is useful to make it possible to speak with a girl. But we can't see her beautiful face from the telephone number
User avatar
Nick
Site Admin
Posts: 15706
Joined: Sun May 06, 2007 10:20 am
Location: West Yorkshire

#21 Re: Going on from simons 45 thread

Post by Nick »

So partial cancellation is a possibility
Its a certainty, but there is a lot in what you posted that I don't understand, and it gives me a uneasy feeling that we are not on the same page, so I am not sure its safe to say much more. Miller and rise time and phase is important, but I just think you are adding more complexity than is needed or is helpful.

A single stage will have a transfer function f1() such that at any time the output will be the result of the input with the function applied out = f1( in), there will be time based aspects, but its just going to muddy the water to consider them. I suspect that because we are talking about harmonics, you are thinking that's a time based thing. Its not, the distortion harmonics are a result of the non linear behaviors of f1(). If you recorded the signal and played it through the amplifier slowed down 1000 times, recorded the output and played it back a 1000 times speeded up so it was back in real time the same harmonics would be generated.

The two stage amplifier will look like out = f2( f1( in )), which is the result of applying the non linear function f1 on the input then applying the non linear function f2 on the output of that. That will look like a combined function fx() that is the combination of f1 and f2, but its not the simple addition of the two functions, so even if f1 is the same as f2 just in reverse the result will never just cancel out.

I think you are thinking out = f1( in ) + f2( in ) such that if f1 is the same as f2 they will cancel.

You can look at the result as partial cancellation, or you can look at it as a different result, it depends on the spin you want to make, but parcel cancellation is not cancellation.
Whenever an honest man discovers that he's mistaken, he will either cease to be mistaken or he will cease to be honest.
User avatar
Nick
Site Admin
Posts: 15706
Joined: Sun May 06, 2007 10:20 am
Location: West Yorkshire

#22 Re: Going on from simons 45 thread

Post by Nick »

the slope of the distortion
I don't understand what that is.
Whenever an honest man discovers that he's mistaken, he will either cease to be mistaken or he will cease to be honest.
steve s
Shed dweller
Posts: 2829
Joined: Wed May 30, 2007 6:19 pm
Location: east yorks

#23 Re: Going on from simons 45 thread

Post by steve s »

Nick wrote: Tue Aug 16, 2022 1:21 pm
the slope of the distortion
I don't understand what that is.
I was trying to explain that I thought if there where partial cancellation, it would be effective at lower signal voltages
Ie reduced distortion would be well within the valves operating envelope not at its limits



Thanks for your detailed reply too nick
You can look at the result as partial cancellation, or you can look at it as a different result, it depends on the spin you want to make, but parcel cancellation is not cancellation
Thanks nick I can see where your coming from,
Your last paragraph is interesting

My question would be ' is partial cancellation better than no cancellation
The tube manual is quite like a telephone book. The number of it perfect. It is useful to make it possible to speak with a girl. But we can't see her beautiful face from the telephone number
User avatar
Nick
Site Admin
Posts: 15706
Joined: Sun May 06, 2007 10:20 am
Location: West Yorkshire

#24 Re: Going on from simons 45 thread

Post by Nick »

I was trying to explain that I thought if there where partial cancellation, it would be effective at lower signal voltages
Ie reduced distortion would be well within the valves operating envelope not at its limits
Well, the distortion is normally dependent on signal level. But I feel as if you are still hung up on thinking of the envelope of the signal (ie how the peak signal level varies over a time) instead of the actual value of a signal that varies with time. The distortion is related to the actual value at any time not the peak value over time. One is the result of the other, but its the moment by moment value I am discussing.

Thing is what you are calling partial cancellation I would just describe as the result of combining multiple distortion sources. But all of this is just as true with two dissimilar valves than as two identical ones.
My question would be ' is partial cancellation better than no cancellation
Maybe. Maybe not, depends on your better.
Whenever an honest man discovers that he's mistaken, he will either cease to be mistaken or he will cease to be honest.
Max N
Old Hand
Posts: 1453
Joined: Thu Aug 16, 2007 6:10 pm

#25 Re: Going on from simons 45 thread

Post by Max N »

On distortion cancellation - an example where this is successfully used is the Aikido with the same valves used in the gain stage and the follower stage
Is that true? I don't think it is because the two points the load is attached (the anode and the cathode) are not equivilant, so I cant see any reason to assume the distortion characteristics would be the same. What you will see is the total current draw of the two stages would be close to constant as equal loads would be swept over equal but opposite voltages ranges so the total current is constant, so you would get improved PSRR which is good. But I don't think either theory or practice means a anode follower followed by a cathode follower with equal load resistances would have zero distortion. AFAIK, the Aikido bit (though I may not be right, not been following that closely) is the injection of power supply noise into the cathode follower to further improve PSRR.
You’re right Nick - I should do some revision before posting :D. There was some talk of distortion cancellation in Aikido amps back in the day, but I can’t find anything definitive. MJ has a bit on distortion cancellation between a common cathode and a cathode follower (4th edition page 181), but he has to jump through some hoops and it’s only really H2 that is significantly reduced. So I don’t really know where I got my idea from, but it’s clearly wrong!
User avatar
Nick
Site Admin
Posts: 15706
Joined: Sun May 06, 2007 10:20 am
Location: West Yorkshire

#26 Re: Going on from simons 45 thread

Post by Nick »

I would guess (without reading) that the only way you would get some cancellation is by making it a bad cathode follower by using a low mu value. I suppose you could cascade a few cathode followers to increase the effect.

Croft does something similar but they apply feedback around the cathode follower after a common cathode stage to make it a buffered anode follower. Its a nice idea, but can't be taken too far (or at least thats what I found when I tried to play with it) as you start to suffer from slew rate problems because you create a virtual null where the feedback and the input join, and the feedback and input resistors need to be large so miller gets to be a problem.

He also uses a mosfet follower for the buffer stage as well.

It is a nice way of providing a low output impedance and reduce the gain of something like a ecc83 down to where it works as a preamp. Oh, and valve noise starts to be a problem as the feedback can only do so much with that.
Whenever an honest man discovers that he's mistaken, he will either cease to be mistaken or he will cease to be honest.
Max N
Old Hand
Posts: 1453
Joined: Thu Aug 16, 2007 6:10 pm

#27 Re: Going on from simons 45 thread

Post by Max N »

MJ used an AC load - and I think the follower was a 7n7

I keep meaning to experiment with local feedback but not tried it yet….
User avatar
Paul Barker
Social Sevices have been notified
Posts: 8863
Joined: Mon May 21, 2007 9:42 pm

#28 Re: Going on from simons 45 thread

Post by Paul Barker »

Paul Barker wrote: Tue Aug 16, 2022 9:19 am The effect on 2nd should be calcable from the curves and measurable.

I have some questions Ive not got a view on but might need an answer.

Take two identical valves in series, ignore loading and coupling for simplicity. I mean the “effect” isnt limited to transformer coupling.

Perhaps Morgan Jones would say to me if he were in my room now, “measure the effect RC coupled” to limit other distortions. He said to me in Chris’s flat about my then 212 amplifire. “Use a high value load resistor for the input valve to use up all that surplus voltage.”

Perhaps another grounding is to find a valve with curves that demonstrate equivalence in 2nd distortion not amplitude but shape. Hold that thought.

Next consider valve one drives valve 2 so valve one’s distortion is multiplied, valve 2’s distortion isnt amplified.

Question 1. Is this the end of the story. Case for coupling identical valves disproved?

Question 2. Not enough gain from two in series. So what happens if we make this 3 series identical valves? Surely we have now obliterated the dual stage theory. To fix this must we make 4 stages?
“ consider valve one drives valve 2 so valve one’s distortion is multiplied, valve 2’s distortion isnt amplified.” This morning I cant remember what I was talking about in this statement.

But the idea I dont get is that the out of phase distortion cancells next stage distortion. But maybe Sukuma never said that? He liked his way but did he ever say it was a distortion cancellation mechanism? That isnt to say its not a good sounding system, neither that it is a good sounding system. Final analysis, does it please you? If you have tried it, what did you think?

Im sure someone in this group way back, said theyd been to Sukuma Cafe and they weren’t impressed. Ive never tried it and never heard anyone elses. So I cant comment on how it actually sounds.
"Two things are infinite, the universe and human stupidity, and I am not yet completely sure about the universe." – Albert Einstein
User avatar
Nick
Site Admin
Posts: 15706
Joined: Sun May 06, 2007 10:20 am
Location: West Yorkshire

#29 Re: Going on from simons 45 thread

Post by Nick »

But the idea I dont get is that the out of phase distortion cancells next stage distortion.
Look at the composite curve of a push pull output stage to see that in action. That happens because the same valves see the same signal (inverted) and drive the same load. The two curves of the transfer function cancel out giving a straighter one hence less distortion. In theory you could do something similar in series instead of parallel. Following that path will lead you to the Amity balanced amp.
Whenever an honest man discovers that he's mistaken, he will either cease to be mistaken or he will cease to be honest.
User avatar
Paul Barker
Social Sevices have been notified
Posts: 8863
Joined: Mon May 21, 2007 9:42 pm

#30 Re: Going on from simons 45 thread

Post by Paul Barker »

which I dont like. Neither did I like Barbaric balanced amp in the end. I liked my final stage only pp amp with a local feedback winding on the phase splitter. Its a great shame I tore it down after Owston and have never found those phase splitters since. C’est La Vis. There shall be more things lost and more found.

Also Clive Messer bought a balanced phono stage at vsac. It sounded crap when we got it back to the UK in his listening room. We both thought that. It wasnt a cheap one.
"Two things are infinite, the universe and human stupidity, and I am not yet completely sure about the universe." – Albert Einstein
Post Reply