Yessir, Bob Stuart sure is one of the good guys. I like reading his papers.
I still remember his two Lecson power amps. For the time, they were well above the norm sound wise, and I imagine they'd still sound good to this day. That is a very serious man who should not be taken lightly.
But not too literally either. Probably partly my fault, I took his words out of context. He was talking about setting design objectives and commented that there are some overkilled amps around, which were unnecessarily complex for what they did.
What concerned me about Bob was when he tried to patent the complementary differential bipolar input stage that I gave to him at his party, in 1972, drawing it on a napkin and giving it to him. It really ticked me off, when he tried to shut me out of my own design. Still, we are amicable when meeting at CES, etc. I DO read his papers, and I recommend that many of you do, as well.
Last edited:
Don`t you think people generally like some type of audio distortions?
Do you think the harmonic structure matters if distortion is low enough?
Perhaps I was not clear enough. I like to use multiple pairs of parallel power devices because no matter what I might be doing, if it's a power amp, it will be expected to be able to deliver amperes of current however you look at it. And I am very suspicious of all loudspeakers as loads. More can't hurt, less can be disasterous.
For the record, I also use by default triple Locanthi type output stages, with a predriver, driver and output device block, obviously, in series. I do so because I find this relaxes the requirements from my VAS very significantly and provides for a more linear performance over its operating range, as well as better stability. And if it's not stable, then it's nothing.
Tubes are cumbersome in many ways, and have to date, and likely to continue to have, lots of excess noise at low frequencies.
But to the extent that you operate triodes in a region of constant mu, with very light plate loading, most of them are nearly distortionless. There are some drawbacks if pushed too hard, like the depletion of the space-charge around the cathode, but this is under fairly extreme conditions. And for the lightest loading, although one can contrive constant-current loads with more tubes and more voltage, transistors can make life a lot easier. Such stages will have a good-sized Miller capacitance, so things will slow down for high source impedances. Cascoding tubes makes things faster but much higher distortion.
I understand that Boyk would have students construct simple small-signal gain stages using tubes and transistors, and have them compare via listening to determine which they preferred. Simple enough, and the tubes will win, even with the stress-producing blind testing regimens.
A very rough pre-feedback equivalent for bipolar transistor stages to achieve good linearity is the application of translinear techniques made most popular by Gilbert. You will need more than one transistor... If you can fabricate devices with intrinsic matching, i.e. on a wafer, areas can be scaled for the desired current gains. Of course the topologies for highest performance are intricate.
Whitlock has opined that tubes have the advantage of no signal-dependent variable capacitances, and I think this is important if not entirely true---but the effects are small. After all, a lot of the interelectrode C's are mostly from the supporting structures, not the active area of the grids, and absent microphonic effects the residual effect is from the amount of electrons in play.
Tubes also have much longer thermal time constants, whereas solid-state devices have thermal behavior well within the audio range, and even well above it, for small-enough high-frequency devices. So although scope amp designers had to develop techniques to deal with tube thermals, for audio we don't usually care about a small shift in a "d.c." baseline, except as it may affect stages downstream.
JC's remarks about parallel structures also brings to mind the technique of compensating for a known load impedance. This involves more parts, but they are assisting a primary structure and not in series with it.
It's all fun. Now if we could explain why there is so little correlation, or even negative correlation, between measured performance and the accounts of reviewers, beyond hand-waving about measuring the wrong things, some of us might sleep a little better.
A prime example of added complexity affecting the "personality" of the
semiconductors is the Baxandall (super-pair)/Hawksford VAS on my last
"attempt".
The Hawksford nearly eliminates the early effect "personality"... and the
Baxandall eliminates load , saturation , reduces Cbb.
All this for just a benign added pole. GOOD tradeoff.
Group delay also rises slightly , but we end up listening to a voltage
stage with no bad "personality" !
OS
All agreed, Brad. The fact that I don't care much for them means just that and only that, it's a personal preference and nothing else, certainly not some higher truth.
To be honest, I have heard tube gear which did sound first rate and which I wouldn't mind owning. Then again, I have also heard SS gear which equalled that tube gear in terms of sound quality.
Overall, God bless good sound, no matter what makes it happen.
Yes, almost. Had one VFA that did not "like" this "fancy" VAS (below).
It had twice the CLG of the CFA's that do well with it.
The effect (below 2) on the current of the baxandall base stopper
was magnified greatly with that VFA.
On the CFA's , nothing .... even on a 100mhz oscilloscope ??
OS
Attachments
That's an easy one: because they look/talk/measure completely different aspects of a unit. It would be a miracle if there WAS correlation.
Jan
My timestep=numsample
numsample is ( .param numsampl=simtime/Freq/((simtime/numcyc)*FFT).
Variable ?
OS
Feel free to sleep, the answer is well-known.
The guys that Jan mentioned, as well as others, have actually demonstrated these performance positives using new topologies, not just recycling well-known circuits by (for example) swapping out a CCS for a resistor (though that can often give great performance as well). I have a phono stage running that only uses tubes for voltage amplification but has silly-low distortion and no Miller effect; Frank and Menno and Guido have likewise done clever things with their circuits to overcome engineering disadvantages of earlier topologies. It's an interesting and wide-open field, but let's be honest, it's not exactly addressing the real problems in audio that need to be solved, it's just doing the same thing as other circuits do, but doing it differently.
A quick question about your power supplies (V1,V2). Do you include any series resistance in those voltage sources?
Would you expect this to change anything ?
Your question - I did not in that post , but have done so now.
When I build something like this , I would have overkill cap multiplier's
"hugging" that buffer.
OS
- Status
- Not open for further replies.