andy_c said:
Well, I have no disagreement there at all 🙂 However if proponents of his design philosophy are going to make remarkable claims about the the PIM credentials of such designs………
G.Kleinschmidt said:
I think you'll find that the justification for these claims is going to be based on subjective observations. I don't think anyone really knows how much PIM the thing really has. Maybe John could ship his amp to Bob and have him measure it with his PIM analyzer
.G.Kleinschmidt said:
Oh, that was just an obtuse comment referring to the "size wars" that often happen here. Hope you didn't take it literally! 🙂
G.Kleinschmidt said:
In order to judge is solution optimal or suboptimal you have to define criteria of optimization, otherwise your jugment is not technical, especially in technical discussion.
Wavebourn said:
Well I for one claim that driving the non-linear input capacitance of a double EF with a ~1.1k impedance is far from optimal, as far as open loop phase modulation is concerned. I don't know how I could explain why any clearer. And I doubdt that the resultant phase modulation (that could be eliminated with Miller compensation) would do any good for measured PIM.
G.Kleinschmidt said:
Well I for one claim that driving the non-linear input capacitance of a double EF with a ~1.1k impedance is far from optimal, as far as open loop phase modulation is concerned. I don't know how I could explain why any clearer. And I doubdt that the resultant phase modulation (that could be eliminated with Miller compensation) would do any good for measured PIM.
You forgot about such factors as available then devices and prior arts. Semiconductors were dirtier, masks were uglier, some technologies did not exist at all. Plus, the designer could have some non-technical criteria that he had to satisfy in order to earn money to continue technically creative work. For example, to reveal why "our" amps are better that "others": particularly, because (for example) they don't need a Miller compensation that together with feedback create phase modulation that "our" competitors even don't know how to measure.
Wavebourn said:
Yeah. Like convincing people that he was some sort of "guru", and that the rest of the world was full of @$#$. Apparently that worked, for a while at least - until people discovered that exactly the opposite was true.
andy_c said:
It works exactly opposite: some people appoint you a Guru and get their profit on it, while others throw in you rotten eggs. Been there, eaten that.
I am serious here. I've observed this stuff for a long, long time. Otala tried to position himself as "the guy who knew what was really going on in the face of ignorant opposition.". His attitude in his AES papers literally reeks of that.
Wavebourn said:
This is just meaningless speculation and pollywaffle. We’re discussing the merit of a particular design and the claims made for it. I couldn't give a toss if Santa Clause designed it.
G.Kleinschmidt said:
I'm not trying to dispute what happens in the closed loop case one way or the other; I believe that it depends on many factors and I don't like to generalize. What I do disagree with you about is this statement you made:
"I don’t buy the “swamping the non-linear load” argument. By bunging a resistor onto the VAS you aren’t getting rid of the non-linear load at all (the driver or pre-driver stage is still connected), you are just bunging a linear load in parallel with it.
The non-linear part of the VAS load is still going to cause distortion, and you’ve just added more due to the drive requirement for that added resistor."
I'm simply trying to show you that a linear device in parallel with a non-linear device can have a more linear parallel combination as andy_c showed mathematically. It is a fact, like it or not.
OK, well we have several different discussions going on here. I was mainly responding to JC's 1M//100M case where he measured a reduction in distortion. Yours may represent a real amp, but perhaps one that does not respond in the same way as the HA-911.
However, your real amp is going to have non-linear device Miller capacitance as I pointed out previously, probably not a 1K load on the VAS, and a real non-linear resistive component load to the VAS. None of which you modelled. You're model *is* highly idealized, sure it does show phase modulation that is obvious.
I don't care enough to model JC's exact case or enough to argue the obvious about paralleled impedances.
G.Kleinschmidt said:
Relax, Glenn; Santa Claus don't exist. But if he did exist he existed long time ago and it is not speculation that motor cars were not available then so he had to ride horses. And it is a historical fact that wise smart people in Great Britain believed that the main problem of 20'th century will be an environmental pollution in big cities: it will be too hard to clean after horses. I can prove that documentally. However, you may argue that they were totally wrong, but you know what happened after.
Edit: it is just an analogy.
PB2 said:
Well I did not dispute this fact, and here is something for you to think about. Suppose a THD analysis of the collector current of a VAS, operating in a closed loop amplifier was performed, firstly with a 100k, non-linear load, and then with a 1k resistor in parallel with that 100k non-linear load. How do you think the THD residuals would compare in magnitude?
PB2 said:
The amp in question *does* have a 1k load on the VAS (a pair of 2k resistors just as in the Otala design). The possible non-linear resistive aspect to the driver inpedance is irrelevant to the point under discussion.
I was only pointing out that an amplifier having a VAS with a high output impedance (such as the Otala design) can suffer from significant open loop phase modulation due to the non-linear load capacitance that may be presented to the VAS. The whole purpose of idealising the simulation was to single out and demonstrate the very real significance of this mechanism of open-loop phase modulation.
If you want to see further sim’s of a complete amplifier that demonstrates my point exactly, have a look at the F5 simulation KSTR mentioned in reply.
And Johns’ HA-911 tests have nothing to my phase modulation simulation in reply to Andy whatsoever! Could we stop mixing these up please?
Rafael.luc said:
1) The collector impedance of the VAS is not determined by the quiescent current.
2) In the Otala amp, there is no form of negative feedback from the VAS collector such as Miller compensation to dramatically lower the collector impedance at HF (such as in a “Blameless” or similar design).
3) There is therefore no pole-splitting effect with the input capacitance of the driver stage.
4) There is a pair of 2k2 resistors loading the VAS output to ground.
5) These two resistors not only load the VAS but set its output impedance (in isolation) to an almost purely resistive 1.1k.
5) 1.1k is not low for a VAS output at the ULGF.
G.Kleinschmidt said:
C'mon Glen, the VAS transistor itself has an output impedance of VA/Ic where VA is the Early voltage... This can be quite low for oldish trannies, e.g. 100V/20mA = 5k, that is 2.5k for a symmetrical configuration. One good reason not to run the VAS to hot, the product of gm*Ro is a technology constant, VA/26mV
Wavebourn said:
Hi Wavebourn,
I'm sorry if I missed a question posed to me.
Let me see if I can provide an answer. First, I assume the C you refer to is the Miller compensation capacitor.
I think that I and a number of others showed why the application of negative feedback to an othersise PIM-free amplifier can create PIM. That process does not depend on any nonlinearity in C and can be shown to occur in a fairly ideal amplifier as long as there is nonlinearity in the input stage.
It is a bit more speculative why negative feedback reduces PIM in the case of a nonlinear C, but it is usually the case that the application of NFB reduces pre-existing distortions, including those related to phase.
The measurements I did for my PIM paper appeared to show a net reduction in PIM when NFB was applied. This suggests to me that the PIM created by the NFB was less than the reduction of pre-existing PIM in the open-loop amplifier.
I hope this explanation helps, and freely admit that what I am saying about the reduction of PIM by NFB is a bit speculative.
Cheers,
Bob
andy_c said:
Not unique to him. In my one run in with the AES I got the impression that the AES intelligencia hold that audio is special somehow and needs a different spin than say other EE heavy dissiplines.