John Curl's Blowtorch preamplifier part II

[Bob Cordell]

Originally Posted by john curl

Now, what is really missing from most hi fi design? What really separates mid fi from real hi fi? Guess what, it is still, OPEN LOOP BANDWIDTH! That is the secret of the CTC Blowtorch, Charles Hansen's designs, and most tube products. It is the almost forgotten ingredient, neglected or even suppressed by many engineers, like the use of salt or real butter is by some cooks, today, to improve the taste of food. We have found high open loop bandwidth to be the key ingredient added to an audio design that makes it more successful than other similar designs. It has to do with the generation of PIM that is equal or even more important than TIM. Listening tests prove this to me, as well as theory, and even measurement. That is my opinion, and it works for me. Your opinions may differ, but please don't berate me about it.

John,

We have been down this road many times. We obviously disagree on just about everything you have said here. It is your opinion; that is all. I cannot rob you of your opinion - it is what it is, and I know it will never change.

However, you are wrong on the factual stuff, and that is where we have been down this road before. There is no theory or measurement that supports what you say.

TIM is not exacerbated by low open loop bandwidth in any properly designed feedback amplifier.

PIM is not exacerbated by low open loop bandwidth in any properly designed amplifier.

PIM, as measured, is often WORSE in a no-feedback amplifier.

PIM, as small as it is, usually exists in no-NFB amplifiers.

Please read my PIM paper again.

Cheers,
Bob

[john curl]

Please Chris, keep it within reason, would you?
By your definition, Charles Hansen, Nelson Pass, and I are by DEFINITION, not design engineers, as we have degrees in PHYSICS.
I might remind you that I STOPPED my technical history, with 3 years of formal analog engineering courses (I minored in engineering) and a degree in Physics at 1969, 40 years ago! I went back to class, in later years, to fill in the rest of the course material that I had previously missed in analog engineering. How do you think that I first met Dr. R.G. Meyer? I took his courses at both the undergrad and later the grad level in analog engineering. He first knew me as one of his students.

[john curl]

Thank you Bob, for contributing. NOW, is the UA741 a 'properly designed' op amp?
Second, your test did not find PIM, because it was too easy, even though it was recommended by Otala in the first place. In hindsight, it is fairly obvious.
Third, quadrature, by definition, invites PIM. Do you deny this?
Fourth, do you consider Gilbert's analysis irrelevant?

[stinius]

John and Bob

Good posts by both of you, my problem is of course that I’m placed somewhere in between.

A simple question: If you have a low TIM, PIM frontend and use a ClassD buffer that only reproduces the signal from the frontend, will that buffer introduce TIM, PIM and other IM’s?

I have attached a waveform showing the input and output of the ClassD buffer.

Thanks

Stinius

[john curl]

Stinius, the Class D output stage adds a completely different problem. It usually shows CROSSOVER DISTORTION, and the output filters ruin high frequency damping and add static phase shift.

[Bob Cordell]

Originally Posted by john curl

Thank you Bob, for contributing. NOW, is the UA741 a 'properly designed' op amp?
Second, your test did not find PIM, because it was too easy, even though it was recommended by Otala in the first place. In hindsight, it is fairly obvious.
Third, quadrature, by definition, invites PIM. Do you deny this?
Fourth, do you consider Gilbert's analysis irrelevant?

Hi John,

The 741 is not a properly designed op amp for audio. It has terrible HF THD, miniscule slew rate, and, yes, plenty of TIM and PIM.

You assert the PIM test is too easy, departing from Otala. This one is new.
When I measured PIM and reported the results 25 years ago, I was just following Matti's protocol and his definition of PIM at the time (you would have dismissed my results if I had not measured it as specified by Matti). To the extent that SMPTE IM is too easy a test, so is PIM.

What test do you suggest for PIM?

What PIM tests have you done?

Barrie Gilbert's analysis is not irrelevant. Bear in mind that I articulated the PIM mechanism 25 years ago in my paper and came to the same conclusion as Barrie (yes, I did indeed show how the application of NFB could cause PIM). I have not read Barrie's paper in over a year, but I'm pretty sure that he did not mention open loop bandwidth. Barrie's paper does not support your assertion about the need for open loop bandwidth any more than mine does, if I am correct in my recollection of his paper.

Cheers,
Bob

[Bob Cordell]

Originally Posted by stinius

John and Bob

Good posts by both of you, my problem is of course that I’m placed somewhere in between.

A simple question: If you have a low TIM, PIM frontend and use a ClassD buffer that only reproduces the signal from the frontend, will that buffer introduce TIM, PIM and other IM’s?

I have attached a waveform showing the input and output of the ClassD buffer.

Thanks

Stinius

Hi Stinius,

Class D amplification has made great strides in the last 10 years, but still falls short in a number of ways for hi-end audio. It also presents some measurement challenges. This is largely due to the usually limited high-frequency bandwidth. For this reason, in-band testing like 19+20 kHz CCIF must be used, combined with full spectral analysis.

Cheers,
Bob

[john curl]

The SINE-Square test, Bob.

[Bob Cordell]

Originally Posted by john curl

The SINE-Square test, Bob.

Hi John,

I assume you are referring to the DIM-30 or DIM-100 test, wherein the results are taken from a spectrum analyzer. This was the original test for TIM proposed by Otala. It is a pretty good test, and is definitely adequately stressful on an amplifier. A bit of a pain in the butt reading off all of those spectral components, but much more practical nowadays that spectrum analyzers are not as exprensive.

However, are you able to distinguish between TIM and PIM with the DIM test?

If so, please explain how.

BTW, my old MOSFET EC amp had almost unmeasurable DIM on the real sine-square test, even though it had high feedback and a low open loop bandwidth (but a 2 MHz gain crossover and 300 V/us slew rate). That would seem to disprove your assertion that wide open loop bandwidth is necessary for low TIM and PIM.

Cheers,
Bob

[john curl]

Yes, fig. 2 in 1976 paper. It is obvious, on reflection.