Bob Cordell's Power amplifier book

Hi Tryphon,
I thought that TIM was proved to be a myth.
Really ? Don't had you read any of the works of Mrs Otala, Leinonem, Jung, Curl, Ensomaa, Raimo etc. mostly published in AES ?
Anyway, Google is your friend.
The feedback happens practically in phase with the input signal. It's not like you have to wait for electron drift.
I don't see the correlation with some feedback delay ... But what happens in a closed loop when the feedback is not able to follow the input signal speed ?
But I let Mr. Cordell answer: it is his thread.

Oh, BTW, when i'm able to "see" a distortion on an Oscilloscope, I begin to really worry ;-)
 
Last edited:
Administrator
Joined 2004
Paid Member
Hi Tryphon,
I'd love Bob to weigh in on this.

As for my reading habits, yes. I have read a fair amount of material on this, especially anything that Walt has written. I do not have access to the AES library, I gave up on them after a certain person made every question a math contest and refused to exchange information any other way. The SMTE folks also made the meetings frustrating. One dense member tended to halt the flow of information for constant debates and explanations.
I don't see the correlation with some feedback delay
The most oft argument I hear is concerning the time delay between input signal and the time it takes for the feedback signal to arrive. This is nonsense in this context.
But what happens in a closed loop when the feedback is not able to follow the input signal speed ?
If you are having bandwidth issues with the input signal being out of band as you are suggesting, time to scrap that design and hire an engineer to design it. The only way you can open the feedback loop is to get close to slew rate limiting, or to clip which does all kinds of other nasty things. You can't have a system having slew rate problems if the input signal is all in-band. That's what the input filter is for. It rolls off the frequencies that might cause slew rate problems. Same for the output filter as that is another valid input to the amplifier. As long as you can filter out out of band signals from the input, output and power supply (good engineering practice), you will not see this problem as long as the amplifier is operating properly (barring a component degradation allowing ultrasonic to RF to exist in the circuit).

I think the issue you're talking about would be seen in the "high end" products using over-simplified circuitry. In my book, incomplete designs. Similar designs were once found inside the very low-end products. In this area, Jutan, Candle and Soundesign products. These were low powered things that didn't have enough power to hurt themselves. They got away with murder with these incompetent designs. They sold thousands upon thousands of this junk. Normally with a turntable mounted on top.
Oh, BTW, when i'm able to "see" a distortion on an Oscilloscope, I begin to really worry ;-)
That was my point actually. Of course, looking at the residuals from a THD analyser with an oscilloscope can show you things of value. So an oscilloscope has value depending on how you use it.

-Chris
 
Really ? Don't had you read any of the works of Mrs Otala, Leinonem, Jung, Curl, Ensomaa, Raimo etc. mostly published in AES ?
Anyway, Google is your friend.
I don't see the correlation with some feedback delay ... But what happens in a closed loop when the feedback is not able to follow the input signal speed ?
But I let Mr. Cordell answer: it is his thread.

Oh, BTW, when i'm able to "see" a distortion on an Oscilloscope, I begin to really worry ;-)

Actually, it is not TIM itself that was proven to be a myth. Rather, the assertion by Otala et al that wide open-loop bandwidth and/or low amounts of negative feedback were needed to avoid TIM was proven to be wrong.TIM is basically high-frequency intermodulation distortion that is caused by slew rate limiting or the onset of SR limiting. TIM was also referred to as Slewing Induced Distortion (SID) by Walt Jung. Amplifiers with very low TIM are routinely built that actually have open loop bandwidth less than 1 kHz and as much as 60 dB or more of negative feedback at 1 kHz.

The key to avoiding TIM or SID is for the amplifier to have adequate linear slew rate, with good margin. A well-designed amplifier with slew rate of, say, 50 V/us is unlikely to have TIM or SID.

If an amplifier has very low THD-20, it is nearly impossible for that amplifier to suffer from significant TIM, since the offending underlying nonlinearity will create both THD-20 and TIM. This is why having low THD-20 is important. It has nothing to do with hearing harmonics of 20 kHz. Rather, THD-20 is a measurable symptom of an amplifier's propensity to suffer from high-frequency distortions like high-frequency intermodulation distortion. 19+20 kHz CCIF IM is also a good measure of high-frequency distortions, and its IM products lie in-band.

Cheers,
Bob
 
Member
Joined 2005
Paid Member
Bob,

Your Hafler update uses a different diode configuration in the positive part of the clamp circuit than Hafler used, could you confirm it’s OK before I connect my output stage back up, the outputs are very hard to replace if I lost them
 

Attachments

  • 72A2DF65-CF75-4BB2-A90D-4E6716901F65.jpg
    72A2DF65-CF75-4BB2-A90D-4E6716901F65.jpg
    979.9 KB · Views: 503
Bob,

Your Hafler update uses a different diode configuration in the positive part of the clamp circuit than Hafler used, could you confirm it’s OK before I connect my output stage back up, the outputs are very hard to replace if I lost them

I'm pretty sure its OK, since the amplifier was actually built and tested, including driving it into clipping.

Cheers,
Bob
 
THX Bob for your last answer.

But I have one more question.

What do you think about placing the compensation capacitor as I show here:

675114d1523717486-bob-cordells-power-amplifier-book-bjt_12-png

This is Miller Input Compensation (MIC), as described in my book. I used it in my MOSFET power amplifier with error correction (see JAES article on my website). This compensation is very effective, and allows very high slew rate. The 50 W MOSFET amplifier I used it on did 300 V/us.

However, the Miller compensation loop formed with MIC encloses 2 stages, and the compensation loop itself must be compensated, usually with a ULGF around perhaps 10 MHz. In the case I used, this was done with some lag-lead compensation. So the complete compensation can be a bit trickier than simple conventional Miller compensation.

Cheers,
Bob
 
Member
Joined 2011
Paid Member
15 days have passed since I mentioned that I got surprising results from googling

Bob Cordell McGraw Hill​

This gives many search hits, of which two are relevant to the 2nd edition of Bob's book. Neither of the two search hits are pirated copies of the book. Not bootleg .pdfs , not low-quality scans on "scribd" . Just web links to web pages.

However the Moderation Team felt (see post #8924) it is better to be safe than sorry, and they deleted my recommendation to google for yourself. So, to save everyone the difficulty and annoyance of running a google search, here are its results. Click for yourself: no copyright violations here. Not even a wispy aroma of copyright violation.

Link Number One

Link Number Two

You can see why the 2nd edition has a different publisher than the 1st edition. Hold onto your copy of the 1st edition, it is now a collector's item.
 
Pioneer SA-8800 NSA amp design and simulation

Hi Bob and others,

Maybe it is too late, but I'll ask anyways, Bob have you included any discussion on the NSA topology in the new book?
I have been helping out a member on audiokarma, fix his Pioneer SA-8800. The original drivers/ops bjts are long obsolete. When they try modern day available parts they cause way too much bias and it oscillates. I simulated the design in ltspice to see how it works. We seem to have the amp with stable bias now.
I will include the sim, models, pdf, if anyone is interested in looking at it and can comment on it.
I could use some help in determining the phase margins and other stability criteria involved.

Cheers
Rick
 

Attachments

  • Pioneer SA-8800 NSA PA(ltspice).pdf
    43.6 KB · Views: 137
  • Pioneer SA-8800 NSA PA.asc
    20.9 KB · Views: 116
  • standard_bjt.txt
    78.7 KB · Views: 137
Hi Bob and others,

Maybe it is too late, but I'll ask anyways, Bob have you included any discussion on the NSA topology in the new book?
I have been helping out a member on audiokarma, fix his Pioneer SA-8800. The original drivers/ops bjts are long obsolete. When they try modern day available parts they cause way too much bias and it oscillates. I simulated the design in ltspice to see how it works. We seem to have the amp with stable bias now.
I will include the sim, models, pdf, if anyone is interested in looking at it and can comment on it.
I could use some help in determining the phase margins and other stability criteria involved.

Cheers
Rick

Hi Rick,

Yes, I will be touching on Non Switching Amplifiers. It is not a really long treatment, however.

Cheers,
Bob
 
T I guess I don't understand the used book market on Amazon.

Cheers,
Bob

Places like Goodwill accept books by the ton -- the one in Summit on Morris Ave has an entire dumpster bin. Some poor soul sorts them and they go on EBay or Amazon.

I got my copy of "Noise in Measurements" van der Zeil for $3.99, "Noise" (VDZ) was a bit more, and Spangenberg "Vacuum Tubes" was around $20, 1964 "Radio Amateur's Handbook" $3.99.

You can get the organic chemistry textbook on Amazon for less than I paid for it in 1969!

Christophe Basso's "Designing Control Loops" $121!