Hi Bob, Thank you for taking the time to provide such a detailed answer. That will certainly help point me in the right direction.
I remember reading something about that in Chapter 2.1 under 'Speed and ft' page 21.
I will re-read over that as well as the chapter on the 'output stage design' chapter 10.
Thanks again for your valued input.
Here is a discussion of this cap where I did some tests to determine what the most logical value would be. My conclusions were that you don't want the cap to store any energy under 20KHz, because that will cause bias pumping in response to audio signals and thus high order harmonics. It's counter-intuitive, but ultimately true. At RF, degenerating the drivers will limit their ability to pump up bias during oscillation or RFI.
http://www.diyaudio.com/forums/soli...s-ab-power-amp-200w8r-400w4r.html#post3502255
http://www.diyaudio.com/forums/soli...s-ab-power-amp-200w8r-400w4r.html#post3502255
This is a good analysis. There are many ways to look at this issue. In hi-end audio, "everything matters", but that needs to be prioritized, and we need to be consistent in our different areas of the pursuit of perfection.
In my view, if we give equal importance to output transistor base suck-out as to other subtle hi-end issues, the driver(s) should never lose control of the output transistor at drive levels up to full power at 20kHz into 4 ohms. If this can be done with dc-coupled techniques, as by running adequate bias current through the drivers, that is a wonderful thing, and the best way to do it while avoiding the risk of subtle reductions in sound quality. But this can burn considerable power in the driver and is only practical when output Triples are used (they should always be used in any serious hi-end amplifier anyway).
If instead we use a speed-up capacitor, we have to deal with the issue of hangover and bias pumping (I like that term) at full power, 20kHz into 4 ohms if we are to be fair in the comparison). It does not take a large number of millivolts of hangover to disturb the bias current of the output stage to lie well away from the Oliver criteria.
With a DC approach to the problem, one thing that can help a bit is to put a diode in series with the resistor connecting the emitters of the complementary pair of drivers (when their bias is set in this way instead of by base-emitter resistors on the power transistors). This merely cuts in half the impedance between the driver emitters when one is off. It is not a complete panacea.
All of these approaches involve compromises.
Fast output transistors help a lot.
Cheers,
Bob
Bob,
Did you cover Error Correcting circuits?
Like this one in Electronics World by William de Bruyn June 1997
If not, would you like to comment on the approach?
View attachment Cordell Diff.pdf
THx-RNMarsh
Did you cover Error Correcting circuits?
Like this one in Electronics World by William de Bruyn June 1997
If not, would you like to comment on the approach?
View attachment Cordell Diff.pdf
THx-RNMarsh
Yes, in fact I had a whole chapter in the book on error correction, specifically my implementation of Hawksford error correction (HEC) for vertical MOSFETs. The Electronics World article shows my implementation of HEC, so of course I think highly of the approach
.Cheers,
Bob
Here is the rest of that article --
Also, in the same issue is a very good investigation/analysis of "RF Effects on AF" by Cyril Bateman.
THx-RNMarsh
Last edited:
Here are the two full Electronics World articles about error feedback quoted above :
K. H. Ellis, december 1996 :
http://iodau.pagesperso-orange.fr/Ellis.KH_'Amp%20err%20corr%201'%201996-12_EW.PDF
W. de Bruyn, june 1997 (2.3 Mb) :
http://iodau.pagesperso-orange.fr/Bruyn_'Error feeback' 1997-06_EW.pdf
Some years ago, I tried to enter in contact with W. de Bruyn (member of this forum) about the quite simple Ellis's circuit, to ask if he tested it. I had no reply.
K. H. Ellis, december 1996 :
http://iodau.pagesperso-orange.fr/Ellis.KH_'Amp%20err%20corr%201'%201996-12_EW.PDF
W. de Bruyn, june 1997 (2.3 Mb) :
http://iodau.pagesperso-orange.fr/Bruyn_'Error feeback' 1997-06_EW.pdf
Some years ago, I tried to enter in contact with W. de Bruyn (member of this forum) about the quite simple Ellis's circuit, to ask if he tested it. I had no reply.
Last edited:
The Ellis amp will SELF-DESTRUCT if overloaded at HF and perhaps also at MF and at lower levels.
What happens is MOSFET Cgs charge up and can only discharge through the 680, 1k and 'R1'. On HF overload, the capacitors charge up to very high voltages until the MOSFETs die or the PSU gives up.
Adding extra Cgs like the lower 220p makes this happen at lower frequencies and/or levels.
The Texan amp this is based on is also a POS .. but for other reasons.
Last edited:
Hi Stuart,
I'm struggling to get it done. Me, the eternal optimist, am behind my own schedule - again
. I am quite close with the text, but still have a few circuits to build and measure. Then after the final manuscript is completed, the process of publication also takes a few months. I'm targeting six months out for it to hit the streets. It will be about 37 chapters and 750 pages.I really appreciate your interest and patience.
BTW, I'll be at Burning Amp.
Cheers,
Bob
Hi David,
Some of the new material includes:
1. A new chapter on building amplifiers based on one of the evolution amplifiers of Chapter 3; a complete amplifier built and measured.
2. A new chapter on Noise that goes moderately deep.
3. A new second chapter on output stage design, covering more material.
4. A new chapter on switching power supplies (SMPS) for audio amplifiers.
5. A new chapter on professional (pro Audio) power amplifiers.
6. Expanded chapter on IPS/VAS design, including a complete lateral MOSFET amplifier with a full-complementary IPS employing the LSK489/LSJ689 dual monolithic JFET complementary devices.
7. Expanded chapter on advanced feedback compensation, including more information on TPC/TMC and a vertical MOSFET TPC/TMC amplifier design and measurements.
8. Use of the LTspice VDMOS ksubthres parameter for modeling sub-threshold behavior in power MOSFETs.
9. Expanded chapter on measuring class-D amplifiers, including some example measurements.
Cheers,
Bob
Wow all that extra detail sounds amazing Bob.
I look forward to this new addition.
Will you be able to expand on power supply design a little further. I found myself looking futher to understand actual design calculations / component selection, soft start circuits and the requirements for high frequency filtering of the supply rails etc.
A big thank you for all the effort that you're going to Bob. Keep up the awesome work. I for one can't wait to own a copy of your book.
Sent from my SM-G920I using Tapatalk
I look forward to this new addition.
Will you be able to expand on power supply design a little further. I found myself looking futher to understand actual design calculations / component selection, soft start circuits and the requirements for high frequency filtering of the supply rails etc.
A big thank you for all the effort that you're going to Bob. Keep up the awesome work. I for one can't wait to own a copy of your book.
Sent from my SM-G920I using Tapatalk
This is a good suggestion. I think I've already bolstered some of that in the expanded chapter on advanced feedback compensation, but I'll also look to see if there are a few things that can be added that would be helpful in the earlier chapter on NFB.
Cheers,
Bob