What can-I answer ? i designed hundred of various amps and preamps in my life for various companies. Some sold with very good success. Do-you believe it should had been possible if i was "not even remotely understanding feedback". What about you ?
Now, you have to understand English is not my native language. And I use to think often "out of the box" with a very personal way to "feel" the things that we cannot see: "Electrons". Not always "academic" ? Right: In 1970, we had to discover by ourselves a lot of things in audio that are now taught in Universities.
A little difficult for you, to understand, while you have your nose glued in your school books and formulas ?
read my sign: "The Only Source of Knowledge is experience, everything else is just information” ©A. Einstein".
So, I apologise if I'm not always as clear and brilliant that i should want. But I can not accept that you question my technical competencies that several managers judged sufficient to pay me to benefit of them during decades, judging on the only thing that matters: results.
And, you know? in this area, the competition was tough.
Last edited:
For that matter, I sincerely do. Of course, you are entitled not to accept this.
Waly doesn't come across as a pigeon (though possibly not entirely potty trained yet)
I built the Alexander design in the need for speed days of the early 90s, anything with a slew rate over 100V/µs. The times when my tongue was still on the floor when I drooled over a Spectral DMA.
Can't say I disagree.
many in audio design seem to be "lone wolf", autodidacts
you can get quite far that way, though the Uni EE curricula is a big boost for those with the passion too
working in a big company in engineering teams, your designs hitting hard tech specs verified by independent test groups...
...having your design torn apart in design reviews by peers with collective man-centuriy of domain specific expertise is another way to learn your relative level of competence, quickly learn what you need to learn
too many who are the only electronic guy in the company may not ever learn what they don't know if their managers, peers aren't competent to judge their technical chops
"market success" is a weak metric for technical competence - especially in as Sy sez "a fashion market"
you can get quite far that way, though the Uni EE curricula is a big boost for those with the passion too
working in a big company in engineering teams, your designs hitting hard tech specs verified by independent test groups...
...having your design torn apart in design reviews by peers with collective man-centuriy of domain specific expertise is another way to learn your relative level of competence, quickly learn what you need to learn
too many who are the only electronic guy in the company may not ever learn what they don't know if their managers, peers aren't competent to judge their technical chops
"market success" is a weak metric for technical competence - especially in as Sy sez "a fashion market"
Last edited:
Why not to try to "understand" what i want to say, instead, when something seems unusual to you ?
By example, my way to analyse what happens in a loop with LTspice is this one:
I look at he bandwidth at the output of the input stage.
Strange, isn't it ?
I explain. Any difference of phase applied to the input signal when it crosses the amplifier's stages, then feedback's path, before subtraction, will result in an increase of the level there. It gives an immediate idea of the open loop+ feedback path linearity, and this with the differences between the two sides of a symmetric amplifier.
And you can see immediately what is your slope at OLG 1 limit in the same time (yes, Nyquist), and the margin.
Exploring this way, you can optimize first the feedback impedance to find the limit where it have no more impact on the phases (with the parasitic capacitance of the base in a VFA or Emitter in a CFA ) in decreasing its impedance, then work on the VAS and output stages to improve their bandwidths, when possible.
Yes, a very strange approach. And not even remotely correct (meaning, integrated in the EE body of knowledge). Your "success" in using it doesn't make it a "method" but more a "hunch" for particular cases.
I always assumed that is when endian-ness came from originally. Although apparantly some nasty DSPS are middle-endian and it's hard to eat an egg that way!
That's just awful. There's no good way to justify that from a programmer's perspective (I don't know what the microarchitecture looks like).
Very interesting remark, jcx.
Yes, most of the famous (now) names i crossed in my professional life were so.
A good reason for this "Audio" was not a matter in universities.
About autodidacts, the inconvenience can be they could had missed things that never came to their curiosity. The positive side is you understand better things that are the responses to your curiosity instead of being imposed to your unique memory. And, searching for answers in a non academic way, you fall on various points of views that can increase your level of understanding.
Last, if you are an autodidact, it is because the matter is your passion. And we all know that passion gives better result than forced efforts.
About working alone, or for an incompetent manager that you could fool, as far as I am in concern, it never happened in my life. For a good reason: you are never alone in your domain of competency, in big companies, and if so, you bore yourself fast. About incompetent managers, hum, they usually prefer to engage good looking and talking guys that i was not ;-) Their opinion about your work don't values you, you feel orphan soon.
Now, the things were very different in the 'good old time'. If nowadays most of the major companies are just managing "engineering industrial processes", copying on each others to produce very similar products, with very few room for real innovations, and no time left for them under the pressure of profitability (Google is an exception ?), putting most of their efforts on marketing and communication, the 70-80, appetite for innovation was at least one order of magnitude more.
Look at the patents. Most of them were on "ideas" in the 70-80, while now, most are on industrial processes.
Last edited:
IMhO.....
I gave the early background of CFA in Linear Audio - Vol 3 (pg 31-41). My independant developemnt of the circuit goes back to late 1970's. I published the topology in The Audio Ameature in 1980. After that publication I got a phone call from designers at Comlinear to my office at Lawrence Livermore National Laboratory asking details about the topology...... were there variations... why that approach... benefits i saw, BW.... speed....... etc Finally, he told me that they had been working on similar concept and had been ready to apply for a patent and since i published already, their patent may be worthless. In 1980 they patented a similar but different enough topology and got it approx 1982 in hybrid form. In 1982 Comlinearand Elantec made the first IC version. My original circuit produced well under .001% THD+N up to 20v p-p in the late 1970's. Get the article and read the footnotes and story of development and a very good headphone amp as well as an example of the topologies potential. That headphone amp is also measured and discussed under my name .. Marsh headphone amp... here at DIYAudio.
All those here who have done recent sim and actually build CFA and measured and listened have it right and I recall many similar thoughts when I was measuring my breadboarded line stage some 35-40 years ago.
In the newer app for PA, this was done many years ago also..... MY CFA PA of yr 2003 didnt change much under varying power output under load, didnt change much under any measured conditions; freq vs pwr vs thd vs Z, etc. Nothing changed (ok... just a very little bit change)
THx-RNMarsh
I gave the early background of CFA in Linear Audio - Vol 3 (pg 31-41). My independant developemnt of the circuit goes back to late 1970's. I published the topology in The Audio Ameature in 1980. After that publication I got a phone call from designers at Comlinear to my office at Lawrence Livermore National Laboratory asking details about the topology...... were there variations... why that approach... benefits i saw, BW.... speed....... etc Finally, he told me that they had been working on similar concept and had been ready to apply for a patent and since i published already, their patent may be worthless. In 1980 they patented a similar but different enough topology and got it approx 1982 in hybrid form. In 1982 Comlinearand Elantec made the first IC version. My original circuit produced well under .001% THD+N up to 20v p-p in the late 1970's. Get the article and read the footnotes and story of development and a very good headphone amp as well as an example of the topologies potential. That headphone amp is also measured and discussed under my name .. Marsh headphone amp... here at DIYAudio.
All those here who have done recent sim and actually build CFA and measured and listened have it right and I recall many similar thoughts when I was measuring my breadboarded line stage some 35-40 years ago.
In the newer app for PA, this was done many years ago also..... MY CFA PA of yr 2003 didnt change much under varying power output under load, didnt change much under any measured conditions; freq vs pwr vs thd vs Z, etc. Nothing changed (ok... just a very little bit change)
THx-RNMarsh
Last edited:
FWIW,
Technics manged 0.002% 200W 20-20K back in -78. Distortion at 100KHz 200W below 0.01%.
IMO, quite good performance. VFA design I might add.
Technics SE-A3
Technics manged 0.002% 200W 20-20K back in -78. Distortion at 100KHz 200W below 0.01%.
IMO, quite good performance. VFA design I might add.
Technics SE-A3
Walt Jung pointed out in one of his articles that this was shown with tubes in the 40's by some aerospace engineers.
No he isn't, ask Waly. In any case Demrow published the current feedback in-amp in 1968. It's right in your article the compound complimentary IS "current" feedback (in quotes for Waly).
Making it complimentary does not change anything in first principle. You can make a diff-in diff-out version of Demrow's circuit with two 844's and the gain resistor between the - inputs all the principles are the same.
Last edited:
Thats good too... esp for a VFA PA with the OPS devices available then. 8 or 16 Ohm load I presume. Who is the principle designer?
I have an A-P graph from 2003 of a PA (CFA) I did for oem.... did <.003 THD+N at 420W/4. And high SR as well. If one know what to design for/to.... most VFA and CFA can get it done. The CFA is just easier and simpler to make and stable. No monkey business fumbling around trying to balance THD numbers at higher AF with stability.
THx-RNMarsh
No he isn't, ask Waly. In any case Demrow published the current feedback in-amp in 1968. It's right in your article the compound complimentary IS "current" feedback (in quotes for Waly).
ask Waly?
Yes, that is my derivation.... unbal to complimentary push-pull/bal amp and direct coupled... that was one of my main reasons to do it that way at first... to make a direct coupled amplifier - no coupling caps-as well.
Modern ss CFA depend on currents for operation and thus lower internal voltages across devices - esp the non-linear C's --- a similar topology using higher Z values can be VFA with voltage controlled operation (tubes) but losses the benefit of current controlled and operated amps.
One of the inherant reasons for CFA low HF distortion is the low voltage across devices and circuitry with limited voltage swings and derived distortion. My original circuit used low Z circuit values and higher than usual (for the time) currents. The circuits Z's were 100-1K max and then I saw the behaviour of the CFA in action.
I am pointing to Waly and others that I and others have known about and used and refined CFA for decades now. Just a bit of history for those who do not know. I am not a complete idiot on the matter.
yes, it is easier in practice to stabilize.
THx-RNMarsh
Last edited:
- Status
- Not open for further replies.