Yes, I'm at UCSD. My focus is way over in the biomedical/materials end of the spectrum, so my circuit skills and controls theory are years old and relatively untested, nor was I particularly clever even then. Which is probably why I ask such sophomoric questions.
Definitely here to learn; actual circuit results are a bonus. Will go have a look at Horowitz's work on QFT (thanks for the reference!) and dig deeper into my old textbooks on control theory while I get a few test circuits built for these TDA's I have to dig deeper.
Thanks,
Daniel
most of the QFT, Robust design stuff happened after I graduated - Kharitonov's theorem hadn't been published in English
and I suspect it still isn't common EE undergrad fare - maybe in a Control specific concentration
Robust Control - The Parametric Approach shows Kharitonov in action
Classical Feedback Control does Classical Frequency Domain a bit differently than usual, has worthwhile insights - see the online chaper 4 for loop gain shaping
and its really only those already interested that care in diyAudio - not all of our design gurus display much mastery of even the basics of feedback theory you would expect a undergrad EE to have as a minimum required class
and you can probably do good enough to be audibly transparent with very basic feedback theory in audio power amps if you start with much faster devices than the power chip amps
and I suspect it still isn't common EE undergrad fare - maybe in a Control specific concentration
Robust Control - The Parametric Approach shows Kharitonov in action
Classical Feedback Control does Classical Frequency Domain a bit differently than usual, has worthwhile insights - see the online chaper 4 for loop gain shaping
and its really only those already interested that care in diyAudio - not all of our design gurus display much mastery of even the basics of feedback theory you would expect a undergrad EE to have as a minimum required class
and you can probably do good enough to be audibly transparent with very basic feedback theory in audio power amps if you start with much faster devices than the power chip amps
Last edited:
Thanks, JCX, for the links.
My younger brother's ME curriculum had much, much more control theory than my EE degree did, but that might have had more to do with my focus. Neither here nor there now, nonetheless.
Agreed on the baked-in compromises (esp in the OPS) of these power chips limiting what you can do with a relatively conservative TPC/TMC to start with sufficient linearity and end with more than enough loop gain at 20 kHz to eliminate any last audibility. There's the obvious compromise between cost/complexity and performance (esp when you move towards high power).
My younger brother's ME curriculum had much, much more control theory than my EE degree did, but that might have had more to do with my focus. Neither here nor there now, nonetheless.
Agreed on the baked-in compromises (esp in the OPS) of these power chips limiting what you can do with a relatively conservative TPC/TMC to start with sufficient linearity and end with more than enough loop gain at 20 kHz to eliminate any last audibility. There's the obvious compromise between cost/complexity and performance (esp when you move towards high power).
This is reasonable because there is a lot of control theory required for Mechanical that is not relevant to Electrical/Electronics.
Mechs often have to worry about Multiple Input Multiple Output situations, say a spacecraft to be stabilized in pitch, roll and yaw, all inter-coupled.
And often seriously non-linear systems, plus other complexities.
Unfortunately there are effects relevant to electronic amplifiers that are not usually considered in "Control Theory".
For instance, it is usually assumed that the control system can be made arbitrarily fast compared to the "plant" (delay is usually lumped into the plant).
Also, Feed-forward thru the control system and load effects of the feedback network are almost never considered.
A result of the omission of feed-forward is that Control Theory's so-called "Bode's Sensitivity Integral" is a complete misnomer.
It is exactly what Bode took immense care to demonstrate is NOT reliably equal to the sensitivity
Horowitz is the only "Control Theory" writer who I have noticed to even mention some of these issues.
Thanks to JCX for his recommendation that I reread the work, otherwise I may not have continued past the somewhat "crackpot" initial impression of Horowitz's last book.
Also thanks for the Kharitonov link.
Seems the PDF is a bit flaky, is that just my set-up?
Best wishes
David
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.