Bob, what typically would cause the closed loop pole to move around? (I am assuming here you are talking about the -3db GBW). A few posts back I showed my power amp distortion curve which was 'flattened' out by loading the VAS with 15K resistors. However, that design was from 2006 and I have moved on and don't load the VAS anymore.
Hi Bonsai,
The closed loop pole location depends on the product of the open-loop gain and the open-loop bandwidth, thus the term GBW. If the open-loop corner frequency remains the same, but the transconductance of the input stage gets modulated by signal amplitude (as in Gilbert's example), then GBW gets modulated, and this in turn modulates the closed loop bandwidth and thus the location of the closed loop pole.
It is notable that when the VAS is loaded with resistors, the input stage must work harder at ALL frequencies, not just the high frequencies. This increases the average amount of signal stress on the input stage, and can actually cause more signal-dependent gm modulation in the input stage, in some cases leading to greater PIM if the PIM results from signal-dependent gm modulation of the input stage. Resistively loading the VAS to get higher open-loop bandwidth is an example of a well-intentioned design approach born of intuition and a failure to understand all of the mechanisms at work.
Cheers,
Bob
The goal is to get reasonably high OLG -3dB corner without loading VAS resistively.
I beg to disagree. The closed loop bandwidth does indeed modulate as well as the open loop bandwidth. This is the 'final' measured result, of course. However, if the gain changes and the slope is -90 degrees in the open loop, then PIM can be generated.
Bob Cordell is not the only engineer to look into this problem. He might dismiss it, but we have a more difficult time, especially with listening experience.
Bob Cordell is not the only engineer to look into this problem. He might dismiss it, but we have a more difficult time, especially with listening experience.
I might say that it was not Otala who got me interested in this design direction. It was Walt Jung and his EDN article (that I basically subscribe to), 'Op-Amp Audio Realizing High Performance Bandwidth Limitations, ED ONLINE ID #7207, Dec.1, 1998.
Second, was Barrie Gilbert's article also found on Walt's website. (I'm pretty sure).
Finally, Mitch Cotter, FM designer, (who apparently first pointed this out to Matti Otala).
Mixing Barrie Gilbert's calculations with Otala's concern about crossover distortion in output stages being the primary open loop gain shifter (something ignored by Barrie Gilbert, in order to concentrate on the input stage) makes the problem a bit more real and obvious to why many audio designs are scrapped at the end of their normal life, and others are lovingly preserved, (Like the Otala Amp).
Second, was Barrie Gilbert's article also found on Walt's website. (I'm pretty sure).
Finally, Mitch Cotter, FM designer, (who apparently first pointed this out to Matti Otala).
Mixing Barrie Gilbert's calculations with Otala's concern about crossover distortion in output stages being the primary open loop gain shifter (something ignored by Barrie Gilbert, in order to concentrate on the input stage) makes the problem a bit more real and obvious to why many audio designs are scrapped at the end of their normal life, and others are lovingly preserved, (Like the Otala Amp).
Cherry is another major contributor. I tend to not go his direction, however.
I want to tell you a little story about Dr. Cherry.
Back in the late 1960's, Cherry and Hooper wrote the definitive textbook on amplifier design. Something like: 'Low Pass Amplifier Design' and it contained almost 1000 pages of 'everything you would want to know about amplifier design' EXCEPT slew rate, slope distortion, distortions related to rate of change, etc, were NOT IN THE BOOK. Both Walt Jung I looked in our own copies of the book.
10 years later, you have thought that he 'invented TIM' along with Bob Cordell, and he wanted to rename it 'Slope distortion'. Well, he would have had the right to it, IF he had put it forth in his textbook. But, alas, he did not.
Personally, I tend to side or at least see the point of people who find problems in the direction that we normally go, in audio, in order to progress.
It is my opinion that both Bob and Ed go with what might be considered the status quo. This is the major philosophical difference between us.
I want to tell you a little story about Dr. Cherry.
Back in the late 1960's, Cherry and Hooper wrote the definitive textbook on amplifier design. Something like: 'Low Pass Amplifier Design' and it contained almost 1000 pages of 'everything you would want to know about amplifier design' EXCEPT slew rate, slope distortion, distortions related to rate of change, etc, were NOT IN THE BOOK. Both Walt Jung I looked in our own copies of the book.
10 years later, you have thought that he 'invented TIM' along with Bob Cordell, and he wanted to rename it 'Slope distortion'. Well, he would have had the right to it, IF he had put it forth in his textbook. But, alas, he did not.
Personally, I tend to side or at least see the point of people who find problems in the direction that we normally go, in audio, in order to progress.
It is my opinion that both Bob and Ed go with what might be considered the status quo. This is the major philosophical difference between us.
Your memory is failing, I (and others) pointed out simply that there are no in-harmonic frequencies because they are impossible.
Here's an excerpt from Vol. 1 No. 1 of the Analog Dialog (1967). Large signal BW/distortion and its relationship to slew rate was well known.
Attachments
Was this in Cherry and Hooper's book? No, I don't think so. OF COURSE, we knew about slew limiting. In 1967, the engineers in my department, Ampex instrumentation division, were complaining to me about the slew rate limitations of the UA709, the best that was available at the time. In fact, I was a near expert on the UA709, in some ways, because I had to measure 100 of them with a special analyzer that we built, with a Fairchild app note, at Friden in 1966. However, slew rate was not what we were most concerned with, on that project, BUT our instrumentation recorders went up to 2MHz. No IC's for us, thank you. We did it all by ring of 3, discrete, using the fastest devices we could get.
I once did a survey of Slew Rate limiting in the 1960's with just about everything available to me and in silicon valley at the time. Very few references to it, except how to measure it, on occasion. It wasn't really a problem until IC op amps became available.
I once did a survey of Slew Rate limiting in the 1960's with just about everything available to me and in silicon valley at the time. Very few references to it, except how to measure it, on occasion. It wasn't really a problem until IC op amps became available.
I don't recall either Cherry or Cordell as claiming they were the first to recognize these types of distortion. I don't think anyone else but you on this forum CARES who was first. I think most of us just want to understand, and we don't care where the info comes from, so long as it is valid. Maybe it is the status quo that follows Cherry and Cordell, rather than the other way around, because they have explained their positions mathematically.
But I hate to go down this road again, because I get tired of the mud slinging.
The heart of these arguments has been the issue of whether or not OLG is a critical issue. The subjectivists say it is, the objectivists say it isn't, based upon their math. While I haven't seen any problems with the math, I haven't seen the issue of phase change between the open-loop pole and the closed- loop pole as being a possible issue that might possibly explain the bridge between these camps. I tried to bring up this issue a year or so ago, but got no response. I'm not saying that it explains anything, but I have never seen the issue discussed. If feedback is applied at a different part of the cycle in different amps, might not that account for a difference in sound that doesn't upset the math?
- Status
- Not open for further replies.