john curl said:
Really?
I thought you were talking about personalities... Let's go back to PHASE INTERMODULATION? 😀
The question is, does feedback create them, or it does not correct them?
I might, once more, try to clarify why I am here, participating in this discussion.
I did NOT start the thread, however it seems that Matti Otala may be deceased, as nobody seems to be able to contact him. This is his legacy.
I happen to own an Otala amplifier and have read the original paper made by Jan L. and Otala, when Matti was at Phillips Research Labs, for a year or so, that describes the operation of this rather strange (for the time) amplifier.
However, in 1975, I was given the opportunity to try this amplifier, and it stood out from virtually everything, commercial, including Marantz (2 models), SAE and Quatre. It just sounded BETTER with STAX electrostatic headphones. I would not have expected nearly so much improvement, and I insisted on buying the unit from Electrocompaniet, even though this was their prototype. I used it, off and on for about 15 years.
This radical design (for the time) had a 20KHz open loop bandwidth, as defined by common acceptance on IC spec sheets, a very high slew rate, and pretty good distortion, mostly low order, due to the relatively high output current bias used. There are many other subtle factors that are commented on in the design paper, but I cannot comment on them here, without re-reading the paper.
For some reason, even though Matti Otala is not present here, people seem to be inclined to take cheap shots at the design, its measurements, etc. I am trying to defend what I have heard with my own ears, and my design experience with a similar approach. The fact that it has degenerated to this, is unfortunate.
I did NOT start the thread, however it seems that Matti Otala may be deceased, as nobody seems to be able to contact him. This is his legacy.
I happen to own an Otala amplifier and have read the original paper made by Jan L. and Otala, when Matti was at Phillips Research Labs, for a year or so, that describes the operation of this rather strange (for the time) amplifier.
However, in 1975, I was given the opportunity to try this amplifier, and it stood out from virtually everything, commercial, including Marantz (2 models), SAE and Quatre. It just sounded BETTER with STAX electrostatic headphones. I would not have expected nearly so much improvement, and I insisted on buying the unit from Electrocompaniet, even though this was their prototype. I used it, off and on for about 15 years.
This radical design (for the time) had a 20KHz open loop bandwidth, as defined by common acceptance on IC spec sheets, a very high slew rate, and pretty good distortion, mostly low order, due to the relatively high output current bias used. There are many other subtle factors that are commented on in the design paper, but I cannot comment on them here, without re-reading the paper.
For some reason, even though Matti Otala is not present here, people seem to be inclined to take cheap shots at the design, its measurements, etc. I am trying to defend what I have heard with my own ears, and my design experience with a similar approach. The fact that it has degenerated to this, is unfortunate.
I have the greatest respect for Dick Sequerra, as I have worked for him and have known him for about 30 years. I suspect that Scott does too, so maybe, Dick will give him enough input from his perspective, so that Scott will give me the benefit of doubt on these questions. I sure hope so. My other critics don't count as much, as to what they may think.
syn08 said:
I did call him a while ago, he does not discuss some of the more controversial stuff which is OK. I have openly asked several folks to be one of those "everyone in the room" to whom these effects are obvious. I am starting to think the karma of a skeptic offsets the delicate balance necessary.
john curl said:
I might, once more, explain to you:
A 20KHz open loop bandwidth spec means exactly **** without specifying at least the open loop gain.
john curl said:
Good, you are slowly starting to learn.
Lesson number two: check the notion of "gain-bandwidth product".
Wavebourn said:
This is a good question. The pencil and paper math that I have seen could all be duplicated with MATLAB or MATHCAD. So is anyone up for running an explicit example? I am thinking of a case in point to start with, an ordinary op-amp say 20MHz GBW with a 10Hz OL 3dB point and with the VAS resistively loaded to make the OL BW 20kHz.
Either program can give exact complex response for THD or IMD.
scott wurcer said:
Scott, it was done here on this forum in spice, matlab, you name it, on AD797, discrete circuits, you name it, nothing goes through. Anything that goes beyond complementary JFET handmatched pairs doesn't compile, it's hopeless.
scott wurcer said:
Sometimes it is good to synthesize things in order to analyze them.
The information you are giving is not sufficient for synthesis.
Ok, let's start from a very simple PIM generator. It does not contain a global feedback; it does not contain a feedback at all because it does not have any active element in it, but it does generate PIM. It's open loop bandwidth is very wide, it's gain is stable and equal to 1, in very wide band of frequencies (well, depending on output resistance of the source and input resistance of the measurement tool).
Now, answer honestly: do we need any feedback to generate PIM?
Now, answer honestly: do we need any feedback to generate PIM?
Attachments
Trevor White said:
As I understand it, Charles Hansen has chosen the path of no NFB because he strongly is in the no NFB camp philosophically. To his credit, he has taken his philosophy to a very high level of performance. In other words, given that he has chosen the no NFB path that he has chosen, he has executed it extremely well.
I think there have been a couple of interviews where he has espoused his philosophy against NFB, but those arguments have never held water in my view.
He may have articulated those views here in another thread, but I don't recall. It would make for an interesting discussion to revisit his reasoning.
Cheers,
Bob
Wavebourn said:
Hi Wavebourn,
Here is what I think the questions and answers are:
Does feedback create PIM? Yes, feedback may create PIM in an otherwise PIM-free amplifier.
Does feedback correct PIM distortion? Maybe, in an amplifier that starts out with PIM when it has no negative feedback.
Does an amplifier with high negative feedback and low open loop gain have more PIM than one with modest feedback and a 20 kHz bandwidth? No, as long as the closed loop bandwidth is the same and all else is equal.
Cheers,
Bob
Ok Bob, please let me repeat:
does this example contain any feedback? No!
Does it have a bandwidth narrower than 20 kHz? No!
Does it generate PIM? Yes!
http://www.diyaudio.com/forums/attachment.php?s=&postid=1856635&stamp=1245179508
does this example contain any feedback? No!
Does it have a bandwidth narrower than 20 kHz? No!
Does it generate PIM? Yes!
http://www.diyaudio.com/forums/attachment.php?s=&postid=1856635&stamp=1245179508
john curl said:
Hi John,
Matti has certainly left a worthwhile legacy, even if some of us do not agree with all of his conclusions about negative feedback. He definitely got us all thinking real hard.
As far as his amplifier goes, I for one have never criticized it.
As far as 20 kHz open-loop bandwidth goes, achieving that will generally FORCE designers to do the right thing in many cases - such as employ generous degeneration in the input stage. However, 20kHz open loop bandwidth is not necessary to achieve equal sound quality.
Cheers,
Bob
Bob Cordell said:
Bob didn't you mean "high open loop gain" in the first example? That was my example where one can show it in closed form. This is also the most common example used by the NFB crowd.
Wavebourn said:
Good point.
I think that any 180-degree all-pass network with the capacitor being a nonlinear capacitance will act as a PIM generator, and it will have unity gain and no amplitude intermodulation distortion.
Cheers,
Bob