The part I'm still trying to get my pea brain around is the case where bandwidth is large compared to signal passband but loop gain is small, for example "local" feedback like degeneration. Experience shows this to be relatively free of the artifacts of Baxandall/Putzeys disease, but no real explanation has, AFAIK, been given.
It's somehow related to bandwidth (obvious in retrospect but took me a while). But *not* in a first-order way that I can see. And *not* directly related to the difference between open and closed loop bandwidth even at high frequencies. But somehow.
And it's probably not itself a linear function. It's probably somewhat of a threshold effect - and with "strong attractors" at more than one combination of bandwidth, loop gain and forward transfer function. Just WAG's of course.
The 'round-the-loop Baxandall/Putzeys transfer function is different below and above open loop break frequency but is there a *general* difference that tells us something interesting? Just wish I had the mental horsepower to get to a more general picture. Oh well, still trying. Maybe this is Mother Nature's way of telling me to learn how to do simulations. Sob.
Thanks,
Chris
Last edited:
The explanation is that the loop gain can sometimes be quite large when degeneration is used, contrary to your assumption. This is why the artifacts can be small: there is sufficient loop gain to suppress them.Chris Hornbeck said:
This is not always true. For example, the cancellation of 3rd when Re=1/2gm - in this case the artifacts are large but happen to be antiphase so cancel.
(Still catching up...)
I'm still laughing!
And yes, I do own a cell phone. It is a "pay as you go" that costs $0.10 per minute. I use it in emergencies, and even then in speaker-phone mode. The operating frequency is about the same as a microwave oven, so I don't really want to press it against my head when it is transmitting...
(Almost caught up...well I actually cheated. I only read the last 20 pages to find out what all the arguing was about. That leaves around 100 pages I've overlooked -- but not missed.)
I'm having a hard time believing that you are asking me this stuff with a straight face.
I re-read your interview with Malcolm recently. Malcolm is very astute and should never be ignored. (That doesn't mean that he is always right, but he is always interesting at the least.) In the interview Malcolm stated his preference for open-loop ("zero-feedback") circuits, as several have done here.
When such an opinion is expressed here, it is inevitably attacked -- sometimes by you. In the interview you calmly accepted what he said and his reasoning. I would have to assume that if you disagreed with his reasoning that you would politely say something. But you did not, and in fact in your post you say that you agree with his reasoning.
But when Nelson makes a post saying exactly the same thing, you mock him and tell him that he better write a letter to the editor to explain exactly why Putzeys is wrong. Why the double-standard? I don't understand.
Gosh, Scott, I was just quoting Malcolm. Perhaps you should ask Malcolm what he meant.
Damn. For a minute I thought it was a compliment.
Nelson, I am only too glad to compliment you!
Thank you for continuing to innovate for over 30 years. You have pushed the envelope for solid-state and keep doing so. You constantly think outside of the box, are able to run in four different directions at once, and you keep a great attitude at the same time.
I therefore dub you the
designer of the 21st century!
So we have two possibilities - you quoted him but disagreed with him. Or you quoted him because you agreed. If the latter then its obvious he was wrong - in the sense that no digital converter has an output rise time of zero. Zero rise time would require infinite bandwith in the following amp; finite rise time requires merely finite bandwidth - still very large but a significant step down from the claim of 'infinite'.
Ah, now I see there's actually a third possibility
You quoted him but really had no idea of what he was saying.
Charles makes digital products for a living. And as all amps are 'finite' in bandwidth, so are any digital signals. However, when you work in nanosecond risetimes with digital, then the digital pulse is faster than almost any input that can be handled by most amps, and the amp will be stressed or overloaded by the ultrafast risetime of the digital pulse. It is obvious to Charles and me, anybody else have a problem with this?
Last edited:
There's an interesting point there. Interviews have different styles- they can be set up as a debate between interviewer and interviewee or they can be set up as a way to let the interviewee express their thoughts and to allow their thought processes to be understood by the reader. Both types of interviews have their place- the latter type, where the interviewer stays out of the way and puts the spotlight on the interviewee is (I think) harder to do and far more interesting. Jan does an all-too-rare excellent job of the latter.
Dear Charles, I really don't get what you are trying to accomplish here. Did I rub you the wrong way sometimes, somehow?
The piece from Malcolm you quoted gives some perfect sensible tech info. For the life of me, I can't find anything wrong with it. And there is nothing in it about open loop being better or not. And I never attack people for preferring open loop or asparagus or Irish beer or blondes, preferences are just that. I do try to ask critical questions if anyone says, for instance, that open loop is better because it has no TIM, as that is a tech statement that I believe is wrong or at least not automatically true in all cases.
What else - ohh right, Nelson. Mocking and double standard. Really Charles, your imagination gets the better of you. I believe Nelson knows me better than you do; I'd like to leave it at that, OK?
jan
Stuart, thanks for your kind words.
What I try to do is to find the area the interviewee has almost an emotional rapport with. If you find that, you get gem after gem. Try to make them feel relaxed and not attacked all the time you get far, far more interesting bits and pieces out of them.
I urge people to read Malcolms interviews, I will put them in my webspace. NOT because I did the interviews, but because of the gold that's in there in terms of ideas and circuit designs. But I would be dishonest if didn't admit I am proud of that interview.
I just read it again after Charles' accusations that Malcolm prefers zero feedback and I let it slip. What I found was this (in a discussion about 'analog jitter'):
"It just might be that simple, open-loop circuits, while having higher large-signal level distortion, potentially have less of these timing nonlinearities, which could explain their very good sound. I would need to get the sums together, but it just might be possible that this is one of the reasons why people prefer those simple, low-feedback amplifiers".
Now if anyone reads this as Malcolm preferring open loop amplifiers and that I should have attacked him for it, I really wonder who's the native speaker here.
Maybe, if you're so geared to finding conflict and fights in any and all statements, you can't recognize anymore a considered, balanced discussion.
Edit: See upper right column here.
jan
Last edited:
I pointed out once that a 5534 will not act as an integrator at HF i.e. the standard first order filter does not not keep the input away from the steps. Dr Small was in the audience and claimed that that can't matter because he invented it. I presented an alternatve (it's on the 797 DS) and spectra to show the effect, waste of time, so I disrupted the Q&A by saying I had no problem with SET amps since it's all an illusion anyway.
Ah - the 2nF cap to ground. Works great for the measurements, tends to make the sound less enjoyable though. Is that why the paragraph is headed 'Professional Audio' rather than 'Audiophile' ?
How could you have overlooked the actual key point that Malcolm stated:
"Modulating the OL transfer function
means you modulate the circuit’s
closed-loop (CL) phase shift. What is
interesting is that it looks remarkably
similar to correlated jitter; they share
a family resemblance (Fig. 3). It also is
similar to what people have been talking
about as dynamic-phase modulation in
amplifiers."
Now maybe I am just dense, but when he speaks of "dynamic-phase modulation" it sounds to me like he is saying PIM in an erudite fashion.
So Malcolm is saying that adding feedback increases the PIM of a circuit. John has been saying the same thing for years, yet he is harassed, mocked, and treated with derision for it -- sometimes by you.
Nelson recently said on this thread that loop feedback is different from using a degeneration resistor. And he was immediately attacked from all sides, including you, who demanded that he start a debate with Bruno in the pages of your magazine.
Now, like Malcolm, Bruno is a very smart guy. And, also like Malcolm, he isn't always right. But if I had to choose between the two, I would choose Malcolm -- especially because my ears tell me very plainly that there is unquestionably a difference. Has Malcolm found the explanation for the difference in sound? Maybe, and maybe not. Even Malcolm doesn't know everything. As Shakespeare said,
"There are more things in heaven and earth, Horatio,
Than are dreamt of in your philosophy."
Scott, I had no idea you were such rabble-rouser! Good on you! (Next thing you know, you will be sin-binned here on diyAudio.)
The datasheet says:
"Capacitor Cf shunts high frequency energy to ground..."
while I assume that you meant:
"Capacitor C1 shunts high frequency energy to ground..."
Ah - the 2nF cap to ground. Works great for the measurements, tends to make the sound less enjoyable though. Is that why the paragraph is headed 'Professional Audio' rather than 'Audiophile' ?
Tends to make the sound less enjoyable? Not in my experience. What type (ie, brand, model, and value) have you tried in this position, and what DAC chips did you try this with?
- Status
- Not open for further replies.