Nelson Pass said:
That's a subjective call. It is the experience of many that bias
current higher than the measurement "sweet spot" sounds
better. I don't doubt there are at least as many in this forum
who will disagree.
The best is as much as possible. That means almost always in class A and almost never transition into B
It would be nice to avoid class B at all.
Re: Question for John Curl
I thought similar when writing a previous post. Perhaps the OLG is rolled off earier than needed due to a fear of NFB, who knows?
This leads to another question - if one really wants to use as little GNFB as possible to meet this purported THX THD spec (whatever it may be) why wrap 34 dB of GNFB around an amplifier whose open loop linearity could probably be an order of magnitude better?
Cheers,
Glen
Bob Cordell said:
I thought similar when writing a previous post. Perhaps the OLG is rolled off earier than needed due to a fear of NFB, who knows?
This leads to another question - if one really wants to use as little GNFB as possible to meet this purported THX THD spec (whatever it may be) why wrap 34 dB of GNFB around an amplifier whose open loop linearity could probably be an order of magnitude better?
Cheers,
Glen
lumanauw said:
Overlooked this post - sorry.
Yes, parallelling would be a problem. If using a FCP one could have Re's in the emitters of the power transistors only, with a slight distortion penalty.
I stabilize quiescent current purely thermally. It does call for a bit of juggling and a few constants change if the output devices are replaced, but I find I can keep the Ico constant over quite a T-range. It takes some time to stabilise (starting with a higher Ico), but then my amplifier is on all the time - allows things to stabilise and extra cost of electricity negligible.
Re: Re: Question for John Curl
Why, indeed?!
And-what-is-this-"fear of NFB"-that-you-speak-of ??!
Seriously, though, assuming that it wasn't done because of some nebulous "fear of NFB" (since that would seemingly be irrational), then what ARE all of the possible benefits, that anyone here can think of, of having the earlier open-loop gain roll-off, and of NOT having to extend the gain crossover point out to 1 MHz or so in order to get the open-loop bandwidth to 20 kHz in order to be able to have 34 dB of NFB all the way out to 20 kHz, and whatever else it entails?
Oh. Are you guys expecting ME to come up with some possible benefits?
I came up with the question. Isn't that enough?!
;-)
OK. Seriously, again (as far as you know): I'M certainly not one of the experts, here. But I can think of at least ONE potential possible benefit that I feel sort-of somewhat-confident about: "Lowered potential for RFI/EMI-induced distortions, RFI/EMI problems in general, and, (maybe way-)fewer HF worries in general (and associated lower costs), all the way through the whole product-development process." (Or some other words, if they would make my answer "sound better".
I'm not sure how SIGNIFICANT any of THOSE "possible" benefits might be, if they even are at all. So maybe none of that will hold water _OR_ fly. (But remember that I didn't want to answer. You guys made me.)
But since there "has to be" a good reason (Doesn't there?!), and since John certainly doesn't seem to be a "technically-challenged" person (quite the opposite), then maybe we should ASSUME, for this exercise at least, that he was NOT motivated by some irrational "fear of negative feedback" (unless, of course, his customers' $$ have an irrational fear of negative feedback, in which case he's still rational, but we can probably scrap the current exercise), and see if anyone here can come up with any more-or-less compelling reason(s) for doing it his way.
Why??! Because we can?! Because MAYBE we can? No!!! Because maybe, just maybe, somebody will come up with something great that no one else here has ever thought of before (or mentioned, yet, at least), and then I... er... WE ALL, can BENEFIT from that knowledge! :-o
Well, I hope that I got my (actually rather minor) point across, despite all of the "humor", which is: Rather than wondering why he didn't do it our way, maybe we should try wondering why he did it his way.
We might even surprise ourselves (or, maybe just me). Many of you who are involved in this thread are both astonishingly-brilliant and very-experienced at this stuff. So I'm guessing that someone here is bound to come up with some plausible, and maybe even good, reasons. (And some of us might even learn some more good stuff, along the way.)
Or... we could just forget all about it.
- Tom Gootee
http://www.fullnet.com/~tomg/index.html
(I guess much of the following message is probably off-topic. Sorry in advance.)G.Kleinschmidt said:
Why, indeed?!
And-what-is-this-"fear of NFB"-that-you-speak-of ??!
Seriously, though, assuming that it wasn't done because of some nebulous "fear of NFB" (since that would seemingly be irrational), then what ARE all of the possible benefits, that anyone here can think of, of having the earlier open-loop gain roll-off, and of NOT having to extend the gain crossover point out to 1 MHz or so in order to get the open-loop bandwidth to 20 kHz in order to be able to have 34 dB of NFB all the way out to 20 kHz, and whatever else it entails?
Oh. Are you guys expecting ME to come up with some possible benefits?
I came up with the question. Isn't that enough?!
;-)
OK. Seriously, again (as far as you know): I'M certainly not one of the experts, here. But I can think of at least ONE potential possible benefit that I feel sort-of somewhat-confident about: "Lowered potential for RFI/EMI-induced distortions, RFI/EMI problems in general, and, (maybe way-)fewer HF worries in general (and associated lower costs), all the way through the whole product-development process." (Or some other words, if they would make my answer "sound better".
I'm not sure how SIGNIFICANT any of THOSE "possible" benefits might be, if they even are at all. So maybe none of that will hold water _OR_ fly. (But remember that I didn't want to answer. You guys made me.)
But since there "has to be" a good reason (Doesn't there?!), and since John certainly doesn't seem to be a "technically-challenged" person (quite the opposite), then maybe we should ASSUME, for this exercise at least, that he was NOT motivated by some irrational "fear of negative feedback" (unless, of course, his customers' $$ have an irrational fear of negative feedback, in which case he's still rational, but we can probably scrap the current exercise), and see if anyone here can come up with any more-or-less compelling reason(s) for doing it his way.
Why??! Because we can?! Because MAYBE we can? No!!! Because maybe, just maybe, somebody will come up with something great that no one else here has ever thought of before (or mentioned, yet, at least), and then I... er... WE ALL, can BENEFIT from that knowledge! :-o
Well, I hope that I got my (actually rather minor) point across, despite all of the "humor", which is: Rather than wondering why he didn't do it our way, maybe we should try wondering why he did it his way.
We might even surprise ourselves (or, maybe just me). Many of you who are involved in this thread are both astonishingly-brilliant and very-experienced at this stuff. So I'm guessing that someone here is bound to come up with some plausible, and maybe even good, reasons. (And some of us might even learn some more good stuff, along the way.)
Or... we could just forget all about it.
- Tom Gootee
http://www.fullnet.com/~tomg/index.html
Re: Re: Question for John Curl
Hi Glen,
I suspect that the answer to the question about the open-loop gain nonlinearity of the JC-1 lies in the exclusive use of JFETs and MOSFETs in the signal path prior to the final output stage. John really likes the sound of JFETs and MOSFETs. I suspect that the use of the MOSFETs meant a significant sacrifice in open-loop linearity, at least in regard to the softer, lower-order nonlinearities.
Cheers,
Bob
G.Kleinschmidt said:
Hi Glen,
I suspect that the answer to the question about the open-loop gain nonlinearity of the JC-1 lies in the exclusive use of JFETs and MOSFETs in the signal path prior to the final output stage. John really likes the sound of JFETs and MOSFETs. I suspect that the use of the MOSFETs meant a significant sacrifice in open-loop linearity, at least in regard to the softer, lower-order nonlinearities.
Cheers,
Bob
Re: Re: Re: Question for John Curl
Hi Tom,
Yes, I really wanted to understand what John's reasoning was here, and am disappointed that John has not chimed in yet. We can all learn something by understanding why someone did something in a way that may not be the way that some of us chose to do it.
Cheers,
Bob
gootee said:
(I guess much of the following message is probably off-topic. Sorry in advance.)
Why, indeed?!
And-what-is-this-"fear of NFB"-that-you-speak-of ??!
Seriously, though, assuming that it wasn't done because of some nebulous "fear of NFB" (since that would seemingly be irrational), then what ARE all of the possible benefits, that anyone here can think of, of having the earlier open-loop gain roll-off, and of NOT having to extend the gain crossover point out to 1 MHz or so in order to get the open-loop bandwidth to 20 kHz in order to be able to have 34 dB of NFB all the way out to 20 kHz, and whatever else it entails?
Oh. Are you guys expecting ME to come up with some possible benefits?
I came up with the question. Isn't that enough?!
;-)
OK. Seriously, again (as far as you know): I'M certainly not one of the experts, here. But I can think of at least ONE potential possible benefit that I feel sort-of somewhat-confident about: "Lowered potential for RFI/EMI-induced distortions, RFI/EMI problems in general, and, (maybe way-)fewer HF worries in general (and associated lower costs), all the way through the whole product-development process." (Or some other words, if they would make my answer "sound better".
I'm not sure how SIGNIFICANT any of THOSE "possible" benefits might be, if they even are at all. So maybe none of that will hold water _OR_ fly. (But remember that I didn't want to answer. You guys made me.)
But since there "has to be" a good reason (Doesn't there?!), and since John certainly doesn't seem to be a "technically-challenged" person (quite the opposite), then maybe we should ASSUME, for this exercise at least, that he was NOT motivated by some irrational "fear of negative feedback" (unless, of course, his customers' $$ have an irrational fear of negative feedback, in which case he's still rational, but we can probably scrap the current exercise), and see if anyone here can come up with any more-or-less compelling reason(s) for doing it his way.
Why??! Because we can?! Because MAYBE we can? No!!! Because maybe, just maybe, somebody will come up with something great that no one else here has ever thought of before (or mentioned, yet, at least), and then I... er... WE ALL, can BENEFIT from that knowledge! :-o
Well, I hope that I got my (actually rather minor) point across, despite all of the "humor", which is: Rather than wondering why he didn't do it our way, maybe we should try wondering why he did it his way.
We might even surprise ourselves (or, maybe just me). Many of you who are involved in this thread are both astonishingly-brilliant and very-experienced at this stuff. So I'm guessing that someone here is bound to come up with some plausible, and maybe even good, reasons. (And some of us might even learn some more good stuff, along the way.)
Or... we could just forget all about it.
- Tom Gootee
http://www.fullnet.com/~tomg/index.html
Hi Tom,
Yes, I really wanted to understand what John's reasoning was here, and am disappointed that John has not chimed in yet. We can all learn something by understanding why someone did something in a way that may not be the way that some of us chose to do it.
Cheers,
Bob
Re: Re: Re: Question for John Curl
Yes, but I was kinda half expecting John to get in with that answer before you though. Nevermind.
So, basically, John needs a fair degree of global NFB to make his amplifiers measure well.
My Irony Meter has deflected off-scale.
Cheers,
Glen
Bob Cordell said:
Yes, but I was kinda half expecting John to get in with that answer before you though. Nevermind.
So, basically, John needs a fair degree of global NFB to make his amplifiers measure well.
My Irony Meter has deflected off-scale.
Cheers,
Glen
john curl said:
Hi John,
Ok, I think I understand. The reason why you did not go to 20 kHz open-loop bandwidth, 34 dB of feedback at 20 kHz, and a closed-loop bandwidth on the order of 1 MHz was so that you could avoid the output coil.
In other words, a capacitive load not isolated by a coil would destabilize a global feedback loop with a gain crossover out as far as 1 MHz.
This makes sense (even if I would have chosen a different tradeoff, i.e., put in a well-done coil and have more NFB at 20 kHz). So the price you paid for not having a coil was less HF NFB and (possibly) the need for 10-ohm base stopper resistors.
Thanks,
Bob
john curl said:
But its fun, John. Nobody is holding a gun to your head to answer our questions. Besides, we'd still speculate about it even if we had the schematic and had measured it
.Lighten up, John. The JC-1 is a darned good amplifier, and you certainly don't need to be defensive about any of your choices on it.
Cryptic answers like "the output coil" just encourage us to speculate.
Here we all were, wondering what kind of secret sauce you used to enable you to get away without an output coil, and now we have a hint that maybe it was nothing more than starving your closed loop bandwidth. Coil-phobia?
Cheers,
Bob
"No coils."
Thanks for that tidbit, John.
So I guess one question, now, might be: Is the benefit of not having to use an output coil worth the cost of doing the other stuff that way (assuming that that's even a fair assessment of the reasoning or trade-offs involved).
I, for one, have been somewhat taken-aback by all of the claims I've been hearing about output coils not affecting the sound, very much, maybe mainly because I have noticed, when simulating amplifiers with LTSpice (yes, with every parasitic I could think of modeled, too, including wires, traces, cables, components of course, and with speaker model loads and with plain resistor || cap loads), that if I have pushed the amp's performance "pretty far up there", some things become extremely sensitive to the output coil's value, often in very bad ways.
So, although I don't have nearly-enough other information to even begin to form any judgement about the tradoffs involved in John's JC-1 design, it seems to me that there might be something that at least tends toward being reasonable (and worth incurring some cost for), in that idea, i.e. avoiding the use of an output coil.
But, yet-again, the difficult-to-answer question is, still, "At what cost?". Or, in terms of optimization, it might be: What benefits are worth what costs? With such a large number of difficult-to-quantify variables, and no clear (to me at least) optimization criteria, it's difficult for me to imagine that there could be a single solution, or even a single class of solutions, that could be declared (or be agreed-on to be) optimal.
At any rate, I'm sorry if my recent posts added fuel to a fire that should have been doused, instead.
- Tom Gootee
http://www.fullnet.com/~tomg/index.html
Thanks for that tidbit, John.
So I guess one question, now, might be: Is the benefit of not having to use an output coil worth the cost of doing the other stuff that way (assuming that that's even a fair assessment of the reasoning or trade-offs involved).
I, for one, have been somewhat taken-aback by all of the claims I've been hearing about output coils not affecting the sound, very much, maybe mainly because I have noticed, when simulating amplifiers with LTSpice (yes, with every parasitic I could think of modeled, too, including wires, traces, cables, components of course, and with speaker model loads and with plain resistor || cap loads), that if I have pushed the amp's performance "pretty far up there", some things become extremely sensitive to the output coil's value, often in very bad ways.
So, although I don't have nearly-enough other information to even begin to form any judgement about the tradoffs involved in John's JC-1 design, it seems to me that there might be something that at least tends toward being reasonable (and worth incurring some cost for), in that idea, i.e. avoiding the use of an output coil.
But, yet-again, the difficult-to-answer question is, still, "At what cost?". Or, in terms of optimization, it might be: What benefits are worth what costs? With such a large number of difficult-to-quantify variables, and no clear (to me at least) optimization criteria, it's difficult for me to imagine that there could be a single solution, or even a single class of solutions, that could be declared (or be agreed-on to be) optimal.
At any rate, I'm sorry if my recent posts added fuel to a fire that should have been doused, instead.
- Tom Gootee
http://www.fullnet.com/~tomg/index.html
I concede that an 0.5 uH, air coil, 14 ga, oxygen free, inductor might be OK. That was the last value that I used. However, Parasound has never used output inductors in any of my amp designs, and I am not about to start. It isn't worth the tradeoff. What I would rather do, is to find a way to increase the open loop bandwidth, without adding an output coil.
gootee said:
Tom,
No reason to be sorry. If material ignites from honest questions, the fault is not with the enquiry, but with the flammable material lying around (a) where it should not be in the first place, or (b) because it has no business to be inflammable.
The laws of electronics applicable here are simple - and they apply equally to all designers. The biases are not. Any optimal solution requires a sober involvement of all electronic means where necessary, without an unreasonable bias toward any single one. If anyone desires not to go that way, it is a free country (sic). If he aquired reasonable success that way, honours to the design; if not to him.
Things matter to me if I can learn from them - moreover, this is called a DIY forum. I have learnt nothing of substance here regarding the avoidance of inductors. Should such a directive come along, I would be the first person to make use of it. Until then .... so sorry. (Now I apologise unnecessarily.)