Hi Andy,
Actually, the net loop gain (minor+major) experienced by the amp., excepting the input stage, is virtually equal to the major loop gain provided by the equivalent double-pole network with the resistor connected to ground.
This is why, everything being equal, I fail to appreciate the benefits of Edmond's modification.
Actually, the net loop gain (minor+major) experienced by the amp., excepting the input stage, is virtually equal to the major loop gain provided by the equivalent double-pole network with the resistor connected to ground.
This is why, everything being equal, I fail to appreciate the benefits of Edmond's modification.
Bob,
You obviously haven't read my post above:
http://www.diyaudio.com/forums/showthread.php?postid=1160803#post1160803
Remember Stuart maintains his arrangement is superior to DPC; it isn't.
Proof below.
You obviously haven't read my post above:
http://www.diyaudio.com/forums/showthread.php?postid=1160803#post1160803
Remember Stuart maintains his arrangement is superior to DPC; it isn't.
Proof below.
Hello Mikeks,
1,
I did ask you to please email me the Baxandall PDF on one of Bob's threads somewhere. Any chance of getting it please? (I would do it for you)
2,
In a email a while back, you said you had written a large paper on amplifier compensation. I remember I confessed I found the issue tricky.
I would really like to hear your views on PLIL compensation published in Electronics World back in '93 by John Ellis. I could probably send you the PDF if you want it.
I mentioned it here
http://www.diyaudio.com/forums/showthread.php?s=&postid=1160790#post1160790
Post 764 onwards..
Best Regards
Kevin
PS. My email -- kevin.dabson(@)gmail.com -- please remove the brackets ()
1,
I did ask you to please email me the Baxandall PDF on one of Bob's threads somewhere. Any chance of getting it please? (I would do it for you)
2,
In a email a while back, you said you had written a large paper on amplifier compensation. I remember I confessed I found the issue tricky.
I would really like to hear your views on PLIL compensation published in Electronics World back in '93 by John Ellis. I could probably send you the PDF if you want it.
I mentioned it here
http://www.diyaudio.com/forums/showthread.php?s=&postid=1160790#post1160790
Post 764 onwards..
Best Regards
Kevin
PS. My email -- kevin.dabson(@)gmail.com -- please remove the brackets ()
mikeks said:
I think the intent of this circuit was to behave more or less like single pole comp, but including the output stage - while eliminating the instability problems of Cherry's proposal of simply hooking the Miller cap to the output. In that sense, I like the idea a lot. It's also good that it's stimulated some discussion and further thought. That keeps things lively and interesting. Cherry's response was simply one of denying any problem. But anyone with LTSpice with its loop gain probe can see that his proposed solution is potentially disastrous. In contrast, this is a constructive approach offering a solution and providing much food for further thought.
It appears to me that the high loop gain advantages of two-pole comp could be realized, along with stable inclusion of the output stage in the minor loop. I touched on a possible approach in an earlier post but haven't developed the idea fully yet. But I think this will take three capacitors. Such a solution could be compared unfavorably with a hypothetical three-pole comp in terms of the loop gain- but would that be fair? I think not.
In fact, I think the extra capacitor is a small price to pay for being able to include the output stage in the minor loop without incurring minor loop instability. In my view, this whole idea deserves lots of study and could lead to improved designs.
Hi Andy,
I would suggest running these sims first, before jumping to any conclusions:
http://www.diyaudio.com/forums/showthread.php?postid=1160809#post1160809
http://www.diyaudio.com/forums/showthread.php?postid=1160892#post1160892
I think, as far as Edmond's arrangement is concerned, you'll find there is no improvement in the net loop gain to which the output stage is subjected.
Moreover, your modification is definitely unstable with respect to the minor loop enclosing the output stage, (i.e. when the major loop is closed).
I would suggest running these sims first, before jumping to any conclusions:
http://www.diyaudio.com/forums/showthread.php?postid=1160809#post1160809
http://www.diyaudio.com/forums/showthread.php?postid=1160892#post1160892
I think, as far as Edmond's arrangement is concerned, you'll find there is no improvement in the net loop gain to which the output stage is subjected.
Moreover, your modification is definitely unstable with respect to the minor loop enclosing the output stage, (i.e. when the major loop is closed).
mikeks said:
The sim I did with the minor feedback loop I showed was of my own amp design - not anything anyone posted here. Furthermore, the network I posted had no component values at all. So I don't think you can reach any conclusive results, because the info I posted was not sufficient to do so.
I'll put a loop gain probe in the inner loop, re-run the sim and post the results tomorrow.
mikeks said:
I would suggest running sims first, before jumping to any conclusions. The same applies to you, Mike!
And remember, the reasonings of a human brain, including yours, are more prone to false conclusions then the the outcome of a sim.
I suppose, you do have a decent simulator and are able to use it in a proper way, do you?
So, go to work and stop your mindless shouting.
Hi, Estuart,
I've been thinking about your topology and relating it to the one that Lars Clausen uses in post#512. The amp in post#512 is not using miller cap around VAS, but uses C6(from VAS) and R24(from output node), and these parrarel point is for feeding the voltage divider to the base of inverting input.
Is it possible that if we use the LC's feedback scheme, we don't need miller cap compensation around VAS anymore?
I've been thinking about your topology and relating it to the one that Lars Clausen uses in post#512. The amp in post#512 is not using miller cap around VAS, but uses C6(from VAS) and R24(from output node), and these parrarel point is for feeding the voltage divider to the base of inverting input.
Is it possible that if we use the LC's feedback scheme, we don't need miller cap compensation around VAS anymore?
mikeks said:
I don't think you are in the position to say things like that.
It's up to you to prove I'm wrong, that is, by means of hard evidence like a sim of a complete amp (the blameless for example), not by false reasoning.
As told before, my sims reveal an improvement of 15dB without any sign of instability. I have provided all the data you need, except spice models, but I suppose do have models, reliable ones of course.
Now, run that sim and don't behave like God or Einstein.
NB: In the event of a negative outcome, most unlikely, it still doesn't prove I'm wrong, rather you give further evidence of lack of expertise on your part.
Hi Edmond,
You are challenging Mike but ignoring my (amatuer) simulations of your own feedback circuit shown above.
Connecting such high a C to the VAS collector is deleterious.
Connecting a C to the output node and introducing a phase shift back to VAS base 'tunes' the internal NFB response to give a measurable steady sine measurement improvement at higher AF.
The global response is not changed, but the inner loop is. This renders the output stage susceptible to transient error and fractional phase shift load induced reverse crossover bias slewing due to phase shift between output node and VAS base before the global loop can re-establish linearity via the differential pair.
Please show us how you established your findings in case there is something else here we need to study.
Cheers .......... Graham.
You are challenging Mike but ignoring my (amatuer) simulations of your own feedback circuit shown above.
Connecting such high a C to the VAS collector is deleterious.
Connecting a C to the output node and introducing a phase shift back to VAS base 'tunes' the internal NFB response to give a measurable steady sine measurement improvement at higher AF.
The global response is not changed, but the inner loop is. This renders the output stage susceptible to transient error and fractional phase shift load induced reverse crossover bias slewing due to phase shift between output node and VAS base before the global loop can re-establish linearity via the differential pair.
Please show us how you established your findings in case there is something else here we need to study.
Cheers .......... Graham.
darkfenriz posted before I could correct.
'Connecting a C to the output node'
should have been
'Connecting a R-C to the output node'
(Ground connection for the intermediate resistor has same effect on dynamics, though a different THD result.)
I tried these schemes in the 70's and whilst it was possible to 'spec chase' as far as THD is concerned, the improved figures were of no benefit to reproduction and could make the sound less natural.
Cheers ........ Graham.
'Connecting a C to the output node'
should have been
'Connecting a R-C to the output node'
(Ground connection for the intermediate resistor has same effect on dynamics, though a different THD result.)
I tried these schemes in the 70's and whilst it was possible to 'spec chase' as far as THD is concerned, the improved figures were of no benefit to reproduction and could make the sound less natural.
Cheers ........ Graham.