Re: Re: hec != hoax ?
Hi Jan,
No, the amount of correction is not limited by the input transistors (I've checked that), but by the limited bandwidth of the EC respectively NFB circuit. See also:
http://www.diyaudio.com/forums/showthread.php?postid=1323328#post1323328
Cheers, Edmond.
janneman said:
Hi Jan,
No, the amount of correction is not limited by the input transistors (I've checked that), but by the limited bandwidth of the EC respectively NFB circuit. See also:
http://www.diyaudio.com/forums/showthread.php?postid=1323328#post1323328
Cheers, Edmond.
@Edmond & Alex:
Thanks for giving these references to the autobias scheme.
@Edmond:
Nice to have that comparison and to see both approaches in one sheet for studying. Also it's good to have all the corresponding values, I will try to sim it LTSPice and see if they come out the same, I *think* I've collected all your posted bjt models to be consistend there. WRT models, is there a an easy way to change the 2SA1407/2SC3601 to 2SA1403/2SC3597, at least on a 1st order approx (that is, fitting the curves to the datasheet to same degree as the 1407/3601 do? Maybe you even have the models and want to share?
Uhm, I have the same question as Janneman, I cannot, with my limited skills, see that the input followers are inside a loop...
- Klaus
Thanks for giving these references to the autobias scheme.
@Edmond:
Nice to have that comparison and to see both approaches in one sheet for studying. Also it's good to have all the corresponding values, I will try to sim it LTSPice and see if they come out the same, I *think* I've collected all your posted bjt models to be consistend there. WRT models, is there a an easy way to change the 2SA1407/2SC3601 to 2SA1403/2SC3597, at least on a 1st order approx (that is, fitting the curves to the datasheet to same degree as the 1407/3601 do? Maybe you even have the models and want to share?
Uhm, I have the same question as Janneman, I cannot, with my limited skills, see that the input followers are inside a loop...
- Klaus
hec != hoax ?
Hi Klaus,
Sorry, I don't have models of the 2SA1403/2SC3597 pair.
I don't think you've 'limited skills'. You only took the wrong conclusion. Maybe I had should have written: although the input transistors are outside the loop, they limit the amount of correction far less than etc, etc.
Cheers, Edmond.
KSTR said:
Hi Klaus,
Sorry, I don't have models of the 2SA1403/2SC3597 pair.
I don't think you've 'limited skills'. You only took the wrong conclusion. Maybe I had should have written: although the input transistors are outside the loop, they limit the amount of correction far less than etc, etc.
Cheers, Edmond.
Re: Re: hec != hoax ?
(cont'd)
Hi Jan,
As I have told already, not all distortion is removed as a result of limited bandwidth of the NFB loop. I would like to elaborate on this subject. Suppose that the THD20 mainly consist of the 2nd and 3rd harmonic, then we could see from the plot below, that the loop gain at 40kHz is 37dB and at 60kHz 33dB (far less than at 1kHz, btw). No matter how accurately the EC circuitry is engineered, mainly these figures determine how far the distortion will be reduced.
Bob reported a 45 fold reduction at 20kHz, that's more than 33 or 37dB. But taken into account that his simulated circuit has 27% more bandwidth, then above figures are reasonable in accordance with his findings.
Cheers, Edmond.
janneman said:
(cont'd)
Hi Jan,
As I have told already, not all distortion is removed as a result of limited bandwidth of the NFB loop. I would like to elaborate on this subject. Suppose that the THD20 mainly consist of the 2nd and 3rd harmonic, then we could see from the plot below, that the loop gain at 40kHz is 37dB and at 60kHz 33dB (far less than at 1kHz, btw). No matter how accurately the EC circuitry is engineered, mainly these figures determine how far the distortion will be reduced.
Bob reported a 45 fold reduction at 20kHz, that's more than 33 or 37dB. But taken into account that his simulated circuit has 27% more bandwidth, then above figures are reasonable in accordance with his findings.
Cheers, Edmond.
Attachments
Hi Edmond,
would the anwer to this problem be, in theory, to have a constant loop-gain reserve, that is, adapt noise gain to open-loop gain over a certain bandwidth, for a constant distance? That would through away useful Aol at low frequencies, though... but assumed we had plenty of it?
- Klaus
would the anwer to this problem be, in theory, to have a constant loop-gain reserve, that is, adapt noise gain to open-loop gain over a certain bandwidth, for a constant distance? That would through away useful Aol at low frequencies, though... but assumed we had plenty of it?
- Klaus
Re: Re: Re: hec != hoax ?
Edmond,
In my own amp I have an excess gain of ohh, around 0.8dB or so. (I try to make it zero). Still, the ec corrects the THD to levels very close to the ec circuit itself, which now becomes the limiting factor. Time and again I need to stress that EC *does not* need excess gain!
But maybe I am not aware of the exact topology of your circuit that shows these almost equal thd figures. Did they only differ in the (local) OPS ec, but with the same ol characteristics, same ol gain/phase/ same fb factor?
Jan Didden
Edmond Stuart said:
Edmond,
In my own amp I have an excess gain of ohh, around 0.8dB or so. (I try to make it zero). Still, the ec corrects the THD to levels very close to the ec circuit itself, which now becomes the limiting factor. Time and again I need to stress that EC *does not* need excess gain!
But maybe I am not aware of the exact topology of your circuit that shows these almost equal thd figures. Did they only differ in the (local) OPS ec, but with the same ol characteristics, same ol gain/phase/ same fb factor?
Jan Didden
Re: Re: Re: Re: hec != hoax ?
Hi Jan,
I was referring to output stages in isolation. The schematics can be found here:
http://www.diyaudio.com/forums/attachment.php?s=&postid=1356927&stamp=1195651662
Cheers, Edmond.
janneman said:
Hi Jan,
I was referring to output stages in isolation. The schematics can be found here:
http://www.diyaudio.com/forums/attachment.php?s=&postid=1356927&stamp=1195651662
Cheers, Edmond.
Re: Re: Re: Re: Re: hec != hoax ?
Yes, but these 'output stages in isolation' are really more than the usual ops, as they have considerable excess loop gain. In that case, the EC 'automagically' turns into a high-nfb loop (view 3). So, in this case, both ops's performance is largely depending on the nfb which depends on the excess loop gain. Would you not actually *expect* similar performance here?
A more usefull comparison would be an ops with excess loop gain and nfb versus an ops with no excess loop gain and EC.
Jan Didden
Edmond Stuart said:
Yes, but these 'output stages in isolation' are really more than the usual ops, as they have considerable excess loop gain. In that case, the EC 'automagically' turns into a high-nfb loop (view 3). So, in this case, both ops's performance is largely depending on the nfb which depends on the excess loop gain. Would you not actually *expect* similar performance here?
A more usefull comparison would be an ops with excess loop gain and nfb versus an ops with no excess loop gain and EC.
Jan Didden
Re: Re: Re: Re: Re: Re: hec != hoax ?
Hi Jan,
Just because you don't believe that HEC is NFB in disguise and don't have NFB loop gain, you are telling me that "these 'output stages in isolation' are really more than the usual ops". You really make me angry. I've spent a lot of time to make my point clear, and you didn't even bother to look at the schematics. The ops at the right is exactly Bob's HEC output stage and the ops at the left is the equivalent based on ordinary NFB .
That IS a very useful comparison, don't you think so?
janneman said:
Hi Jan,
Just because you don't believe that HEC is NFB in disguise and don't have NFB loop gain, you are telling me that "these 'output stages in isolation' are really more than the usual ops". You really make me angry. I've spent a lot of time to make my point clear, and you didn't even bother to look at the schematics. The ops at the right is exactly Bob's HEC output stage and the ops at the left is the equivalent based on ordinary NFB .
That IS a very useful comparison, don't you think so?
Edmond Stuart said:
Hi Edmond,
Quite frankly, this is exactly what went through my head about you when you refused to accept that any view of HEC other than your own feedback view had any value.
Vanderkooy and Lipshitz showed the validity of the NFB view of HEC more than 20 years ago, and I have never disputed it as a valid view. I think the alternative circuit you have shown makes a valuable comparison, but there was never a need for you to prove the validity of the NFB view.
Cheers,
Bob