Evolve Non Global Feedback Distortion Cancellation Power Amplifier

Status
Not open for further replies.
Jump to Latest
EUVL

EUVL

Member
Joined 2003
xen-audio.com
#1
A while back, someone brought up this circuit :

Distortion removal circuit

I thought it was very clever and interesting. With my limited knowledge of Japanese and amplifier design, and with the help of Babbel Fish, I put together a not-too-perfect translation of the article, and hope that it will stimulate some discussion.

If a native speaker spots any mistakes, please feel free to correct. And I hope the the original author has no objections to me posting it here.

😉


Patrick
 

Attachments

#2
For simplicity here also the experimental circuit.

Patrick
 

Attachments

  • Experimental Circuit.gif
    Experimental Circuit.gif
    17.4 KB · Views: 976
#6
EUVL said:
A while back, someone brought up this circuit :

Distortion removal circuit

I thought it was very clever and interesting. With my limited knowledge of Japanese and amplifier design, and with the help of Babbel Fish, I put together a not-too-perfect translation of the article, and hope that it will stimulate some discussion.

If a native speaker spots any mistakes, please feel free to correct. And I hope the the original author has no objections to me posting it here.

😉


Patrick
Click to expand...

The problem here is always the efficient and effective adding of two signals
at the amp output - at low impedance levels. That is shown in the data, which is orders of magnitude worse than a regular nfb amp.
Hawksford has written about this and one option is to add the error signal not at the output but at the input - which is much easier. This also has - in theory - total error nulling.

My error correction amp as described in both Elektor April/May 2008 and in AudioXpress this year uses this method.
These are intresting circuits to play around with!

jd
 
#11
EUVL said:
Maybe the author would be kind enough to send a copy to a fan ??
I think he might not need approval for that .....

😉


Patrick
Click to expand...

This is the basic circuit, courtesy of AudioXpress magazine.

The trick is that if a+b=1, the error is cancelled. So you can either add the error to the output, (a=0, b=1), to the input (b=0, a=1) or something in between.

jd
 

Attachments

  • ax p1 f1.jpg
    ax p1 f1.jpg
    20.1 KB · Views: 680
Last edited:
#15
janneman said:
Officially, no. The publisher owns them. But I'll ask.
jd
Click to expand...
Here two circuits of me - but without simulation results.

Unfortunately I haven't perform detailled investigation by simulation - especially the comparison between a normal "stripped naked" push-pull power follower and this one together with a correction network according the PAX article in elektor - both in hard running mode and in class A/B with only 20-50mA through the last stage. Perhaps one of you can do this with LT-spice.

But the DC values there are present in the schematic.

Hallo Jan,
here an off-topic question: what do you think about the Circlomos topology? - go to
http://www.diyaudio.com/forums/solid-state/154388-its-cheap-its-n-its-dirty-its-circlomos.html
 

Attachments

Last edited:
#18
janneman said:
The problem here is always the efficient and effective adding of two signals
at the amp output - at low impedance levels. That is shown in the data, which is orders of magnitude worse than a regular nfb amp.
Hawksford has written about this and one option is to add the error signal not at the output but at the input - which is much easier. This also has - in theory - total error nulling.

My error correction amp as described in both Elektor April/May 2008 and in AudioXpress this year uses this method.
These are intresting circuits to play around with!

jd
Click to expand...

Geez. I go away for 5 months and when I come back you seem to have forgotten everything we discussed about this.

It is misleading (IMO a lie) to claim that NFB can "null" an error totally, nor reduce an error to zero even theoretically. I thought I had proved that to you, and you had agreed.

I realize that you had already published your articles before we had our discussion. I suppose for reputation's sake you are compelled to back up your own article. This is my only explanation as to why you persist.

Not something I approve of, sorry. It's not engineering and it is misleading for amateurs trying to learn. That's why I point this out.

Also, this thread is about true error cancellation using feed-forward so I wouldn't want anyone to confound the two.
 
#19
Bonsai said:
The APAX looks like a classic CFA amp, minus the R to gnd at th e junction of R13 and R14 that I would expect to see. I'm trting to understand how coupling the i/p signal to the VAS Vbe multiplier is cancelling the error. Are you sure this is right?
Click to expand...
It doesn't cancel anything. It is just another NFB circuit. This has been proved ad nauseum in this forum.
 
Status
Not open for further replies.
Top Bottom