Intermediate stage distortion caused by negative feedback

I find these tremendously valuable in amplifier design, especially in multistage amplifiers that have a single feedback loop. The feedback circuits reduce the distortion at the output, but transfer the distortion to the intermediate stages. There they are hidden, but they can be measured with the right equipment.

It is possible to have .01 percent distortion on the output of an amplifier while there is an intermediate stage running at 1 to 4 percent or more due to a negative feedback loop attempting to modify the signal to make it appear linear at the output. A straight line is no straight line at these intermediate points. The same intermediate stage without overall negative feedback may be running at well below a quarter of a percent distortion.

Harmonic distortion measurements at these intermediate points are thus extremely valuable in detecting hidden multistage amplifier problems. If you want to unhide these types of problems you may need one of these distortion analyzers to redesign the part of the circuit that is causing the problem outlined above.

John Fassotte
Alaskan Audio
 

halojoy

On Hiatus
2002-11-05 6:14 pm
Here it all STARTED - a blessing or an evil?

alaskanaudio said:
I find these tremendously valuable in amplifier design, especially in multistage amplifiers that have a single feedback loop. The feedback circuits reduce the distortion at the output, but transfer the distortion to the intermediate stages. There they are hidden, but they can be measured with the right equipment.

It is possible to have .01 percent distortion on the output of an amplifier while there is an intermediate stage running at 1 to 4 percent or more due to a negative feedback loop attempting to modify the signal to make it appear linear at the output. A straight line is no straight line at these intermediate points. The same intermediate stage without overall negative feedback may be running at well below a quarter of a percent distortion.

Harmonic distortion measurements at these intermediate points are thus extremely valuable in detecting hidden multistage amplifier problems. If you want to unhide these types of problems you may need one of these distortion analyzers to redesign the part of the circuit that is causing the problem outlined above.

John Fassotte
Alaskan Audio

Here is the Patent, where NEGATIVE FEEDBACK
was first introduced.

Some say as A BLESSING
Others say as AN EVIL

In those days were only Tubes.
Yet the Tube people are the most negative to
this way to correct the output signal.

:cool: halojoy likes feedback - but not a massive amount of global :cool:
 

Attachments

  • nfb_pat.gif
    nfb_pat.gif
    26.7 KB · Views: 281
It's logic..

It fully logical that if output stage distorts the signal up to a 1%, for having a 0,001% THD at output you'll have to put a distorted signal with 1,001% THD at the input, but with all of its harmonics 180º out of phase, so inital and generated harmonics will cancell.

This is also ok for any stage.

There are many issues about if it must be blessed or it's evil, yes... Once I made an amp, and an error left 2nd stage unbiased (bad soldering). it was incredible because It sounded (it sounded like crap, but it was working! Unbiased 2nd stage and class B output)

Ive made an amp without any feedback, and it sounds special, cool, very detailed, but someway you can hear 2nd order distortion and a lack of dynamics... I think a little bit of NFB is neccesary
 
Intermediate THD

John,

I agree with PCP, that a logic result of the dist reduction at the output is a skewed, so to say, internal signal to compensate for it.

But measuring the internal distortion gives you a good idea of the amp's open loop distortion. (If the "original" dist of the intermediate stage is much lower, of course). Making circuit changes with an eye on the internal THD is much more revealing then looking at the output THD only, especially if you get to the point with a couple of zeros behind the decimal point.

Jan Didden
 
I don't see the logic in the statement that global feedback causes distortion within intermediate stages.

If the amplifier is properly designed, all stages will have minimal distortion with the output stage generally having the most distortion due to operation at AB or B. Negative feedback
attempts to correct for this by creating in effect counter-distortion
to cancel nonlinearities within the feedback loop.

But in the meantime the various stages are still as linear as the designer could make them; if negative feedback caused those stages to distort, then feedback would attempt to correct for
>that<, and so on and so forth. I think this would be some kind of oscillator.

No, you're just not looking at it correctly. Distortion checks of intermediate stages have to be done without global feedback.

As with all good things in moderation, feedback has to be used with an appreciation of its limitations and pitfalls, and there are
a few.

How many cases are there of amplifier designs with nested feedback loops, aside from local feedback techniques? I suppose a complementary feedback pair might count in this group? Or not?
 
Intermediate etc

Damon,

I'm with you, I think we are just having a semantic problem, me being not too clear. When I mentioned "intermediate distortion" I did mean the distortion you see looking internally in the amp when the feedback loop is closed. You are right that that is not the distortion of the intermediate stage, but rather the "compensating" signal resulting from the subtractive action of the feedback to get the output signal right. And as such it is a good indicator of the open loop distortion of the amp, do you agree?

Jan Didden
 

halojoy

On Hiatus
2002-11-05 6:14 pm
alaskanaudio said:
I find these tremendously valuable in amplifier design, especially in multistage amplifiers that have a single feedback loop. The feedback circuits reduce the distortion at the output, but transfer the distortion to the intermediate stages. There they are hidden, but they can be measured with the right equipment.

It is possible to have .01 percent distortion on the output of an amplifier while there is an intermediate stage running at 1 to 4 percent or more due to a negative feedback loop attempting to modify the signal to make it appear linear at the output. A straight line is no straight line at these intermediate points. The same intermediate stage without overall negative feedback may be running at well below a quarter of a percent distortion.

Harmonic distortion measurements at these intermediate points are thus extremely valuable in detecting hidden multistage amplifier problems. If you want to unhide these types of problems you may need one of these distortion analyzers to redesign the part of the circuit that is causing the problem outlined above.

John Fassotte
Alaskan Audio
;) I hope John comes by, sooner or later.
;) And gives his views and answers to this.
;) This was my intention by bringing this thread alive again.
Not because I know so much about this,
but I'd like to know more.
By sharing what the more experienced say.

I myself is in favour of several stages with local feedback.
At least instead of global feedback over 4 stages or more.
When using global, I try to keep feedback below 40dB (<=100).
This means using gain-limiting resistors in every stage.
Even local resistive loads.

/halo - has his ways, of reducing negatives of feedback
 
I agree with Alaskan's idea but not the nomenclature. NFB doesn't, necessarily, change the distortion of intermediate stages. Rather it makes it easier to see the error signal at these intermediate stages. It tries to reveal "what is the input signal to this stage that will result in a perfect signal at the amps output". Of course this depends upon the accuracy of the differential stage and to what extent open-loop distortion can be corrected by NFB. Also, it may be with some designs that intermediate stage distortion is deliberate and so minimizing it may be counter-productive. But in general this is a worthwhile way to investigate a circuit, IMO.
 
Damon Hill said:
I don't see the logic in the statement that global feedback causes distortion within intermediate stages.

If the amplifier is properly designed, all stages will have minimal distortion with the output stage generally having the most distortion due to operation at AB or B. Negative feedback
attempts to correct for this by creating in effect counter-distortion
to cancel nonlinearities within the feedback loop.

But in the meantime the various stages are still as linear as the designer could make them; if negative feedback caused those stages to distort, then feedback would attempt to correct for
>that<, and so on and so forth. I think this would be some kind of oscillator.

No, you're just not looking at it correctly. Distortion checks of intermediate stages have to be done without global feedback.


We don't mean that NFB causes distortion in intermediate stages. (by them selveS) and there's no oscillation. You know that the amp's gonna generate harmonics, well you put at the input those harmonics but in opposite phase.

Try to make a very distorting amp with spice (use resistor dividers to make it more distorted), and then measure input differential voltaje. It's distorted even though you get 0,1% at the output and an ideal voltage source at the input.

Or do try to make a class A amp and meassure THD just before O/P stage.