Importance of the quality of the DC-blocking feedback capacitor?

Jump to Latest
#161
gabdx said:
When I talk to the amp designer, I said I would upgrade this part first, based on logic and what others say about the feedback loop.

He told me that it would not make any difference because the feedback was self correcting... we talk about 22 db + here.
Click to expand...
It is sad that there are people described as "amp designers" who do not understand basic electronics, such as how feedback works.
 
#162
I am not an amp designer, this is my hobby,

If the feedback is 500kh bandwidth, I don't see why it wouldn't correct errors in the feedback loop too, there are ways to calculate the nano seconds that the feedback takes to apply a correction, well,

DF96, great engineers ignored the functioning of amplification feedback for years.

AND guess what ... we live in a world where LOCAL feedback is PROMOTED WOW,

so, who is the smartest now? anyone with decent knowledge of feedback knows that global feedback is better than any local feedback, it is mathematical, it is proven, it is scientific.
 
#169
gabdx said:
If the feedback is 500kh bandwidth, I don't see why it wouldn't correct errors in the feedback loop too, .
Click to expand...

Nothing to do with bandwidth. It was explained at least three times and ways in this thread why errors in the feedback loop cannot correct themselves. Why didn't you read those?

Let me give you - again - one of the simple explanations: the feedback network sets the amp gain. So any errors in the feedback network muck up the amp gain, causing what we usually call distortion.

Mark gave a nice example, like a feedback resistor that changes its value through the signal cycle, maybe it has an unusually high voltage coefficient. That means that the resistor values gets lower with higher signal level. If this is the feedback resistor to ground, the amp gain gets higher with higher signal level and vice versa. This is expansive distortion, 3rd harmonic in this case.

Since the distortion is caused inside the feedback loop that sets the gain, there's no correction.

Jan
 
Last edited:
#171
Yes. There are often cases where it is the performance of the fb network that limits the amp performance. Voltage coefficient of the feedback resistor being one, as it carries the full (almost) output level.

Bruce Hofer, 'Mr. Audio Precision', had a nice trick. If you have say a fb network of 20k and 1k for a gain of 20, make that 20k from 20 resistors identical to the 1k to use in the ground leg. In this way, the voltage coefficients of all resistors move in unison and the effect cancels at the fb point.
Amazing what you can do if you know what's going on!

Jan
 
#172
How much NFB is enough in Class A and AB amplifiers?

Now that you talk so much about NFB, in class A and AB when it is too much? As much as necessary?

But if the design implies the need for so much NFB, will not it affect the quality of the sound? I refer to parameters that can not be measured such as three-dimensionality/depth and correct tonality (acoustic instruments of course).

Yes, assuming that we are listening to excellent recordings and, if possible, with acoustic instrumentation and barely processed in the recording studio, something almost impossible to find in current commercial music.

If the music is highly processed in the studio (and at home? ) and the instrumentation is computer generated then the amount of NFB used may not matter, I suspect.
 
#173
maty tinman said:
Now that you talk so much about NFB, in class A and AB when it is too much? As much as necessary?
.
Click to expand...

Yes of course.

What does feedback do? It makes the amp output look more and more like the input. Then ask yourself: do I want the amp output to look as much as possible like the input? Or do I want the amp to add some pleasant coloration, some booming bass?
Then you can answer the question how much should I use?

Everything else is a smokescreen.

Jan
 
Last edited:
#174
maty tinman said:
Now that you talk so much about NFB, in class A and AB when it is too much? As much as necessary?

But if the design implies the need for so much NFB, will not it affect the quality of the sound? I refer to parameters that can not be measured such as three-dimensionality/depth and correct tonality (acoustic instruments of course).

Yes, assuming that we are listening to excellent recordings and, if possible, with acoustic instrumentation and barely processed in the recording studio, something almost impossible to find in current commercial music.

If the music is highly processed in the studio (and at home? ) and the instrumentation is computer generated then the amount of NFB used may not matter, I suspect.
Click to expand...

Bruno Putzeys has a nice paper on it. TLDR: You should use as much as you can without building something that is unstable or conditionally stable.

https://linearaudio.net/sites/linearaudio.net/files/volume1bp.pdf


The amplifier is here to reproduce the input as faithfully as possible with gain. NFB helps us do this. If you want a tone control, then I suggest you implement it in your preamp or DAC so you can turn the knobs.

Everything Jan says is correct here, I would listen to his advice 🙂.


Personally, I don't distinguish between acoustic and any other type of music. I do think that chasing low levels of distortion is an academic exercise and the thresholds for the majority of people are higher than most posters here think. It doesn't mean you shouldn't though. If you are going to DIY something you might as well do it right.
 
Last edited:
You must log in or register to reply here.
Top Bottom