NFB - good, bad?

[Jorge]

There is a lot of arguments re NFB being a good thing or a bad thing.
Some vote against it, some vote for it.

To be able to define if a given amp is good or bad due to the use or not of NFB one must to know the amp's open loop distortion and decide if NFB is 'icing on the cake' or is used to straighten the bent donkey's leg.

Below we have a CCIR (19k & 20kHz) 80Vpp in 8 ohms trace of a simulated amp.

Just by this trace, this amp could be classified as a decent one (all other parameters, such as freq response, stability, etc are good).

[Jorge]

Is this frightening waveshape!

Here, NFB is used to straighten the donkey's leg.

[Jorge]

Here's the distortion trace of a well designed amp, all else very much equal.

i bet the sound of these 2 amps will not be the same...

[Jorge]

Just to make it clearer, a zoom of the bad distortion trace:

(it should have been senoidal!)

[mzzj]

What are NFB parameters for both of these amps? How many db of NFB lets say at 20khz? or what are actual differences between these amps? some sort of schematic illustrating differences between two of these would be intresting.

[DSP_Geek]

I noticed the clean trace amplitude runs between +- 2.5 volts, while the dirty trace runs between +- 14 volts. What happens when the outputs are both +- 2.5 volts?

[mikeks]

Originally posted by Jorge
NFB - good, bad

Good.

[darkfenriz]

Is power supply a good thing or a bad thing...?

[Mr Evil]

Originally posted by darkfenriz
Is power supply a good thing or a bad thing...?

Bad, no doubt about it. You can prove it easily: Measure distortion with PSU attached, then measure again without power... all distortion gone!

[Jorge]

mzzj

Both have about the same NFB factor - between 55 and 60 dB at 20K (so they are of the high NFB type)

DSP_Geek

Both amps have the same input, output voltages.
The distortion traces scale have been multiplied by 1000 just to make it easier to compare.
The distorted waveshape is larger just because OL distortion is higher, but full of spikes, this being the main point here.

Here's the circuit of the 'bad' amp (the generators are used for the many simulations I've done - loop gain, FR, input disto, output disto, stabilty).