no mercy distortion killer circuit

Status
Not open for further replies.
Jump to Latest
#147
thanh said:
have you known a amp which can cancel distortion absolutely?
why don't you try to simulate with my topo or unitgain?
Click to expand...

I've tried to simulate my topo for a very basic Class B amp. It didn't work, inderstandably. When the input crosses 0V, both halves of the bridge can do very little about correcting each other's crossover distortion. It might work better in Class AB, especially if the two halves are differently biased (their xover points will be in different places so they could help each other in reducing this kind of distortion). I'll try to do Classes AB and A later.

This whole circuitry thing is kinda new to me, I'm learning my way as I go, so please bear with me if I don't make much sense.
 
#148
Bernhard said:
@ the moment I make my 4 x S1 DAC.

Distortion is not zero because it can not be.

The error amp has its own distortion and even the sim seems to know that :xeye:
Click to expand...

I figure that, if the error amp's distortion is level-independent at low levels (Class A?), then the sum of dB distortion figures of the main amp and the error amp will give the figure for the whole contraption, more or less.

Error amp's output is main amp's distortion plus any distortion of the error amp itself. It is subtracted from main amp's output, which leaves us with the error amp's distortion only. But since error amp's output is very small, the resulting distortion is hopefully even smaller.

Does it make sense?
 
#149
unitgain said:
Does it make sense?
Click to expand...

I hope.

Main amp and error amp both are classA.

Error amp voltage swing is small.
So error amp distortion should be small also.

Error amp current swing is same as main amp.
Distortion may rise with current too 😕

As the error amp reamplifies only the distortion of the main amp, the total distortion on the load is smaller.
 
#150
This works (sort of) (in simulation).

Here are FFTs of the response of the same (very basic) amp with and without correction.
 

Attachments

  • fft.jpg
    fft.jpg
    93.5 KB · Views: 236
#156
Bernhard said:
unnitgain,

the distortion goes up again above 10k.

This circuit which you posted may give problems because distortion is reamplified, maybe I tried that before, tried so many ...


tanh,

parallel does not work.
Click to expand...

I think it goes up because the amp is too slow, and no good at all to begin with. The distortion should be re-amplified, but the amp does it slowly and not very accurately. With a faster amp the figure will go up too, but at much higher frequencies, perhaps 100k or more. I'll try it with other designs when I get some free time.

Also note that, though it goes up, at 20k it's still 15dB or so lower than the uncomensated figure.
 
#158
the distortion output is the distortion of error amp if the cancelling is absolute.But the error amp is also normal amp so it also has got distortion .I want a topology which can cancel the distortion of the error amp
I think there is a killer amp on simulated sofware.There is not a killer amp in the real world .Because components are different or not similar.
If the future amp operate in class A,i will put it into rubbish bin and i don't need the feedforward .Because it is so hot
 
Status
Not open for further replies.
Top Bottom