T
Tim Williams
Guest
<tabbypurr@gmail.com> wrote in message
news:fa1510bf-6028-4255-b703-469a7aa432d5@googlegroups.com...
So you want low distortion, but it's a precondition that it has high
distortion?
You can't have it both ways. This isn't even about the having and eating of
cake, but the simultaneous having and not having of it!
Well... mine doesn't
When I have the option of preventing it, at least.
As JL mentioned, a lot of opamps are well behaved.
I'd like more opamps to expose their internal compensation node so their
outputs can be clamped externally, but alas, there are many things that I
like that just ain't gonna happen (for worse and for better, admittedly!).
A classic example: using two opamps to control a linear power supply in
voltage or current regulation mode, with the outputs diode-OR'd. Whenever
one is active, the other is just moping around hugging the rail. So when it
comes time for it to kick into gear, it has to sit up and climb all the way
off the rail to the operating point. Integrator windup. In a power supply,
it looks something like diode recovery, but terrifically slow (10s of
microseconds, milliseconds even, if limited by compensation capacitors).
Semantically-simple solution: clamp the inactive opamp's gain node so it
doesn't wander off into the weeds.
Another example, the precision analog rectifier. When reverse biased, the
opamp rails; this can be partly ameliorated by strapping a diode from output
to -in, so the amp follows its input, and only needs to swing two diode
drops to return to forward bias.
There are some far-out techniques to compress or expand, and predistort
signals, with the result that the final distortion cancels out and the
output is linear. But this is mostly done with RF amps, where the added
complexity is justified by the cost (capital -- RF transistors are pricey --
and operating electricity).
At low frequencies, just do what musicians do: use a bigger fucking amp.
;-)
Tim
--
Seven Transistor Labs, LLC
Electrical Engineering Consultation and Contract Design
Website: https://www.seventransistorlabs.com/
news:fa1510bf-6028-4255-b703-469a7aa432d5@googlegroups.com...
I'm looking at putting together something similar to a Class AB/B audio
amp, but it will be driven outside its linear range into saturation a lot
of the time.
So you want low distortion, but it's a precondition that it has high
distortion?
You can't have it both ways. This isn't even about the having and eating of
cake, but the simultaneous having and not having of it!
That's all well & good but for one thing: wrapping nfb round saturating
outputs doesn't work too well as it takes time for output devices to
unsaturate, and the nfb effectively overreacts, adding distortion.
Well... mine doesn't
When I have the option of preventing it, at least.
As JL mentioned, a lot of opamps are well behaved.
I'd like more opamps to expose their internal compensation node so their
outputs can be clamped externally, but alas, there are many things that I
like that just ain't gonna happen (for worse and for better, admittedly!).
A classic example: using two opamps to control a linear power supply in
voltage or current regulation mode, with the outputs diode-OR'd. Whenever
one is active, the other is just moping around hugging the rail. So when it
comes time for it to kick into gear, it has to sit up and climb all the way
off the rail to the operating point. Integrator windup. In a power supply,
it looks something like diode recovery, but terrifically slow (10s of
microseconds, milliseconds even, if limited by compensation capacitors).
Semantically-simple solution: clamp the inactive opamp's gain node so it
doesn't wander off into the weeds.
Another example, the precision analog rectifier. When reverse biased, the
opamp rails; this can be partly ameliorated by strapping a diode from output
to -in, so the amp follows its input, and only needs to swing two diode
drops to return to forward bias.
Keeping distortion low matters here. What tips would you recommend to keep
unwanted distortion minimised?
There are some far-out techniques to compress or expand, and predistort
signals, with the result that the final distortion cancels out and the
output is linear. But this is mostly done with RF amps, where the added
complexity is justified by the cost (capital -- RF transistors are pricey --
and operating electricity).
At low frequencies, just do what musicians do: use a bigger fucking amp.
;-)
Tim
--
Seven Transistor Labs, LLC
Electrical Engineering Consultation and Contract Design
Website: https://www.seventransistorlabs.com/