0
\$\begingroup\$

What is the best way to limit an audio signal for ADC input (to 0V - 3.3V) with out disrupting it harmonically? Ideally, I'd like the conditioned signal entering the ADC to be a signal uniformly spanning 0V to 3.3V. That is, I'd like to strip the original signal of its envelope and have a flat envelope at the output. I've spent the better part of a month attempting to get simulations of different VCA circuits to respond to these specifications using variations on the original input signal's envelope as the control signal. I know near-blind trial and error through simulation is not the most elegant way to go about design but I'm early in my EE studies and it's all I have.

Is there a standard scheme for this kind of signal conditioning?

Is designing a circuit, whose response is consistent across the audio band (or a good part of it), which attenuates/amplifies a signal, varying from 0.1Vpp to 6Vpp, to a uniform 3.3Vpp signal with negligible change to the harmonic content (negligible meaning that an FFT would still faithfully report the first several harmonics of the original signal) a realistic goal?

\$\endgroup\$
  • \$\begingroup\$ It sounds like what you are trying to do is compress the dynamic range. Have you already searched on google using "dynamic range compression?" If not, give that a try. \$\endgroup\$ – mkeith Dec 28 '15 at 21:45
  • 1
    \$\begingroup\$ thatcorp.com/datashts/dn00A.pdf \$\endgroup\$ – Daniel Dec 28 '15 at 21:57
  • \$\begingroup\$ Thank you, Daniel. This looks very helpful. I'll read it tonight. \$\endgroup\$ – Malcolm Regan Dec 28 '15 at 22:01
1
\$\begingroup\$

What you are asking for is a non-linear operation, so will add distortion.

The details are all in the tradeoffs you haven't told us about, particularly response time. For example, suppose there is a loud passage, then it gets quieter. What metric do you use to decide this quiet isn't part of the signal that you want to keep, like a person pausing between words? Are you allowed to look ahead (store signal, then play back later)? If not, how long is long enough to decide that quiet is the new normal? Notice that you don't get to say "instantly", since audio is iherently AC, which will have regular zero crossings.

People have been dealing with these issues and making these tradeoffs in various ways for a long time, before there were even transistors. Look up something called AGC (automatic gain control).

Otherwise, without a real spec, there is little we can help with.

\$\endgroup\$
  • \$\begingroup\$ Right, of course. It wouldn't amplify something below a certain threshold and would have a certain attack time. I understand these limitations but I won't be able to define them exactly without a little experimentation. All I'm after now is a general framework. I will look up AGC. Thank you. \$\endgroup\$ – Malcolm Regan Dec 28 '15 at 22:11
  • \$\begingroup\$ the amount of distortion will depend on the compression attack and release times. i.e. how quickly the compressor responds to level changes with gain changes. if the response is slow, with the exception of the initial down-change of gain, the distortion should be low. \$\endgroup\$ – robert bristow-johnson Dec 28 '15 at 22:15
  • \$\begingroup\$ I see. That makes sense. Thank you. What is Initial down-change, though? \$\endgroup\$ – Malcolm Regan Dec 28 '15 at 22:20
  • \$\begingroup\$ Whoever downvoted this, what do you think is wrong? \$\endgroup\$ – Olin Lathrop Oct 3 '18 at 17:25

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.