Non-linearly vary resistance with guitar envelope

[fNNR]

Hi everybody

I am doing a control circuit that uses the envelope of my guitar signal to vary certain resistances in the signal circuit (like an envelope filter). I'm using a H11F1 optofet. The resistance in the H11F1 should change linearly with my guitar signal, that is if I double the envelope of my guitar, the resistance should double as well (right?). (Well decrease half-fold, due to the inverse relationship between resistance and current through the LED in the H11F1)

What I want to do is to make the resistance vary non-linearly, I want it to vary logarithmically (or similarly to logarithm). That is, when the guitar signal is small and small changes are being made, the resistance changes should be approximately linear (because first off the logarithm function is approximately linear) but as the envelope gets higher then the resistance change should decrease.

So what I am wondering is how to implement this thing, some sort of logarithmic amplifier. I think could make a piecewise linear amplifier, say for example that it would have gain 1 in the range of 0 to 50mV and then a gain of 0.5 for 50mV and up but that seems kinda... well not as neat of a solution as I would hope to get.

Isn't it possible to make an amplifier with logarithmic gain? http://en.wikipedia.org/wiki/Operational_amplifier_applications#Logarithmic_output this is ideally what I would need, but it doesn't quite seem to work in practice? It seems to work just as a smooth clipper...

Any thoughts?

Thanks

[Cliff Schecht]

Typically to get an exponential response out of a transistor, you have to cut down the signal to very small levels (as I recall, 60mV per volt is the proper scaling for Volts/Octave scaling) and bias the transistor outside of it's linear region (near saturation). If you bias the FET outside of its linear region, you're going to get a square law deal - i.e. the output is the inverse square of the input.

These circuits are usually voltage in-current out, although there are tricks to get voltage in-voltage out. I haven't played around much with the operational amplifier based log amps (at least not in this simple of a form) outside of simulation but take a look here: http://www.play-hookey.com/analog/logarithmic_amplifier.html (the circuit at the bottom is a simplified implementation).

[soggybag]

R.G.s Nuetron filter has option to use the H11F1. This might provide some ideas.

[Cliff Schecht]

To correct myself earlier, the proper scaling is 18mV/Octave.

[alanlan]

I would have thought you could make the diode op-amp cct work.  It may be a matter of correct scaling.  It would be a good idea to understand the function of each block before you stick them together i.e. work out the working range of the log amp and then look at the output of your envelope detector.  Then ask yourself how to match the two together.  That's my advice.

[fNNR]

Thanks for your replies guys, I'll take a better look at this tonight and see what I can make out of it...