Simple compressor uses a few 2N7000's and one Op Amp

[rring]

I have cooked up a new compressor design using mostly junk box parts and it works well.

I am using a 2N7000 as a voltage controlled resistor. It works fine, but cannot tolerate more than 80 mV or so across the drain to source. To solve this problem, I use shunt feedback in my first amplifier stage. This results in a cancelling voltage subtracting from the input signal at the VCR node. I end up with 30mV max across my MOSFET VCR with a 2 volt or so input. Just about any general puprose NMOS FET should work. I'll post a sound clip soon

Also, I use a LED as a peak detector diode. I did this because it lights up with varying intensity, proprtional to compression. This provide a nice visual compression level indication. Of course, you can use any generic silicon diode. The schematic is annotated with relevant design details. Also, there is more description at: www.circuitsalad.com

Schematic Link:
http://circuitsaladdotcom.files.wordpress.com/2012/10/simple-compressor.gif

[PRR]

Looks like it will work (of course).

The LED peak-flasher is brilliant. So is the MOSFET reference voltage.

> I use shunt feedback in my first amplifier stage

Yes, that reduces the signal there, but it also reduces the effect of the to-ground resistor.

And there are trade-offs between input impedance and resistor hiss voltage.

The 2N7000 can go down to an Ohm. R1 (series pad) could be down near 20K for 10dB lower hiss. That would of course need a buffer for e-guitar, an added complication.

The best loading for this type attenuator would be infinite, not the low Z looking into Q1 with 1Meg series NFB. You could perhaps re-allocate Q1 to an input buffer, get all the recovery in IC1A.

> cannot tolerate more than 80 mV or so across the drain to source.

Won't tolerate? Or distorts badly?

In J-FETs there is a trick. The real problem is that the channel resistance responds to the _average_ voltage fron gate to channel, but one end of the channel bops up-and-down with signal. When signal becomes a significant fraction of bias+CV, gain changes over the cycle, distortion.

In J-FETs one trick is to add half the signal voltage to the gate. Then the gate moves with the average channel voltage. It doesn't work miracles. Max signal voltage is still sub-Volt. But THD may be very greatly reduced up to that higher point. If the pad impedance is very low, two 1Meg resistors can do this. In other cases the correction must be buffered; the "-" side of the recovery opamp has a low-Z version of the pad FET voltage.

I do NOT know if the same thinking works with MOSFETs. To some degree a FET is a FET; OTOH most MOSFETs are burdened with parasitic diodes and other quirks which do not hurt the switching-chores they are marketed for.

[rring]

Yes the problem is distortion of going from triode region to saturation. The JFET would be better, but I would need an inverted signal with a Pchannel(pos to neg) or a negative control voltage (going in the right direction) for a N channel. The mosfet simplifies the design but it may be worthwhile to look at a JFET. You are correct about the resistance creating noise. It may be worthwhile to rework the ratios here, currently the noise level does not seem problematic for guitar applications. I am familiar with the linearization technique with the JFET.  It may help here, but early experiments showed it not it not to be of much benefit. I found just keeping the voltage across it low did the trick. Also, I am biasing it to just barely be on and only driving it over a 1 volt or less  of swing so it stays mostly in the middle of the triode region , where the I/V curve is reasonably straight. The compression ratio ends up  a little better than 3:1 max. This of course has to do with the total amount of control loop gain. One thing that was interesting was the peak detector, which changes level abruptly - so I had to add an integrator to the input of the mosfet to smooth transient spikes of the peak detector's fast response which created a great deal of noise. You can't use just any op amp here either as it needs to operate near ground and tolerate driving it below ground.