Dyna Comp Envelope Detector Diodes

[Stratomaster]

Hello all,

I was wondering what the purpose of the diodes in the Envelope Detector was--and if there was modding potential here.  The circuit analysis on Electrosmash says they're merely protection diodes from negative voltage.  An old thread on here about the JC Maliet 'Hi-Fi Mod' talks about removing a cap to create a half-wave rectifier to eliminate the potential for clipping by the diode pair to ground.  In that thread, someone had mentioned replacing the diodes with LEDs and not getting much audible compression.

It would seem that the logic is sound-- rather than resort to half-wave rectification to eliminate the possibility of clipping--simply use a diode with a higher forward voltage to raise the clipping threshold. However, it seems like this can be taken too far and raise the threshold of the compressor to where the signal never compresses (see LED comment above). 

My question boils down to this:  can I swap the diodes here to intentionally reduce the possibility of clipping and raise the threshold of compression simultaneously?  Or does this not compute?  Meaning a high enough forward voltage simply bypasses the envelope detector and shuts off compression without any 'in between'?

[Rob Strand]

I was wondering what the purpose of the diodes in the Envelope Detector was--and if there was modding potential here.  The circuit analysis on Electrosmash says they're merely protection diodes from negative voltage.

The older oranger Ibanez compressors CP102 (?) didn't use them but most other units did.

Those diodes are quite tricky. The rectifier only activates while it is peak detecting.  Without the diode it relies discharge through the 1M to get a another peak detect.  This means the rectifier doesn't activate much on cycles following the first peak.  With them the negative cycle resets the cap through the diodes and so it forces another peak detection.  Another effect is the voltage on the cap in the negative cycle removes the base-emitter threshold (but only after the first negative swing).

My question boils down to this:  can I swap the diodes here to intentionally reduce the possibility of clipping and raise the threshold of compression simultaneously?  Or does this not compute?  Meaning a high enough forward voltage simply bypasses the envelope detector and shuts off compression without any 'in between'?

For this circuit the threshold is an output threshold.  As you dial the gain up it reduces the input threshold.

You could increase the threshold by adding diodes in series with the emitter or base of the two rectifier transistors.
The other way is replace each 10k phase-inverter resistors with 2x4.7kin series then wire the centre points to the rectifier.  The output would remain off the emitter.

[Edit:
Despite all the above I suspect the reason the diodes were added was to stop negative swings on the base.
It was a common thing to do on transistor circuits.]

[PRR]

At 9V, the negative base swings are not fatal.

As Rob says, without the back-diodes, the thing desensitizes an unpredictable amount as C6 C7 charge-down from feeding a diode (the transistor). With the back-diodes, it is diodes both ways, it works "proper".

Adjusting the threshold of limiting *should* be done by adjustment of audio gain from input to detector. This circuit gives no easy way to do that; it "happens" that a 0.6V output level is a bit too hot for guitar-cord work, the Output pot trims that, and the Sensitivity pot trims gain and thus input threshold.

If you want less knob interaction, you need more parts.

[Transmogrifox]

If you don't mind flying-leads mods, then I would just add a second buffer transistor off the output of the CA3080 for the output.

Distortion from driving the diodes will not be in the output signal path and you will be buffering off a part of the circuit that isn't effected by that in any appreciable way.

As an answer to your question, the diodes can be modded and get audible changes in the way the circuit behaves, but my intuition says you will cause side-chain ripple to increase by increasing the diodes, which will make it sound more distorted.  If you do anything with these diodes I would think that going for lower-drop diodes (Schottky diodes) would tend to improve things if it makes any audible difference.

Another mod that could help buffer the signal from distortion would be to add a 10k resistor in series with C7.  Then the distortion from reverse biasing the diode will be significantly less audible.  The single 10k will be easier to shoe-horn into an existing board -- C7 footprint with C7 and 10k meeting in the air.