Hi Kris,
Fred, if you have a chance would it be possible to get a little walkthrough of the sidechain circuit.
Yeah I can do that.
The gain reduction element is effectively a variable resistor, whose resistance is inversely proportional to the current driven into the LED. This resistance, then acts to pad down the input to the output stage.
Yes, the LDR acts as a variable shunt element as in a L-pad, working against the series resistance created by R43 and R44.
The threshold control appears to be a level adjustment on the sidechain input buffer amp. After this, the signal is paralled and one side inverted then the two signals are rectified by the small signal diodes. Out of curiosity, why not use an absolute value, or precision rectifier type circuit here? Maybe the forward diode drop isn't significant given the response time of the compressor/Vactrol...or maybe you're using the forward drop to set the threshold of compression (thats what I'm currently thinking, and would explain how you can have a threshold control without a comparator to see if the level is over or under threshold).
Yes, the threshold control is a simple level control that adjusts the signal level feeding the side chain. U5A is configured for some gain (appox +24 dB) to allow the compressor to function at low signal levels. U5B is a unity gain inverter. The outputs of U5A and U5B are rectified through D9 and D10.
The reason I used this scheme to create a full-wave rectifier instead of the more commonly encountered precision full-wave rectifier is simple; I wanted to use D9 and D10 to prevent C24 from discharging into the very low source impedance of U5A/B. While a precision rectifier could be built with one opamp, its output connection would cause C24 to discharge very rapidly preventing a long release time setting. Making C24 larger to try to improve the release time would be problematic. The only drawback to my approach is that signal levels lower than one diode drop (about 0.6v) will not trigger the compessor. Adding gain to U5A took care of that problem.
The main thing to think about from this point on in the side-chain is that everything after D9/D10 is DC.
So what happens is that the output of D9/D10 is a series of positive going AC waveforms. C24 charges to the peak value of the outputs of D9/D10 and discharges through the release time pot and R26 to ground. The input impedance of U6A is very large compared to the pot/R26 path to ground, so it doesn't play much of a role in the release time characteristics. R26 determines the quickest release time. The 500k pot determines the longest release time.
U6A buffers the release time R/C network from the Attack time R/C network and provides enough drive current to charge the attack time cap quickly. It also help reduce interaction between the attack and release time controls.
C25 and the output charactistics of U6A determine the fastest attack time (if we ignore the LED/LDR time constants) when the 500k attack time control is not creating a resistance in series with C25 (fully CCW). In this case, C25 will not add to the release time characteristics because it discharges rapidly into the output of U6A. As we increase the attack time by moving the pot CW, C25 does take longer to discharge, so it is important to use a fairly small value C here. R27 is there to provide a ground ref to the input U6B and needs to be large so as not to attenuate the drive to C25.
U6B is a unity gain buffer that provides a good high current drive to the LED/LDR cell(s). The input of U6B is also a good point to passively sum drive signals for a multi-channel compressor.
The ratio control looks like an variable current limiting resistance on the output of the sidechain, which makes sense to me as it will change how hard the LEDs are driven.
I notice too that the circuit is set up as a feedforward compressor....I wonder how hard it would be to tweak to be feedback....I guess I'd want to pad down the threshold an amount equal to the amout of make-up gain I'm applying in the output stage (may need a two deck switch).
Yeah, that's how the ratio works. Another way to do that is to place a variable R in series with the LDR (to ground), and for a long time I did it that way, but it makes is difficult to design a GR metering circuit that tracks the amount of gain reduction as you vary the ratio. If one were to check the tracking on other opto compressor design (as a function of ratio control setting), you'd find that almost none of them (none of them?) track properly. It does on my design, if you use my metering circuit (once you get it calibrated), at least well enough to be useful.
With regard to feedback/feedforward-- my desire was to design a reasonably simple side-chain. There is a bit of feedback and a bit of feedforward on this design, at least with regard to how the side-chain functions. The point at which the signal to the side chain is picked off determines this. If you were to reduce R43 to zero, you'd have an all feed-forward design (and you'd have to increase the value of R44 by 4.99k). If you were to reduce R44 to zero, you'd have a feedback design. I chose values for the R43/44 ratio by trail and error. You might have fun playing with these values to see what effect it has on the design.
It is possible to use this design as a gate instead of an compressor. The LDR would be connected in series with the signal (like R43/44 are now) and you'd need a resistor to ground where the LDR is now. You would want the side-chain pick off point to be prior to the LDR, therefore it would be a feed-forward design. Max attenuation would be determined by the ratio of off resistance to the shunt R resistance (and may vary as a function frequency, so be aware). Min attenuation (gate on) is determinded by the ratio of the LDR on resistance and the shunt R.
Anyway, that's kind of the idea. I hope my explaination makes sense and that it addresses your questions.
I wish a great new year for all of you.
