Well, as Wayne (mediatechnology) also pointed out, there's also the benefit of reduced CV feedthrough. -Many VCAs will allow a small amopunt of the modulated control voltage to 'bleed-through' into the signal. With a muting circuit for example, this manifests itself as a quiet 'thump' behind the signal. In a compressor the DC excursions are usually smaller, and the fact that the audio doesn't suddenly disappear also tends to rather 'mask' the problem.
With the cross-pole VCA approach, the effectively reversed control voltages on half the VCAs mean that any slight leak-through is basically canceled... In the same exact way that a variable-mu compressor (which has colossal feed-through in any single pole) cancels out when the two poles are combined. -In the case of variable-mu, each pole's Control voltage appears massively on the signal, so any misadjustment of the balance tends to introduce this low-end 'thumping' modulating with varying GR. In a VCA the CV bleed-through is already very low and often ignored, but this circuit helps quell any feedthrough still further.
The benefits summed up are:
Improved distortion -perhaps by as much as an order of magnitude. -DEFINATELY easily measurable.
Improved input CMRR, since now both input legs have the exact same impedance... In the original Gyraf version, the cold leg has an effective impedance of 44kΩ and the hot leg has an impedance of 22kΩ... in the paralleled version, both legs have an input impedance of 15kΩ.
Improved output current drive. -The output should be able to provide the same performance into an impedance which is lower by a factor of two than the original design... in case you want to go leathering a hot output signal into a 600Ω input impedance for example.
Noise should be slightly lower. -Perhaps a 3dB improvement, though we may be into the realm of diminishing returns... the unit shouldn't really be noisy to begin with (unless you used 2001 VCAS perhaps).
I think those are the major points... but I do like to see people's eyebrows raise when I tel them that my compressor has 34 VCAs in it, and they're ALL being used, except for when it's in 'Gyraf' sidechain mode, when only 33 are being used!
Basically, when I put this in the bench, it exhibits mind-numbingly tiny readings of noise & distortion... amongst the very best of ANY unit I've ever measured.
However...
To the people who've been using this for the last 3 months, none of that means anything, nor has it been anything that they've concerned themselves with... -They LOVE the switchable sidechain. -They can't get over the fact that it changes the way the compressor operates. The Alan Smart doesn't do what it does, neither does anything else that they have, or anything else that I can think of.
The unit is worth building this way, even if only for the two sidechain approach.
Hey, Rob, -did you do that one yet? -if not, just wire up an SPDT switch to interrupt the sidechain signal of the slave at the summing resistor (next to the omitted one on the board) and switch it to either continue normally on it's way ("Oxford" SSL mode) or to meet up with the same point on the master unit, ("Aarhus" GSSL mode) and revel in the pleasure.
The single biggest demonstration of the difference in behaviour comes when you pan a tone from side to side on a mixing console. -In Oxford mode it compresses less in the center. IN GSSL mode it compresses more in the center.
THAT my friends, has been the revelation of this project. The difference is wonderful. THAT is the reason I can't prise the unit out of my friends' cold, clammy hands... and it's worth more than any distortion or noise measurements. -And it's why I reckon a couple more people should build it this way, because of the power it brings...
Nobody complains that the GSSL is noisy (unless they used 2001s and are doing very quiet music, perhaps) and nobody says that it's harsh or distorted sounding. Most of the test numbers that I foind I'd improved upon, I think are more than good enough in the original Gyraf version, so don't feel you HAVE to do this or yours will be unlistenable... but DO please take away that the behaviour of a twin-sidechain unit is markedly different, and THIS may be something that you will really enjoy.
Rob, I'd really welcome your comments on how you feel about the sidechain behaviour.
Keith