Understanding the gssl auto circuit

[fandusss]

Hi all,

I was hoping someone could help me conceptually grasp the inner workings of the stacked RC circuit the GSSL uses when release is set to auto. A tall order, perhaps.

I will outline my understanding, and if I’m wrong, I’d really appreciate being set on the right path.

The circuit appears to use shared capacitor(s) for both attack and release phases. There’s an attack resistor feeding the cap during the charge phase, and a release resistor to handle the discharge phase. This would also explain why these units do not have any settings that would result in the release time being shorter than the attack time - they are acting on the same capacitor and such a setting would result in any charge being instantly drained.

Assuming so far I have not made any horrific errors, enter the stacked release. Would I be correct in thinking that the intended attack time set by the pot would apply only during the duration it takes for the smaller capacitor to charge, during which both capacitors would be charging at this rate while current is passing through the capacitors. Once the small capacitor is charged, the larger capacitor would then switch into a new, longer attack time to charge the rest of the way, as it now has the small caps release resistor acting in series with the main attack resistor?

I would assume there is no sudden switching, but rather a gradual transition between these extremes as charges build.

Am I thinking about this all correctly?

Thanks for reading, if I’ve said anything stupid please correct me, I am just trying to learn.

 

[Rob Flinn]

In a nutshell it has 2 time constants, one fast one & one slower one. This is because it has two pairs of caps & bleed resistors with different values stacked on top of one another. One discharges quickly & the other more slowly, so the initial release is quick but isn't a full release.

 

[abbey road d enfer]

Would I be correct in thinking that the intended attack time set by the pot would apply only during the duration it takes for the smaller capacitor to charge,

Yes.

during which both capacitors would be charging at this rate while current is passing through the capacitors. Once the small capacitor is charged, the larger capacitor would then switch into a new, longer attack time to charge the rest of the way, as it now has the small caps release resistor acting in series with the main attack resistor?

The larger cap gets charge more slowly than the small one.
There is virtually no effect on the way the control voltage develops, because the charge path is capable of delivering more current than the discharge path.
The difference is at release. In addition to a fixed discharge path, there is an additional path, that goes to the larger cap; depending on how the larger cap is charged, the release time will vary.
If the signal has been short, the larger cap is not significantly charged, hence the release time is short, but if the compressor has been subjected to a long signal, the larger cap is fully charged and the release time is longer.
Operationally, a transient will be caught as expected, an the comp will return very quickly to no GR; alternatively, if the signal is permanent, the release time will be high, which reduces LF distortion.

I would assume there is no sudden switching, but rather a gradual transition between these extremes as charges build.

Indeed.

 

[JohnRoberts]

I am not familiar with GSSL innards, and too lazy to search for a schematic... that said I know a little about dynamics side chains.

Perhaps not obvious but when a dynamic processor varies the gain of an audio signal it introduces distortion, effectively multiplying the audio signal by the side chain gain control voltage (or current). In an ideal world we want these gain changes to be fast enough to control transients while slow enough to audible distort the audio signal. An obvious but extreme example of this compressing a LF sine wave with fast release. The processor tries to start releasing the gain reduction during the LF waveform causing visible distortion. This is usually masked by complex waveforms but demonstrate the tension between fast and slow gain changes.

The hard part is how to transition between fast and slow time constants... there are more ways to skin that cat, than cats, so I won't bore you with my list.

JR