Why is SSL style compression so popular? [general question]

[gyraf]

..if you try to pull current from the timing capacitor, you discharge it. The buffer amp is just an "isloating" unity gain amplifier, making it possible to source at least some current / load it/ without disturbing the capacitor timing voltage (as the opamp is very high input impedance)

 

[altfidelity]

I was reading about different VCAs, and the most simple approach (from my slightly limited point of view) would be using FET vca.

The gain reduction stage from Dan Electro Orange Squeezer circuit seems pretty minimalistic. Would it work on line level applications? Are FETs difficult to match and generally hard to work with, especially for dual/stereo purposes?




What's your opinion, what the most elegant solution regarding VCAs?

 

[JohnRoberts]

I was reading about different VCAs, and the most simple approach (from my slightly limited point of view) would be using FET vca.

The gain reduction stage from Dan Electro Orange Squeezer circuit seems pretty minimalistic. Would it work on line level applications? Are FETs difficult to match and generally hard to work with, especially for dual/stereo purposes?

What's your opinion, what the most elegant solution regarding VCAs?

FETs are crude but cheap...much worse than real current ratioing VCAs.

Elegant would be the latest generation of THAT corp IC VCAs.

JR

 

[altfidelity]

Elegant would be the latest generation of THAT corp IC VCAs.

So, ok good start would be right at the core of GSSL, right?



Basically, the safest bet would be to keep whole schematic of Jakob's GSSL, apart from CV/sidechain, which would be take after GR stage, right?
But then I'd have to have additional make gain stage?

 

[richiyobs]

i like Jfet cause easy to build and experiment with but distortion and noise is the first rule...
i m still a beginner but experimented/played a lot with different kind of VCA ... it so funny

Gssl, for VCA , is still the best design i ve built ! very very clean and great sonic flavour
" super Gyraf , always thanks for this great device! "

i ve simulated a lots of schemos to learn a bit , take a look at Urei, Trident and D&R for Nch-jfet if you want to experiment with this typology

if you wanna do experience with VCA look at THAT Design Notes, the site section is full of great example to follow, but Gssl rules!!!!
Best

 

[altfidelity]

At this point I should probably rename the topic to "Dumb person asking dumb questions loosely related to GSSL schematics" but:


question 1


I am having problem understanding log/antilog/decilinear math/logic behind DBX202 equivalent circuit.
If the CV is near 0, the amplification factor of THAT2181 is I guess 1:1 or 0 dB right?


What's the needed voltage to achieve lets say reduction of -20dB? 1volt right?


question 2

Can I replace following/circled part of the curction with an isolator 1:1 transformer + buffer stage?




Thanks for helping me out.

 

[dmp]

If someone truly wanted to make an innovative compressor in 2023, they probably wouldn't be messing with capacitors, resistors, and semiconductors.

They'd be working with lines of code!

Yes, like multiband compression or frequency dependent compression (i.e. de-ess) - in hardware you'd have a rack of EQs and compressors to do something a plug-in elegantly does.
I was looking over the shoulder of a "pro" engineer working on drums and he was doing this with really impressive results. But having the ears to know what needs to be done and the technical skills to accomplish it are key.
You might want to think if the hardware is doing something that cannot be done with a plugin and focus on that.

 

[Script]

So you mean transformers, inductors, tubes and diodes ?

 

[JohnRoberts]

It really helps to know what result you want...

JR

 

[dmp]

So you mean transformers, inductors, tubes and diodes ?

Maybe - I'm not going to pretend I know the answer or that it isn't perhaps different for everyone.
But in my experience, my SSL clone (Mixbuzz) isn't that distinctive. Not like my LA2A.
The complexity of some components are more difficult than others.

 

[altfidelity]

It really helps to know what result you want...

JR

The trend in hardware is that all compressors aim to achieve one specific compression action. I would like to try to make compressor, that would be as flexible as possible and have sort of chameleon character.

When mixing, I (personally) do not think "in compressors", this needs a bit LA2A levelling, or DBX spank , or SSL glue, or VARImu thump etc.
I think, "I need to clamp this transient", "I need to enhance the transients", "I need some slow pump action to enhance the rhythm..." etc.

And most of compressor have very "fixed" sidechain, that doesn't encourage experimenting and have few side chain controls (attack, release, ratio). And those design that experiment with this kinda stuff are very expensive (example GML 8900...).

The idea of blending secondary envelope running parallel to peak detector (with variable integration time and amplitude), which would integrate peak detector, should/could provide enough space for further experimenting and shaping the sound.
With greater amplitude of secondary envelope, combining with feedback topology, the classic VCA response compression action could mimic some of the program/time/dependencies found in opto levellers, and used for mixbus & mastering duties.
With lesser amplitude of secondary envelope, it wound act as a standard VCA peak compressor used on drums, vocal etc...

That's the theory (though tested in digital).

In reality, I have far to few experiences to test this theory in practice! it frustrating

 

[Script]

those design that experiment with this kinda stuff are very expensive (example GML 8900...).

Wasn't the 8900 conceived to yield 'invisible' compression ? Strikes me as a prime example for code. Maybe not

 

[altfidelity]

SPL is also doing something similar in KULTUBE compressor.

 

[gyraf]

I wouldn't read too much into this, seems more like marketing to me

 

[Script]

Their Progressive Time Control (PTC) could be as little mysterious as THAT's "Non-Linear Capacitor" (Design Note 114, I think). Maybe a mix of ATT/REL and NLC.

 

[gyraf]

https://www.experimentalistsanonymous.com/diy/Datasheets/SA571 AN.pdf - p.777 has a variation of this

..and this NE570 application note is generally a fine intro to compressors..

/Jakob E.

 

[altfidelity]

Just when I was about to re-invent the wheel...

 

[gyraf]

If you want to try something new, I've had this idea for some time, and haven't gotten around to realize it:

What if we, for compressor A/R timing, skip the very-traditional first-order R/C filter stage - and in stead use a higher order active filter. Like a state-variable filter or such. Even perhaps with variable Q, resonating around cutoff..? I have a feeling that this could make compressor timing much "tighter" without distorting, more timed and less program dependent, with perhaps new musiological possibilities..?

Might also just be crappy, but ymmw

/Jakob E.

 

[altfidelity]

I also experimented with band-passing sidechain with high Q filtering in sidechain also. But results were kinda mixed, the CV oscillates around certain frequency range, which sounds like distorting/emphasizing certain portions of signal. I do firmly believe in hi-passing sidechain signal though.

The most usable approach the yielded nice result, was using parallel CV envelopes. Speaking simplified, using one very fast envelopes in parallel with one or two slower ones. It remainded me of varimu compressors, when compressor initially grabs quite quickly and aggressively, but the sort stays up in GR zone leveling with secondary envelope (which kicks in later on).

Example of attack and release curves using my so-called inertia parallel approach.

without inertia



with inertia



The trick, I think, is in timing of those secundary envelopes. I choose one at 1sec integrating, the third at 5sec.
The third on at 5sec, probably rarely really kicks in I guess. But It adds certain weight on mix bus applications.

EDIT: I'am also proud that I've implemented push-pull principle for gain reduction cell, not just digital liniear multiplier/attenuator.
So occasionally when overloading I get crossover distortion.

 

[david moran]

seriously ignorant question from one who does not keep up: what do 'modern' compressors do, euphoniously or otherwise, that 40yo dbx 16_ compressors don't or didn't ?