Not all compressors are born equal. You have probably already heard of different types in regards to the gain reduction element (here is a great article about FET, VCA, optical, vari-Mu and diode bridge compressor types) but although those have a big effect on how the compressor sounds, the signal flow's architecture (the so-called topology) can also make a big difference!
Feed what? Topology?
So what's with all that feeding, who's so hungry all the time?!
The expressions "feed-back" and "feed-forward" are slightly more confusing than the male and female connectors.
So let's start at the beginning. A compressor is, broadly speaking, made out of 4 basic building blocks. Those are: Input stage, Gain reduction element, Output stage, and Sidechain. They form the "compressor topology" as follows:
Input and Output stages are pretty self-explanatory, the first one welcomes the signal from the input connector and prepares it for internal processing, and the other one makes sure that what's coming out of the box is sends it off back to the real world no matter what you connect to the other end of the cable. Easy.
The gain reduction element is what gets the job done. If the signal is below the threshold, this block should not do anything, and when it's above the threshold, it should attenuate it by lowering its gain.
But who tells the gain reduction element when to change the gain and by how much?
It's the sidechain! This part of the topology is a big part of the compressors' functionality and sonics. It looks at an audio signal at one end and outputs the control signal for the gain reduction element on the other.
This control signal tells the gain control element by how much it should reduce the gain of audio. Parameters like Threshold, Ratio, Attack, and Release are applied and their effect plays a major role in what kind of control signal is generated and how the actual compression sounds as a result.
But where does the sidechain listen in on the audio that's passing through the compressor to generate the output signal? The answer to that is the feed-back vs. feed-forward mode and that's what sets these different types of topology apart.
We can tap into the signal path just before the output stage, after the gain reduction element. Since that means that we're returning to the previous parts of the circuit, it is called the feed-back topology. The thinking process for this kind of arrangement is "when the output starts to get too loud, I'll turn down the volume" - it's reacting to change without anticipating it.
This way of compression is well-suited to general thickening and level riding with colorful timing constants, but can also be made extremely fast, especially if using the FET or VCA as a gain reduction element.
And who are the representatives of this type at mixanalog?
... or we could try and anticipate when the output will get too loud and prevent that from happening in the first place! We can go about that by looking at the input signal and calculating what will happen in the next stages, then reacting appropriately. This means feeding the input signal forward to the sidechain, so it can adjust the control signal in advance of the gain reduction element - hence the name feed-forward.
This type of architecture usually sounds more "snappy" or "grabby" and is a preferred one when you want to tame fast, sharp transients. But as with the previous one, it can be made to sound more "laid back" by dialing in slower timing constants and gentle ratios.
The most well-known compressor of this type is without a doubt the SSL mix bus compressor. An augmented, more versatile version of this classic design is available as a part of RACK1 by the name of "Gold Can" VCA compressor.
And there you have it - another nerdy term demystified, in a way (we hope) is understandable to all. Happy feeding, err, compressing!