After an especially long time of us promising analog parallel compression and a mid/side matrix, they are finally here! We worked our asses off during the last two weeks and even managed to summon the "white smoke of death" during some circuit tests - but now all of that is behind is and it is (a)live in mix:analog!
What is it?
Parallel compression has been around for decades as an industry secret on getting a lot of tone and sustain from tracks without sacrificing dynamics. When you use a compressor like the Fairchild 670 or Gold Can VCA, you can get awesome tone from it, but the final result can get mushy, squashed and hard to listen to.
By using parallel compression, a copy of the original signal without compression is added along with the processed signal, bringing back the dynamics while retaining the tone.
With a mid/side matrix, the traditional left/right stereo signal is broken up into two new kinds of channel: mid and side. The mid channel contains audio that is the same for left and right speakers, that is, tracks and sounds that you are keeping straight in the middle. The side channel contains all the other audio.
This is useful if you want to process sounds in the centre differently to the other ones, usually because the former has a lot of bass and energy which reacts very differently to compression. Similarly, the side information can traditionally be made more energetic in the high frequencies without making the overall sound peaky and shrill.
Why analog MS and parallel mixing?
Unlike analog compression and EQ where the audio engineer is usually looking for a colourful, interesting palette of sounds to make a mix or master, parallel compression and mid/side processing is supposed to stay out of the way and not colour the sound too much. This is usually relegated to digital signal processing (DSP) where the least amount of colour can be achieved.
We decided to do it a little differently by making both mid/side and parallel compression in the analog domain with electronic circuits instead. Here's why:
What if I want only one processor in the whole chain to work with mid/side?
If you've been reading our previous posts, you're probably already familiar with our fanaticism to minimise AD/DA passes when building chains of gear, for which purpose we designed a 1000-relay monster of a digitally controlled patch bay. And in keeping with that philosophy, we're not cutting corners here either. So we designed an analog add-on board with switchable MS encoding/decoding loop that "encapsulates" the compressors and EQs, so this kind of operation can be engaged "per processor" without adding another AD/DA pass that would be required with the usual DSP approach.
To ensure the best possible M/S separation, precision-cut summing network resistor op-amps were used.
On the same board, there is also a 4-channel mixing circuit. Being all-analog, forget about sample accurate alignment with AD/DA loop delay compensation - this is as accurate as physics go! At the "fully wet" setting, the whole mixer is relay-bypassed and the number of active stages minimised. All internal active stages operate at a fairly high impedance, so there is no need for bulky (and distorting) DC offset blocking electrolytic capacitors in the signal path, only high-quality film type caps are used. In a nutshell, we went for a no-compromise, mastering grade design.
At the time of writing, the GoldCan VCA compressor in Rack1 and Fairchild 670 got the full treatment, with Elysia museq and Pultecs joining them later this week.