Vintage compressors reverse engineered with ReaComp
Posted: Wed Mar 02, 2016 10:25 am
Edit 2022. For the record. I have recently incorporated the subsequent information and much more into a written document on the subject that I have published earlier. This document is about using compressors to instill some sense of loudness early in the mixing state. It uses Reacomp as a versatile compressor engine, and I have worked out several basic task-oriented compressor models besides looking at the vintage icons below and more. Against all odds, I finally published ed.2, and it has become eminently useful. You may want to behold it here. The subsequent information is thus redundant and likely obsolete. It may remain there as a reference and for consitency. .
This is for you folks to try out.
I find compressors hard to understand. They are powerful tools that demand to be understood, or you are at the mercy of boutique plugins or presets.
I suspect very few people understand them thoroughly and thus resort to those units, that are proven to work, such as 1176 and La-2a types. It is ironic that those are half a century old.
Unfortunately, buying those elusive plugins does not make me wiser.
I went to find out what the difference in those units is and what makes them so special. This is what I found till now:
#1) don´t let yourself be fooled: aesthetics
#2) some hard to define analog quirks like tube saturation and subtle coloration. For those who hear the grass grow.
#3) hard and fast electric peculiarities as follows.
* LA-2A uses an optical element (light bulb) to excite the gain reduction. This is very slow. How slow in milliseconds is hard to say, but this acts as an averaging element. The signal is averaged over time. LA2A has a fixed attack time of 10ms, but seems to vary up to 80ms. Auto release seems to work well. A slim knee of <1dB is in order.
The side chain is derived from the main input (detector input).
Edit: the LA-2A seems to be of feedback type too.
Note that in the comparison the threshold has been set some dB lower, since the detector circuit "sees" less.
*1176 uses a FET that is very fast (sub-ms). Also a knee of <1dB (hard knee on 1:20). The detector is fed from the output (feedback type), but this time the signal is not averaged. It is not as fast as a VCA (peak = 0ms), so a few ms are in order.
This units uses classic attack, meaning the curve is heading towards 1:1 at higher excitations again.
This exercise would be meaningless if it was not compared to an "official" 1176 unit. I used the free audiocation compressor that is said to be much like an 1176. Use the same settings and compare for yourself. This unit makes this easy, because you see the figures beside the controls. All tests used a drum part made by EZD.
Note that the audiocation unit cheats (like most others) with their presets according to the motto "louder is perceived better". Aim to set the output compensation painstakingly correct for subjective equal loudness, or the comparison will be unfair.
I also tried Stillwell´s Rocket, also an acclaimed 1176 clone, but this is difficult to compare, since (like before) the presets are hyper-inflated in volume, the controls are unreadable and ambiguous and the meters too slow. Fancy it is though.
So many of those retro-looking devices are flawed by the fact, that the controls are unreadable on a first look.
You always have to read those. For an equalizer this means, you have to read all of them to see where the tweak is. Centuries of optimization for easy readability and space age handling improvements have bypassed those units.
We humans need a mental map engraved in our brains so to find the relevant parameters easily and efficiently and to recognize differences in one look. All other things may look fancy, but are nothing but distraction if you pay attention.
On top of that, there was no standardization on what knobs show, like attack. They are basically labelled "min to max", which makes them hard to compare, let alone simulate with other units.
Those Rea plugs compete well in this respect and are very powerful.
Verdict: I can hear no darn difference between the 1176-ish setup and the "official" emulation!
However, the La2A does sound different (I have no free vst unit to compare). This is mainly down to the different RMS computation window. Play with that, this makes the biggest difference.
It is to be expected (or hoped) that expensive emulations make something better, but if reacomp gets me close to them, I´m in!
Attached you will find two screenshots that show all settings for both presets.
Let us know, what you find, fellow mixers.
have fun,
-helmut
(to be continued...)
Addendum:
A guy here emulates an SSL-G with reacomp:
This is for you folks to try out.
I find compressors hard to understand. They are powerful tools that demand to be understood, or you are at the mercy of boutique plugins or presets.
I suspect very few people understand them thoroughly and thus resort to those units, that are proven to work, such as 1176 and La-2a types. It is ironic that those are half a century old.
Unfortunately, buying those elusive plugins does not make me wiser.
I went to find out what the difference in those units is and what makes them so special. This is what I found till now:
#1) don´t let yourself be fooled: aesthetics
#2) some hard to define analog quirks like tube saturation and subtle coloration. For those who hear the grass grow.
#3) hard and fast electric peculiarities as follows.
* LA-2A uses an optical element (light bulb) to excite the gain reduction. This is very slow. How slow in milliseconds is hard to say, but this acts as an averaging element. The signal is averaged over time. LA2A has a fixed attack time of 10ms, but seems to vary up to 80ms. Auto release seems to work well. A slim knee of <1dB is in order.
The side chain is derived from the main input (detector input).
Edit: the LA-2A seems to be of feedback type too.
Note that in the comparison the threshold has been set some dB lower, since the detector circuit "sees" less.
*1176 uses a FET that is very fast (sub-ms). Also a knee of <1dB (hard knee on 1:20). The detector is fed from the output (feedback type), but this time the signal is not averaged. It is not as fast as a VCA (peak = 0ms), so a few ms are in order.
This units uses classic attack, meaning the curve is heading towards 1:1 at higher excitations again.
This exercise would be meaningless if it was not compared to an "official" 1176 unit. I used the free audiocation compressor that is said to be much like an 1176. Use the same settings and compare for yourself. This unit makes this easy, because you see the figures beside the controls. All tests used a drum part made by EZD.
Note that the audiocation unit cheats (like most others) with their presets according to the motto "louder is perceived better". Aim to set the output compensation painstakingly correct for subjective equal loudness, or the comparison will be unfair.
I also tried Stillwell´s Rocket, also an acclaimed 1176 clone, but this is difficult to compare, since (like before) the presets are hyper-inflated in volume, the controls are unreadable and ambiguous and the meters too slow. Fancy it is though.
So many of those retro-looking devices are flawed by the fact, that the controls are unreadable on a first look.
You always have to read those. For an equalizer this means, you have to read all of them to see where the tweak is. Centuries of optimization for easy readability and space age handling improvements have bypassed those units.
We humans need a mental map engraved in our brains so to find the relevant parameters easily and efficiently and to recognize differences in one look. All other things may look fancy, but are nothing but distraction if you pay attention.
On top of that, there was no standardization on what knobs show, like attack. They are basically labelled "min to max", which makes them hard to compare, let alone simulate with other units.
Those Rea plugs compete well in this respect and are very powerful.
Verdict: I can hear no darn difference between the 1176-ish setup and the "official" emulation!
However, the La2A does sound different (I have no free vst unit to compare). This is mainly down to the different RMS computation window. Play with that, this makes the biggest difference.
It is to be expected (or hoped) that expensive emulations make something better, but if reacomp gets me close to them, I´m in!
Attached you will find two screenshots that show all settings for both presets.
Let us know, what you find, fellow mixers.
have fun,
-helmut
(to be continued...)
Addendum:
A guy here emulates an SSL-G with reacomp: