Calling the SSL bus comp a hybrid between FF and FB is kind of a misnomer. Technically the signal is split before the VCA to feed the detector circuit - which looks like a feedforward compressor, but the detector itself is made up of a VCA in a feedback loop - causing the compressor to act as a FB compressor.
You can see it here: http://www.gyraf.dk/gy_pd/ssl/ssl_sch.gif
In the third row, you'll see the "sidechain VCA," and at the end of the fourth row you'll see the TL072 that's feeding the "CV to VCAs." This is a classic feedback loop.
I'll dig into the details more later.
Presswerk patch/emulation area + inits
-
OneRoomStudios OneRoomStudios https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=94231
- KVRer
- 26 posts since 13 Jan, 2006
-
- KVRAF
- Topic Starter
- 3817 posts since 8 Mar, 2006
^ I understand (but almost no clue when it comes to schematics 
)
I set the detection to FF and compensated with a small (+ 0.50 ms) positive amount of sidechain delay (this way, we get that fit snap associated with this compressor)
RMS value for the SSL bus comp? (SSL G4000 bus)
)I set the detection to FF and compensated with a small (+ 0.50 ms) positive amount of sidechain delay (this way, we get that fit snap associated with this compressor)
RMS value for the SSL bus comp? (SSL G4000 bus)
-
- KVRAF
- Topic Starter
- 3817 posts since 8 Mar, 2006
BTW guys, seems like the best -init- setting for the sidechain DELAY for a 1k sine tone is around -0.25 to -0.30
I tested this with a sudden level increase on a hard knee, max ratio, minimum time (attack, release (rel atapt too), RMS)
The recorded results present an "un-crippled" attack with that setting for the sidech delay, so it's good to know especially when using fast time settings
I tested this with a sudden level increase on a hard knee, max ratio, minimum time (attack, release (rel atapt too), RMS)
The recorded results present an "un-crippled" attack with that setting for the sidech delay, so it's good to know especially when using fast time settings
-
OneRoomStudios OneRoomStudios https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=94231
- KVRer
- 26 posts since 13 Jan, 2006
It should be FB, not FF - like I said, it operates as a Feedback design.
The RMS window should be 0 - the SSL 4000 G bus compressor acts as a peak detector.
The side chain delay might sound better with a negative value (and probably does - that's why a lot of plugin compressors use "look-ahead"), but that's not how analog hardware works. The ugliness you might hear on the attack will happen in the analog realm as well - that's what happens when transients fall victim to phase-shift and control delay. If you're going for clean, emulating analog is the wrong way to go.
The RMS window should be 0 - the SSL 4000 G bus compressor acts as a peak detector.
The side chain delay might sound better with a negative value (and probably does - that's why a lot of plugin compressors use "look-ahead"), but that's not how analog hardware works. The ugliness you might hear on the attack will happen in the analog realm as well - that's what happens when transients fall victim to phase-shift and control delay. If you're going for clean, emulating analog is the wrong way to go.
-
- KVRAF
- Topic Starter
- 3817 posts since 8 Mar, 2006
FB detection for the SSL one, right (got a bit messed up with all these emulations, sorry )
...anyway, will upload the patch asap for testing!
about the delay, thanks for the confirmation, I actually haven't used an analog compressor so the pseudo-experience with one is based on audio demos and feedback on the internet.
Was only pointing out the optimal value for a default all-rounder.
After these tests I'm a bit more conservative now with negative values for analog emulations... positive ones seem to be OK and required in getting some shapes...
...anyway, will upload the patch asap for testing!
about the delay, thanks for the confirmation, I actually haven't used an analog compressor so the pseudo-experience with one is based on audio demos and feedback on the internet.
Was only pointing out the optimal value for a default all-rounder.
After these tests I'm a bit more conservative now with negative values for analog emulations... positive ones seem to be OK and required in getting some shapes...
-
OneRoomStudios OneRoomStudios https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=94231
- KVRer
- 26 posts since 13 Jan, 2006
Here's a good starting point for the SSL 4000 G bus compressor. This is for a "glue" setting. I'll make a "smash" setting as well.
https://drive.google.com/file/d/0B4-39s ... sp=sharing
https://drive.google.com/file/d/0B4-39s ... sp=sharing
-
OneRoomStudios OneRoomStudios https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=94231
- KVRer
- 26 posts since 13 Jan, 2006
Here's a starting point for the SSL 4000 G "smash" setting. This is an ugly sound - almost like the famous "talkback compressor" used on Phil Collins records. It sounds great if you use the wet/dry mix, or if you like distortion.
https://drive.google.com/file/d/0B4-39s ... sp=sharing
https://drive.google.com/file/d/0B4-39s ... sp=sharing
-
- KVRAF
- Topic Starter
- 3817 posts since 8 Mar, 2006
Before I have a look, here's what I was fine-tuning referring to the NI Solid Bus Comp (seems that it has different release settings than the original; example: instead of 300ms there's a 400ms for the release)
So I modeled mine after: 10 ms attack, 0.4 s release, 4:1
https://app.box.com/s/c0et25r73ovses6ecrb5
edit: whoops, seems I forgot the threshold... no biggie.
So I modeled mine after: 10 ms attack, 0.4 s release, 4:1
https://app.box.com/s/c0et25r73ovses6ecrb5
edit: whoops, seems I forgot the threshold... no biggie.
-
- KVRAF
- Topic Starter
- 3817 posts since 8 Mar, 2006
Your Glue patch seem very cool!
added: yes, your "smash" patch works well blended with the dry/wet level. I suggest you won't take the time values that literally because the envelope shapes are of course different and need a bit of compensation.
For example, instead of 1ms of the attack, try a tiny bit bigger value like 2.5 perhaps... anyway, might be wrong since I'm trying to emulate an emulation...
about my patch:
btw, it seems that the NI Solid Bus comp is a hard knee (almost a perfect one) and the lower I bring the ratio, the threshold seems to be shifting down progressively, -simulating- a soft knee, but it's still a hard one.
Also, the attack time is a bit slower than 10 ms (about 15ms + 0.30 positive sc delay) to compensate for the shape.
I also have half of release adapt since I couldn't get the release shape without it..
added: yes, your "smash" patch works well blended with the dry/wet level. I suggest you won't take the time values that literally because the envelope shapes are of course different and need a bit of compensation.
For example, instead of 1ms of the attack, try a tiny bit bigger value like 2.5 perhaps... anyway, might be wrong since I'm trying to emulate an emulation...
about my patch:
btw, it seems that the NI Solid Bus comp is a hard knee (almost a perfect one) and the lower I bring the ratio, the threshold seems to be shifting down progressively, -simulating- a soft knee, but it's still a hard one.
Also, the attack time is a bit slower than 10 ms (about 15ms + 0.30 positive sc delay) to compensate for the shape.
I also have half of release adapt since I couldn't get the release shape without it..
-
OneRoomStudios OneRoomStudios https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=94231
- KVRer
- 26 posts since 13 Jan, 2006
The NI Solid Bus comp sounds nice, but it doesn't quite sound like a 4000 series quad compressor to me, at least, not in the way that I would usually use one (I have one right in front of me that I built).
So...if the DBX 160 VU and the SSL 4000 G buss compressor are covered, what next? In VCA land you still have the SSL channel compressor, the API 2500, the Empirical Labs EL8 "Distressor," the Neve 33609, or a weirder one like the Aphex 661.
So...if the DBX 160 VU and the SSL 4000 G buss compressor are covered, what next? In VCA land you still have the SSL channel compressor, the API 2500, the Empirical Labs EL8 "Distressor," the Neve 33609, or a weirder one like the Aphex 661.
-
- KVRAF
- Topic Starter
- 3817 posts since 8 Mar, 2006
Yeah, probably a good idea to download a different/better emulation to reference against, although the NI Bus Comp sounds cool, it seems to have a bit of an exaggerated character like the rest NIs maybe?!
Next, a bit problematic since I don't have emulations to reference against (the Waves has an API 2500 emulation but can't demo that)
Has a play some time ago, still remember it and I have some info and video/audio demos on it.... so I guess that's next. Will get started right away. What RMS value does it need?
Next, a bit problematic since I don't have emulations to reference against (the Waves has an API 2500 emulation but can't demo that)
Has a play some time ago, still remember it and I have some info and video/audio demos on it.... so I guess that's next.
Will get started right away. What RMS value does it need?-
- KVRAF
- Topic Starter
- 3817 posts since 8 Mar, 2006
Just had a look at the API 2500 manual.... seems that if we want a -loud- "thrust", We set the saturation warmth to +15db tilt (aka maximum)
-
OneRoomStudios OneRoomStudios https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=94231
- KVRer
- 26 posts since 13 Jan, 2006
Inspired by your mention of the "Thrust" circuit, I made a "New York-style" parallel-compression preset based on the API 2500. If you follow the directions in the preset description, this setting will do terrible (in a good way) things to your drum bus.
Tomorrow I'll do a much gentler "mix bus" setting based on the API 2500.
https://drive.google.com/file/d/0B4-39s ... sp=sharing
Tomorrow I'll do a much gentler "mix bus" setting based on the API 2500.
https://drive.google.com/file/d/0B4-39s ... sp=sharing
-
- KVRAF
- Topic Starter
- 3817 posts since 8 Mar, 2006
RMS size for the API 2500 seems to be about 10ms according to the sudden volume increase test while attack and release are at minimum. (0.1ms : 50ms)
added: also, when thrust is engaged, there seems to be a sidechain HP filter cut (did a sinesweep test and the GR dipped accordingly) ... set in Presswerk to around 200 Hz on 6 db/oct slope.
Just checked out your patch, cool "woolly-dark" saturation you got there....
Even if in hard knee mode there seems to be a tiny bit of soft knee.
medium knee seems around 3 db
I had a play with the Waves emulation for comparison.
added: also, when thrust is engaged, there seems to be a sidechain HP filter cut (did a sinesweep test and the GR dipped accordingly) ... set in Presswerk to around 200 Hz on 6 db/oct slope.
Just checked out your patch, cool "woolly-dark" saturation you got there....
Even if in hard knee mode there seems to be a tiny bit of soft knee.
medium knee seems around 3 db
I had a play with the Waves emulation for comparison.
-
- KVRAF
- Topic Starter
- 3817 posts since 8 Mar, 2006
here's my API 2600: https://app.box.com/s/ilfmkjbppjuhjx13r7li
As always, needs more fine-tuning. (getting sick of saying that after hours of fine-tuning a patch!
)
Brought RMS value down to 1.0 ms to keep the feed-forward aggression and fast release... the attack seems to require more time to fit the shape... for example, for 1ms on the API, about 10ms on the Presswerk.. something like that.
This is one "organic" compressor... always seem to change loads of stuff while changing a parameter, the character has a significant shift if changing the ratio for example...
I'm thinking to make an "old" and a "new" patch one using FB and one using FF since there's loads of juice to extract from this one...
As always, needs more fine-tuning. (getting sick of saying that after hours of fine-tuning a patch!
)Brought RMS value down to 1.0 ms to keep the feed-forward aggression and fast release... the attack seems to require more time to fit the shape... for example, for 1ms on the API, about 10ms on the Presswerk.. something like that.
This is one "organic" compressor... always seem to change loads of stuff while changing a parameter, the character has a significant shift if changing the ratio for example...
I'm thinking to make an "old" and a "new" patch one using FB and one using FF since there's loads of juice to extract from this one...