SSL Turbo Mod added - Blind test - Sound files - have fun :)

[SSLtech]

Keith, it's all true what you said.

The only MOD I am missing now is how to match the levels when switching between them.
I really hate this jumps in volume/GR levels when switching. Has anyone thought of any way
to make it balanced/matched?

You have to appreciate that what you're asking for cannot be done...

Try this to illustrate it:

Patch signal through ONE CHANNEL ONLY. Now, no matter HOW you set it, it will ALWAYS be matched.

Now parch a mono signal into BOTH channels. -Now the Turbo will ALWAYS compress less hard than the GSSL.

Now patch an opposing-polarity signal into both channels. Now the Turbo will always compress INFINITELY more than the GSSL.

So it depends on the program material and their 'width'. -The above tests are specifically chosen to illustrate that.

It also hopefully suggests how the program material can influence why one might sound "better" than another, depending on the material's stereo content.

Keith

 

[Purusha]

Keith, I was thinking in a term to MOD the threshold pot when switched to TURBO for example...
... as from my point of view, but I am still more or less a beginner in DIY  , I see that I need
to crank the threshold much more in the Gyraf version to get same GR as in TURBO version...
... so why not adding a certain resistor value to the THR. pot in series in one of these versions
to get closer in GR values?

 

[radiance]

:

 

[SSLtech]

Again... Do the test and you'll see that what you're asking for won't work.

If it reduces the sensitivity in GSSL by 6dB then things will match when you feed absolute MONO into both inputs. -But then stereo will be off. And single channel only will no longer match, as they do now.

Basically what you're asking for CANNOT be done automatically. It is program-material dependent.

Because the program material is different, the AMOUNT will be different. -Honestly, DO what I said in terms of a test, and you'll see exactly what I'm talking about, live and in real-time.

Keith

 

[radiance]

Maybe two separate threshold controls, one for each mode.....

 

[Purusha]

Maybe two separate threshold controls, one for each mode.....

Exactly what I had in mind. 

 

[clintrubber]

Maybe two separate threshold controls, one for each mode.....

As I understand it this still won't give you what you want, see the Keith-underlined stuff above (& quoted again here below):

Basically what you're asking for CANNOT be done automatically. It is program-material dependent.

 

[radiance]

Well, you can adjust both threshold controls acording to the material....THEN you can switch between modes without to much compression difference making a comparison possible.

This is indeed NOT automaticlly...

 

[SSLtech]

Exactly why I wrote 'automatically'.

I see that I need to crank the threshold much more in the Gyraf version to get same GR as in TURBO version...

Perform the tests which I described previously, and you'll see why this is not necessarily true.

Two threshold pots? -No problem. -Is that more of an idiosyncratic nuisance keeping these two thresholds 'tweaked' every time the program material changes (solo has less mono content etc)... almost certainly.

I thought about the two threshold controls as a solution... but I dismissed it, since I don't see it as being any MORE convenient than simply nudging the threshold control a little each time you switch... specially when the 'difference' is likely to change with the instrumentation.

The problem with 'matching' the GR between the two is that they're both compressing DIFFERENT things. The GSSL compresses a mono mix of the two channels. The Turbo compresses two channels independently... Make that match when the balance changes? -You can't really do it. -The whole POINT is that it works differently.

Keith

 

[clintrubber]

Sound like we've found yet another board/gizmo/feature that could be added to those cases:

the M-S matrix  (from L-R to M-S & back again) Anybody properly done that yet for the GSSL ?


BTW, I expect that in say a few hundred years from now, one of the future generations will dig us up
and that they'll immediately understand the relation between the lack of a healthy amount of good music
from these years AND the overwhelming amount of overfeatured gear  8) 

Bye,

  Peter

 

[livingnote]

Hey Peter, did you get that mail I sent you? I'm working on it...

 

[clintrubber]

Hi Lukas,

Backlog of mail here, not collected it all of the past days yet. M-S already ? Oops & nice, perhaps something like
confused minds think alike ? (OK, I sure realize this goes back to at least the Fairchild)

Cheers,

  Peter

 

[livingnote]

Ah yeah, no rush - I'm currently working on a 5532 balanced I/O with CnB functionality and cross-coupled output, as well as FnM M/S matrix, all in one thing. Methinks CnB belongs in the bal/debal circuit to begin with.

And while doing so, I'm upgrading the internal CnB to Rev3, which will have the FnM option on-board.

 

[SSLtech]

Sound like we've found yet another board/gizmo/feature that could be added to those cases:

the M-S matrix  (from L-R to M-S & back again) Anybody properly done that yet for the GSSL ?

Peter, I think I discussed this a while ago with someone, and the thing is that -in both turbo and GSSL modes- it's probably a significantly less useful idea, if you give it some thought.

For one thing, the GSSL will simply sum both channels to one... M/S is basically L+R on one channel, and L-R on the other channel. The GSSL sidechain mode sums both together and compresses that. Therefore, Since L+R+L+R = 2L, only signals on the LEFT channel will be compressed, the right channel signals will pass through completely uncompressed.

So that rules out any real usefulness for M/S with the GSSL... but while the Turbo version will work rather more 'typically', I still contend that you shouldn't M/S compress with a stereo compressor.

What you CAN do with M/S is two use two COMPLETELY DIFFERENT types of compressor: one for the 'mono' part of it (*which is basically what the GSSL does anyway*) and a second one for the 'width'. -What this will do (and you need to be wary of this) is to artificially widen any narrow parts, and 'narrow' any LOUD parts to either side.

Using a DIY SSL clone in 'Turbo' mode through an M/S matrix will squash the signal when the mono-sum gets loud, but ALSO when anything loud gets panned AWAY from center... Not necessarily greatly different from the GSSL mode in practice for most of the time, since anything panned away from the HARD edge will ALWAYS sum louder in L+R than it will in L-R, and thus the L+R channel will dominate the compression. Only (usually random, occasional) opposing-polarity stereo acoustic clues will sum higher in L-R than L+R, and these make a phase meter go completely batshit bezerk, and you'd usually notice even without a phase meter that something is 'wacked' if these were louder than the main center signal...

It might be a useful effect for certain applications, but myself, I find that M/S works brilliantly well for EQ and similar processing... though the whole POINT of M/S is to process the two channels DIFFERENTLY... otherwise, there's NO difference compared to just running it through a non-M/S piece of gear.

-Been there, studied that, found that it didn't work. -Remember that on a Fairchild 670 there were independant threshold controls for EACH channel... VERY important that you don't overlook that.

Keith

 

[clintrubber]

Hi Keith,

Sure, certainly, it'd need two sets of controls (at least the threshold) to make sense for the Turbo-mode.
And no claims or anything silly implied with my post, despite it's a bit less common, M-S-processing of this'n'that is of course nothing new.

Best regards,

  Peter

 

[Purusha]

I uploaded new samples HERE.

It was unprocessed material from my friend. Great musician and singer BTW.

 

[livingnote]

Hey Keith, just a thought, but there's something I have floating around my noggin here - you mentioned that the turbo compresses two channels independently - what would happen if you extend the turbo's reach beyond point C, i.e. you would double the timing, threshhold and makeup elements rather than summing CV's post-detector, getting a true independence of VCA activity rather than either channel compressing both? I guess that would make it something of a stereo-linked dual mono compressor, but would this be to any sonic advantage on certain program materials? Do-able with the Lorlins for certain, and with my LED blitz would necessitate some serious relay activity, but now that's got me wondering. Are there compressors out there that operate like this to advantage or would it just throw things off kilter?

 

[SSLtech]

Hmmmm... I'm not sure I understand the question.

As it is, it operates as a dual mono compressor, but one which shares the DC parts (time constants DC voltage, threshold DC voltage, makeup DC voltage, Meter DC voltage). -Could you make it a true dual-mono switchable to stereo, -like a Fairchild 670 or sim'lar? -Yes of course.

This in fact is precisely what the Alan Smart version does. When you switch it to mono, it uses one control for the threshold and sends the DC voltage to BOTH detector VCAs. It uses one makeup control and sends the same DC voltage to both signal VCAs. It uses one set of time constant controls, and sends the same DC voltage to both signal VCAs... but it operates in 'turbo' mode when switched to stereo, just as the original SSL design does. -It doesn't allow two different sets of threshold, makeup or time constant settings for each channel, but if you REALLY wanted to, you could just defeat some relays and make it do that...

Not sure if that's what you were asking though.

Keith

 

[livingnote]

Well if I understand correctly, what is happening in the turbo edition is that you have two detectors which then sum their CVs into one at point "C", and feed that through the time constant, controlling both signal path VCAs with that single CV coming out of the timing/threshhold/makeup network. Would it be of any benefit to make both entire CV paths separate, so that instead of having one CV controlling both signal path VCAs, having two independent CVs set at the same timing, threshhold and makeup settings controlling the signal path VCAs completely independently? Or would this make a mess of the stereo image because now L and R are compressing with respect to their own transients only?

 

[SSLtech]

Or would this make a mess of the stereo image because now L and R are compressing with respect to their own transients only?

Yes.

Now a loud signal on one side will cause that side to turn down but not the other... thus causing equal-level center-panned signals to "shift" to the other side, then 'slew' back to center at a rate determined by the release control. What you're describing is basically two mono compressors.

Keith