Help needed understanding SSL Mixbus RATIO circuit

Hi All,
I have been studying the SSL Mixbus compressor sidechain ratio switch circuitry, and I just haven't been able to figure out the operation of the resistor network and diode used for the 1:2 ratio setting.

Can somebody here explain to me what the network, especially the diode, is doing in the circuit?

It's SW1 on this except from Gyraf's GSSL schematic (below):
< img src <http://216.77.188.54/coDataImages/p/Groups/207/207065/folders/247331/2475423sslratiocircuit.jpg" width="470" height="204"> /img src>

(Please forgive my ieptitude with attaching graphics to posts) :

And does the 1:2 ratio diode and resistor network do anything when the ratio switch is set to 1:4 or 1:10?

Thanks,
mr coffee

mr coffee said:

Hi All,
I have been studying the SSL Mixbus compressor sidechain ratio switch circuitry, and I just haven't been able to figure out the operation of the resistor network and diode used for the 1:2 ratio setting.

Click to expand...

This network provides some kind of kink in the transfer function of the sidechain. Seems like the ratio starts with a higher initial value and then gets reduced somewhat. Check attached graph. In cyan is the amplitude of signal, in red the resulting control voltage. For comparison, I have added a CV that follows the same overall slope but without the kink (green).
I must admit I cannot really explain why the designer has chosen to do so; I guess only him can answer that.

And does the 1:2 ratio diode and resistor network do anything when the ratio switch is set to 1:4 or 1:10?

Click to expand...

No.

Pls find attached the CV graphs for 2:1, 4:1 & 8:1

Its interesting that there are 2 rectifying opamps, one producing a positive output and the other negative. Its like the ratios switch balances the output of one against the output of the other... its a bit of a mind bender!

...oh and notice in Abbey's second plot the onset of compression occurs at different thresholds, just as the SSL Bus Comp is supposed to. (I understand this was supposed to mean you could switch ratios and still maintain a consistent level of gain reduction.) Those SSL chaps are a cunning bunch!

Abbey,
Thanks very much for the graphs. I get the idea it's some kind of soft knee circuit perhaps?

I think I'd understand what was going on better if the 620K resistor to + 12 volts wasn't there. Seems like it's out of the circuit because of the diode polarity except when the output amplitude is  high enough to exceed the diode Vf turn-on voltage. Still confusing to me.

@MikeClev
I think the two op amps are part of a full-wave rectifier producing a negative output voltage, and that the ratio switch changes 1) the quiescent bias (which is clamped 1/2 volt positive at the output of the 2nd op amp) creating a threshold for onset of compression, and 2) changes  the gain of the FWR being fed to both the sidechain VCA and the signal VCAs.

BTW - Abbey - Really Like your sig. Great quote.

mr coffee

I know this is an old topic, but SOOOO interesting and Not understood. I've been wanting (and searching) for and answer to this for ever...

Please, a fresh look???

How does the ratio circuit work? How to change the ratios? How to add ratios? How to change the knee?

These have all been asked and have largely gone unanswered.

For what I see, the threshold change when ratio change, it's not the only compressor which do that, the 1176 do it too, in the switch you have ratio and bias (bias work as threshold I understaand)

JS

So is the threshold being changed? If so, how? I have also heard the knee changes... how is that happening? If someone wanted to add or change the ratio which resistor would change?

I don't know, just guessing from the plots... but a sim will showit, the next week maybe I'll do it...

JS

MikeClev said:

Its interesting that there are 2 rectifying opamps, one producing a positive output and the other negative. Its like the ratios switch balances the output of one against the output of the other... its a bit of a mind bender!

Click to expand...

No. It's a textbook absolute value rectifier that operates on a quite simple concept. The 1st opamp produces a half-wave positive rectification, which is turned into a current via the 10k res. It is summed in the second opamp with the input current, with a factor of 1/2. The resulting combined current is then transformed into a voltage by the 2nd opamp.

jwhmca said:

So is the threshold being changed? If so, how?

Click to expand...

This is easy to answer. The 2nd opamp recives a DC bias coming via the "A" connection from the switch that goes to 3 different resistors to -12v. This creates 3 differents negative input currents, against which the positive output current of the rectifier has to fight to start producing an effective control voltage. That's a quite common technique in compressors, where the rectifier fights against a reference to trigger the control voltage; here it is done in current mode, in a dbx compressor it is done in voltage mode, but it's essentially the same principle.

I have also heard the knee changes... how is that happening?

Click to expand...

It has to be analysed in the broader context of how CV controls gain. In the SSL comp, it is not a linear dB/V process, so the transfer curve is not decilinear, meaning the more above threshold, the higher the compression ratio. That's quite common on non-VCA based compressors. Optos and FETs do that naturally. They generally are included in a FB configuration that also changes the CV-gain control law.
In the case of the SSL comp, the control port of the VCA is the typical decilinear dbx/THAT type (i.e. a constant dB/V constant - about 6mV per dB), which is intended to receive a decilinear control voltage from the rms detector, but here, they use linear detection, which results in an increasing ratio. The consequence is that ratio can be changed by changing the level; however, doing that also changes the threshold (as seen by the user). There's total interaction between ratio and threshold, just like an 1176; in order to make that transparent to the user, changing the ratio also changes the gain-cell threshold.

If someone wanted to add or change the ratio which resistor would change?

Click to expand...

You would need to change one of the resistors that go to -12V. You may probably find that it would be relatively easy to find values that would span the 1:4 to 1:20 range, but below 1:4, you may have to use the diode addition that some obscure engineer at SSL has lost hair in designing. I bet a lot of experimentation and fiddling went on at the time.

I'm making the sim now... all nubers aprox. I dont know if it is wright... *I forgot to load the output diode at the sim, it affects at threshold at least and dont know at the rest of the sim...

With max ratio you have -15V output (max reduction) with 0.5V input, (30V/V)
with mid ratio you have -15V output with 1V input, (15V/V)
with min ratio you have -15V output with 5V input. (3V/V)

Without the diode you would have linear response (like 2 others) but higher ratio, also with a short instead the diode, but this time with a ratio near the original.

The effectof the diode is something like a higher ratio from 0 to 0.25V input giving 0 to -2.5V (10V/V) output, then you have 5V output at 1.15V (4.3V/V) and for a 3V input 10V output (3.3V/V)

So, the ratio isn't linear and decreses with amplitude, the first (0V to 0.25V input) the highest part, then falls quikly (at 0.25V input) and then slow keeps falling the rest of the range.

Hopefully it's usefull for you JW, what is the exact effect of all this in the audio, the knee and threshold I still don't know, but a deeper sim and thinking should help. For that I need more time I don't have now, but next week probably will... I will post future results over here, it's an intresting thing to analise this type of things. Seeing it quikly seems to be a "harder" knee, cause at first it compress more and then turns less compressing in low ratio, the higher ratios are quite linear.

(the diode load seems to made a low ratio at the first part of this, then higher and then lower again... maybe there is the soft knee, tell more later, running out of battery)

Off topic, but is SSL buscomp FF or FB... I mean, the signal to sidechain is befor VCA but have a FB with the sidechain VCA.

Regards.

JS

Thanks guys! This is really interesting.

joaquins said:

Off topic, but is SSL buscomp FF or FB... I mean, the signal to sidechain is befor VCA but have a FB with the sidechain VCA.

Click to expand...

It is definitely FB, but with a particular structure where one of the VCA's carries "un-noble" audio but makes sure that the resulting CV is convenient for driving the ones that carry the "noble" audio.

Can you see any reason the resistors would interact with each other when making ratios and such. What I mean is: It seems like when I change a resistor value it makes the other ratios/levels change...

jwhmca said:

Can you see any reason the resistors would interact with each other when making ratios and such. What I mean is: It seems like when I change a resistor value it makes the other ratios/levels change...

Click to expand...

Yes. The 610k, 1M8, 270k, 3M9, 68k, 620k and the diode are always connected more or less directly. Only the 1M2 is not "interactive" because it is connected only in the last position of the switch.

yeah, I can see that and thats what i thought... so, really, in my quest to add a 1.5 ratio I need to do a "balancing act" to get everything the same after the included ratio...(?)

Also, it seems from what you were saying before, I would need both the 2 and 1.5 ratios to have diodes? Or is it just the last ratio in line?

you can add another network between +12 and -12 without touching the other networ so it won't interact with original values... You can put the first part with 620K + 68K and put another higher ratio "soft knee" and mid ratio hard knee for example...

But in any case I dont see much worth, 1:2,4,10 seems enough, you won't get much higher than 10 and is dificult to go lower than 2... maybe you could want something like 1:4 soft and 1:6, maybe, don't know...

If you get 1.5 show results, there you should use soft knee...

JS

I've tried in so many ways to get this to work! ughh.

I must be missing something. It seems no matter what I try the bottom ratio gets stuck at 2:1. Could someone show me how this might work? If I see a working example it will all pop into my head...

Here is what I have tried so far in adding a 1.5 ratio.

When I do this, and testing the way SSLtech recommended... Setting a gain reduction then increases input etc.

I get about 2:1

And really no matter how I change the resistors around it seems that when I reset everything it goes right back to 2:1 (ish)