VCA Compressor Gain Reduction Limit

[JR.]

Without steering you toward a specific circuit it's pretty common in side chains to establish thresholds by adding a current to or subtracting a current from an opamp inverting input then comparing that current to a second varying one with the opposite polarity. A steering diode from the output of the opamp can clamp it to prevent response above or below the threshold depending upon what you want.

I have used a handful of opamps each doing something different in a side chain but working together masage the CV as needed. In the log domain it's is simple to scale or offset all kinds of things.

JR

[mediatechnology]

One suggestion when operating in the side chain is to scale the voltages up so you are not trying to parse single digit mVs

So true. If you do this, remember that the Ec driving the VCA must be from a low impedance source. If you expand the dB/V relationship be sure to attenuate it at the VCA actively. I see a lot of designs where this is done resistively and the source impedance driving Ec is too high resulting in the input signal modulating Ec. This raises THD.

Most of the THAT app notes operate at higher dB/V scalings in the threshold and ratio amps and attenuate in the ratio control voltage divider and final Ec buffer. The scalings are typically 6X to 10X the native 6.1 mV/dB.

Tom - I see no reason why you couldn't do this in a GSSL at the same location.

[emrr]

Doug: Are there many (any?) vintage pieces that aren't feed-back topology? Without the gain cell having linear dB control, or even "linear" linear it seems like feed-forward topology would have trouble with gain control ballistics.

Another Doug question: Have you ever looked at the Texar Audio Prisim aka Gentner Audio Prisim? It had a unique gated AGC. (Another cool thing it had was a Cds cell in the shunt of the input U-pad. That wasn't the working attenuator they just put it there for protection. It was impossible to overload.)

I think you've mentioned the Audio Prism to me before. I can't find any real info out there. Love to see a working link to the manual.

I'm not spotting any definite feed-forwards, which raises the question: what was first? Everything seems to be variations on feedback control. I have one of those Poor Man's 660 amps in progress, and it was noted that one could connect it feed-forward; no one has tried it or warned of any issues to date. I'll have to see what happens.

How would you define the LA-2A? It's a opto cell driven by what I would think of as a feed-forward side chain amp. But maybe I'm looking at that incorrectly. I suppose the opto shunt effect may also drag down the input signal hitting the side chain, giving feedback effects.

I was also thinking of the Langevin AM-5301 Leveline compressor, which looks like a standard PP vari-mu type with low ratios, but has two steps in the curve. I've heard mention of a bump in the bias curve of the 6ES8 tube, and maybe that is the cause, or maybe it is an issue of steam power.

Kinda wrong in memory on the Level Devil. The short description is:

Below the expander threshold it is a linear amp, at expander threshold the gain increases 10 dB, above this level it acts like a conventional peak limiter. At a no signal condition it doesn't return to full gain, but 10 dB less. The truly weirdo thing about it is it uses 12AX7 for rectifier in the limiter section, and 12AT7 as rectifier in the expander section.

There's another weirdo Gates limiter, made from 1938 to 1947, that uses a tiny bit of feedback vari-mu control on PP 6J7's, followed by a variable bridge shunt consisting of two light bulbs and two resistors. This bridge exists between two separate transformer coupled amps, and the side chain is pulled from a separate winding on the 1st amps output transformer. I do not know if the shunt action triggers at such a point as to drag down the output of the 1st amp and side chain input, but I suspect it may. You can pull the rectifier and find not more than about 6 dB control from the feedback path. I suppose that sort of passive shunt exists outside of either classification.

[JR.]

I wouldn't expect too many feed-forward designs before decent VCAs with log control.

I have tricked up log control for OTAs but they were so noisy why bother?

Feedback comps can use any old gain law and the feedback keeps them OK.

JR

[mediatechnology]

I think you've mentioned the Audio Prism to me before. I can't find any real info out there. Love to see a working link to the manual.

I'm still looking myself. I only have partials.

I'm not spotting any definite feed-forwards, which raises the question: what was first?

I'm wondering that too. Could it have been dbx or someone else using a truly linear VCA gain element?

How would you define the LA-2A? It's a opto cell driven by what I would think of as a feed-forward side chain amp. But maybe I'm looking at that incorrectly. I suppose the opto shunt effect may also drag down the input signal hitting the side chain, giving feedback effects.

Made me look. It's definitely feedback by virtue of the sidechain's input from the right side of R6. The sidechain input is post-attenuator.



Move it to the left side of R6 and it's feed-forward. But it wouldn't be an LA-2 and might not work well.

Maybe some of your other examples are. Now I'm curious who was first.

[mediatechnology]

OK JR. That's what I was thinking too. Decent gain law conformity by virtue of VCA. It seems that to get a FET, CdS cell or tube to work well in feed-forward would require a lot of linerization.

[emrr]

Yeah, I see even in limit with R7 active it's close enough to zero Z to not count, putting SC input at point of reduction.

[TRW]

Tom - I see no reason why you couldn't do this in a GSSL at the same location.

Excellent, will play and get back to you guys. Simulator first I think.

Good suggestions from John too. Thanks.

Interesting little tidbits of compressor history too. Thanks Doug and Wayne.

Personally most of the compressors I love are feedback. But the Pico seems like an interesting tool indeed.

-Tom

[Script]

Hello,

I have been reading this thread with a lot of interest and wondered whether anyone has tested this clamp successfully yet. I tried it briefly on a test board but all it produced was heavy distortion -- probably due to some mistake of mine. I should try again...

But I wonder whether the schematic below would achieve a similar result. Or does that mess with impedances? Anyway, I tested that one on a comp with extremely soft-knee characteristics , but -- at first listen -- all it seems to do is act like a second ratio control. When settings are compared, I do not hear much of a difference between an extremely high ratio setting plus clipper in the sidechain as opposed to a lower ratio setting without the clipper in the sidechain (and meter movements too look the same, relatively that is). I used a 10K pot for Rl to ground, but did not combine V(thr)+V(ref). Maybe that's why? Or is it because the clampig works better on hard-knee, or because I should try harder to understand the Jung schematic , or probably clean my ears better before intensive listening tests


Sorry, can't make the image bigger [http://www.circuitstoday.com/diode-clippers]

Regards,

[JR.]

I'm not sure I understand your question, many of us here have successfully used diode clamps and diode break amps for various functions.

Since this is a long and wandering thread I am not sure what exactly you are trying to accomplish,

I diode clamp in the side chain of a limiter between the calculation circuitry and the VCA has the effect of stopping further gain reduction beyond a finite amount, so the transfer function returns to 1:1 beyond that.


JR