Compressor/limiter for DIY vocoder?

[electrospeaker]

I'm working on a vocoder (the ETI/Powertran 14-channel vocoder from around 1980/81)


...and having experienced how easy it was to distory the mic input of a Roland VP-330 vocoder I was wondering if adding a compressor/limiter to the mic-level inputs would be a useful modification?
I need to rebuild the input amps again anyway, so I might as well get them redesigned if that'll be an improvement. Any opinions/comments on tis? Here's the vocoder block diagram and the existing input amplifier circuitry (same circuit for both speech and excitation).




NOTE: for some reason the images got resized when uploaded. Hopefully their reduced sizes are still readable. Sorry about that.

[L´Andratté]

There´s only one way two find out-

http://www.thatcorp.com/datashts/dn107.pdf
http://users.ece.gatech.edu/mleach/papers/limiter.pdf
http://www.forsselltech.com/media/attachments/SideChain_B.PDF

Nice project btw!

[L´Andratté]

Or the simpler version for not losing focus on vocoding:

http://www.thatcorp.com/datashts/dn125.pdf

[electrospeaker]

Interesting... but confusing as to which kind of compressor/limiter to use etc.
(I have just about no practical experience with compressors/limiters).

To avoid distortion I suppose a "brick wall" limiter would do the trick. If the input level is exceeded I suppose the audio will be quickly lowered to the maximum permitted level, but wouldn't that also result in perceptually big difference in levels?
I suppose circuitry wise I would have to put the compressor/limiter right after the input jacks, and before the pre-amp IC?

Another thing: I'm confused about how the jack socket switches work in the original circuit and how to modify the circuit for adding two additional jack socket inputs.
It would be nice to have line and mic level inputs on the rear panel of the vocoder (for "permanent" inputs that go to a separate audio patchbay unit) but if anything is plugged into either the front-panel line and mic level input jacks then either of the rear-panel jacks should be bypassed. And regardless of rear or front panel connection, one of the two available inputs (line or mic) should be bypassed so that not at any time should it be possible to have audio coming in from any of the 4 jacks at once.

Opinions on how this could be done (now that I'll probably have to redesign the PCB in Eagle anyway, in order to accomodate the compressor/limiter circuitry)?

[LFLab]

You could try the Forssell Sidechain compressor, an optical compressor using a vactrol. Modifications to the original input circuitry can be minimal, just insert it somewhere after the input opamp. Pretty simple circuit too, and well designed.
You can find more information on the groupdiy.com forum (dedicated to studio electronics), better microphone opamps can be found there as well (maybe not needed for the excitation input).
Or maybe one of the PCB's on www.pcbgrinder.com (the guy running that site is also on groupdiy.com).
But then again one knob squeezer (from the THAT4301 application note) is perhaps enough.

[electrospeaker]

The simpler, the better!
Several options to choose from here. From what I understand the THAT-4301 IC might give the best result with the least amount of work/components and complications. Out of the schematics linked here I suppose the very last schematic for the "one knob squeeze" (screenshot below) is the one I should look at?

Would that allow me to limit any signal above a certain threshold without affecting anything below, not changing or affecting the audio signal in any way otherwise? Does the VR1 pot simply adjust the threshold level? I suppose I could just use a trimmer internally for that and calibrate it once and for all instead of on the front panel as a regular compressor. How about R10 "symmetry" -what does it do?

LFLab: I had never read about optical compressors before, but found out that they are more "musical" than other solutions (but then again maybe this isn't really taken advantage anyway in a vocoder). Perhaps the optical ones aren't as simple to build as the IC based one mentioned above? You said the compressor should be inserted AFTER the input op-amp, but isn't the whole point to limit the signal so it won't distort the input amp? I was under the impression that the audio signal should enter the limiter first, then be amplified by the vocoder's input amp.
The vocoder input amp uses a 4558 and a 1458, but I read that the 5532 should be a better replacement for both as they would result in less noise (hiss). So apart from adding the limiter circuitry and replacing those two op-amp ICs, are there other changes I should look into for the vocoder input amp schematic?

---

one_knob_squeeze.png
Description:
Filesize:	108.03 KB
Viewed:	627 Time(s)
This image has been reduced to fit the page. Click on it to enlarge.

---

[LFLab]

Is it overdriving the input opamp already? I thought the compressor kept the signal from overdriving downstream stuff.
If it is the input opamp you are concerned about, the passive gain reduction of the forssell might be the ticket, because you might otherwise overdrive the THAT4301 input. But taming a hot signal can be done with a passive resistor pad, just add some gain somewhere else.
The one knob squeezer is pretty good, expensive IC though, but I believe some people on the freestompboxes forum also did some compressors based on this schematic (for guitar).

[L´Andratté]

I don´t think it´s expensive as it gives you a very good compressor/limiter for a few coins. You can use the one knob data sheet circuit right away, you don´t have to use THAT! transistor arrays, use garden variety transistors, or use the diodes of the earlier example in the data sheet.

I have built my share of optical-, FET- and VCA compressors and they´re all worthwhile, but you can get pretty lost

optical: very nice sound, low noise, but eccentric "coloured" behaviour, maybe not fast enough for reliable limiting, high current draw
http://moosapotamus.net/images/LA-Light%20comp.gif

FET: also nice punchy sound, very fast, but not simple to get to work right without too much distortion, FETs not easily available any more
http://users.ece.gatech.edu/mleach/papers/limiter.pdf

VCA: more neutral sounding, fast, sometimes not even noticeable while working reliably, simple to set up.

Don´t forget you´re building a vocoder