Stereo Compressor Link

[JohnRoberts]

I believe Ted Fletcher used another approach in some of his units - he converted the stereo signal to MS before the opto compressors and back again afterwards

Microprocessor control of the sidechain is used in the RNC and others

Nick Froome

A microprocessor could measure and manage the gain law of a LDR very nicely but far from what anyone would ever call elegant. While I will never say everything that could be done has been tried already, LDRs are a mature technology that have been well explored.  I kicked the tires decades ago and moved on, while Peavey used them in a tube comp/lim to target a specific niche market.

I was involved in one digital controlled analog path for a mostly analog company that wanted to preserve their analog path brand identity but that project was abandoned as their analog market was eroded by all digital products.

JR

PS: One could use a a HF chopper to alternate a single LDR between  a DC resistance measure and control loop and an audio path. This too would introduce new problems to solve, and make Rube Goldberg smile while being more analog (if that matters).. 

 

[ruffrecords]

I have had a thought about overcoming the opto variation. If we use a dual opto,(like the 5C3/2 where one led drives a single opto resistor that is split into two equal regions, there's a good chance the two halves will match quite closely. So if we use a pair of these in each channel (one for +ve and one for -ve signal cycles) we can use both halves in each channel in dual mono mode. When linked we can swap one opto resistor from each of the 4 opto  to the opposite channel so the left and right each have equal numbers of matched opto resistors. Switching is more complex but tracking should be much better. The attached schematic shows what I mean.

Cheers

Ian

 

[joaquins]

Ian, first is true that dual cells track better, at least that's what specs says.

Second, I know you are all tubes and no SS, but as you are already using LEDs but adding a bridge rectifier to each side, LED at ± ports and feed the AC to the AC ports. Then you only need one dual cell on each side, and as bridge rectifiers or 4 small diodes are much cheaper than vactrol this may help you in doing this more poor men friendly.

JS

 

[ruffrecords]

Ian, first is true that dual cells track better, at least that's what specs says.

Second, I know you are all tubes and no SS, but as you are already using LEDs but adding a bridge rectifier to each side, LED at ± ports and feed the AC to the AC ports. Then you only need one dual cell on each side, and as bridge rectifiers or 4 small diodes are much cheaper than vactrol this may help you in doing this more poor men friendly.

JS

Yes, that is an interesting idea and I think I had a schematic for it on the poor man's tube compressor thread. The only problem with the bridge rectifier is it raises the threshold by another 1.3 volts (two diode drops) so it no becomes 1.5 (LED) + 1.3(bridge) = 2.8volts. To pass 2mA through the usual 2K series resistor requires another 4V making a total of 6.8 volts or nearly +16dBu. OK, maybe not so bad as I thought since the current circuit requires +14dBu to reach 2mA. I guess we could always use germanium diodes if necessary. As you say it is cheaper and 4 x 1N4148 do not cost much.

Thanks for the input.

Cheers

Ian

 

[joaquins]

Tubes are pretty good to use as current controlled, rather than voltage, why to use the 2k resistor? The isolation voltage appears to be 2500VRMS for VTL5C2/2 and 5C3/2, so I don't see a problem using them over the anode, could be used at the cathode but I don't know. You could use a smaller resistor at the cathode to improve stability and avoid or reduce trimming needs between different tubes.

JS

 

[ruffrecords]

Tubes are pretty good to use as current controlled, rather than voltage, why to use the 2k resistor? The isolation voltage appears to be 2500VRMS for VTL5C2/2 and 5C3/2, so I don't see a problem using them over the anode, could be used at the cathode but I don't know. You could use a smaller resistor at the cathode to improve stability and avoid or reduce trimming needs between different tubes.

JS

I have looked at this in some detail. My very first design was based on this technique. It is basically a dc amplifier. The big problem is setting and maintaining the quiescent dc conditions in  the tube. Basically the problem revolves around the fact that you need the tube to be very close to cut off when there is no signal and then reduce the bias to allow it to conduct when there is signal. It turns out that the cut off point varies a lot between tube samples. After a lot of experimentation I decided on an SRPP stage working in an 'H' bridge with a couple of resistors across the HT supply and with the LED connected as the horizontal arm of the bridge. The SRPP always produces a centre voltage close to half the supply voltage so it is easy to trim out with a variable resistor in the other arm of the bridge. This approach is more appropriate for a top of the range compressor where the attack and decay are determined by a separate rectifier and a fast opto is then driven by the resultant dc control voltage.

Cheers

Ian

 

[ruffrecords]

Here's the cheaper version of the split opto idea incorporating a bridge rectifier to drive a single LED per channel as suggested by joaquins:

Cheers

Ian

 

[PRR]

> basically a dc amplifier
> you need the tube to be very close to cut off when there is no signal

That's obviously not a happy path.

I think you want the tube to be an AC amplifier.

With a final like the attached scribble, the V/I ratio is pretty well defined, the output Z is much higher than an LED, and LEDs may be stacked to reasonable numbers.

You do lose the voltage-threshold of an LED. But note that the LA2a has no real threshold in the electro-optical chain. The threshold is the lamp current which drives the LDR down to the 33K(?) resistance of the R-LDR L-pad.

However--- any really-simple stereo solution WILL do a wrong thing on some unlikely stereo signal. All-in-Left, or L/R totally out-o-phase, or something.

 

[PRR]

And HEY! Why is a tube-man using LEDs? They are even younger than transistors, about as old as chips. Way too new for you.

The EL Panel in the LA2 is 1960s, less-too-new than LEDs.

There's lots of older ways to make light. Neon-bulbs are classic and used to be cheap (though LEDs have driven them off the mass market and the residual market neons may be junk). There's also hoards of old Russian bar-graph neons which are plenty consistent for your project. Neons have a definite (if large) threshold (but also huge hysteresis).

Magic-Eye (phosphor) and Nixie (neon) tubes give electric controlled glow.

 

[L´Andratté]

Thanks for the rundown on 47k!

Regarding dual vactrol- tracking, there´s a lot of evidence that they still have less than excellent tracking, because the variation is in the LDRs not the LED. And the matching of two dual units is bad as with singles. There is (at least some) well informed talk to be found here: http://search.retrosynth.com/synth-diy/.

There is hand matching: expensive and bothersome, but possible.
There are fancy servo circuits: complicated and IMO the beauty of vactrol sound IS their eccentric response curve (to be matched), else a vca could be used... As well as their zen-like simplicity of use.
And there is the mid/side processing, which is easy, expensive transformer based or cheapo solid state. Sounds interesting, could work really well with your actual idea.

OTOH I used a lot of 5C3/2 for voltage controlled filters and never had grave tracking problems, which would have resulted in way off q factor. So it can only be found out by experiment, I guess!

And HEY! Why is a tube-man using LEDs? They are even younger than transistors, about as old as chips. Way too new for you.

Why not get an assistant to work the faders? Real vintage! And maybe cheaper...

 

[joaquins]

I've found more than decent match between two discrete LDRs, even impressive in a first shot once. In any case, how much compression do you want for stereo tracking? maybe matching them with an external resistor and using in a wind of narrow range would give decent tracking I think, maybe adding a resistor in parallel to match them at some point (3dB reduction?) and trust them for the rest of the day, since you will be at the ball park most of  the time. The nice thing of dual cells is the temperature tracking which may be a problem in some applications, since the tempco is really bad. The other thing is that maybe, just maybe, would be just a really good dual mono compressor, not so good for stereo, I can live with it, I don't need every single compressor I have to have a matched pair for stereo, this will probably be a really good option for vox tracks, I'm fine with mono there (I've used some stereo VOX tracks but probably wouldn't match a tube opto comp there for what I've used it.) I'm not saying to give up, just I wouldn't break my head long on this, I would make my best without braking my head, make it usable as I said for example trimming for a given point and see how it works, if it's good enough for some uses I would be happy, I don't know if I would point to a mastering opto comp with this approach where really low shifting between channels would be a problem, for a guitar bus instead I probably can live with a bit of shifting.

The other thing I don't know yet is about timing, I don't know if the dual cells are as slow as we are used in LA2A.

JS

 

[L´Andratté]

The dual ones I tried personally are 5c3/2 (the go-to cell in synth world) and 5c4/2. I found the former to have a more pronounced attack and "precise"(sounding) response, the latter more slewed or blurry (which could be cool for soft compression).
For (mono) compression I liked most the 5c1 (overall fast, maybe too fast without RC network) and 5c10 (fast attack, slooow decay) but these are not available as duals. I can´t comment on the 5c2/2.
It´s all a matter of taste as I don´t bother with measurements and stuff (as I´m too stupid). One could rather use a prototype with sockets for testing the optos by ear (I made a unit with switchable cells ).

Oh no, yet more facts and fiction on vactrols have just been added to the pitiful interweb...  :-X

 

[pvision]

I think if the project is a "stereo compressor" then the tracking of the two channels is an issue.  If the compressor will be used as a buss compressor then it is definitely an issue

Bearing in mind that Ian is a valve man, and valves imply transformers, perhaps the mid-side approach might be a good alternative if the proposed solutions don't track sufficiently well?

Nick Froome

 

[joaquins]

I think if the project is a "stereo compressor" then the tracking of the two channels is an issue.  If the compressor will be used as a buss compressor then it is definitely an issue

Bearing in mind that Ian is a valve man, and valves imply transformers, perhaps the mid-side approach might be a good alternative if the proposed solutions don't track sufficiently well?

Nick Froome

I like that idea, still not good if you are looking for stable image, going wither and narrower if there is something going on, is less problematic than going from side to side.

JS

 

[ruffrecords]

> basically a dc amplifier
> you need the tube to be very close to cut off when there is no signal

That's obviously not a happy path.

I think you want the tube to be an AC amplifier.

If you are happy for the compressor characteristics to be entirely defined by the opto characteristics then an ac amplifier is fine and this is exactly how things are at the moment.  I currently use and SRPP stage (because I have it to hand) but your simple CC stage is likely just as good and uses only half a tube.

If you want some control over attack and decay, for example, you need to include some form of detector to define these and follow it with a dc amp drive of the LED(s). Singled ended dc amps I have found to be problematic for the reasons stated earlier but an SRPP in a half bridge seems to work well.

Cheers

Ian