Some basic opto comp questions - LA-2a schematic

[NewYorkDave]

Then you stuff the output into a capacitor (NO DC please!) and then into the LED...

The simple approach will work but will, of course, trigger GR off peaks of one polarity only. This may or may not be a problem in practice--people seem to love those Federal limiters, after all.

I'd use a transformer, centertap grounded, outer legs to rectifiers, output of rectifiers joined together and connected to LED. The tube stage driving it has to be of fairly low source impedance--not so much for the sake of pumping big current to light up the Vactrol, but to avoid low frequency rolloff when working against the primary inductance of the transformer. Again, something like the One-Bottle driving a 4:1CT xfmr would do it, I'd say.

 

[NewYorkDave]

Sure, that would work, but I'm not seeing the advantage of it at the moment. The mark-space ratio of the light pulses would be halved, which as I comprehend it right now (late at night, lacking sleep, not at 100% brainpower) would mean that the two paralleled photocells would not provide any linearity advantage at low frequencies since they have twice the amount of time in decay between light stimuli, and at best would even out compared to a single photocell driven by twice the frequency (as in the fullwave rectified arrangement). But I'll look at this again in the morning after a good night's sleep.

 

[analag]

[quote author="kiira"]
Thanks for the help. No I haven't seen Rowan's design I don't know anything about it but he usually designs hybrid tube stuff, which I'm not into. In a way I'd rather NOT see it if it is similar to the ideas I have because I'd like some way to try and design/cargo cult my idea so I could learn some more. Problem is, there's lots of books and stuff on the web for designing an amp but I havn't seen anything about designing compressors that a math impaired beginner could understand.

Kiira[/quote]

Not hybrid at all...no problem not showing it to you. Go ahead with the designing vibe,I totally encourage it.

Rowan

 

[Kit]

Wayne speaks the truth.

I dont know about any "ripple reduction" though.......if youre talking about "audio CV ripple", then thats just a matter of ldr decay time.

But yes, using two ldr (one fast one slow) in parallel does give program dependent release characteristic.

 

[SSLtech]

Revision of thoughts:

VACTROL. -If you're using this, there will in fact need to bew a little more than I mentioned: -I forget that the Vactrol is a fast-on-fast-off variable resistance, as I recall. -I was thinking about the LA2a LDR behaviour, with the "inbuilt time constants" as it were...

Not a frightful amount more complicated, but you will need to add some time constants and rectification.

Keith

 

[Boswell]

[quote author="mediatechnology"]What about making the LEDs the rectifiers? (Driving two paralleled photocells.) You'd want some rev polarity diodes, maybe in series with the LEDs, but they'd operate in the current domain. Use a resitive load to set the current so the LEDs have something to drive. Sure there's no integration (other than the photocells) but the LA-4 doesn't have any either other than the "afterglow" mod. I know it's kinda different but might be fun for kiira to try and we could learn from her experiment.

And what would the effect of a "rectifier" (LED) filter cap be? More current at the crest of the peak e.g. peak to average ratio responding? (aka crest factor?)[/quote]
With anti-parallel Vactrols, you're back to the matching problem again, only this time mismatches give you different amounts of compression on positive and negative sidechain waveform peaks. It would be better to rectify using diodes and have one Vactrol per channel, plus it gives you the opportunity to modify the attack and decay times rather than always use the inherent times of the Vactrol (which are current-dependent). For more precision, I would recommend buffering the rectifier output with a transconductance-connected op-amp, so that rectified volts drive Vactrol current.

 

[Gus]

A guess

Maybe a trimpot 500ohm? wiper as input

one end to one LEDs anode

other end to the other LEDs cathode

cathode and anode other sides connected as the other input

then one series R to that module
The pot is used as a trim for balance LED light output.

Like I said just a guess I have not seen one

 

[kiira]

jeezy peezy thanks for all the help everyone. I did a quick schematic not as a final or even potentially working circuit but more to help me organize my thoughts and to also help me visualize what is going on and work from there. Also as a reality check in case I have misunderstood what I have been trying to learn about opto comps.

Instead of designing the makeup amp and to get things going faster I used one of Dave's line amp designs, I was pretty much thinking two gain stages follwed by a CF anyway, seems sensible. I would try and use 6SN7s too because I have them around. I didn't do any of the maths for the line amp using 6SN7s yet. Also the plate voltage of the sidechain cathode follower should be around 230 volts I think. I wasn't sure of some things that I could not interpret from the LA2 schematics so something will be missing in the sidechain stuff I am sure. THe vactrol is just a guess. If this is correct as far as it goes I think that the xformer/rectifier idea is where I wouldwant to go, if that will work... I guess basically this won't work as is. Rowan would know ha ha. :razz:

EDIT: I'm totally into trying the two LED thing too. Ijust want to make sure that I have the basics down and understand what is going on. I could actually get a lot of this together this weekend I think. OMG I'm doomed!

I hope this is sized right for the forum. A fullsize image is at

http://www.2tough.com/~kiira/optocomp.jpg

oh wow I hate putting up my schematics. :?

Kiira
[/img]

 

[NewYorkDave]

Step 1: get rid of that volume control between V1A and V1B, replace with a fixed resistor. Take the signal for the sidechain circuit off the primary of the output transformer.

Alternate plan: keep the volume control between V1A and V1B but get rid of the global feedback. Take the sidechain input off the plate of V1A, before the volume control.

In either case, I'd use a fixed resistor in series between the input xfmr and the Vactrol, and place the input gain control (if needed) after the Vactrol.

The LED in the Vactrol needs reverse-polarity protection. I'm still not crazy about the "one polarity only" detection scheme.

(Well, I assume you wanted opinions. Am I right? :razz.

 

[kiira]

[quote author="NewYorkDave"]Step 1: get rid of that volume control between V1A and V1B, replace with a fixed resistor. Take the signal for the sidechain circuit off the primary of the output transformer.
[/quote]

OK. did that.

The schematics have been revised now. Ooops! OK just realized that the input to the sidechain should probably come after the cap off the plate of V1A but before the resistor. I'll change that on th next revision.

Alternate plan: keep the volume control between V1A and V1B but get rid of the global feedback. Take the sidechain input off the plate of V1A, before the volume control.

I was taking the sidechain input off of V1A, the schematic was not too clear so I cleaned it up to read better.

In either case, I'd use a fixed resistor in series between the input xfmr and the Vactrol, and place the input gain control (if needed) after the Vactrol.

OK. did that too.

The LED in the Vactrol needs reverse-polarity protection. I'm still not crazy about the "one polarity only" detection scheme.

Well this is just the basics and to get help like you are doing :razz: If this will work in theory then I would like to improve it with something maybe more experimental like you and Wayne have been talking about or use another transformer and rectifiers to drive the led. I'm not too clear on how that would look though. Something which will work better than the basics would be totally ok.

(Well, I assume you wanted opinions. Am I right? :razz.

Yup. You bet. :razz: Thanks Dave!

Kiira

 

[SSLtech]

[quote author="NewYorkDave"]Step 1: get rid of that volume control between V1A and V1B, replace with a fixed resistor. Take the signal for the sidechain circuit off the primary of the output transformer.[/quote]
Hmmm... I don't like the fact that the makeup gain affects the threshold... But then I don't like seafood and I eat too much red meat.

[quote author="NewYorkDave"]Alternate plan: keep the volume control between V1A and V1B but get rid of the global feedback. Take the sidechain input off the plate of V1A, before the volume control.[/quote]Ah. I like that much better. The Beefatarian in me is happy.

[quote author="NewYorkDave"]The LED in the Vactrol needs reverse-polarity protection.[/quote]Not according to analag. I also made that point, and a combination of Rowan and PRR persuaded me otherwise. -I do PREFER a back-strapped diode for protection, but Rowan was quite insistent that it's not essential, and the diode will self-protect. -Like you, I'd still include ot for 4 cents worth of protection and a gram and a half of solder.

[quote author="NewYorkDave"]I'm still not crazy about the "one polarity only" detection scheme.[/quote]Oh you're crazy alright. -It just has nothing to do with the half-wave detector! :wink:

You seem to be raising the same concerns that I had over analag's first offering, and I have to say that if analag likes his, then I'm inclined to say build one with some of the same chartacteristics... half-wave detection may very well be a nice 'character' contribution... hey who knows?

Otherwise, just build a 1N4148 bridge around the vactrol diode, and add a small zener in reverse-series, to try and provide a hard threshold step, and make it into more of a limiter, if you so fancy.

Keith

 

[kiira]

[quote author="SSLtech"][quote author="NewYorkDave"]Step 1: get rid of that volume control between V1A and V1B, replace with a fixed resistor. Take the signal for the sidechain circuit off the primary of the output transformer.[/quote]

Hmmm... I don't like the fact that the makeup gain affects the threshold... But then I don't like seafood and I eat too much red meat.

[/quote]

I changed that and revised the schematics. I like it better now because Iam not happy about the makeup gain afecting the threshold either. I'll just lose the negative feedback.

[quote author="NewYorkDave"]I'm still not crazy about the "one polarity only" detection scheme.

Oh you're crazy alright. -It just has nothing to do with the half-wave detector! :wink:
[/quote]

I don't get it. :shock:

You seem to be raising the same concerns that I had over analag's first offering, and I have to say that if analag likes his, then I'm inclined to say build one with some of the same chartacteristics... half-wave detection may very well be a nice 'character' contribution... hey who knows?

OK... so this WILL work as drawn? Iguess I better build it fast and see but I thought that there were some theoretical bad things about it.

Kiira

 

[NewYorkDave]

Looking much better... You just have to move that series resistor on the vactrol. It should go between the xfmr secondary and the vactrol AND the grid. It acts as the upper leg in a voltage divider, with the Vactrol as the lower leg.

 

[NewYorkDave]

Since the grid draws negligible current--unless driven positive of the cathode--the answer is no.

 

[CJ]

You can run AC on an LED.
For a while. It is a slow degeneration, depending on the voltage.

 

[NewYorkDave]

Keef: Of course! Why didn't I think of that? I've worked with enough telephone stuff that it should have been apparent to me right away.

Certainly, there's no need for the LED to be grounded at either end, it doesn't care as long as there's a path for the current.

 

[NewYorkDave]

Yeah, you'd use a coupling cap. Otherwise that poor LED would be hurtin'.

 

[PRR]

http://www.2tough.com/~kiira/optocomp.jpg

Is that a point-two-two cap driving the LED? If so: always write ZERO-point-two-two, so the leading Zero alerts us to the decimal point.

And the naked LED is about 30-100 ohms when lit-up good. You would need 100uFd for full bass signal to the LED. AND a beefier driver.

Note that a Cathode Follower is NOT always the best way to drive a high-current load. You get exactly the same current in the Plate side.

There's a couple ways to go. To use the LED voltage as a limiter threshold, you want a fairly high-power low-Z output. Put the LED inside a bridge rectifier, I like that. Threshold will be around 1.7V+1.2V= 3V peak. For 6dB rise, we want several mA of LED current at twice that level, 6V peak. Say 6mA at 6V. Put in a series resistor to drop 3V at 6mA, or 500 ohms. Now we can use a 20uFd coupling cap. If the driver tube is resistance coupled, it needs to idle at a lot more than 6mA.... 20mA at a minimum. Maybe you could fudge the LDR series resistor, work the LED at 1mA-2mA peak, and get away with the driver you have.

To get a very stiff limiting curve for speaker protection, I've used ~100 ohms in series with a bridge around a couple LEDs (the second one worked the meter).

The simplest VCA for your purposes is a JFET, such as used in the UREI or the Shure. Yeah, it was made with transistors and chips, but it would work fine all-tube except the one JFET. The Gate drive is lovely high impedance, over 1 Meg release resistor. For reasonable attack times, the series resistor could be as high as 10K, easily driven from a resistance-coupled tube. I don't know if NYD has published his little vari-Mu design, but that sidechain plus a small negative DC supply would make a fine FET driver. Note that the signal at the FET must be held below 100mV or 10mV for low THD, so you need a chunk of gain between that point and your Line Output.

The opto has the decided advantage that you can see it work. Note that this can be tricky: room-lights will drive the LDR to minimum gain, and the glow from the LED may be dim at limiting threshold.

 

[NewYorkDave]

This seems to reinforce the case I was making for a stepdown xfmr, if an LDR is being driven by a tube sidechain amp. (I'm assuming just for the sake of discussion that Kiira is sticking to her no-transistor policy, although a JFET as a variable resistor is an attractive arrangement for the reasons you describe).

The transformer alleviates the requirement for high plate current--replacing it with the need for a high voltage swing, more easily managed. And we should also remember that any single-ended stage (common-cathode or common-plate) will not push and pull equally well.

 

[NewYorkDave]

This brings me back to what I was describing earlier in the thread, shown here in its skeleton form. (Add feedback, response shaping and other refinements as needed).

With the second stage biased for a plate current of 7mA or more, and its cathode bypassed, a 10K (or 15K):600CT transformer would probably do. Add a -FB loop and it's essentially the "One-Bottle" preamp.

If it must be a Vactrol and it must be a tube sidechain amp, the choice is either to go for the transformer or run the tube pretty hot. One might opt for the latter to save the cost of the transformer, although as long as Edcor is around the transformer option can still be pretty affordable.