The Simplest Opto Compressor Imaginible

[Meathands]

That's what I'm going for here. I'm working on an opto compression circuit for a larger project of mine called "Colour" which a search will turn up. I basically hacked away at the What Compressor / Weak Joe circuit until I thought I couldn't take anything else away.

There are no controls, since the input signal is coming from the Colour's master input drive, and ratios, attack, and release are fixed or left up to the opto cell.

My questions, besides welcoming your criticisms / ideas / disdain, are:
-Why did "DeeT" use an active rectifier like the one attached rather than a passive bridge?
-I see some opto sidechains, such as Fred Forsells, do not bias the LED from the power rails. Is this just a matter of getting the LED bright enough without having to amplify the sidechain signal?

 

[PRR]

"Simplest"?

All those parts??

Attached is a simpler opto limiter which served well in a public venue for many years and many idiot operators. If you don't have a Crown DC-150, you can use an LM380 with a fairly good supply voltage. IIRC the pot was needed to scale the 34V peak output to the ~~5V peak needed by the diodes; with a smaller rig you instead pot-down between the ~~5V raw peak output and your 1V-2V line output.

Time-constants are selected by picking LDRs, and are generally short.

> do not bias the LED from the power rails

That's offset, threshold. So the LED don't light at all until the precision signal reaches a specific calibratable point. But when saying "Simplest", say fooey to that. The 1.6V of an LED is offset a-plenty. (4.5V-5.0V since you need an indicator and a diode bridge.)

 

[PRR]

Point of design: 680 seems a low value for the LDR feed. Many LDRs don't go much below 1K, and 100r is very-very low for an LDR. That gives only -4.5dB to -18dB reduction. The latter is fine, if a suitable LDR falls to hand; the former is hardly useful. Anyway there is no objection to a much higher resistor, since you budgeted a buffer. And since your signal source is strong but maybe not "pristine clean", the noise-voltage of 100K maybe 1Meg isn't offensive. Values 10K-50k seem sweet.

Likewise IF you do need to slam the LDR below 1K, meaning more than a few mA in the LED, the voltage drop across a 1K feed resistor makes loop voltage rise from ~~4.5V at just-glow to >10V at full-slam. Soft action. And seeing that the LED is fed from an opamp chip, you are unlikely to burn-up the LED in any case. The spec may be 16mA, but we know it is a 20mA LED (they all are), and they can take 40mA for long periods. And if your resistances allow the LDR to do its job, it will turn-down huge inputs and never be over 10ma for long. While I do throw a little R in there (specially with a DC150!), with chippy-work it isn't vital, and should be small.

 

[ruffrecords]

Far too complicated PRR, all those diodes. All you need is a pot divider consisting of a small series incandescent lamp  and a resistor to ground.

Cheers

Ian

 

[Meathands]

Many thanks for the great ideas, PRR.

I've drawn up your circuit with an opamp instead of a power amp. I took out the indicator LED and made the rectifiers germanium so the IC wouldn't have to work as hard. Set the opamp for a gain of -2 to get +4dBU around 2.2v peak.

It occurs to me that even at +10dB RMS, it may be asking a lot of the IC to drive the LED without assistance (~30mW at 20mA), so the What Compressor's current gain transistor may be a good fail safe.

Far too complicated PRR, all those diodes. All you need is a pot divider consisting of a small series incandescent lamp and a resistor to ground.

Like a Zen master, Ian points the way to true simplicity. What kind of current needs do you think we're looking at with a small incandescent lamp?

 

[Kingston]

Like a Zen master, Ian points the way to true simplicity. What kind of current needs do you think we're looking at with a small incandescent lamp?

Have a look at the univibe lamp driver. Works with various types of small bulbs like 12V/25mA. That's a maximum "headroom" current that your zen-compressor obviously doesn't require. Has a trim so you won't immediately burn it. www.geofex.com/Article_Folders/univibe/vibeupdate.pdf

 

[PRR]

> too complicated PRR, all those diodes

Later recollection suggests, instead of the FWB, I used back-to-back LEDs to get full-wave action.

Whether two costly LEDs is "simpler" than one plus four cheap diodes is debatable.

Incandescent has always seemed too slow to me.

> made the rectifiers germanium so the IC wouldn't have to work as hard.

Bah. Let it work.

Anyway you only shave 0.8v off a 4.8V total. If you can't do 4.8V, 4.0V may be a stretch also; in practice neither is difficult.

Also Ge production is slim to none (though excavation continues in Russian warehouses). And Ge is very precious to fuzz-pedal builders and crystal-radio fans. Prices gone up 100X in my lifetime. Don't stir the market.

 

[Brian Roth]

Schottky diodes have much lower forward voltage drops compared to silicon.

 

[ruffrecords]

Like a Zen master, Ian points the way to true simplicity. What kind of current needs do you think we're looking at with a small incandescent lamp?

Lol.

You can get grain of rice bulbs that require only 15mA at 1.5 volts to illuminate at which point they would have a resistance of just 100 ohms. So you could probably make a pot divider from one of these and a 10 ohm resistor and drive the thing from an 8 ohm speaker output or even a headphones op amp.

http://www.rapidonline.com/Electronic-Components/TruOpto-Rice-Grain-Meter-Bulb-81587

Cheers

Ian

 

[L´Andratté]

On a sidenote and against the thought of simplicity, I find the -3db HP slope most useful, at least in my use of compressor for instrument (don´t know bus/summing), as it gives really good frequency response (eg. with bass, notoriously pumping). But most, even very good and expensive comps seem mostly to use 6db sidechain filters(which improves a bit, but still uneven distribution, having tried 3db, there´s no doubt for me, it sounds best), I wonder why?

 

[leigh]

OP: Check out this thread from a few years back for a super simple opto design, which includes lots of valuable feedback from other board members:

http://www.groupdiy.com/index.php?topic=29986.0

(I just went through and updated all the image links I had made which had broken...)

 

[Meathands]

Interesting turn of events. I built up PRR's circuit I posted above with an LDR and a 3mm red LED.

The output has a low-pass rolloff that's level-dependent and the slope varies per octave! The output is also delayed so that it reaches opposite polarity of the input around 10kHz.

What could be going on?

My best guess is that the LDR has some varying parasitic capacitance, but surely this is unusual in an opto circuit!

Any thoughts would be much appreciated.

 

[L´Andratté]

Thanks Leigh, I bookmarked that fred!

Meathands, is it on breadboard? They can also be a source of weird capacitance...

Amplitude AND frequency dependent filtering? I think you should make a synth module out of it! :

(EDIT)Quote from WIKI:´Parasitic capacitance limits the frequency range of the photoresistor by ultrasonic frequencies.´

 

[kato]

Hey Peterson, Mictester's "Really Cheap Compressor" might be a source of inspiration. It's designed for guitar but a doubling of the RC networks' cap values would extend the bass response.

 

[L´Andratté]

Those compressors seem to grow like weed.




http://www.aronnelson.com/gallery/main.php/v/johan_0/LA_Light_last_one_001.gif.html

http://experimentalistsanonymous.com/diy/Schematics/OOP%20Japanese%20Electronics%20Book/LED-LDR-comp.gif

http://geograffff.chat.ru/schemes/comp/whispCom.gif

http://sslstudiodesign.free.fr/DIY/COMP/comp.gif

http://www.diystompboxes.com/smfforum/index.php?topic=89110.0

and many others...

But for very simple I prefer the "rectifier bridge style", like Hollis-Flatline et al.

 

[sebpmtl]

"Simplest"?

All those parts??

Attached is a simpler opto limiter which served well in a public venue for many years and many idiot operators. If you don't have a Crown DC-150, you can use an LM380 with a fairly good supply voltage. IIRC the pot was needed to scale the 34V peak output to the ~~5V peak needed by the diodes; with a smaller rig you instead pot-down between the ~~5V raw peak output and your 1V-2V line output.

Time-constants are selected by picking LDRs, and are generally short.

> do not bias the LED from the power rails

That's offset, threshold. So the LED don't light at all until the precision signal reaches a specific calibratable point. But when saying "Simplest", say fooey to that. The 1.6V of an LED is offset a-plenty. (4.5V-5.0V since you need an indicator and a diode bridge.)

10 years later.. Thanks, this thread helped me make my first opto limiter for my modular synth. I've been learning electronics and try to use parts I never used before to see how they work. That was the first time using a bridge rectifier. Also, I recently discovered how great vactrols are, making my own with a led and photoresistor. So, for this circuit I ended up using a mix between yours and OP's circuit. I do the limiting between two op amp buffers, really simple. I also wanted to soft clip the signal, so I put some 1n4148's in the non-inverting feedback path.

 

[ruffrecords]

This is still a fascinating and little explored topic. I have done some work in this area myself. I have found that many off the shelf opto-couplers have a breakthrough issue if the LED is driven directly by the audio. Some form of coupling takes place which means a proportion of the very distorted diode waveform appears in the audio. Often this simply sounds like "typical compressor distortion" but if you look at it on a spectrum analyser the distortion can reach very high levels. So now I favour designs which first rectify and then smooth the input signal in order to eliminate this breakthrough. On the other hand, others who have used LEDs driven by audio in their own home made opto-couplers do not seem to suffer from this so perhaps it is something in the construction of the opto coupler.

Cheers

Ian

 

[Script]

When using a diode bridge, bias is required for stepless precision. With a synth module or a simple PA protection circuit this might not matter. For an FX compressor too it might not matter.

@Ian
Your "A tube opto compressor" paper from 2021 is a highly inspiring read. I recommend to anyone looking into opto comps -- but I don't have the link at hand.

 

[ruffrecords]

@Ian
Your "A tube opto compressor" paper from 2021 is a highly inspiring read. I recommend to anyone looking into opto comps -- but I don't have the link at hand.

Copy attached. I really should find a place for it in the DIY section of my website.

Cheers

Ian