Opto Limiter

[Albert H]

Hi all. I found these cheap analogue opto-couplers - the LCR0202. They contain an LED on one side and a Light Dependent Resistor on the other. They are widely available for pennies each from the usual Chinese suspects.

There is a bit of a production "spread" on them, so it's worth buying 10 or 20 of them, and matching them in pairs for stereo use. The easy way of doing this is to power the LED side of the package with a known voltage through a resistor (I used 12V and a 5.6kΩ resistor) and then measure the resistance of the LDR side. You just have to find pairs that give the same resistance (to within a few tens of ohms).

The circuit is dead simple - an inverting op-amp audio stage with unity gain through it. The feedback resistor on each channel is paralleled by the LDR in each LCR0202. The audio level is resistively combined to mono, then rectified by a couple of op-amps, and the resulting DC voltage is used to charge and discharge a timing capacitor. A couple of cheap transistors are then used to apply this filtered variable voltage to the LEDs in the opto-couplers. The louder the sounds, the brighter the LEDs will be, reducing the gain of the audio paths below unity, thereby controlling the level. The threshold level at which the LEDs start to illuminate is adjustable by a preset on the PCB. You could wire that to an external pot if you wanted to, but DJs usually have a tendency to twiddle any controls they see, so it's probably best as an internal preset!

The transistors can be any NPN and PNP general purpose parts - just be careful of the pin-outs!

The circuit is:

Audio-Limiter Stereo Opto.jpg

You can build these on Veroboard:

New Stereo Opto Limiter.png

Or you could build them on a single-sided PCB:

New Stereo Opto-Limiter PCB.png

The attack and decay times are (mostly) down to the response of the LCR0202s, but the timing components shown in the diagram above give a fast attack and slower decay - subjectively they're OK for music and speech as shown.

This isn't going to replace an "Audio Prism" and may suffer from "pumping" on bass-heavy material, but it will stop you going wildly overmod! Because of the attack time limitation of the LCR0202, it's probably best to follow this limiter with a clipper. I'll put up a suitable circuit shortly.

It's just another way to limit, and it's reasonably transparent.

[1608cc]

Think RC time constant of C5+R12 is attack, and C5+R11 is for decay. Worth to put trimms in series with resistors (with lowered values) and then You will achieve usable range of regulation. If You have oscilloscope, you can easily determine time response of this circuit (attack and decay times) due to initial step, fg. low frequency square wave - procedure like for a PID regulator.

[Albert H]

You're right, but I just wanted something cheap, simple and easily thrown together, which would give reasonable results.

Practically - in broadcast - you need the fastest possible attck time, and the recovery time needs to be pretty slow. As shown, that's what you get.

I'll put up a simple clipper / filter circuit for use with the above limiter soon.

I've also been working on a cheap and simple split-band limiter (<250Hz and >250Hz - 15kHz), which sounds really good. I've just built a couple of them for use in Nightclubs, to curtail the over-enthusiastic DJ's wild levels!

[radionortheast]

Does seem nice and simple, does amuse me theres so many users for leds, like the guy painting over his leds , the only problem I have with home made equipment is rf and hum getting into them, end up with a buzz box!. I even have problems with smps noise, very low level, I hooked up a half frequency oscillator I made up to a noisy smps it was dead quiet, alot of other transmitters I could hear it weird!, not that I want to take this off topic, also high frequency whistles too.

[Albert H]

Northeast: The ways to prevent the dreaded noises and hum:

Try to avoid SMPSU if you can. The traditional transformer, bridge rectifier, smoothing capacitor combination is always MUCH quieter. Linear voltage regulation - using 78 / 79 series 3-terminal regulators is also easy. Just be certain to RF bypass both the input and output of the regulator ICs - I usually use 10 - 100nF as close to the regulator pins as possible. It's also worth adding a smaller electrolytic across the output of the regulator too....

As a quick aside - as a "rule of thumb", use about 1000 µF / Amp drawn. It's much more smoothing than theory recommends, but you'll be sure that it'll be clean!

The next thing is pretty obvious, but many forget it: You MUST use "single point" earthing. It's usually easiest to put a central earthing terminal in the middle of the base of the case you're using, and radiating the earths to the various parts from there. The incoming mains earth also needs to go straight to that point. Do not earth audio (in particular) sockets to the case as well as running wires to them - you'll create earth loops, which will always cause hum.

Next - and most obvious - use metal casings! Don't consider plastic cases - or even part-plastic cases. RF goes through plastic easily, but has much more trouble with metal!

Many of our constructions use internal metal boxes - for example it's often worth screening the PLL from the rest of the transmitter to keep the digital switching noises out of the modulation.

[Hugh]

Good one Albert, looking forward to your added limiter/clipper.

[tvtv]

Is this better than the pira comp / limiter ?

[tvtv]

I’ve got a TL081C can I do anything meaningful with these ? Also got vag of NE5532 and bag of TL072

[Albert H]

Is this better than the pira comp / limiter ?

Better? Probably! Certainly lower noise and distortion, and really cheap.

In answer to your question about op-amps - the NE5532 is a really good dual op-amp, with high slew rate and very low noise and distortion. I make lots of use of the '5532 (or its LM833 equivalent) in mixers, tone controls, and other audio circuits.

[tvtv]

IMG_2507.png

IMG_0896.png

Albert, with reference to pira, there is another. I’ve seen and wondered how that was rates?
They look very similar, the robosem uses TLC272 which I know little about.

[Albert H]

The TLC272 is a CMOS version of the TL072. It has lower current consumption, but has a (slightly) worse noise specification, but a higher slew rate. In the "Pira" circuit, you could probably use either the '272 or the '072 with little or no change in results.

The biggest problem with the "Pira" circuit is matching the FETs and transistors if you're trying to use it for stereo. The channel to channel interlinking only works for the limiting (actually compression above a threshold, but near enough), but doesn't work for the clippers (you'll also have to match the LEDs accurately for Vf). This could compromise the stereo image!

[tvtv]

The TLC272 is a CMOS version of the TL072. It has lower current consumption, but has a (slightly) worse noise specification, but a higher slew rate. In the "Pira" circuit, you could probably use either the '272 or the '072 with little or no change in results.

The biggest problem with the "Pira" circuit is matching the FETs and transistors if you're trying to use it for stereo. The channel to channel interlinking only works for the limiting (actually compression above a threshold, but near enough), but doesn't work for the clippers (you'll also have to match the LEDs accurately for Vf). This could compromise the stereo image!

Thanks