Super Simple Limiter

[scanlory]

I don't know if anyone is interested, but for a class we had to come up with a simple control systems project.  I made this feed-forward signal limiter, trying to keep it as simple as possible.  Just set the threshold (the 10k trim) and it's good to go!  I don't know what you could actually use it for, but who knows...it may be useful somewhere:



Any ideas for what it could be used for?  Or any improvements....let's be honest, it's not the greatest design. 

[brett]

Hi
nice idea.
A few improvements might make it more practical.  Firstly, you'll need to increase the values opf those resistors so that the input impedance is higher.  You need at least 470k on each input line (for a total resistance of at least 235k).
Don't forget the basics like an input cap and an output cap. 
You might want to add some DC bias to the +ve input to the LED-driving opamp.  This would let you set the signal strength at which the LED lights.
You might also want to rectify and average (wrt time) the signal into the LED driver.  A germanium diode before the resistor, followed by a small cap to ground would be good.
If you want the ultimate in rectification, use an "ideal" diode, as similated with a pair of opamps (search the web).

Also, for your photoresistor, do you mean 10 Mohms dark?  Otherwise I can't see how the circuit will work (but then again, I am very tired).
good luck

[Mark Hammer]

The drawing looks simple, but that's largely because all of the supporting and input/output circuitry (like input/output caps to block DC, and the entire biasing structure) is missing.  Once you put all that stuff in, it starts to be as complex as something like the DOD280.

[valdiorn]

Nice idea, but I'd really half-wave rectify the control signal and put it through a low pass filter to smooth it out a little bit more (I know there is some hysteresis in the photoresistor but probably not enough, you might get a very "sharp" sound).

Also, where to get photocells with such a small "on" resistance?? Most of the sources I know of only have cells that go to about 5k ...

[moosapotamus]

Search "LA-Light" to see something similar.

~ Charlie

[valdiorn]

also, I was reading the Silonex datasheets (seems to be one of the largest manufacturer of photoresistors), and can anybody tell me what the heck does "ftc" stand for??? I see stuff like "R @ 2ftc (Kohm)" and it gives some value of resistance, problem is I don't know what ftc is...

[biggy boy]

also, I was reading the Silonex datasheets (seems to be one of the largest manufacturer of photoresistors), and can anybody tell me what the heck does "ftc" stand for??? I see stuff like "R @ 2ftc (Kohm)" and it gives some value of resistance, problem is I don't know what ftc is...

Foot Candles a unit of measurement

[snap]

http://www.mediacollege.com/lighting/measurement/

[scanlory]

Thanks for the comments, keep it coming! 

I think that I'll draw up the bias network to run on a single 9V rail, and add a 4.5V reference to the signal op amp (w/ coupling caps on the input/output of that stage).  When I was running the demo in lab, I was using -10V to 10V rails..not very practical in the world of stompboxes.

The photoresistor values are kind of strange, b/c all the stockroom had were a bunch of unlabeled LDR's.  The particular one that I used had a light resistance of ~2 kΩ, and a dark resistance of 20 MΩ  .  For sake of time, I just threw a 10kΩ resistor in parallel w/ the LDR to improve performance a little (otherwise release time was unbelievably slow...maybe 10 seconds or so).  If I have a little bit of time today, I'll try redesigning it a bit w/ some more 'real-world' components.  I'll also try redesigning the control circuit w/ some of the suggestions here, thanks!

[scanlory]

After a little bit of a redesign, here's what I came up with.  How's this look?



I feel like the sidechain should be isolated from the signal path via a coupling cap, would that be correct?

[Mark Hammer]

Depending on the particulars of the LDR involved, very often it is unnecessary to use full-wave rectification for optical-based limiters/compressors.  FWR is used to provide greater "smoothness" to the controlling voltage, which is important if the control element is fast-responding OTA or JFET.  LDRs, however, are more sluggish than either of those two other control elements, so the "smudging" that one would normally use FWR for is actually accomplished by the LDR's lag.

The simple way to do this is "the Anderton way".  Drop the FWR stage, stick the current limiting resistor before the LED rather than after, and run an electrolytic cap from the resistor/LED junction to ground to add a wee bit more lag.  The LED itself provides the unidirectional envelope following (i.e., passes only one half-cycle of the signal), and the cap to ground reduces any remaining ripple.

I can't actually see this at work, but here is what I believe to be a link to the EPFM compressor, showing what I describe: http://i197.photobucket.com/albums/aa211/bajaman002/CraigAndertonCompressor1980.jpg

[Eb7+9]

Depending on the particulars of the LDR involved, very often it is unnecessary to use full-wave rectification for optical-based limiters/compressors.  FWR is used to provide greater "smoothness" to the controlling voltage, which is important if the control element is fast-responding OTA or JFET.  LDRs, however, are more sluggish than either of those two other control elements, so the "smudging" that one would normally use FWR for is actually accomplished by the LDR's lag.

the time constant associated with the smoothing cap (or photo-element time constants) is way longer than any signal period you're going to pump in there - the original OS and Dyna are proof enough of this ... the bumpyness one experiences in comp circuits has more to do with fluctuating "string" envelope than its frequency content and missing half-cycles thereof in the charging circuit

[scanlory]

A few more redesigns:



Capacitively coupled sidechain, and a level control stage as the output of the first stage (when dark: i.e. not limiting) will be about twice that of the input.  Also removed the rectifier..although I would like to try it both ways; see if there is any discernible difference.

[brett]

Hi
one thing that is a pain in the butt to us DIYers is that everyone chooses a different LED and a different LDR.

Any chance you could design this thing with the super common, super cheap 4N25?  It will have a couple more components, but will be garaunteed to work as described and will therefore get a lot more builds. 
cheers

PS  The 4N25 also has a significant technical advantage.  The Vf of the LED is very low (about 1.2 to 1.5V ??), so you won't need much amplification on the side chain.  This can help you to achieve more subtle, natural compression.   The maximum If is also large (>50mA), so you can also seriously amplify on the side chain and totally crush the signal (if that's an option you want).

[scanlory]

Any chance you could design this thing with the super common, super cheap 4N25?

Hmmm....I've never worked with phototransistors, but I like the idea.  Do you of any good resources where I could read up on some info about them?

[Mark Hammer]

Depending on the particulars of the LDR involved, very often it is unnecessary to use full-wave rectification for optical-based limiters/compressors.  FWR is used to provide greater "smoothness" to the controlling voltage, which is important if the control element is fast-responding OTA or JFET.  LDRs, however, are more sluggish than either of those two other control elements, so the "smudging" that one would normally use FWR for is actually accomplished by the LDR's lag.

the time constant associated with the smoothing cap (or photo-element time constants) is way longer than any signal period you're going to pump in there - the original OS and Dyna are proof enough of this ... the bumpyness one experiences in comp circuits has more to do with fluctuating "string" envelope than its frequency content and missing half-cycles thereof in the charging circuit

Quite true, especially if your strings are not new.  Deformations produced via repetitive fret-to-string contact produce perturbations in the free vibration of the string.  Many of those bumps and lumps in the string envelope are in the sub-audio or low audio range (e.g., 15-40hz), and are one of the principle sources of undesirable ripple during the decay phase of the envelope.

I guess the point to be made is that, while FWR can help, it is not a panacea, and if it increases the complexity, footprint, and current draw of the circuit, it may provide little return on investment compared to addressing the string envelope.

[brett]

Hi
what's wrong with a high-pass filter and half wave rectifier before the EF storage? 
fc=60Hz? 
Then you should be able to bend, slap and vibrato to your hearts' content without an over-full envelope follower.

or am I missing something?

[scanlory]

Another update:



Following SPICE analysis, it seems as if the DC bias on all of the op amps are not where I want them to be (and therefore the circuit is not working correctly at all).  I am a little unfamiliar with single-rail op amp design; should I add coupling caps between all of the op amp stages, as I want to use 1/2 Va as the virtual gnd?

W/ a 200mVpp 1 kHz input, i get maybe a 1/10 of a uV swing on the output, and not centered around 0V.  Any help is appreciated.