Using LEDs to prevent a PT2399 from distorting [renamed]

[midwayfair]

EDIT: renamed the topic. Problem solved, now let's make it better!

I'm trying to limit a signal to a PT2399 from ever spiking above ~2.3V. I don't have a zener on hand with a value that low to experiment, but will that work? Or will it cause the audio signal to distort anyway (though maybe it'll be nicer than what the PT2399 does)?

And is there a way to use multiple zeners of a particular value (I have 4.7v for instance) to fake it so I can find out?

[Seljer]

Three or four regular diodes will be in that range. Or maybe and LED. But it's still going to be distorting.

A failsafe way would probably to just attenuate the signal in front (two resistors) and then apply gain afterwards. If you want to get all fancy about the noise levels and such maybe look into using a compander chip (look at the PT80 delay on generalguitargadgets)

[R.G.]

Clippers - that is, all diodes, zeners, LEDs, etc. - cause distortion.

Nondistorting limiting is done by compression. What you're looking for is the NE/SE571 compander chip, to compress or limit the incoming signal and expand it after it comes out.

[midwayfair]

Three or four regular diodes will be in that range. Or maybe and LED. But it's still going to be distorting.

A failsafe way would probably to just attenuate the signal in front (two resistors) and then apply gain afterwards. If you want to get all fancy about the noise levels and such maybe look into using a compander chip (look at the PT80 delay on generalguitargadgets)

Plenty of gain afterward. And the signal is pretty severely attenuated before it gets into the PT2399 -- it's coming off the buffered out of a transistor and it only distorts when I feed it about 15db, which is decent but not perfect.

Physics is not on my side here, I know ... But a diode distorting only under extreme conditions would probably sound better than the PT2399's distortion. I'm under the impression that the Zener would conduct only if it hits 2.xV, not all the time. I'll try the three diodes (or just two LEDs, probably ...) at least.

[garcho]

Thomas Henry has one of his books on the compander if you decide to look into that route.

You don't want to drop everything down, just limit the peaks, right? Otherwise a drop-down resistor would work easily enough, wouldn't it?

[midwayfair]

Actually, two back to back LEDs on pin 15 [woops] 7 seem to be working ... they aren't changing the readings I get with "normal" sized signals, but they light up when the signal gets huge (over 20db at least). No audible distortion. Pretty crude limiter, just like in the Lovesqueeze ...

I'll probably go this route. Can't fit a compander.

[Kesh]

Actually, two back to back LEDs on pin 15 [woops] 7 seem to be working ... they aren't changing the readings I get with "normal" sized signals, but they light up when the signal gets huge (over 20db at least). No audible distortion. Pretty crude limiter, just like in the Lovesqueeze ...

I'll probably go this route. Can't fit a compander.

clipping of transients isn't usually heard as a distortion

[midwayfair]

The only drawback seems to be that the mix needs to be turned up for louder input signals (signals above about 20db won't get above unity anymore), because the PT2399's output will be clamped.

The question remains: Is a single Zener of appropriate voltage better than the back-to-back LEDs? Will the zener actually do what I want (conduct super large signals for a very short period of time), such that it would be worth ordering one? I'm not sure I like the idea of using a very, very rare part.

Edit: I almost can't believe how well this works.

[therecordingart]

I have some THAT 2150A VCAs for sale if you are wanting to build a compressor. 

[midwayfair]

I have some THAT 2150A VCAs for sale if you are wanting to build a compressor. 

I have one, thanks. It's one of my favorite toys.

[EATyourGuitar]

what you are doing with the clipping diodes before the delay is exactly how some big manufacturers do it. I was never sure if it was intended to prevent damage or maybe have distortion on some settings as a new feature.

[midwayfair]

what you are doing with the clipping diodes before the delay is exactly how some big manufacturers do it. I was never sure if it was intended to prevent damage or maybe have distortion on some settings as a new feature.

But it's not before the delay, exactly. It's on the current control chip of the PT2399 (at least, I think that's what CC stands for in the datasheet). Do you have a schematic where it's used? I'm genuinely curious to see if it's done any other ways.

[slacker]

For what you're doing a zener would do exactly the same as the LEDs. The only advantage would be it would do it at a known voltage where as different LEDs will do it at different voltages, not a problem for a one off build but worth thinking about if you share the design.
Sounds like a neat trick, I'll have to try it.

[greaser_au]

Nondistorting limiting is done by compression. What you're looking for is the NE/SE571 compander chip, to compress or limit the incoming signal and expand it after it comes out.

See Philips application note AN174...  www.sss-mag.com/pdf/AN174.pdf  I found a couple of errors in the 1V pk-to-pk hard limiter in figure 12:  the comparator inputs are all reversed,  and there is a missing +15V connection at the junction of R15/R16. Works well once you find/fix those!!!  

[midwayfair]

For what you're doing a zener would do exactly the same as the LEDs. The only advantage would be it would do it at a known voltage where as different LEDs will do it at different voltages, not a problem for a one off build but worth thinking about if you share the design.
Sounds like a neat trick, I'll have to try it.

Awesome. I will definitely add the Zener as an option in the build document with instructions for plugging it into the board.

The only place I can find a Zener of appropriate voltage is Mouser, and no one's going to have a 2v zener just lying around, so the boards and etch layouts will still going to have the LED pads there -- it's easier to put one Zener standing up than to fit two LEDs into a single diode slot.

[samhay]

Jon - You would usually use parallel LEDs as clippers (like in the Love Squeeze), whereas you usually use Zener's back-to-back to clamp an AC signal. It will then clip at the Zener breakdown voltage + 0.7V (Si diode forward drop). As a result, LEDs and Zeners will not be interchangeable in any given layout. I don't think you can find 1.6V Zeners anyway, but I may be wrong.
I would stick with LEDs - they have a reasonably soft knee and clip right where you want them.

[midwayfair]

Jon - You would usually use parallel LEDs as clippers (like in the Love Squeeze), whereas you usually use Zener's back-to-back to clamp an AC signal. It will then clip at the Zener breakdown voltage + 0.7V (Si diode forward drop). As a result, LEDs and Zeners will not be interchangeable in any given layout. I don't think you can find 1.6V Zeners anyway, but I may be wrong.
I would stick with LEDs - they have a reasonably soft knee and clip right where you want them.

I think it just needs to be a single, reversed Zener (anode to ground) -- just like what we do with a zener to protect a charge pump from overvoltage with a 9.1 to 12v Zener.

There might be a concern about the zener CONSTANTLY seeing too large a signal, though. It takes about +25db boost to do that in the circuit I'm working with. How long can a zener handle, say, 1v over its rating? I'm not sure what to look for on a datasheet ...

EDIT: more appropriate datesheet for through-hole:
http://www.mouser.com/ds/2/68/1N4614-4627-14375.pdf

though I could certainly buy some and subject them to a stress test.

Edit edit: Hah. 1.8v Zener is the only through-hole, sold as singles, diode at Mouser. So it will be identical to what the LEDs are doing.

[R.G.]

It's also worth noting that LEDs have different forward (clipping) voltages depending on the exact recipe used to make them.

Old red LEDs were 1.2-1.4V, modern blue LEDs are as much as 3.5-4V. Orange, Yellow, and Green are in the middle.

You may have to test and specify an LED color or forward voltage to get clipping where you want it.

But then, you can.

[midwayfair]

It's also worth noting that LEDs have different forward (clipping) voltages depending on the exact recipe used to make them.

Old red LEDs were 1.2-1.4V, modern blue LEDs are as much as 3.5-4V. Orange, Yellow, and Green are in the middle.

You may have to test and specify an LED color or forward voltage to get clipping where you want it.

But then, you can.

Yellow and Green will work best. My red LEDs all measure 1.7v, never below 1.6, which seems high from what other people say, but they're all modern production. They'd work in a pinch.

[~arph]

But it's not before the delay, exactly. It's on the current control chip of the PT2399 (at least, I think that's what CC stands for in the datasheet). Do you have a schematic where it's used? I'm genuinely curious to see if it's done any other ways.

Yes it does stand for current control. It says it in the ds I have. You might be on to something.. I never saw anything else beside the standard caps to ground from pins seven and eight.