Clipping LED's

[Goodrat]

I've noticed no one ever mentions a part number when discussing a good sounding LED for clipping in for example a Crunch Box.
All that is said is "red diffused". Sometimes one might mention a Vf of 1.9 or 2.0 or 2.1 volts.
Well, that is not enough info. The Vf may be a spec at 20mA and have nothing to do with the 2mA encountered in a circuit.
I measured the current for a Fluke meter on the diode test setting and it is 0.2mA.
I've measured a TLHK4600 3mm Red LED spec'd at 1.9v @20mA and a meter will measure 1.70V.
With variable DC applied:
1.76@2mA, 1.80@5mA, 1.83@10mA, 1.87@20mA. I can conclude that in-circuit clipping will be roughly 0.1V higher than a meter test.
In-circuit this TLHK4600 LED clips at 1.75, compared to my Joyo clone that clips at 1.84 in-circuit. (and 1.68 on a meter diode test).
I have a 5mm that is closer. (part# MVS754A, 2.0V@20mA spec)
My point is, if someone finds an LED that sounds good, please tell the part number.
"Diffused" I don't think will make a difference because I suspect the plastic has nothing to do with the diode junction behavior
and brightness does not necessarily make a difference either. The spec at 20mA may have little to do with what it does at 2mA.
Finally, 5mm vs. 3mm, forget it. I have examples of high  Vf on a 5mm and low on a  3mm and vice-versa. There is no correlation.
Part number!!!

[midwayfair]

LEDs, like all diodes, regardless of part number, vary in their Fv. Knowing a part number won't change that.

Also, a .1v difference in an LED is about 5%, which is a pretty normal tolerance for parts.

Also, almost no parts supplier gives a part "number" for a 3mm led. Why? Because the color indicates the likely Fv. If there is a part number, it's irrelevant.

Diffused are cheaper. That's why diffused is mentioned usually.

In any case, I really think you're over thinking this. A 5% difference in a diodes Fv will not make a meaningful difference in the sound of a circuit. If your pedal doesn't sound good, I'd look elsewhere.

[gritz]

I wouldn't worry too much about it.

A particular red led might start flowing enough current at e.g. 1.5V or 1.6V to start making itself felt and heard in a circuit, but the effect will depend more on other values within the circuit than on the led itself. From there any led will exhibit a vaguely logarithmic response of forward voltage with respect to current - i.e. if a tenfold current increase from 50 microamps to 500 micromps might cause a jump in forward voltage from 1.6V to 1.7V, then a further tenfold current increase will get you to 1.8 volts. In reality there is a little internal resistance in series with the led that makes Vf increase a little more than this pure log approximation, but the point I'm trying to make is that all of your red leds behave in a very similar manner - there is no "magic" led that bends sound in a unique way.

I think it's telling that although folks all over the internet will tend to argue the merits of their favourite component, there is no flamewar on this subject - and neither is there any argument about whether e.g. red sounds better than green (green has a very slightly higher forward voltage, but obeys the same principles), etc. With any clipping distortion there are other (possibly far more influencial) things to think about, like the frequency response of the circuit and the symmetry of the clipping applied to the waveform.

[Goodrat]

Yes, I was going to add that the tolerance is a factor and the Vf will vary even with the same part number. But it is a starting point. It wouldn't hurt to mention part numbers.
I have red LED's (different part #'s) for example, with a Vf of 1.6 and 2.0. They will sound quite different and for good reason. But no matter how many you buy, one  will always have a lower Vf than the other. One may have a tighter tolerance than the other for typical Vf and minimum. One may always start to conduct earlier or have a certain non-linearity at the low current end. We  spend time getting the frequency response and gain just right, we just can't throw in a part due to it's color. That would be like saying, make R5 a low resistor, brown ones usually sound nice.
I should mention that I'm making a clone. I don't want to make changes to gain or response.

[Mark Hammer]

In principle, your goal is appropriate.  I think the problem is that the universe of "red LEDs" is so vast, and often under-described, as to make it tricky for folks to aim for a specific LED from a specific maker. Moreover, there are more than enough units from manufacturers and distributors we've never heard of that focussing in on a few with known characteristics cuts some folks off from something they actually have in hand, even if it is from a busted toy.

Instead, "red" is a starting point that either leads to measured forward voltages in the desired range...or not.

At the very least, people should realize that the LED does not have to light...or NOT light...for it to function as a clipping element.

[Gus]

Just do a search for LEDs
https://en.wikipedia.org/wiki/Light-emitting_diode    look at the color and materials section

IIRC Craig Anderson wrote about LED clipping diodes in an older guitar player article.  IIRC one point was about the junction capacitance.

So not only does the color matter but the size of the LED(not the package) as well, think of the pf value cap you will often see in the feedback path. This can matter with the selection of the feedback resistor values

[Goodrat]

Yes, one design I was working on had LED's in the feedback and It drove me nuts wondering why I was not getting the calculated upper frequency response.
But then I put the led on a cap meter and I was enlightened. The 5mm I was using had I think 100pf.
Better to use 3mm in a feedback which seem to be around 20pf, unless intentional capacitance is needed.
Here is the project I'm working on at the bottom of this page:
http://rickviola.com/electronics.htm
Mark, it originally had the wstc tone, but something sounded a little different. So I opted to short the two leads of the tone control. Maybe it was the pot I used at time i was testing and full clockwise wasn't quite zero ohms.

[gritz]

This topic piqued my interest - I occasionally use clipping LEDs, but for amplitude limiting, rather than as an artistic effect. I'd never noticed any significant issues introduced by the inclusion of those LEDs into a particular circuit, so I shuffled off to the bench to see if I'd missed anything...

I decided that rather than just measuring the capacitance of various LEDs I would build something resembling a Tubecreamer clipping stage and measure it's frequency / amplitude response. I figured that to accurately assess capacitive effects I would have to keep signal levels below the level at which the LEDs would conduct.

I use two types of LEDs for clipping (because I have stacks of each of them!) - a red 3mm low current type similar to Boss power indicator LEDs and a green rectangular low output type. Subbing pairs of both into the circuit it was clear that both types were low in capacitance - a pair of reds was worth about 10pF and the greens about 15 pF. The effect was measurable, but hardly significant (the TS feedback cap is 51pF but I was using a 47 'cos that's what I had). I wasn't surprised that I'd never noticed an affect from my clipping LEDs, which would normally find themselves in rather lower impedance setting anyway.

I had a scout around for pairs of different LEDs (it's not something that I use a lot of) and I found a couple of tricolour high brightness types. These are just a separate red and green die in the same package, sharing a common cathode lead. A pair of the reds back to back was equivalent to 50pF and the greens were worth 160pF. This was obviously a very significant amount of capacitance in this application. The effect on the TS circuit also tallied with the capacitance figure reported by my meter - i.e. it's nicely predictable. I didn't notice any measurable effect from reverse leakage current.

So, how do you pick low capacitance LED's? Short of measuring them with a capacitance meter, or testing them in circuit it's probably not easy. Package size is no indicator - a manufacturer's 3mm, 5mm etc. LEDs of a particular electrical / optical spec. may well contain the same die. I don't know enough about LED construction (i.e. I don't know anything...) to speculate about bandgap widths, die areas and all...

It is possible to mitigate the effect by putting a junction diode in series with each LED, but at the cost of extra forward voltage. Capacitors in series!

[samhay]

This topic piqued my interest - I occasionally use clipping LEDs, but for amplitude limiting, rather than as an artistic effect. I'd never noticed any significant issues introduced by the inclusion of those LEDs into a particular circuit, so I shuffled off to the bench to see if I'd missed anything...

I decided that rather than just measuring the capacitance of various LEDs I would build something resembling a Tubecreamer clipping stage and measure it's frequency / amplitude response. I figured that to accurately assess capacitive effects I would have to keep signal levels below the level at which the LEDs would conduct.

I use two types of LEDs for clipping (because I have stacks of each of them!) - a red 3mm low current type similar to Boss power indicator LEDs and a green rectangular low output type. Subbing pairs of both into the circuit it was clear that both types were low in capacitance - a pair of reds was worth about 10pF and the greens about 15 pF. The effect was measurable, but hardly significant (the TS feedback cap is 51pF but I was using a 47 'cos that's what I had). I wasn't surprised that I'd never noticed an affect from my clipping LEDs, which would normally find themselves in rather lower impedance setting anyway.

I had a scout around for pairs of different LEDs (it's not something that I use a lot of) and I found a couple of tricolour high brightness types. These are just a separate red and green die in the same package, sharing a common cathode lead. A pair of the reds back to back was equivalent to 50pF and the greens were worth 160pF. This was obviously a very significant amount of capacitance in this application. The effect on the TS circuit also tallied with the capacitance figure reported by my meter - i.e. it's nicely predictable. I didn't notice any measurable effect from reverse leakage current.

So, how do you pick low capacitance LED's? Short of measuring them with a capacitance meter, or testing them in circuit it's probably not easy. Package size is no indicator - a manufacturer's 3mm, 5mm etc. LEDs of a particular electrical / optical spec. may well contain the same die. I don't know enough about LED construction (i.e. I don't know anything...) to speculate about bandgap widths, die areas and all...

It is possible to mitigate the effect by putting a junction diode in series with each LED, but at the cost of extra forward voltage. Capacitors in series!

Nice work.
One way to mitigate against this is to reduce the reistance of the feedback network - if you use a 50k c.f. 500k gain pot and then you would need a 500pf cap to get the same treble roll-off... You would also need to reduce the any other relevant resistors by 10* to maintain the same gain. As a bonus, you also get lower thermal noise.

[R.G.]

I decided that rather than just measuring the capacitance of various LEDs I would build something resembling a Tubecreamer clipping stage and measure it's frequency / amplitude response. I figured that to accurately assess capacitive effects I would have to keep signal levels below the level at which the LEDs would conduct.
...
The effect on the TS circuit also tallied with the capacitance figure reported by my meter - i.e. it's nicely predictable. I didn't notice any measurable effect from reverse leakage current.
...
So, how do you pick low capacitance LED's? Short of measuring them with a capacitance meter, or testing them in circuit it's probably not easy. Package size is no indicator - a manufacturer's 3mm, 5mm etc. LEDs of a particular electrical / optical spec. may well contain the same die.

It's worse than that. The capacitance of the reverse-biased junction varies with the voltage applied to it.  Some diodes are specially set up to enhance this effect, and used for voltage-tuning RF stages.

[R.G.]

It just hit me as I clicked "submit" that there's another comment to be made.

Philosophically, one should assume that everything varies - constants aren't and variables do.

What's important is control. That which cannot be controlled must be made irrelevant.

I guess that is what happens in politics as well as electronics. 

[gritz]

What's important is control. That which cannot be controlled must be made irrelevant.

I guess that is what happens in politics as well as electronics. 

Indeed! Or, if you can't control or even predict it then just call it mojo...

Samhay also makes a good point about designing around the problem too, but with classic designs some values may be non-negotiable for some people. Attempting to scale to a necessary degree may introduce it's own problems also, although being mindful of thermal noise and stray whatnot I try to stick to sensible resistance values...

@ R.G. - I did think of the varicap thing and did even wonder if it could be exploited in some way for audio if I could rustle up enough capacitance and mitigate modulation via changing signal level. I remember building an FM bug when I was a kid that relied on the variable Miller capacitance of the reverse biassed base / collector junction for modulation. I think. It was a long time ago. I know there's a square-root relationship between junction voltage and capacitance, but if your clipper has two similar leds back to back the change in capacitance with applied voltages cancels *to a degree*. If not, then it might get interesting...

Anyway, I notice that Kingbright specify capacitance on their datasheets (even if no-one else does).

[Goodrat]

I ordered a 3mm dual Red/Yellow LED (Bivar 3BC-R/Y) with a Vf  of 2.0 in both directions. This would be very handy and should sound good. I'll report back soon.
Interesting article here about LED light:
http://www.cliftonlaboratories.com/led_vf_vs_if.htm

[Goodrat]

I got the red/yellow led. Interesting results. Although the Vf spec is 2.0/2.0 @20mA, in-circuit I'm getting 1.72 (red)/1.84 (yellow) at whatever current a crunch box is feeding them, so I am a little non symmetrical.
Which seems to be a good thing. I get nice harmonics, but good clarity. I may go with this plus it saves space.
BTW, without LED's, when ever I build these, I always get more clipping on the negative cycle if I turn up the input on a signal gen and drive it to the rails.
If I change the bottom resistor of the 100K pair of the Vb divider to 120K, I hit the rails the same. Does anyone know why? Because I usually change the bottom resistor to 120K.

[gritz]

I got the red/yellow led. Interesting results. Although the Vf spec is 2.0/2.0 @20mA, in-circuit I'm getting 1.72 (red)/1.84 (yellow) at whatever current a crunch box is feeding them, so I am a little non symmetrical.
Which seems to be a good thing. I get nice harmonics, but good clarity. I may go with this plus it saves space.
BTW, without LED's, when ever I build these, I always get more clipping on the negative cycle if I turn up the input on a signal gen and drive it to the rails.
If I change the bottom resistor of the 100K pair of the Vb divider to 120K, I hit the rails the same. Does anyone know why? Because I usually change the bottom resistor to 120K.

Hello Goodrat - you've found that opamps (cmmon-or-garden bipolar output ones at least) can't swing quite as close to the negative rail as the positive rail, so they tend to clip a bit asymetrically.

[Goodrat]

So, bad idea, or good idea to make the bottom resistor 120K (top one 100k)? Because, then I get symmetrical clipping, although maybe it will not be driven so hard anyway.

[Goodrat]

I think I will do what I always do with 100K's for the bias voltage and make the bottom resistor 120K. This enables both peaks to always clip at the same time.
It sounds a little smoother too. When the LED's start to clip, the result is symmetrical.

[gritz]

There was a thread a few weeks ago regarding this and the upshot (including some tests by me) was that some LEDs have significant capacitance that can affect the higher frequency output of a typical clipping circuit at low signal levels (i.e. before clipping sets in). This can be a good or a bad thing depending on what you want your circuit to sound like.

With the limited number of LEDs I had to test I found that some greens tended to exhibit highest capacitance and in a Tubescreamer (for instance) stage they might at least double the 51pF capacitance that sits in the opamp feedback path. This might give rise to a noticeable loss of treble at low guitar levels. In contrast, low current red LEDs (similar to the ones used in Boss pedals as status indicators) exhibited only a very small (~15pF) capacitance and had very little effect. Bear in mind that as soon as the clipping LEDs start to conduct this capacitance is effectively shorted.

It's worth listening out for when experimenting though!

[Gus]

Diodes that might be interesting to try
http://en.wikipedia.org/wiki/Varicap

[mistahead]

There was a thread a few weeks ago regarding this and the upshot (including some tests by me) was that some LEDs have significant capacitance that can affect the higher frequency output of a typical clipping circuit at low signal levels (i.e. before clipping sets in). This can be a good or a bad thing depending on what you want your circuit to sound like.

With the limited number of LEDs I had to test I found that some greens tended to exhibit highest capacitance and in a Tubescreamer (for instance) stage they might at least double the 51pF capacitance that sits in the opamp feedback path. This might give rise to a noticeable loss of treble at low guitar levels. In contrast, low current red LEDs (similar to the ones used in Boss pedals as status indicators) exhibited only a very small (~15pF) capacitance and had very little effect. Bear in mind that as soon as the clipping LEDs start to conduct this capacitance is effectively shorted.

It's worth listening out for when experimenting though!

This is interesting - but I have a newbie question - is that "flat" capacitance that is present due to the LED being used or is it related and variable to the state/operation of the LED, does it shift as the led is driven to clip I suppose I'm asking?