what clips harder than an LED ?

[Rectangular]

I've been playing around with some turbo rat designs, and have been trying out all the diodes in my weird collection. so far LEDs are the harshest thing I've got. I tried a lot of fancy germanium diodes, but the clipping is really soft, not my thing at all. the best runner-up to an LED seems to  be these weird "two pin transistor" (they look like a can package transistor with two pins)  looking diodes I scrapped from somewhere (maybe a radio) , they're ok.

does anything clip harder than an LED, or is that beside the point ?  will everyone tell me to just cascade multiple leds in series to achieve a harsher sound ?  are a bunch of  1n4148s in series the exact same as an LED, as far as clipping characteristics are concerned  ?

[Ripthorn]

You could always try either a schottky barrier diode or you could try a low voltage zener diode (I'm not sure if zeners will work, but they definitely come in higher reverse breakdown voltages).  I don't do a lot of diode clipping stuff, so that is all I got.

[alanlan]

Have you tried different colour LED's?  They all have differing forward voltages.

[Rectangular]

my understanding was that red leds were the hardest of the led family. I thought it was like green, yellow, red

if I had some blues I could test those.  has anyone ever tried a super bright led ? I imagine the higher current drain does...something

[Rectangular]

I completely forgot about schottkys . I've never actually used them before, I dont really understand where they're important, or I've never had a circuit that called for them specifically.

[R.G.]

A high feedback circuit like an opamp is about the hardest clipping device, when overdriven.

[Guitar_Ninja]

A lawnmower!!!

[amptramp]

Following what R. G. said, there are active diode and active clamp circuits shown in the National Linear Applications handbool as AN31.  Check out:

http://www.national.com/an/AN/AN-31.pdf

on page 17.

[David]

I completely forgot about schottkys . I've never actually used them before, I dont really understand where they're important, or I've never had a circuit that called for them specifically.

How are you doing your clipping?  Are the diodes in the feedback loop of the op-amp, or are they bridged between the op-amp output and ground?  The latter is what I've been led to believe is hard clipping.  Of course, as R.G. said, if you raise the gain of the op-amp much past 100 or so, clipping is pretty much a given, and the diodes will matter a lot less...

[jimbeaux]

Jack Orman's website has a few schematics that may help.

Using Mosfets and Zeners
http://www.muzique.com/lab/zenmos.htm

2-Stage Distortion using Zener Diodes
http://www.muzique.com/schem/louis.gif

and the MOSFace
http://www.muzique.com/schem/fuzzface4.gif

[edvard]

I always thought that popular opinion said LEDs clipped softer because the higher forward voltage meant it actually clipped less of the signal. 

That said, the wisdom behind using Schottky diodes can be found from a quick perusal of Wikipedia

The most important difference between p-n and Schottky diode is reverse recovery time, when the diode switches from non-conducting to conducting state and vice versa. Where in a p-n diode the reverse recovery time can be in the order of hundreds of nanoseconds and less than 100 ns for fast diodes, Schottky diodes do not have a recovery time, as there is nothing to recover from. The switching time is ~100 ps for the small signal diodes, and up to tens of nanoseconds for special high-capacity power diodes. With p-n junction switching, there is also a reverse recovery current, which in high-power semiconductors brings increased EMI noise. With Schottky diodes switching instantly with only slight capacitive loading, this is much less of a concern.

So, lower noise and much quicker conductance recovery (read: 'harder' clipping) as well as a larger range of forward voltages to choose from.
How's them apples?

[Mark Hammer]

I think there are about 20 different ideas being thrown together like Colonel Sanders secret herbs and spices mixture here.

Everything clips "hard" if you feed it a hot enough signal, and everything clips "soft" if you feed it a wimpy enough signal.

But, assuming that the signal is of an amplitude high enough to be supra-threshold for that particular clipping device, what some folks mean by "hard" is that the onset of conduction takes negligible time.  In other words, there is no rounding of the corners of the resulting waveform because the diode went into conduction instantaneously and not gradually as the signal approached the threshold of conduction.

Now, there is what shows up on a scope - "yup, there it is, I see it now" - and there is what is audible.  Is the amount of time it takes for a diode of any type to go into conduction, that slow that it has an impact on audible bandwidth?  I don't know, and eagerly await other more informed opinions.  What I will say, though, is that if the turn-on/turn-off time is on the order of nanoseconds (let's say <100, to be safe), and if diode is being used in a circuit that trims off treble to avoid fizziness, in conjunction with an amp that uses 12" speakers intended to provide a "creamy" tone, I am not at all sure that any measurable diode turn-on differences of the sort I describe would be audible under anything other than clinical studio conditions with the sort of listening circumstances that rarely present and the sort of equipment that few can afford.

[jacobyjd]

also--what is 'hard' clipping?

I would guess that the hardest clipping I can think of is a square wave fuzz, but it doesn't sound like that's what you're after.

It sounds like you're looking for something that sounds harsh, which would be a different direction entirely apart from hard/soft clipping. Maybe you need to try some asymmetrical options.

[Paul Marossy]

Now, there is what shows up on a scope - "yup, there it is, I see it now" - and there is what is audible.  Is the amount of time it takes for a diode of any type to go into conduction, that slow that it has an impact on audible bandwidth?  I don't know, and eagerly await other more informed opinions.  

I can say from my experience that just because you can see it on a scope doesn't mean that you can hear it. I've looked at various circuits with my scope - compared waveforms from different opamps, etc. and most of the time, I could not honestly hear a definitive difference between the different whatevers. Any differences that I did hear were usually so subtle that I started second guessing myself as to whether or not I was really hearing any difference(s) at all.

[Rectangular]

josh, that's kind of what I'm going for. maybe I forgot to distinguish between the hard clipping from an engineer's perspective, and "harsh"-sounding clipping, which is a subjective thing

for everyone that's responded, I'm talking specifically about my Turbo Rat circuit, obviously there are completely different circuits and circuit arrangements that will sound a lot louder/harsher/etc. the turbo rat does not use diodes in the opamp feedback path, they come after the opamp, and before the output FET

mainly I'm asking because when I flip through all my different clipping diode arrangements, the RED LEDs are the loudest and "harshest" of all the diodes I've tried. when I use germaniums or 1n4148s, the audio actually sounds muffled, the output volume is audibly different, when compared to the LEDs.

I may look into some asymmetrical clipping arrangements, that might get me somewhere

also, lm307s are awesome in this circuit. the gain seems to be even higher, and the treble is much more pronounced.

[JDoyle]

mainly I'm asking because when I flip through all my different clipping diode arrangements, the RED LEDs are the loudest and "harshest" of all the diodes I've tried. when I use germaniums or 1n4148s, the audio actually sounds muffled, the output volume is audibly different, when compared to the LEDs.

The LEDs are louder because they allow more signal out of the opamp to get through unclipped before they turn on. Ge diodes clip at around 0.2 to 0.3V, Si at 0.5 to 0.7 and LEDs anywhere from 1V to 2.8V.

So, the total output swing of the signal will be 0.4V to 0.6V peak to peak (p-p), for Si 1V to 1.4V p-p and for the LEDs 2V to 5.6V p-p.

As you can see there is a significant difference and that is the difference in loudness that you hear - keep in mind that due to the way we hear, to get something to sound twice as loud, one needs to change the signal size by a factor of ten, not two.

In terms of the 'harshness' of the sound, Mark Hammer has written several great posts about what he calls 'proximity to clip' which is basically: how much larger is the the size of the signal than the threshold voltage of the diode? The larger the signal, the more the diode simply slams from 'off' to 'on' - a 'harsh' sound. As the signal gets closer to being equal in size (for each polarity, by the way) to the diodes threshhold, the more the signal is 'rounded' rather than clipped.

Another factor is the curvature of the transfer function of the diode itself; i.e. the amount of voltage it takes to cause the diode to go from 'off' to 'on'. Ge diodes are pretty much impossible to turn all the way on, at least with the signals and currents we use in FX, whereas Si and LEDs go from off to on very rapidly, producing a harsher sound as they 'clip' more than 'squash' like Ge diodes. Keep in mind 'proximity to clip' - the size of the signal INTO the diodes is probably the most important factor.

One way to think about it is this: Consider a voltage divider, the 'upper' resistor is the impedance of the signal and the 'lower' resistor is the diode. Until the threshold is reached, the diode is effectively an open circuit having a resistance of many megaohms - therefore the 'lower' resistor of the divider has no effect on the signal and Vin=Vout. BUT, when the signal is large enough to turn the diode on, the diodes resistance drops from that 'open circuit' many megaohm range, to that of just a few ohms, and thus the voltage divider action is severe - almost all of the signal is sent to ground. Yet - between the many megaohm/open circuit condition, and the few ohms/short circuit condition is a very small range where the effective resistance is between those two extremes - this is the 'knee' or curve of the diode. In that range the divider action is dependant upon the signal voltage across the diode - the larger the ratio of those two things, the quicker the signal traverses the 'knee' and the harsher the sound.

Regards,

Jay Doyle

[George Giblet]

opamps

[slideman82]

You could try pairs of diodes in series... or, why don't you try the diode arrangement present in Marshall's Silver Jubilee preamp?

I think the feedbak cap is too small, use a 220p cap instead of the stock 100p one. Probably will reduce harshness

[amptramp]

While we are talking about LED's, be aware of one thing: they also act as efficient (but leaky) photodiodes.  So if you are planning on putting them where you can see them (so you can see where clipping starts), be aware that any ambient light may change the clipping level.  If you have daylight on them, not much effect other than higher reverse leakage.  But if you have AC lighting on them, expect some hum due to modulation of the reverse-biased diode.

I used to be at a company that manufactured photodiode-based touch screens and we found that LED's worked as photodiodes.  Therefore, any circuit application where they are reverse-biased requires them to be kept away from ambient light.

[JDoyle]

Forgot to mention you may want to try regular bipolar transistors wired as diodes - if you tie the collector to the base the diode formed is closer to the theorhetical 'ideal' than two pin diodes.