What makes clipping diodes "sound" different?

[Fancy Lime]

Hi there,

I've been experimenting with different clipping diode arrangements. What has bugged me before and did so again: Different Si diodes with very similar voltage drops sound different (compared in the same exact arrangement). 1N4001's sound different from 1N4148's. The body diodes BS170's sound different again, as do the B-E or B-C diodes of 2N3904's or the non-zener diodes of zener diodes (you know what I mean... what's that called actually?). At least they seem to sound different to me. A little. I think. But knowing how good the human brain is at hearing what it wants and at not giving a flying frog about what the ears have to say about it, I ask myself: Am I crazy? Are the knees actually different? Does the open-resistance make that much of a difference in common hard or soft clipping arrangements? Is it the capacitance of the diodes that makes the difference? You guys can hear it too, right? Or do I just need to get more sleep?

Disclaimer: My comparison method is far from scientific. I just swapped diodes or connections on the breadboard. But I thought I'd ask if a full on blind test is even worth the trouble.

Thanks,
Andy

[iainpunk]

i am really curious as well, it probably has something to do with the ratio magic smoke to black plastic... and maybe the physical distance between the ends of the leads

[willienillie]

I'm sure I've heard subtle differences between similar Si clipping diodes.  To really know, it's best to be able to switch between them quickly, in the time it takes to swap a pair on the breadboard, your head (ears) will have moved, etc etc.  You can use an actual switch obviously, or you can set up a number of pairs on the breadboard and make all but one connection, and just move one jumper wire from pair to pair.  That's how I selected the Ge clippers in my Klone, some really noticeable differences with those.

When we measure Vf with our DMMs, it's at a specific current, which I just assume is something quite different (larger?) than what the diodes will experience in the clipping circuit.  So maybe types that measure about the same are farther apart elsewhere on their curves.  If that makes any sense.  I also suspect other factors like capacitance come into play.  If you want to hear something very different, try to find some of those metal can power diodes that came in 70's Fender amps (I forget the actual type).  At some point while experimenting, I tacked one of those in parallel with the standard clippers in a Tube Screamer, and it changed the tone of the pedal quite a bit, rolled off highs like a cap.

[Rob Strand]

In reply #8  I posted a heap of normalized measurements,
https://www.diystompboxes.com/smfforum/index.php?topic=121964.0

A very simple way to see softness is from the exponential diode equation,

Id  = Is exp( Vd / (n Vt)  )

Where,
Id = diode current,
Vd = diode voltage,
Is is a constant which contains information about the size and material of the diode,
n is the ideality factor of the diode
Vt = k T / q     ; k = Boltzman's constant, q = charge on electron, T = temperature in Kelvin

The ideality factor is what makes one diode softer or harder than another.  Typically the value lies between 1 and 2.

If you differentiate the diode equation you will get the slope of the diode curve,

         rd  =  n Vt  / Id

The way to interpret rd is if the diode is currently at a point on curve with V=Vd and I = Id  and you change the current by a small amount  "Ichange"  then the change in the voltage is "Vchange",

      Vchange = rd * Ichange

So if you have two diodes both at the same Vd and Id but one diode has a larger n than the other the diode with the larger n will show a larger change in voltage than the smaller n.   If you think about hard clipping as no change in voltage for a change in diode current then the diode with the smaller n is "more clipped" as the output changes less for a given change in current.

The small n diodes have a sharper knee.    For a clipper, the sharper knee means it will be cleaner for small signals as the diode has not kicked in whereas a softer knee starts to have an effect earlier, on albeit a small effect, which adds a bit more hair to the sound.

Transistors, Body diodes of MOSFETs, *forward* biased zeners tend to be around n=1.    Silicon signal diodes and silicon power diodes tend will be anything from just above 1 to 2 it varies.

[antonis]

The way to interpret rd is if the diode is currently at a point on curve with V=Vd and I = Id  and you change the current by a small amount  "Ichange"  then the change in the voltage is "Vchange",
Vchange = rd * Ichange 

So, we're actually talking about diode transconductance (1/rd = slope of ΔI/ΔV..)
(never faced a diode concernig this parameter before..)

[Mark Hammer]

As I understand it, the behaviour of some diodes is a function of the current in the signal reaching them.  An interesting experiment to try would be to use one of the many designs employing a diode pair to ground with a fixed resistor leading up to the diodes, such as the Proco Rat, DOD250, MXR Distortion+, and listen to what changes to that series resistance does to the diodes' function.

[Fancy Lime]

As I understand it, the behaviour of some diodes is a function of the current in the signal reaching them.  An interesting experiment to try would be to use one of the many designs employing a diode pair to ground with a fixed resistor leading up to the diodes, such as the Proco Rat, DOD250, MXR Distortion+, and listen to what changes to that series resistance does to the diodes' function.

Funny you should say that. I was thinking the same thing. Someone must have tried that, no? It should act as a "knee softness" control. Should only make a difference with diodes that have an inherently soft knee. I would have already breadboarded it, if I had any Ge diodes at hand.

Andy

[amz-fx]

Someone must have tried that, no? It should act as a "knee softness" control.

http://www.muzique.com/news/changing-diode-vf/

regards, Jack

[Fancy Lime]

Someone must have tried that, no? It should act as a "knee softness" control.

http://www.muzique.com/news/changing-diode-vf/

regards, Jack

Thanks, Jack! I could have guessed *who* must have tried that already 

[amptramp]

A transistor with base connected to collector has a much sharper knee than a B-E junction of a transistor since the current through the B-E junction is multiplied by the hfe of the transistor.  We used to use that on occasion where a sharp knee was required.

[PRR]

> talking about diode transconductance (1/rd = slope of ΔI/ΔV..) (never faced a diode concernig this parameter before..)

Trans-conductance is short for Transfer Conductance. Voltage on some pair of pins affect conductance on some other pair of pins. V(b-e) affects I(c-e).

In a diode this is just Dynamic Resistance.

And this *tends* to follow a Standard Relation, as Rob explained in #3 above. There are varying "constants" on different diodes but these can mostly be "corrected" by adjustment of Gain before and after.