Multiclipper!

[mictester]

As an experiment, in a recording studio recently, I had a box with me that allowed switching between different types of clipper. The basic circuit was a low noise Tubescreamer type (circuit below), with the option of differing diodes, MOSFETs, transistors and LEDs for clipping. The clippers could either be connected to the inverting input of the first op-amp or to the mid-rail (which is AC ground because of the large electrolytic from the midpoint of the bias potential divider).

I had several musicians and engineers "audition" the effect, and we did "blind" testing of the different means of clipping. In each case, the levels were carefully adjusted to remain the same, so that "loudness" wouldn't be a factor in their comments about the sounds. We found that there were three discernible types of clipping, and all the rest sounded much the same as one of these three:

1 Diodes in the feedback loop of the op-amp (like a basic Tubescreamer)

2 Diodes to ground after the op-amp

3 Asymmetrical diodes in the feedback loop (very marginal difference to the first).

It didn't really matter what kind of clipper you used!!!! In most cases the only differences are the point at which distortion starts - germanium diodes clip at 0V2 p-p so will seem more "sensitive" than silicon one which clip at 0V65 p-p. LEDs are even less "sensitive" because the take even higher voltage to conduct. There was a slight difference audible with asymmetrical clipping (though, on average, only three out of ten listeners could hear the difference!).

After the experiments, I was really quite disappointed - I felt sure that there would be a noticeable difference in sound between differing types of clippers (not just the point at which clipping starts, which is what I found). The most important part of the over all sound of the effect is the frequency response before and after the clipper stage (a low pass filter before the clipper makes your effect really smooth), and the best we can really do is to put a really effective tone control circuit after the clippers. We also need to keep the noise down!

Multiclipper.png (7.61 KiB) Viewed 1725 times

(Sorry about the pencil scan from my workbench!)

It proves (at least to my satisfaction and to the satisfaction of several other music professionals) that 95% of the claims for the "sound" of all these dirt boxes is nonsense. You can build a generic box, and by selection of just a couple of components, you can emulate a huge proportion of the expensive boutique rubbish out there!

I plan to repeat the experiment with a set of "Big Muffs", "Fuzz Faces" and "Tonebenders", and see how much variation people find!

[MoonWatcher]

I was really quite disappointed - I felt sure that there would be a noticeable difference in sound between differing types of clippers (not just the point at which clipping starts, which is what I found).

I was actually a bit "happy" when I determined something similar. It now allows me to decide on the cumulative forward voltage of diodes that I chain in series, to get that onset of clipping within a certain point on the drive knob. It's nice to be able to concentrate on other things instead of specific diode types, unless I just happen to insist on a singular pair that have a forward voltage that I prefer (i.e. laziness).

As a very generic rule of thumb, I tend to prefer forward voltages somewhere in the range of 850mV to 1100mV, either to ground or in a loop. But that is probably just personal preference. The subtle difference between asym and sym is a bit lost on me personally, for all but a few designs. I find that it tends to be masked at the amplifier at "nominal" volumes.

One test that was slightly revealing some years ago was to use a bridge rectifier for clippers. It was clunky, but allowed for quick switching. I would orient it to get either two series pairs, or simply short out a pair for the "standard" clipping setup. I was using a DPDT to short out a pair of diodes. One day, I inadvertently forgot to connect the shorting point at one "end," and ended up with just a crossover short at the series junction. To my ears, the results were slightly more subtle and usable than the asym. setups that are so popular.

...I have since replaced the bridge with singular diodes for cost/space reasons. It's only real limitation (to me) is that I do not like the audible differences in clipping when mixing different diode types to get specific forward voltages. My perceived result of it is that there are unpleasant clipping characteristics with the crossover short in place when mixing diodes - "traditional" asym. actually sounds preferable in this instance. I need to get it on a scope and see what is going on, to determine if it's worth pursuing any further to get mixed diodes to work.

I did this some years ago with a big box with full sized pots, before it was fashionable to build stuff in tiny enclosures. It was clunky looking, but the results were what mattered. I also would tailor the ranges of the controls around whatever I had in the way of pots, which were typically those used for guitar - 250K log, 500K log, etc. I actually prefer 250K, since I would typically yank those from my Fender guitars.

It dawned on me at some point that Paul C.'s design also uses this "crossover short" for his diode setup, so he must hear something beneficial as well. I like that it works with different diode types, as long as all 4 are the same. The next step is to use 3 series pairs with two crossover shorts. I will probably start with BAT42's for the lowest cumulative forward voltages, and move up to 1N34A's/BAT41's, and then possibly to something with a singular forward voltage just a touch higher than that. Probably won't reveal much, but should be interesting just the same.

[MoonWatcher]

...ended up with just a crossover short at the series junction. To my ears, the results were slightly more subtle and usable than the asym. setups that are so popular.

Edit: meant to say that the results were slightly more discernible and usable than the asym. setups that are so popular.

I also wanted to address asym. clipping just a bit. I personally do not find that mixing diodes that create "big" ratios of asymmetry to work well, including the ever so popular singular 1N419 in parallel with a pair of the same. I mentioned my preference for 850mV vs. 1100mV. So in many cases, it might be worth trying an asym. ratio of 75% to 80% paired with the "100%" side. Or 100%/130% if we flip it around...

This is probably why I have never cared for the pairing of 1N419/LED. If the 1600mV LED could be paired with something that gives about 1200mV or so, it might sound fine.

Again, these are my personal preferences. But they mainly center around at what point the clippers play a role as opposed to specific types, and any "knee" or such that they might have. Such subtleties might be great for the bedroom player, or the guy with a Dual Showman with D130F's in it. Or someone who goes D.I. into a recording rig. Otherwise, the output transformer and speaker(s) are already contributing enough asymmetry on their own that it doesn't need to occur at the clipping diodes.

[cpm]

i think i reckon that "crossover short" thing. It's subtle but seemes like i could notice some different sound from a plain series diodes.

Sometimes I like a strong asymetric clipping, i usually put 2 mosfet in series (in opposite orientation). This would have a quite high clip thershold (symmetric), and a switch shorting one of them gives a strong asymmetric that i like quite well, and then having two clearly different sounds within a switch.

[MoonWatcher]

i think i reckon that "crossover short" thing. It's subtle but seemes like i could notice some different sound from a plain series diodes.

Maybe Paul C. or mictester can elaborate on what is going on when there is this short just in between the series pair.

I've never put it on a scope, and I'm waiting to get mine back from a friend.

I've just never really bothered to try and understand what is specifically happening. I like how it works, so I never went beyond that point.

[Hides-His-Eyes]

i think i reckon that "crossover short" thing. It's subtle but seemes like i could notice some different sound from a plain series diodes.

Maybe Paul C. or mictester can elaborate on what is going on when there is this short just in between the series pair.

I've never put it on a scope, and I'm waiting to get mine back from a friend.

I've just never really bothered to try and understand what is specifically happening. I like how it works, so I never went beyond that point.

I've always been confused by this, and have asked on this board before to no avail.

If we label the diodes

Code: Select all

   A - B
  -| x |-
   C - D

say that the top row is pointing right and the bottom row pointing left.

then look at the situation in the wave where the gain is at full (100?), the input is at 30mv, so we have 3V across the clippers. 3-0.9 through diode D gives us 2.1V in the middle, and C and B see this at their anode.

B has 2.1V at the anode and 3V at the cathode, so doesn't conduct. C sees 2.1V at the cathode and 0v at the anode (All relative to vref of course) and so clips as well.

I can't see where joining the diodes in the middle would make a difference. Either I can't see it, or it's some non-ideal aspect of the diode I don't know about.

[earthtonesaudio]

Mictester:
Shouldn't there be a current limiting resistor between the op-amp and the diodes when they're shunting signal to ground? If you did your testing without one, I'm not surprised the different diodes sounded the same, but I am surprised you didn't let out any of the magic smoke!

[mictester]

Mictester:
Shouldn't there be a current limiting resistor between the op-amp and the diodes when they're shunting signal to ground? If you did your testing without one, I'm not surprised the different diodes sounded the same, but I am surprised you didn't let out any of the magic smoke!

There IS one - a 1k! I missed it out of the sloppy scribble in the notebook. See the corrected version below:

Multiclipper.png (7.75 KiB) Viewed 1608 times

Sorry about that!

[earthtonesaudio]

I figured it was just a typo. I bet that circuit would be quite the useful widget in a small studio.

[mictester]

I figured it was just a typo. I bet that circuit would be quite the useful widget in a small studio.

It's quite a useful dirt box. It's interesting to feed it from an equaliser and follow it with another equaliser - you can easily emulate EVERY boutique TS-alike!

[stringsthings]

interesting!

any specific reason for using the LM833 chip?

[mictester]

interesting!

any specific reason for using the LM833 chip?

Yes - two good reasons. One was that they're pretty low noise, and the second was that they have useful current drive capability. The main reason was that I have a boxful of them!

You can use pretty much any op-amps you like. The circuit works fine with TL072, NE5532, LF353 and 4558 (tried all those myself). There's little change from IC to IC, but you might prefer the "sound" of one over another. Use an IC socket, and try several!

[Gerry de la Sel]

We also need to keep the noise down!

There is a heavy noise penalty in having the volume pot after the second opamp. You could eliminate this by putting the pot before the second opamp, or better still, arrange the second opamp as a non-inverting stage and use a pot to vary the feedback. That has the added bonus that the effect would be non-inverting overall!