Diode switching

[Willthebold]

I was wondering what's a good way to switch diodes in a circuit like the Tube Screamer.  What I want to do is be able to switch the third diode in and out to go from symmetrical to asymmetrical clipping.  Any ideas?

Will

[Mark Hammer]

I've used two methods, and both work just fine.

1) Shunting any diode (i.e., adding a piece of wire in parallel) renders that diodes essentially invisible or nonexistent to the circuit.  You can easily do this with a single SPST switch to "convert" a 2+1 SD-1 type diode complement to a 1+1 TS-9 diode complement.  If you want to get fancy, get a 3-position DPDT on-off-on toggle that you can use to toggle sets of diodes.

2) A pot in series with a single diode reduces the clipping impact of that diode.  I made a TS-808 clone with a single 50k (or 100k, I forget) pot, wired as variable resistor in series with one of the diodes.  Turn it to zero ohms and its a stock TS clipper.  Set it to around 11:00 and its an SD-1.  Turn it higher and its a different pedal.

Diode switching is an audible, but not in-your-face sort of change.  In that regard, neither of these produces a less effective version of that change.  The choice to use switches or pots or both is up to you, your creator, the amount of money you're willing to shell out for knobs and pots, how much panel space you have to work with, and what matters to you musically.  I urge you to read the doc on warp circuits/controls in the labnotes section of Jack Orman's site for some excellent material.  If you don't walk away from that article with some useful ideas, something's wrong with your sense of creativity.

[aron]

I always likened different diodes to more "feel" than clearly audible changes.

The truth is, a simple tweak on a tone knob is much more drastic than one diode vs two on one side of a feedback loop.

But you can feel the difference for sure.

[Mark Hammer]

Agreed.  There is a tonal difference, but the difference emerges (or not) depending on how you play.  AS you rightly point out, a tone control change makes a difference regardless of what or how you play.  Moving from 2 to 3 to 5 diodes is something you won't really be able to easily detect unless you really dig into a note.

[brett]

After using switches, etc a number of times, I've recently decided to simply hard-wire for asymetric clipping. ie two 1N4148s on one side and a single 1N4148 on the other.  More "open" and "tubey" than symetric clipping, and I figure that about 99% of people like the sound better.  
When I was using a switch, I used the "diode shunt" method described by Mark.

[aron]

I agree Brett. I like more diodes than the usual one or two, but I agree. I don't really switch between diodes anymore.

[Willthebold]

I was fiddling around with the TS-808 I've had built for a while working with jumpering the extra diode.  I seemed to notice just a little signal loss when it was shorted.  Is that normal?  I assume so, but I haven't messed much with different diode configurations.  

Will

[Mark Hammer]

Willie,

Yes, you *will* hear a bit of apparent signal drop when shunting the 3rd diode.  People who A/B compare a Tube Screamer with a Boss SD-1 will talk about the TS being more "compressed" than the SD-1.  

Here's why.  

The diodes set a ceiling on how much signal comes out of the op-amp.  The more diodes, the higher the ceiling.  The diodes do not change the *gain* of that stage, merely how fruitful that gain will be with any given input.  If the input is  low enough, and the gain modest enough, and the "ceiling" high enough, then the signal will be boosted, but not clipped.  In fact, this is what normally happens with any gain stage that is set up for "clean" sound; push it hard enough though and you run out of headroom, resulting in clipping.  

The diodes in the feedback loop of the TS/SD-1/clones alter that headroom so that it doesn't take much signal at all for that lowered ceiling and applied-gain to result in running out of headroom and producing clipping.  (I like to refer to this difference as "proximity to clip" but that's just my little quirk, not an approved industry term for it.)  When you have a different number or type of diode in one direction than the other, one half cycle of the input signal runs out of headroom before the other does (or rather, the side with more diodes gets to keep more headroom than the other).

Apply an input signal whose level after applying the gain of that stage does not drastically exceed the clipping threshold/ceiling of the side with *fewer* diodes and you won't hear any fuzz plus you won't hear any difference whatsoever between 2-diode and 3-diode version.  The 2-vs-3 difference emerges when the input signal times gain results in a product where one half cycle clearly butts its head against the ceiling created by the single-diode and the other half-cycle has more room to move by virtue of 2 diodes in series (or whatever you use that generates a diminished clipping effect for that half cycle).

You will hear this discrepant amount of clipping for each half cycle as an increase in apparent dynamics, since you can pick harder and still hear the difference.  Moreover, since one half cycle is less restricted in its amplitude, you will also hear this as a louder output, even when the gain and settings of all other stages in the pedal are held constant.  As was discussed above, though, that difference will NOT be perceptible unless you aim for it in your picking.

In general, the apparent output levels of a number of pedals can be changed by swapping diode types.  For instance, the Jordan Bosstone has a pretty hot output.  If you want to smack an amp's input stage silly, switch the diode pair to ground on the Bosstone (which will limit the signal passing by that point to whatever the voltage drop of the diodes is) to a pair of red LEDs (which at least double the clipping threshold and subsequent output level).  Similarly, switch the diodes in an MXR Distortion+ from germanium to silicon and you will also notice an output level increase.

[WGTP]

Mark, do you think that using 2 LED's rather than 2 SI's gives the opamp more of an opportunity to clip since the LED's threshold is 1.8v and the SI's are .57?  I measurse some I had and came up with those values.  The Jfet MPF102 is over .7v a 2N7000 Mosfet is about the same as an SI and the GE's are around .27v.

I seem to like an LED/SI or FET combo which should generate a lot of even and odd ordered harmonics.