Any recommendation for diode-connected MOSFET?

[Shea]

Geofex's Musical Distortion Primer says that a diode-connected MOSFET "makes for a 'diode' that has about a two volt forward drop, but a "knee" that is very softly curved."

Intrigued, I tried this out with a pair of MOSFETS I got at Radio Shack. I thought they sounded about the same as silicon signal diodes. Then I found that they had the same forward voltage drop as a silicon diode. It occurred to me that they must have had built-in g-s diodes.

So, can anyone recommend a good MOSFET for this purpose, or at least one that doesn't have a built-in diode?

Shea

[Steve Conner]

Hi Shea

I experimented with this back in 1998. In fact I like to think that R.G. Keen might even have got the idea from me ;-) I got good results with a tiny BS108 MOSFET with its gate connected to its drain. Bear in mind that you get the soft knee in the forward direction (drain positive wrt source) but a regular hard knee in the other direction (drain negative) due to the parasitic diode in there.

Now, you can get MOSFETs with the parasitic diode disconnected, but they're difficult to track down. So to get bidirectional clipping, just connect two of these MOSFET diode thingies in series, with one reversed. Or put a single one inside a 4-diode bridge, like I ended up doing in my Toaster amp:

http://www.scopeboy.com/overdrive.html

The BS108 is a European part that you might not be able to buy easily. You could try any small FET like the VN10LP.

I went through a phase of trying every non-linear device I could get my hands on to see what its clipping behaviour was like. One other thing I found that has a nice soft knee is the high voltage diode from a microwave oven. I tried the grid-cathode diode of a 12AX7 tube too, but it turned out to have a very hard knee.

[Ray Ivers]

Steve,

Did you ever find any correlation between soft-knee clipping and a good overdrive sound? I've never been able to myself, but I haven't experimented with as many devices as you have. Every diode clipper I've tried ends up sounding somewhat 'garbled' to me (as in, I can't hear every note of a three-note power chord clearly), and they also don't seem to respond as well as tubes to a reduction in pre-overdrive bass response to try and 'clean up' the garbling a bit, at least IME.

Ray

[Mark Hammer]

My gut sense is that any apparent association between pleasingness of a clipping circuit and use of "soft-knee" devices is illusory.

I say this because the harmonic complexity of the input and output signal is such that the shape of the "knee" is a very very small player in terms of the output content. Much much smaller player than the filtering. IF we were dealing with analog synth or steady-state oscillator signals where ONE tone comes in and exits with a slightly different harmonic structure, yeah maybe. But the number of factors at work in the average clipping pedal context is so large as to make the manner in which the onset of clipping occurs have about as much impact as one person in a crowd of 10,000.

[Shea]

Steve, thanks for the info. The Toaster looks extremely cool.

Ray and Mark, the reason why I was interested in trying soft-knee clipping is to get rid of the click-click-click sound that I hear in every overdrive or distortion box I've ever tried. Filtering can take it out, but then the tone is too muffled if I try to clean up a little by playing lighter or backing off the guitar's volume. If soft-knee clipping isn't a good fix for that, then maybe it would be better to try a variable low-pass filter using an operational transconductance amplifier.

Shea

[Ray Ivers]

Shea,

The 'clicking' sound you mentioned sounds like a pick-attack thing - does that sound right to you? If so, you might try pre-distortion filtering; I'd suggest 12db/octave @ 10kHz F3 for starters, and 18dB/octave might be even better (a single 6dB/octave cap really won't help much, as you know - when you make it big enough to help, you lose most of your high end). Something between 100pF and 470pF across the first 100K plate resistor is the approximate ballpark you're looking at, with additional filter sections scaled accordingly - if this is something you'd want to try, let me know and I can work up the numbers for you; the computations are a little complex as both series and shunt resistances must be taken into account if the filters are to work correctly. Then once you're in the range you want to be in, by-ear testing would finalize the values. I've used caps across the first plate resistor and following shunt grid resistor for 12dB/oct, sometimes with an additional cap across the second gain stage's Rp for extra-edgy single-coils.

My Pearce amp uses an OTA following a semi-parametric mid control. It also used clipping diodes in the first gain stage's feedback loop, but once I took those out - and replaced the first op-amp with an LF353 - the OTA was getting hit a lot harder, and I noticed a real improvement in overdrive sound.

Ray

[Shea]

Shea,

The 'clicking' sound you mentioned sounds like a pick-attack thing - does that sound right to you?

No, I was referring to the hard edge on clipped signal. It sounds to me like a stream of very fast clicking. I can't pick that fast.


I noticed that in tube-screamer-derived circuits, sometimes when I strum hard, I get an ugly blatt or it sounds like the clean signal punches through and overwhelms the clipped tone briefly. In my breadboard experiments, I avoid that by following the non-inverting gain stage with an inverting stage, and put the feedback diodes in the inverting stage. I guess when you put the diodes in a non-inverting gain stage, the point at which the signal breaks over the diodes is not the reference voltage plus the diode's forward voltage, as one might suspect, but actually the input signal plus the diode's forward voltage. In an inverting stage, it's the reference voltage plus the diode's forward voltage.

Shea

[Ray Ivers]

Shea,

When you pick a single note, you get a series of rapid clicking sounds? That's a first for me - I'd say that something is definitely wrong.

In a non-inverting stage that consists of an op-amp, input shunt resistor to ground, and two clipping diodes back-to-back from the output to the inverting input, the output should operate in "undefined-but-basically-unity-gain" mode below clipping, and then hard-clip at 1.2V peak-peak or so. With an inverting circuit - series input resistor, grounded + input, and diodes as before - the circuit would operate open-loop (max op-amp gain) below clipping. That's why you always see an additional feedback resistor in both circuits - to firmly define unity gain in the non-inverting configuration, and to limit maximum gain in the inverting circuit. Sometimes an additional resistor & cap are added to the non-inverting configuration (from the inverting input to ground) to define a higher-than-unity resting gain (I know - Op-Amps 101 - just including it for completeness' sake).

Your clicking sounds could be open-loop artifacts in the inverting stage (perhaps caused by the op-amp output oscillating and/or slamming hard against the power-supply rails during through-zero waveform transitions) which might explain their disappearance when the diodes were moved to the non-inverting stage (and the diodes replaced by a resistor in the non-inverting stage, I imagine). Adding a feedback resistor in the inverting stage (or lowering the value of an existing one) might have eliminated the problem there.

It sounds to me like you still might be a candidate for pre-distortion filtering, especially since you mentioned a 'blatting' sound (which I call "splatter") which I've found to be caused by high-amplitude leading-edge transients - i.e., pick attack - hitting a hard-clipping stage... but YMMV.

I wasn't sure what you meant by the 'reference voltage'.

Ray

[Shea]

When you pick a single note, you get a series of rapid clicking sounds? That's a first for me - I'd say that something is definitely wrong.

I guess I'm going a poor job of explaining myself. I'm just saying that the clipping in every single distortion or overdrive effects pedal that I've heard -- regardless of who is playing through it, or how hard they pick -- sounds unnaturally harsh to me, and "clicking" is my way of describing the harshness. That's what every little clipped wave sounds like to my ears.

The other stuff I mentioned about inverting stages vs. non-inverting stages and blatting and whatnot was really unrelated, but something you said earlier reminded me of it.

Shea

[Steve Conner]

Hi Ray, Shea, all

Ray: I found that the soft knee clipper circuit just got rid of the peaks without changing the sound much. I could really drive it pretty hard before it even started sounding overdriven, and when it did start to break up, it didn't sound aggressive or edgy at all. This is partly because I used a symmetrical circuit that doesn't modulate the duty cycle of the clipped wave according to how hard it's driven, like regular tube overdrive does. Not much use as a guitar overdrive effect, but I like it when I use the thing as a bass amp. It helps get rid of the slapping and popping peaks without harsh distortion.

I believe that this duty cycle modulation effect is what gives overdriven sounds their personality, far more so than softness and hardness of knees. It's all explained in Russell O Hamm's famous "Why tubes sound different than transistors" paper, IIRC. But in particular, if you allow the bass to modulate it too much, you get those nasty farting and blatting sounds.

Shea: Thanks I built the Toaster back in 1998-2000 with help from you guys on the old Ampage. It's been a great amp and has seen lots of jams and gigs as both a guitar and a bass amp. Though when playing bass through it I really wish it had nearer 500w than 50 After that I heavily modded an old Selmer Treble'N'Bass 50, to try and make a "Mr. Nice" that complemented the nasty heavy tones I had already. Next I'm planning to try a hybrid bass amp with a tube preamp and big SS power amp.

[R.G.]

Hi, Steve. I didn't get it from you, unless you wrote the article in EDN ( I think...) that I saw it in. That's a possibility, though - it was about that time. Did you submit that to Electronic Design or EDN?

I didn't generate it fresh. But I liked it!

I agree with you - it can make the onset of distortion be very hard to notice, not a sudden threshold. But that has its uses.

[Ray Ivers]

Steve,

Wow, I've taken my time responding to this post, eh?

I've noticed a correllation between duty-cycle modulation and overdrive-circuit performance I don't like, i.e., circuits that maintain close to a 50% duty cycle sound better to me than ones that go all PWM on me during heavy overdrive, all else being equal.

I haven't given up on the single-stage symmetrical-clipping idea, especially since the Zener-bounding circuit I mentioned in other threads may help get rid of the problems I was having with it cap-coupled (if, not, I'll go direct). I couldn't get it to clip as symmetrically as I would have liked (the negative output peak was still pretty rounded, although the cutoff peak was flat as a pancake, as you'd expect), but it was still a huge improvement IMO over conventional single-stage clipping behaviour - and the duty cycle stayed at about 50% even during pick-attack.

Ray

[JohnGreene]

Hi, Steve. I didn't get it from you, unless you wrote the article in EDN ( I think...) that I saw it in. That's a possibility, though - it was about that time. Did you submit that to Electronic Design or EDN?

I didn't generate it fresh. But I liked it!

I agree with you - it can make the onset of distortion be very hard to notice, not a sudden threshold. But that has its uses.

It was Electronic Design Magazine. 'member?

--john