op-amp biasing for distortion

[brett]

Hi.  After an enquiry on the forum last week, I made up a distortion+ type circuit and replaced one of the 10k Vbias resistors with a pot.  The op-amp I used was a TL071.  The results with a 1kHz test signal were that at about 0.6k resistance (Vbias ~ 10 x 10/(10+0.6) ~ 9.4V) there was nice-looking clipping on the old o'scope.  Interestingly, the effect wasn't just on one side of the waveform - it seemed to reduce the amount of amplification on the non-clipped side too.

So I tried it with the guitar.....and it sounded very ordinary.  Some fairly nasty harmonics and plenty of "roughness" that got worse with more clipping (ie as Vbias approached the positive supply rail).

So mis-biased op-amps DON'T produce very nice distortion. :cry:  In hindsight, I can see that it were that simple, lots of circuits would already be doing it.  

But has anybody tried other tricks like starving the op-amp of current by putting a resistor in the supply line?  Any other tricks?

[Peter Snowberg]

You may want to try again with some different opamps. R.G. has some great info here:

http://www.geofex.com/Article_Folders/TStech/tsxtech.htm

The opamp tone secret looks like it's how the chip recovers from the overdrive. The TL071 isn't on the "recovers gracefully" list.

I really like some of the tones from a misbiased MOSFET. I'm guessing that's because it recovers instantly from overdrive without odd artifacts.

Take care,
-Peter

[brett]

I agree that there's probably some significant differences between op-amps.  I'm going to re-try with a 741 - which will probably clip the furthest from ground and the positive rail.  Hopefully it will be more graceful than the TL071.

But it's also worth noting that the op-amps that will perform well in these tests may well be completely different to those that perform well in the tubescreamer (eg NE5532).  The distortion I'm chasing is due to running out of power because the signal is approaching the positive supply rail - in the tubescreamer it's due to the rapid increase in negative feedback when a diode starts conducting.

I'll try some different op-amps tonight (and try running them at lower gain - I was using a 4.7k ground resistor and a 47k feedback resistor for a gain of 11)

cheers

[Peter Snowberg]

Recently somebody posted a design that ran a 741 from a +3/-1.5v supply (or was that a 1.5/3v?). Sorry, I don't remember where but I sure somebody does.

That would sure collapse the headroom on you.  :wink:

Take care,
-Peter

[Brian Marshall]

I would think there has got to be some comercial pedal out there that uses opamp clipping.  It is just so easy to do.  I would think that if you filtered it right after the clipping it would make a nice fuzz

[gez]

You could try using an op-amp with a CMOS output.  They still sound like dogs in my experience as the rest of the chip, even if MOSFET, isn't CMOS.

Lower voltages, as mentioned, would probably get you partly towards your goal.  Perhaps something with really low slew rate too?

OTAs non linearity can shape a triangle to sine, so perhaps with the right amount of signal you could get soft clipping?

[aron]

So mis-biased op-amps DON'T produce very nice distortion.  In hindsight, I can see that it were that simple, lots of circuits would already be doing it.

But has anybody tried other tricks like starving the op-amp of current by putting a resistor in the supply line? Any other tricks?

OK, you basically found what I found a while ago.

I did pretty much what you did and found roughly the same results.

Starving the op amp OTOH does work.

See the Shaka HV. I used a voltage regulator and adjusted to voltage from dead to 33V.

All kinds of stuff in between!

[Mark Hammer]

One of the quirks of the MXR Distortion+ is that the voltage divider used to provide the Vref employs ridiculously high value resistors (1meg).  Look around at all the other clones, and you'll see they use something much less.  For instance, the DOD 250 uses a pair of 22k resistors to accomplish the same task. I don't know what role variations in current to the inverting pin play but that is also something worth fiddling with.

The post-humously published project-article from Charles R. Fischer in Electronic Musician used current regulator to variably starve a 4049 invertor section as the clipping element,  That might be worth looking into as well.  It,s posted somewhere, but I forget where.

[brett]

Just to clarify the biasing arrangement that I'm using - it's started life as a pair of 10k resistors, with the centre tapped by a 1M resistor (to input) and a 47uF cap (to ground).  I replaced one 10k resistor with a 10k pot, which I reduce until I get the desired amount of clipping (e.g. 0.6k gives 10V x 10/(10+0.6) ~ 9.4V).

Last night I substituted a 741 and an NE5534 for the TL071 that I'd been mis-biasing.  As expected the 741 started clipping a lot further from the supply rails (about 8V).  It also sounded significantly better (but not good enough to be useful on its own).  The more interesting result was the NE5534, which also sounded better than the TL071, but it kept working all the way up to the positive supply rail (10V!).  There was still quite a reasonable sound coming out of the circuit even when there was absolutely no positive headroom (and presumably only the negative half of the signal was coming out).

Mark H
Thanks for the note about starving CD4049s.  In my 4049-based fuzz I use a 220 ohm resistor on the supply line.  I think this probably helps it give quite a soft fuzz.  Someday I'll have to try smaller and larger resistors.  Thanks for your many suggestions re 4049-based fuzzes.  I REALLY like mine (gain of 11 in a non-inverting TL071, 2 stages each with feedback of a 100k resistor and 220pF cap).