How does a sine wave look like after some clipping stages?

I've read in several places of the soft clipping nature of a diode-connected MOSFET transistor (drain and gate shorted as anode, source as cathode). So I wanted to see what was really happening, and compare differents setups. The MOSFET I used was a 2N7000 made by Fairchild, with a threshold at about 1.9v (the internal diode had a 0.55v threshold).
I did this little experiment to see the effects of different MOSFET-based clipping stages on a 600hz sine wave. The stage setup is the standard non-inverting opamp (LM358) with a variable resistance in the feedback path for gain setting. The clipping devices where also in the feedback path. Each stage was tested at two gains, calibrated to obtain 16% and 26% THD readings, so it can be seen how the waveform looks when providing comparable amounts of distortion.

The stages I tested where:
1) Opamp alone, with no clipping device
2) Diode-connected 2N7000 (this equals two back-to-back diodes, the "soft" one when the drain/gate is more positive than the source, and the internal protection diode backwards)
3) Diode-connected 2N7000 plus a 1N4148 in series in each path
4) Diode-connected 2N7000 plus a 1N4148 in series with the soft diode path, and a 5mm diffused red LED in series with the internal diode path
5) Two diode-connected 2N7000 in series but in opposite directions. This equals to one soft diode and one internal diode in series on each direction. (The MOSFETS where matched for the same Vt at 1.9v at the soft side, and 0.55v at the hard side)
6) Two back-to-back 5mm diffused red LEDS (Vt around 1.4v)
7) Two back-to-back 1N4148 diodes (Vt around 0.55v)

Test setup

Soundcard based signal generator (Cool Edit's Generate Tone function, 600hz sinewave), routed from the right channel output to the input of the stage in test (built on a breadboard). The output of the stage goes to the left input of the soundcard, and is analized with Zelscope (soundcard based osciloscope). The soundcard is a M-Audio Audiophile 2496.

Let the images speak

Stage 1) Opamp alone, with no clipping device. 16% and 26% THD.

Stage 2) Diode-connected 2N7000. 16% and 26% THD.

Stage 3) Diode-connected 2N7000 plus a 1N4148 in series in each path. 16% and 26% THD.

Stage 4) Diode-connected 2N7000 plus a 1N4148 in series with the soft diode path, and a 5mm diffused red LED in series with the internal diode path. 16% and 26% THD.

Stage 5) Two diode-connected 2N7000 in series but in opposite directions. 26% THD.

Stage 6) Two back-to-back 5mm diffused red LEDS. 26% THD.

Stage 7) Two back-to-back 1N4148 diodes. 26% THD.

I think the soft clipping action of the MOSFET can be seen in the graphs, though they are not as informative (and fun) as seeing the waveform crushed in realtime. Perhaps I should have normalized the amplitude to allow for a better comparison.
Pay attention to the different clipping thresholds and the resulting asymmetry of stages 1, 3, and 4. Stage 5 can be asymmetric two, if the MOSFETs are not matched.

How does all this sound? You'll have to build it and hear for yourself!
If you have comments, corrections or suggestions, please send me an email!
Good luck.
Miguel.

Related stuff

To keep reading:
MOSFET page at Wikipedia.
A Musical Distortion Primer at GEOFEX.

v1.0 09/09/2008

Miguel Canel, Buenos Aires, Argentina bioroids.miguelNOSPAM@yahoo.com.ar (Quitar NOSPAM) http://www.bioroids.com.ar