I started out wondering: "How variable are JFETs? How many parameters must be measured when selecting them?" I ended up with a grungy overdrive/distortion: the Stupid FET Trick.
Looking at the 2N5457 datasheet from Fairchild, you can see that drain saturation current (IDSS) and gate cutoff voltage (VGS(OFF)) are correlated: I've seen other methods on the web for selecting FETs to use in audio, and they generally involve selecting for gate cutoff voltage, since this affects the operating point. IDSS is much easier to measure, or at least to select for. In my case, I was designing a circuit using the stock CircuitMaker model, which has an IDSS of 3.859 mA at 15V. (The range of values in the datasheet are 1.0 mA to 5.0 mA.) The model's behavior was determined with this little test circuit simulation:The device test circuit is not much more than that. It needs a supply voltage somewhat greater than the reference voltage in the datasheet (I have an 18V regulated supply), has gate and source tied together, and has a drain resistor set to drop the supply voltage to about the reference voltage with saturation current flowing:
You don't really have to have a close-tolerance on the resistor; just measure its actual resistance and use that to select the range of voltages it will drop for the range of currents within your design tolerance. I needed five 2N5457s that were within 5% of the Circuitmaker model. I bought 100 (for 8 cents apiece at Mouser), and I tested 40 before I found six that were within my range. Most of the others were under and a few were over (not surprising since the selected value was in the upper portion of the range). My circuit was nothing earth-shaking, just a few ideas taken from typical tube guitar preamp design with the JFETs substituted in, and component values adjusted to get a meaningful operating point at battery voltages. I didn't do a thorough scientific verification of the gain in the real circuit, but the measured operating points were all within 10%. (Even though I selected the JFETs at 5%, the resistors were 10%). It doesn't really sound like a tube amp, but it is better than the last distortion unit I built (some 25 years ago), the Tube Sound Fuzz from Craig Anderton's Electronic Projects for Musicians. I continued my tradition of natural aluminum enclosures, and aside from the name of the unit, I only used pictograms for the legend (Skull & crossbones = gain; heart = tone; quaver = volume). I'll have to design the next JFET circuit around a device with a lower IDSS (and VGS(OFF)) to use up the ones I have left.
Coming soon: Interstellar Overdrive!