[sdiy] jfet driver

David G Dixon dixon at mail.ubc.ca
Wed Feb 27 02:29:59 CET 2019

```> Is the -10V shut off voltage for the 2N4391 the same as the
> "Voltage - Cutoff (VGS off) @ Id" characteristic?
>
> If so I'm seeing N-Channel FETs on Digikey with a VGS(off)
> anywhere between 180mV and 10V.
>
> Something like J106FS-ND has a VGS(off) of -2V typical and is
> through-hole.
>
> Mike

Yes, I believe so.  For 4391, V_GS (cutoff) is minimum -4V, maximum -10V.
One of the problems with JFETs is that you really cannot rely on any of the
numbers to be at some sort of average value, so you just have to assume that
it might take -10V to cut the damn thing off.  5485 specifies V_GS (cutoff)
of minimum -0.5V, maximum -4V.  I've also used 3819, which is maximum -8.0V
(it doesn't specify a minimum).

Bottom line, taking the Gate to the negative rail (or nearly) is the most
reliable way to shut off JFETs.  However, the 5485 is a nice option for -5V
square waves, which is why I use it for all my saw shaping and similar
circuits.

In Multisim, the 4391 has a Vto (which I assume means "turn-off voltage") of
5.803.  It is completely shut off at -5.803V, with an infinite source-drain
resistance.  It doesn't quite get down to the specified on resistance of 30
ohms.  At a V_GS of 0V (grounded gate) the on resistance is about 35 ohms
according to the simulation.  However, it is only 100 ohms at a V_GS of
about -3.9V, so it has about a 2V active range.

In Multisim, the 5485 shuts off at -2.2776V, has a grounded-gate resistance
of 200 ohms.  Interesting.  I guess it pays to really understand what a
simulation is telling you, because you could use one of these based on a
simulation result and then find that they don't all work because they were
closer to their maximum cutoff values than the simulation suggested.

```