series jfet as pulse current limiter

Speed is no problem with an RF JFET, but the IDSS spread is huge and the drain impedance is the pits.

Are you trying to make a bidirectional current limiter? First choice would be a depletion MOSFET across the DC terminals of a Schottky diode bridge. That 40V is a bit of a problem--maybe use two in series per leg. You could do that with 4 x BAT54S.

Alternatively, a two-terminal BJT current limiter (you know the one--auxiliary BJT steals base drive from main BJT when sense resistor drops a V_BE) inside the bridge. Better Early voltage, better accuracy, lower voltage drop, almost as simple.

Cheers

Phil Hobbs

Mark

There's no issue in one direction; I'd be happy to see it passing. I'm familiar with the IDSS spread, but figured the source resistor biasing the gate would tame this to some extent - as a function of Vp.

In the sim you should see a spike in the first 10s of nanoseconds of limiting in the negative-going pulse, before the jfet settles. As you can see there, the circuit is convoluted enough.

I'm trying to guarantee triggering a possible 1V hysterisis spread, centered around 1300mV, with only 500mV of solid positive-going active signal, that must still function when the amplitudes hit 20V. Impedance of the target pin is poorly defined, but the literature assumes ~high, even though it's supposed to sink and source up to 3mA at the clamping voltages of the sim.

It has to do the former to simply start, and the latter to not go kaboom. I'm becoming increasingly kaboom-shy, in my dotage......and anyways; what is modeling for, after all?

RL

I'm worried about pulse response, rather than tolerance. As long as an XmA limit is assured, reductions of an order of magnitude are probably acceptable as a unit to unit variation.

However uncontrolled pulses prior to limiting are as bad as no limiting at all.

RL

Really sounds like a job for an LM319 rather than a JFET. Only a few cents more, no midnight phone calls, and you get a second channel for free.

Cheers

Phil Hobbs

Fortunately, the source is of sufficiently low impedance to get results using simpler diode clippers and filters. Just had to step back and figure out what was needed.

Thanks for taking an interest.

RL