JFET used as a diode

Yep, should work fine. The JFETs are tested for leakage current, one would hope lots of other diodes are just as good in reverse blocking (but would require sorting). Tested transistors, with B-C shorted, might do well - and were discussed here not long ago.

The old PN4117A reverse leakage spec was one picoamp (if they were available). That's a selected part, though.

What sort of leakage current do you need? And how much V_rev. ? I've used BE and BC junctions in transistors.

George H.

Not B-C shorted--it's the base-collector junction that has the low leakage. You just leave the emitter open. BFT25As are great for that, but of course won't take 20 mA.

Those had a lot of series resistance though.

Cheers

Phil Hobbs

Jfets make terrible diodes. Series resistance and capacitance are both high. And jfets are expensive.

The c-b junction of a small transistor is a much better diode. BFT25 c-b is around 0.35 pF and has leakage that's hard to measure, below 10 fA.

Most gumdrop bipolars will leak well below 1 pA.

If the "low voltage" requirement is low enough and OP doesn't need the full reverse voltage capability the c-b junction of a particular transistor provides the b-e junction should be even better.

But whaddya do with the unused terminal, though? Just leave it hanging around? Does it act like an antenna?

It's not a good part, though, for 20 mA; if you want the c-b diode, with emitter open, you want something that allows 20 mA base current.

Speed? The bc or be junction of a 2n3904 would probably work. B. Pease "Bounding an clamping techniques" has a nice discussion of diodes... I can't find a good copy on the web. :^(

George H.

If all you need is a few volts reverse, connect the base to the collector. Then the forward drop is reduced by the transistor beta effect, 100 mV or so. For more reverse voltage, leave the emitter open, or connect it to the base; doesn't matter.

BFT25 must be a tiny chip; it's only good for 30 mW dissipation. Something hunkier would be appropriate for 20 mA, but getting pA leakage is easy.

Capacitance is not important here, but resistance might be, thanks for pointing this out.

MMBFJ202 costs 7 cents at quantity 10, 4 cents for a full 3000 reel. Would say it's dirt cheap.

I've seen them used in this role on several Tektronics schematics. Why didn't they use a bipolar instead?

So again, how about its reliability? The base region isn't very famous for its rugged physical structure. No long-term degradation?

Best regards, Piotr

If you don't need fA leakage, go for something bigger than the BFT25. You can get SOT23 transistors that are good for amps of collector current, which implies a lot of base current.

The jfet is OK if you like the forward conduction curve, but you should maybe test it. It will have a biggish ohmic component with a large positive TC.

There are low leakage p-n diodes, like CMPD6001. They are typically way better than their specs.

There's this one, but it's a bit blurry.

Cheers

Phil Hobbs

None, milliseconds would be fine.

Thanks for the reference, George.

Best regards, Piotr

Since when there has been base and collector in an JFET?

When other options do not work, the PAD-1 seems to be still around. A pair of them performed beautifully in a fA electrometer.

The referent was "tested transistors", i.e. BJTs, not JFETs. If you short the gate of a JFET to drain or source, it stops acting like a diode.

Cheers

Phil Hobbs

Right I've got the same blurry copy on my computer. I've got an older paper copy some where.

George H.

Yes, but high Rs, and expensive.