Max current for a JFET connected as diode (2023 Update)

Jfets make pretty terrible diodes. The PAD1 series of picoamp leakage diodes are actually jfets inside, just expensive. Jfet diodes have a lot of equivalent series resistance.

Idmax could well blow out the gate. Try it.

A transistor c-b junction can be a far better diode. Some of the smaller transistors leak femtoamps but still behave like PN diodes.

So which bjt's have such low leakage? Since Im thinking of it how much current can the collector base junction take without failing?

- D

in AoE3, chapter 5, H&H like the PN4117 JFET. John's advice has been immortalized , p.294. footnote 4. "BFT25".

Even dumb bipolar gumdrops can have pA leakage. PNPs seem to be better for some reason. I recall that 2N3906 is pretty good c-b. But it could vary by manufacturer.

I missed that AoE reference. Cool. I'm on p 360 too!

It's not an easy question, low-leakage isn't a high test priority for BJTs. At a guess, you'd want low-gain audio (low-frequency) parts, which have relatively high base doping.

How about a zener diode? Should be good for high currents, certainly specified for 'em.

Small geometry RF transistors seem to have fA reverse currents when used as diodes. I'd expect audio parts to be much bigger.

Note that my diode connection exploits reverse beta to keep the base current down, so it should be good for rated max Ic. That will limit max reverse voltage.

The best ones I know about are BFT25As (RIP, sniff sniff), which are small-geometry 5-GHz NPNs.

Cheers

Phil Hobbs

See, e.g. this thread from a dozen years ago:

"Types of diode-connected BJTs"

Cheers

Phil Hobbs

True, but there's the 'how much current' question to consider; small geometry isn't always a good tradeoff, if it's a protection diode application.

The tradeoff with RF transistors includes low Rbb, so keeping base current down might not be an issue. Low-noise audio requires high current gain forward, which implies low base doping and high Rbb is acceptable. Those wouldn't be a great choice for protection.

Any guesses about SiGe transistors as diodes? I should get some and try them. One figure of merit is capacitance * reverse current.

I miss JT's harassment. He was good at it.

Re BFT25 (without the A, has higher Noise Figure, lower f_T) about 1700, left in stock at DigiKey

Though I assume if one chose another of similar construction & specs, they should see similar results (?) if not in the ballpark... = RS

But when Fields got involved, it went to a whole new level. ;)

Cheers

Phil Hobbs

Oh, I have three reels of BFT25As in stock. Doesn't help much with customer designs.

Cheers

Phil Hobbs

Haven't tried it. I should hang some on my nice shiny HP 4145B and find out. I use BFU520As to replace BFT25As for actual three-terminal jobs.

Cheers

Phil Hobbs

A zener is too leaky for this application alone. It is a Zener circuit but where where I'm adding the extra forward voltage drop of a low leakage diode in series so the zeners leakage current isnt causing error on the signal line below the zener clamp voltage. Jfet clamps are very good at staying out of the way in terms of their leakage. I've used them for many microphone or other audio signal line protection but never had to pay much attention to the current through them when conducting. - Just that it does when called for. This application is a little different. This is to protect a "universal" voltage input where the range of signal can be 0 to 10 mV or a large as 0 to 10V. But also prevent damage when some dope connects the +24V power supply rail to it.

AofE does have the closest thing I've read that discusses some of the question but doesn't really speak to the question of the limit of current through the clamp. Logically is would be below the max rating for the device but how far below?

It's possible for it to oscillate, though!

The normal diode-connected transistor runs out of beta at low currents, too.

Cheers

Phil Hobbs

Thanks for including that thread here. I very much appreciate everyone's useful comments on this topic.

Don

Idiot data sheets seldom specify s-params in the reverse beta mode!

(repeat usual rf rant)