Would you specify a P-Channel JFET for a new design?

[TimNJ]

Greetings,

I have a particular problem that (I think) would be neatly solved by a P-channel JFET with minimal parts and PCB area. (At power up, before VCCs are established,  I want to pull a node to ground until another node has reached a certain voltage -- A "normally closed" device.) A crude solution is fine (+/- 1 or 2V tolerance on Vgs is OK). A P-channel JFET seemingly would be an easy solution...something like MMBF4393 would do fine.

But the supply seems very limited, and dwindling, especially in the P-channel department. Generally does not give me a good feeling, and that probably answers my question.

Do you think I would be playing with fire, say, 10 years down the road to specify a P-JFET in a new design? Do they have an important "modern" use elsewhere that would keep them alive?

Thanks,
Tim

[Zero999]

I would've thought they'll be available for some time to come, even though some are going obsolete, as is the case for the part you've suggested. Try the J177, in the surface mount package instead.

https://www.mouser.co.uk/ProductDetail/onsemi/SMMBFJ177LT1G?qs=vhu28TeoNRioevgN%2FJPZtw%3D%3D&gad_source=1&gclid=CjwKCAiA1MCrBhAoEiwAC2d64Y-y01iVMIsyyC2hsxueWzpcMVsmrRHa8YUQTLUenTAP5djRgg3NChoCkY0QAvD_BwE

[SiliconWizard]

The MMBF4393 is flagged obsolete by Mouser, but doesn't seem to by Onsemi. So, I don't know. But availability of the MMBF439x series is close to none at the moment indeed.
I don't know what the future holds in ten years from now. I don't think discrete JFETs are going to disappear, but they may indeed be harder to source. I had absolutely zero problem sourcing the MMBF4393 by thousands a few years ago.

The Jxxx series is an alternative as Zero999 suggested. You also have the MMBFJ211 available in large quantities at Digikey.

You could otherwise consider using a depletion-mode MOSFET instead. Infineon for instance has a number of references, the BSS159N is available at digikey as well in large quantities.
https://www.infineon.com/cms/en/product/power/mosfet/n-channel-depletion-mode/bss159n/

[TimNJ]

Ah, I should've included the full PN I was thinking about: MMBF4393LT1G (which is surface-mount).

I was just thinking that they come in handy (frequently enough) that there would be more options and suppliers available. But, guess I'm wrong, or else the market would indeed answer that demand. Seems like Onsemi pretty owns the market. Is JFET manufacturing process very unique compared to other common proceses? No one wants to keep a line up, except maybe onsemi for some reason?

[magic]

You could otherwise consider using a depletion-mode MOSFET instead. Infineon for instance has a number of references, the BSS159N is available at digikey as well in large quantities.
https://www.infineon.com/cms/en/product/power/mosfet/n-channel-depletion-mode/bss159n/

This says "n-channel" right in the URL...

Not sure if anyone makes P-ch depletion MOS at all. I know that some people specifically looked for them and found none.

[Kleinstein]

I have also not found p-channel depletion mode MOS. P channel Jfets are not very common, but there are a few applications for them on the process should not be that special or complicated. There are only a few types available ( mainly MMBFJ270 and MMBFJ175-177), but still several manufacturers.  It could be tricky with a THT type though.

If a specific MMB4393xxx goes EOL this is only one specific version, there are plenty of others of the very common N channel type.

JFETs in general have quite some spread in the parameters, which may be an issue.

[tszaboo]

For what you describe, I would use a SOT23 reset controller. Same PCB area, same function, readily available replacements parts.
Jfets are nice, but I would only consider them for high end analog stuff.

[ArdWar]

Supervisor/Reset ICs are neat, easy and dirt cheap, but beware of potential glitches at very low voltage level. It normally should not be an issue for logic circuits, but idk what kind of signal you actually want to shunt here or what glitch you can tolerate.

[magic]

It looks like those chips should be using depletion FETs internally to avoid such problems

Not sure if CMOS can do it easily, and I suppose BiFET may be too "exotic" to be economical in such things.

[mawyatt]

Supervisor/Reset ICs are neat, easy and dirt cheap, but beware of potential glitches at very low voltage level. It normally should not be an issue for logic circuits, but idk what kind of signal you actually want to shunt here or what glitch you can tolerate.

(Attachment Link)

Good recommendation

We just used an APX803S (SOT23) to sense the input supply voltage from MeanWell PS (nominal 12VDC) and issue a POR_bar until/if this is above/below predetermined levels (also has a build-in delay) for a production product. You can conveniently scale the "trip" voltage with 2 resistors.

Nice, simple, effective IC in a small SOT23 package!!!

Best,

[TimNJ]

Thank you all for your input, and good notes on possible power-on glitches of supervisory ICs. APX803S looks excellent for lots of applications. Curious what that peak "glitch voltage" might be. <0.5V probably?

In my particular case, the node in interest is across the bottom resistor of a high impedance divider. If powering APX803S  from this same node, a Vcc current of 10uA is too high by about a 100x. Could possibly be compensated for.

[TimNJ]

And I should note that I don't particularly need a JFET...just a depletion-mode P-channel device, which as far as I can tell, has never(??) been manufactured in the MOSFET variety.

[sadboi]

So I'm looking to do something similar (normally closed switch until powered), but I need the reset output to be guaranteed low for Vcc=0 V. Are there any supervisory ICs that function down to 0 V?

[ArdWar]

How about using NJFET or depletion NMOS instead of using the increasingly rare P devices, and devise some negative charge pump to bootstrap down the gate drive.
Just like how everyone uses positive bootstrap for normal NMOS instead of using PMOS .