12

In a MOSFET, the obvious cause of asymmetry is the fact that the source is tied to the body. But in a JFET, no such obvious answer exists, as the body, gate, and drain are all the same piece of silicon with no material differences between them (as far as I'm aware). And indeed, most JFETs seem to be symmetric.

But there are some JFETs on the market have specifications that would imply asymmetry. See for example the UJ3N120070K3S that I just used in a design; it's rated for a gate-to-source voltage of no more than -20 volts, but a drain-to-source blocking voltage of 1200 volts. But if \$V_{ds} = 1200\ \mathrm V\$, and \$V_{gs} = -20\ \mathrm V\$, that leaves \$V_{gd} = -1220\ \mathrm V\$! Even if \$V_{gs} = 0\ \mathrm V\$, \$V_{gd}\$ is still vastly higher than the rated \$V_{gs}\$.

If the FET were symmetric, that would imply that the gate-source and gate-drain ratings should be the same, but they clearly can't be here, as if they were, the drain-source rating couldn't possibly be higher than the gate-source rating, and applying rated \$V_{ds}\$ would fry the gate junction.

What about its physical construction makes this JFET, and others like it, asymmetric? Why can \$V_{gd}\$ be so much higher than \$V_{gs}\$?

Hearth
  • 27,177
  • 3
  • 51
  • 115
  • It wreaks of being a depletion MOSFET rather than a JFET. Just an opinion. Nice device though. – Andy aka Jan 09 '21 at 18:32
  • 1
    @Andyaka While troubleshooting the circuit I used it in, I was able to measure that the gate was indeed a diode junction. The forward voltage didn't seem much like a SiC pn junction though, now that you mention it... Perhaps a SiC Schottky junction, though. I didn't think to test if it had the characteristic body diode of a MOSFET, but I can check that on Monday. – Hearth Jan 09 '21 at 18:38
  • @DKNguyen That would be my guess, some kind of graded doping might explain it. I'm not sure what the *shape* of it would do, but I'm also not an expert on this sort of thing. – Hearth Jan 09 '21 at 18:47
  • Are you sure it's not one of of those hybrid devices where it's a JFET and something else (I forget if it is a MOSFET or IGBT) connected together? I mean the datasheet says it's a JFET and the symbol says it's a JFET, but every time I go looking for a power JFET those are the devices I find, not true power JFETs. – DKNguyen Jan 09 '21 at 19:00
  • I think these are what I am talking about: https://www.powersystemsdesign.com/articles/normally-on-sic-jfets-characteristics-and-applications/138/14198 Because if what you found is a true power JFET then I must have it because as $12 a pop that's pretty cheap. I don't even care if it's actually a depletion MOSFET. I needed a bunch of devices like that a couple of years ago. – DKNguyen Jan 09 '21 at 19:02
  • @DKNguyen Hmm, that circuit there would imply that it would be normally off, though, since that's an enhancement-mode MOSFET in the schematic. Though of course if you had the same thing with a depletion-mode MOSFET... – Hearth Jan 09 '21 at 19:16
  • I think we might need a @jonk in here. – Marcus Müller Jan 09 '21 at 20:32
  • @MarcusMüller Shine the J-light at the sky – DKNguyen Jan 09 '21 at 21:37
  • 1
    @DKNguyen [I did](https://electronics.stackexchange.com/questions/541745/model-train-incandescent-wheat-grain-bulb-to-led/541765?noredirect=1#comment1405302_541745). He points at [this](https://patentimages.storage.googleapis.com/ab/77/58/1867d15a0bea80/US9076880.pdf) patent. – Marcus Müller Jan 10 '21 at 17:12
  • 1
    @MarcusMüller Funny thing: when you wrote asking me, it was entirely by coincidence that I'd read the patent a few years back. I had no real need at the time. I just enjoy reading about new ideas and there was a news item or something that triggered my interest. Since I had a bookmark handy, I was able to respond instantly when you mentioned this question. I don't feel as though I know enough to do a credible job of debating/discussing the details. So no answer from me. I've no problem writing the inventors. But Marcus, I think you may be in a better position for that. Why not? We'd all learn! – jonk Jan 12 '21 at 08:10
  • I think I'll pass that on to @Hearth; it's his question after all. – Marcus Müller Jan 12 '21 at 10:56
  • @MarcusMüller I'm not a he. I'll look at that patent later--yesterday was too busy and today it's too early in the morning for deciphering patentese, but thank you for the reference! – Hearth Jan 12 '21 at 13:59
  • @Hearth sorry! That's a bit embarrassing for me :( – Marcus Müller Jan 12 '21 at 17:22
  • @MarcusMüller No trouble! Apology accepted. – Hearth Jan 13 '21 at 02:22
  • One thing that's immediately apparent from the UJ3N120070K3S datasheet is that the drain terminal is connected to case. That's a substantial physical asymmetry, though I don't know exactly how it leads to the device's characteristics. – Theodore Mar 23 '21 at 13:58
  • 1
    @Theodore I can't see how that asymmetry would affect breakdown voltages, though. It would certainly affect high-frequency performance with capacitance differences and such, but breakdown voltages depend on the characteristics of the junction itself. – Hearth Mar 23 '21 at 14:20
  • @Hearth: I wasn't thinking that it affected semiconductor breakdown so much as dielectric breakdown in the package. – Theodore Mar 23 '21 at 14:46
  • 1
    @Theodore The semiconductor breakdown occurs far before package dielectric breakdown, I do know that much. In fact, you'd see the air around the package breaking down before dielectric breakdown in the package happens, in most cases! (MOSFETs and IGBTs being a notable exception, with their very thin gate oxide) – Hearth Mar 23 '21 at 15:37

1 Answers1

1

Many JFETs are symmetric and this is sometimes indicated in the datasheet such as the

enter image description here Source: https://shop.micross.com/pdf/LSM_2N4117A_TO-71.pdf

If it is indicated then you know its symmetric otherwise you would have to assume either.

I think this is question is backwards:

What about its physical construction makes this JFET, and others like it, asymmetric? Why can Vgd be so much higher than Vgs?

If there is asymmetry you can infer this from measurements or from the datasheet because if the physical construction were symmetrical there would be no difference.

A better question would be: Can I infer that a JFET is asymmetric if there is a difference between Vgd than Vgs. I believe the answer is yes, but you could never be sure unless you actually check the physical construction or a patent. Also if there were Cgd and Cgs listed and they were different I also think you could assume asymmetry.

This J308 "N-Channel JFET" has a different Cgd and Cgs.

enter image description here Source: https://www.digchip.com/datasheets/parts/datasheet/078/J308-pdf.php

Voltage Spike
  • 75,799
  • 36
  • 80
  • 208
  • 2
    Yes, but what I'm asking is specifically what makes *some* JFETs asymmetric--I'm well aware that most of them are symmetric, but the ratings of specifically the U3JN120070K3S and others like it indicate asymmetry (and considering I've actually blown one by exceeding its gate-source voltage, measurements also indicate asymmetry). – Hearth Mar 05 '21 at 01:40
  • 1
    In short, I'm asking what in the physical construction could possibly make a JFET asymmetric, because I can't seem to understand how a JFET could possibly have *that* different Vgs and Vgd ratings. – Hearth Mar 05 '21 at 01:41
  • One is the body might be tied to one of the terminals (might account for differences in capacitance like in the below example), the other is that the gate is not symmetric and not equal amounts of material to the gate from the 'source' or 'drain'. Now that I think about it, you could check with some kind of impedance measurement for asymmetry, if it wasn't physically symmetrical the impedance measurement could not be the same – Voltage Spike Mar 05 '21 at 03:46
  • The old Siliconix data books were brilliant. They included metallization patterns. Have a look at p182 of this link and you will see that source and drain geometry are not symmetric: https://archive.org/details/bitsavers_siliconixdixFETDatabook_24133064/page/n181/mode/2up – 10ppb Apr 10 '21 at 01:28
  • @10ppb p182 is a mosfet though... the JFET on page 186 is symmetric – BeB00 May 06 '23 at 02:57