Author Topic: SiC cascode  (Read 1463 times)

0 Members and 1 Guest are viewing this topic.

Offline JohanHTopic starter

  • Frequent Contributor
  • **
  • Posts: 620
  • Country: fi
SiC cascode
« on: October 27, 2022, 09:50:27 am »
Just found out about SiC cascode devices that combine a SiC JFET and regular MOSFET in one package, with some advantages over traditional MOSFETs. They have standard gate characteristics, so they are a "drop-in" for regular MOSFETs and IGBTs.

E.g. UnitedSiC UJ3C065030T3S https://unitedsic.com/datasheets/DS_UJ3C065030T3S.pdf

Application note about the concept here: https://unitedsic.com/appnotes/USCi_AN0004-Cascode-Configuration-Eases-Challenges-of-Applying-SiC-JFETs.pdf

They also have good DC SOA, so I'm thinking they could be used in a linear (HV) DC power supply or electronic load.

Thoughts?
 
The following users thanked this post: Circlotron

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 14076
  • Country: de
Re: SiC cascode
« Reply #1 on: October 27, 2022, 11:25:58 am »
For linear use, I would consider the SiC Fet in a seprate case, not a combined part. With the one case version one still has the thermal instability problem when using more than 1 device in parallel. As seprate the MOSFET would see rather little heat, even hat high current.

The gate drive should be even easier than with just a regular MOSFET, as there is little drain to gate couling effect.
 
The following users thanked this post: JohanH

Offline Slh

  • Regular Contributor
  • *
  • Posts: 115
  • Country: gb
Re: SiC cascode
« Reply #2 on: October 27, 2022, 05:39:32 pm »
The jfets are pretty awesome. We've got them in a simple constant current source and they just work. I've also been playing with them as high voltage amplifiers and they behave really nicely.

I'm hoping that more folks bring out SiC jfets. I can think of a decent number of uses for them in switching and linear applications but am a little wary of single source devices.
 
The following users thanked this post: JohanH

Offline Slh

  • Regular Contributor
  • *
  • Posts: 115
  • Country: gb
Re: SiC cascode
« Reply #3 on: October 27, 2022, 05:41:20 pm »
I suppose I should also mention that we've been using the cascade ones in switching converters and they've been pretty good but they do switch awfully fast (faster than I'd trust a 500MHz scope to be truthful about) so play with care...
 
The following users thanked this post: PartialDischarge

Offline PartialDischarge

  • Super Contributor
  • ***
  • Posts: 1611
  • Country: 00
Re: SiC cascode
« Reply #4 on: October 27, 2022, 05:46:31 pm »
The jfets are pretty awesome. We've got them in a simple constant current source and they just work.
What voltage/current is that current source running at?
 

Offline Slh

  • Regular Contributor
  • *
  • Posts: 115
  • Country: gb
Re: SiC cascode
« Reply #5 on: October 27, 2022, 05:57:59 pm »
Up to 1000V, 100mA. It's only running for around 100ms max in the application but it fixed a terrible bodge on the first prototype.
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 21609
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: SiC cascode
« Reply #6 on: October 27, 2022, 10:24:33 pm »
Maybe worth noting that SiC dies are tiny.  The current density can be impressively high; or, not so much the current density but voltage (SiC is a bit worse than Si on mobility, but several times improved in breakdown voltage), but you see the product of both as switching area and power dissipation, of course.

That's not to say the power ratings are pitiful.  Equally impressive is the thermal conductivity, so, as long as the chip can be made quite thin, and the backing plate is solid copper (or maybe AlN), overall power dissipation can still be reasonable.

(I guess that means I don't actually have a point, then. :P It's fascinating, though.)

I haven't looked at them for linear purposes before; the usual SOA warnings apply, of course.  (I gather the JFETs generally offer fairly unrestricted SOA, then?  That's nice.)  I would guess for most purposes, the lower cost of Si will dominate the decision, but I wonder if the lower capacitance offers higher bandwidth for certain applications?  Could be a performance edge that way, for the same reason it's an edge in switching applications.

I don't know that cascodes wouldn't track well; they're certainly not going to be worse than discrete parts on a common heatsink.  Arguably maybe even better, as the bottom transistor doesn't suffer from self-heating, maybe even its RthJC is better than the Rth(J1-J2) as it were.  I don't know offhand how they're constructed (two die pack? stacked? surely they're not monolithic..).

The biggest, disappointment I guess, and more generally, is the kinda failed promise of high Tj(max) -- they're limited by packaging materials (150-175°C) same as everything else in plastic.  So the only exceptions are the few metal and ceramic packages (mostly MIL spec $$$).  That said, it's not necessarily a big loss, as some MOSFETs (JFETs too maybe, not sure offhand?) have truly abysmal tempcos up there (like 4-5x Rds(on) at 300C or something, I forget how bad exactly?).  But some others are fairly flat and I'm not sure what the difference is (SiC allotrope? transistor design? just more substrate thinning to remove internal resistance and accompanying tempco?).  Anyway, that's not a comparative thing, just a reaction to, earlier hype I think.  Overall, compared to Si, modern types at rated temperatures are indeed as impressive as their specs show. :)

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Online magic

  • Super Contributor
  • ***
  • Posts: 6733
  • Country: pl
Re: SiC cascode
« Reply #7 on: October 28, 2022, 05:53:57 am »
The biggest, disappointment I guess, and more generally, is the kinda failed promise of high Tj(max) -- they're limited by packaging materials (150-175°C) same as everything else in plastic.  So the only exceptions are the few metal and ceramic packages (mostly MIL spec $$$).
I can't help but wonder what happens if you just don't give a toss about the plastic >:D

Perhaps not the brightest idea for an industrial product, but some DIY amplifier or dummy load on the other hand...

I remember cooking some 78L05 for many minutes at 300~400°C and it kept working afterwards.
Obviously, that's not the same as cooking for many days or years.
« Last Edit: October 28, 2022, 05:59:22 am by magic »
 

Offline JohanHTopic starter

  • Frequent Contributor
  • **
  • Posts: 620
  • Country: fi
Re: SiC cascode
« Reply #8 on: October 28, 2022, 06:23:50 am »
The jfets are pretty awesome. We've got them in a simple constant current source and they just work.

This was exactly another use case I was looking at. In a simple constant current source with a resistor they work from almost 0V up to max specified voltage. I can't find any other simple component/circuit that does that. I.e. common regulators and LED drivers don't work under 7-12 V for instance and old JFETs have low max voltage rating.
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 21609
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: SiC cascode
« Reply #9 on: October 28, 2022, 06:39:23 am »
The biggest, disappointment I guess, and more generally, is the kinda failed promise of high Tj(max) -- they're limited by packaging materials (150-175°C) same as everything else in plastic.  So the only exceptions are the few metal and ceramic packages (mostly MIL spec $$$).
I can't help but wonder what happens if you just don't give a toss about the plastic >:D

Perhaps not the brightest idea for an industrial product, but some DIY amplifier or dummy load on the other hand...

I remember cooking some 78L05 for many minutes at 300~400°C and it kept working afterwards.
Obviously, that's not the same as cooking for many days or years.

Y'know, it's weird, too -- I've had MOSFETs desolder themselves, melted breadboards with BJTs, and lord knows we've both soldered no end of plastic encapsulated parts whole, without mind for how hot they get (like reflowing SMTs).

I've also blown transistors almost exactly at the point where TJ exceeds, 160, 180C, something like that -- at least, based on given data.  Maybe the real internal temp was higher, who knows, but that's what the data said.

I don't actually know what the difference is.  Is it the gradient?  Does that cause differential expansion (especially above T_G of the resin -- which doesn't make up much of the plastic, it's mostly silica gel IIRC, but still, that's the most marked change in expansion of the material) and, like, crack off bondwires and then the whole game's over?  Surely it's not linear expansion alone (heating the whole part doesn't cause failure, at least not after several cycles, and for modest durations (10s of mins?)).


This was exactly another use case I was looking at. In a simple constant current source with a resistor they work from almost 0V up to max specified voltage. I can't find any other simple component/circuit that does that. I.e. common regulators and LED drivers don't work under 7-12 V for instance and old JFETs have low max voltage rating.

Depletion MOS are fantastic as well -- great for certain niches, like overvoltage protection, including input cross wiring or surge protection, power input load dump, etc.  Biggest downside is the huge spread in Vgs(th), or Vgs(off) for JFETs.  Also, the knee voltage is pretty gross (a couple volts), would be nice sometimes to have one tighter (a Vbe say, or even less like just some 100s or 10s or single ~mV), and as a consequence the Rds(on) is kind of a mess (can be hard to use with, say, a precision voltage divider?).

It's funny that HV Si JFETs used to exist (e.g. "FETRON" vacuum tube substitutes), went utterly extinct, and then DMOS brought them back, in PogMOS form.  But go figure; if they can leverage CMOS processes or whatever to make parts like these, I'm not complaining. :)

Tim
« Last Edit: October 28, 2022, 06:41:36 am by T3sl4co1l »
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 
The following users thanked this post: JohanH

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 14076
  • Country: de
Re: SiC cascode
« Reply #10 on: October 28, 2022, 06:45:17 am »
As a simple, relatively low power current source (or better current limiter) there are also depletion mode Si Mosfets for relatively high voltages (e.g. 800 V). Different from most JFETs they even have relatively small variations in the threshold.
 
The following users thanked this post: JohanH

Offline JohanHTopic starter

  • Frequent Contributor
  • **
  • Posts: 620
  • Country: fi
Re: SiC cascode
« Reply #11 on: October 28, 2022, 08:22:53 am »
Thanks, you are right that depletion mode MOSFETs should work also as constant current source with a wide voltage range.
 

Offline Circlotron

  • Super Contributor
  • ***
  • Posts: 3168
  • Country: au
Re: SiC cascode
« Reply #12 on: October 28, 2022, 10:52:25 am »
With those voltage, current and RDS(on) figures, a pair of them would make a great basis for a pretty serious class D audio amplifier.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf