Electronics > Projects, Designs, and Technical Stuff
short circuit protection of IGBT?
<< < (5/7) > >>
T3sl4co1l:
If it's not due to latch-up (device continues carrying fault current despite Vge << Vge(th)), then it's due to thermal runaway alone (for all the various mechanisms that have been mentioned).

Typically the critical time frame is ~10us, so you want to be much faster than that in control response.  It is somewhat crazy for a thermal phenomenon to be quite so fast, but we are talking hundreds or thousands of amperes, and hundreds of volts, across a die much smaller than a typical MOSFET, say.  There's a metric shitload of heat (10s, 100s kW) across not much silicon here!

The desat detector I designed for the industrial projects, responded in about two microseconds, the slowness being intentional to avoid nuisance faults under hard switching.

The same protection strategy can be used on MOSFETs of course, but the expected voltage drop is usually higher (I set it at 50V for the industrial system), and the allowable fault duration is usually longer (less short-circuit current density, much bigger die).

SuperJunction MOSFETs weren't around back then; now that they're mainstream, I wonder what the allowable fault duration is.  They don't seem to be rating them [yet?] so it must not be too critical (over 20us? over 100us?).  They're much smaller dies than the previous generation, in high voltage ratings (I used 800 to 1200V FETs there), so the fault duration should be correspondingly shorter.

Interestingly, SJMOS also has an intrinsic current limiting characteristic, typically experiencing current saturation for any Vgs > 7V or so.  (This is true of ST's MDmesh II at least.)  This is comparable to small-die IGBTs offering modest current capacities.  Large-die IGBTs do seem to offer considerable short-circuit current capacity though, so beware.

Example: compare a 30A 1200V, TO-3P packaged IGBT being good for maybe 100A short circuit, to a 300A module being capable of >3kA short circuit.

Tim
T3sl4co1l:
Here's an example:
https://www.seventransistorlabs.com/Images/STP19NM50N%20Drain%20Output%20Curves.png
The curves for Vgs > 7V are pretty much identical so I didn't bother to measure them.

I haven't played with SiC JFETs so that's good to know.  Apparently there are two-ish styles of MOSFETs: one make has a small tempco, the other has a tempco massive enough to trigger thermal runaway (in switching applications).  I forget which is which (Cree/Infineon/etc.)...

Tim
coppercone2:
i am confused because blueskull says its damaged by the time this happens (what mechanism) while teslacoil is saying that there is something like 10uS to respond.

do you have enough thermal energy built up in there by the time its considered off that it just breaks after? why is it not broken then, are the atoms vibrating loose ?

does anyone have a GIF of what is happening in the transistor with fields, carriers and heat? when it breaks by energy shock. and whatever actually breaks (i guess it cracks or expands or oxidizes or diffuses or something else?) like the motion of the structure as its its being destroyed
coppercone2:
i cant find it free
\
but with what tesla says, if there is a fault current thats very high, can you run a current limited SCR to shoot the rail down so the IGBT gets something manageable?

maybe not all the way but at least take the edge off.
BravoV:

--- Quote from: coppercone2 on December 29, 2018, 06:04:01 am ---i cant find it free

--- End quote ---

Not my stuff, as its way too academic for this noob hobbyst :-[, a sneak peek.  :P
Navigation
Message Index
Next page
Previous page
There was an error while thanking
Thanking...

Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod