Winfield said:
Terry Given wrote...
. mtw6n100 (2)
.
. D --+--- d s --+- d s --+-- S
. | g | g |
. | +-|<|-' +-|<|-+
. | 220p | 220pF | 10V | 1.7kV 2A
. +--||--+--||--, | | use with care
. '- 1M -+- 1M -+-- | ----'
. 1W 1W |
. G
nice. technically its 2 FETs though, but best nobody look too hard at
whats inside a FET (or an IGBT)
Of course any self-respecting IGBT will beat the pants off it...
Not much safety margin for the latter two.
Ultimately it boils down to the reduction of stray inductance. I read an
article (IEEE PELS IIRC) not so long ago on the use of (Eupec?) 6.6kV
IGBTs at 3.3kV, about 4650Vdc. 2000V is a fair old bit of margin.
But when you consider the switching times (1us or so), the kA current
and the relatively high packaging inductance, 2kV can disappear real
fast. 100nH is a pretty small inductor, and 1kA switched in 1us thru
100nH is 100V. Thats a breeze to deal with.
Alas when things go wrong and the IGBT desats, the current is more like
10kA, and voila - there goes a kV or two, ka-Boom.
OK, so I downloaded this IGBT datasheet:
http://www.eupec.com/gb/2_PRODUCTS/2_1_ProductRange/pdf/db_fz600r65kf1.pdf
call it 500A, with 0.5us switching times. package inductance is 18nH so
a half-bridge is at least 40nH. Because of the high voltage, it will be
difficult to keep total inductance below about 100nH. Its a seriously
less than useful datasheet, no fault current info at all, but 10x is a
good guess.
So the previous calc is in the ballpark for this IGBT, meaining you
could build a 3.3kVac inverter with it running from a 4.7kV DC bus, but
you'd better pay close attention to minimising L and dealing with
desats, which will eventually happen (eg when one IGBT die becomes
sufficiently detached from the baseplate to overheat & fail short-circuit).
If the desat current is more like 20kA, then for 100nH the IGBT will be
over-voltaged and fail pretty much immediately upon turn-off.
so 2,000V aint that much headroom at all
690Vac gives a nominal 1kV DC bus, during regen it might pump up to say
1200V. But its a low enough voltage that the DC bus inductance can be
extremely small (1nH), so the limiting factor is simply the device
interconnects, which tend to be around the 20nH mark (eg SKiM, SEMiX)
the fault currents are more like 1-5kA, but switching times are around
0.2us, which gives 100V - 500V spikes. Once again they will work fine
during normal operation, but require attention under fault conditions,
especially during regen.
Cheers
Terry