Also possible I read you wrong...
MOSFETs are heavily interdigitated too, of course; but having positive tempco Rds(on), this doesn't matter in saturation.
I think with IGBTs in saturation, as long as the Vce(sat)s match up close enough, and the thermal resistance between points is good enough, they're plenty stable. In other words, the tempco and incremental resistance aren't so bad. On die, those obviously are easily met -- now, discrete devices in parallel, especially if the thermal design is crap (e.g., separate heatsinks?), is where you can get problems.
Assuming saturated operation, that is. So it's going to be worse if it's linear.
10A at IGBT levels (Vce(sat) ~ 2V) would certainly be of some thermal concern, and could be much worse if it's linear.On the upside, if the voltage is low (say, 24V or less), the peak power will never be ridiculously high, easily managed by a few BJTs (I won't say
2N3055, but MJ15010 and such are quite fine) without much concern over 2nd breakdown. BJTs being the more traditional linear power amplification device. I guess MOSFETs are supposed to be more expensive than BJTs, which might be true watt for watt, but the difference probably isn't much anymore these days. You'll spend more on the heatsinks and hardware either way.Yeah. IRF740 for instance used to claim a DC SOA curve (back in the IR days, ca. 1992 or so I think). Today's Vishay (IR or Siliconix) datasheet doesn't give this.
On the upside, I tested one the other day at almost double the power rating it should've failed at. And this at high voltage. So that's pretty cool. The die inside is also surprisingly beefy. I doubt you can buy so much finished, functional silicon so cheaply in any other family of devices.
Newer transistors I've tested and blown almost exactly at ratings, which is kind of comforting to know, I guess.
Wow, that's yeechy!
I've seen audiophools using them in amplifiers, which probably doesn't create too big of a problem at low voltages (under 100V, say).
Yeah, even the worst I've seen (among those which still provided DC SOA) are still doing a sort of useful amount, say 10-100W. Out of a 500W device, not very useful, but... you'd only be using that particular device in this way if it was the last one in your parts box. :)
Usually true of BJTs as well, though I've had some cases where, for instance, I needed a PNP to source only a few mA in the 100s of volts, and even a 450Vceo device just wouldn't do it.
Reminds me, I should invest in some of those depletion mode FETs -- good for current limiting, active pullup/down, amplification and switching (for those odd jobs where you simply need a tube instead of a FET, but can't afford the heater power :) ), etc.
Tim