Modern go-to switching devices

[John123]

What are the modern go-to power MOSFETS/IGBT's these days? You know for the usual SSTC or high voltage driver duty, back in 2010-2012 IRFP460 was usually recommended for mains stuff whilst IRFP260 was generally recommended for lower voltage drivers (even though the 200v drain-source breakdown voltage was a bit wimpy imo). Is 55 milliohm on resistance still considered low for a high voltage switching fet?

Also are IGBTs much better these days? Would be interesting to see what's changed over the last 8 years, the last "exotic" 600v, 33 milliom MOSFET I tried had a nasty drain to gate capacitance.

[klugesmith]

Veering slightly off topic...
Am reminded of continuous improvement in the thyristors (SCR's) used for High Voltage DC transmission by utility companies.


That snip is from a 4-year-old article on a web page by the ABB company.
https://www.abb-conversations.com/2015/11/thyristors-the-heart-of-hvdc/

It mentions a commercial order for devices rated to conduct 6,250 amperes and block 7,200 volts.  That's for a single die, which is round because it fills the silicon wafer.  You still need to stack dozens in series.  From a business point of view, the equipment is intended to be maintained and operated for decades.

[T3sl4co1l]

For what purpose?

There is an endless supply of types available.  Choose what's suited to the application.

For informal HV circuitry, I would dare be more worried that newer devices are more likely to fail.  Not at all because they've gotten less robust -- quite the opposite; but because they may be harder to use, because performance is so much higher (easily 5x better) than what those ancient types offered.

Typical case, VHF oscillation.  Modern SuperJunction transistors offer significant gain in the 100-400MHz range; not that they can be used this fast for large signals, but they are quite capable of oscillating this fast during transitions.  Which can contribute to spooky behavior (RF transmissions interfering with other circuitry), increased peak voltage or heating, or poorer current sharing.

This is, in part, a consequence of the greatly reduced capacitance, which isn't really all that reduced when comparing at low voltages (Vds < 20V say), but the difference is orders of magnitude at higher voltages.

And the faster switching allows higher dV/dt, dI/dt, and issues with snubbing (or the lack thereof).  MOSFETs still don't take kindly to repetitive avalanche breakdown.  In short, tight layout is much more critical.

For a given V, I rating, the die area is significantly smaller, so the power dissipation and thermal time constant are lower.

And simply using a lower Rds(on) part may invite higher peak currents and subsequent destruction.  Most HV circuits have no protection whatsoever, no way to address this.

There are also more materials on the market.  SiC offers high voltage ratings (at the expense of higher gate drive voltage); GaN offers insane switching speeds in extremely compact devices (making them exponentially more sensitive to layout).  Both perform better than Si on a per-area basis, i.e., the dies are extremely tiny -- making them that much more sensitive to excessive power dissipation.

Tim

[Mads Barnkob]

There are also more materials on the market.  SiC offers high voltage ratings (at the expense of higher gate drive voltage); GaN offers insane switching speeds in extremely compact devices (making them exponentially more sensitive to layout).  Both perform better than Si on a per-area basis, i.e., the dies are extremely tiny -- making them that much more sensitive to excessive power dissipation.

That being said, SiC can perform truly amazing when designed right.

Here Steve Ward is pushing some 10-15 kW of power through 6! *SIX*! TO-247 devices.

[T3sl4co1l]

Oh, no doubt.

Tim