IGBT versus FET

Say same current and power rating..which is best, and why?

Reply to
Robert Baer
Loading thread data ...

You need to specify switching speed, and how often you want the device turn on and off.

IIRR MOSFETs are faster, and IGBT's better at coping with higher currents.

--
Bill Sloman, Sydney
Reply to
Bill Sloman

IGBT you can buy voltage margin cheaply. With MOSFETs it costs you Rds(on) and/or dollars.

IGBT are slow (maybe up to 10's of kHz some may be able to go close to

100kHz), MOSFETs can be very fast if you drive them hard enough.

IGBTs can be short-circuit rated (even at 600V). I don't think that's common with MOSFETs. That means you can protect a motor drive against some dufus wiring it up wrong.

IGBTs may require more complex drive circuitry (negative bias).

You can get complementary MOSFETs, same is not true with IGBTs I think.

You can get very low voltage drop from a MOSFET if you are willing to pay for the chip area- IGBTs drop close to 2V at full current IIRC. So, perhaps more power dissipation and associated costs.

For low voltage fast switching MOSFETs rule and IGBTs drool.

For big stuff runing off industrial mains especially, IGBTs start to look attractive.

--sp

--
Best regards,  
Spehro Pefhany 
 Click to see the full signature
Reply to
Spehro Pefhany

IGBTs have higher current density (higher than BJTs), so they are smaller (die), and cheaper per VA of switching capacity.

They're almost all made for 330V (plasma TV driver), 600V (240V mains) and

1200V (400/480V mains), and higher ratings too, for medium voltage induction heating and traction control.

Because of the higher current density, and more than complete lack of temperature compensation, they are utterly useless for linear applications. A lot of switching MOSFETs are, too, for the same reasons, though not as badly so as with IGBTs.

That sounds like old information -- they're available above 200kHz, at least discrete (up to TO-264).

What they actually do is, produce speed grades. Speed and Vce(sat) trade off, so you can pick the best combination of cost and losses for your application.

What they're varying is the ratio of BJT-ness to MOS-ness. The fast ones have an internal beta of 3, maybe, so a large fraction of the load current is majority-carrier (i.e., MOS-like), which turns off quickly, and causes the minority-carrier (BJT-like) part to turn off quickly too (with help from the same doping that makes superfast junction diodes do their thing).

The slower ones have a beta of 20 or more, and turn off very slowly (large >1200V devices take some microseconds).

(None of them are as slow as SCRs, as far as I know. But SCRs (and GCTs) are still king for absolute highest single-device VA capacity.)

Ehh, I think this is more for "just in case".

MOSFETs don't draw exponentially more current in saturation, only linearly so. So a fault will draw much less current from a given size of MOSFET, than it will from an IGBT. Both exhibit a linear (constant current) range, but the IGBT levels off at 5 or 10V, while the MOSFET might remain resistive until 20 or 40V. The MOSFET also has a much larger die, for the same current flow in the linear range.

I don't think you see MOSFET short-circuit ratings, because it's implied by the other ratings: you're hard pressed to draw more than the peak current rating (at least for nominal gate voltage), and dissipation is given by the SOA and thermal response.

IGBTs are much more power-dense, so it's a matter of life and death, whether they can survive even 5 or 10us. The MOSFET might be able to simmer for 25 or 50us.

In both cases, you can deal with this quite excellently with a desat protection circuit. I've used this on both types, and have had zero failures while the protection is functioning. The only difference is using a voltage divider for the MOSFET drain sense, to accommodate the higher "on" voltage.

Yes. This is necessary especially on the large and slow types. It may not be necessary for slow applications, or smaller (discrete) types, but proceed carefully in that case.

I think I once saw Toshiba P-IGBTs, years ago; but yeah, they're *really* obscure. Anyway, most places you'd use IGBTs, you wouldn't want a different drive voltage anyway -- if your gate driver is isolated, you might as well copy and paste it for all the channels!

Typical Vce(sat) ranges from 1.1V for 330V parts, up to 2-2.5V for fast grades in typical ratings (as you'd use in a switching converter for commercial or light industrial applications), up to maybe 4V for the largest (up to 4600V) modules.

Yup. A 400-550V MOSFET is basically equivalent to a 600V IGBT; if you need lower voltages, you're better with MOSFETs. (There are 300V IGBTs, but you aren't saving much by then, and can probably save more power by spending a few extra cents on a larger device.)

IGBTs are also great in pulsed operation, if you need a tremendous amount of power from a small package, with modest speed (t_r ~~ 100ns). That explains the 300V plasma display application.

Stanford and others have used them to drive magnetic pulse compressors, to generate high energy pulses for physics applications (particle accelerator deflection, plasma physics). There's a paper out there about how to get the most speed and power from your IGBTs without nuking them. They pulse the gates at something like 60V to make them move. :-)

IGBTs and motor control go together like PB and J. :-)

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
 Click to see the full signature
Reply to
Tim Williams

Do you have a link by chance?

Bye Jack

Reply to
jack4747

Mmm, this is related I think:

formatting link

Similar keywords as "igbt pulse drive slac" (no quotes, mix and match keywords and related) should show how others have been doing it. :)

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
 Click to see the full signature
Reply to
Tim Williams

Below 200 V, FETs look pretty good. They can be paralleled, switch fast, handle overloads well, you can get saturation voltages as low as your wallet can afford. Forward drop is a resistance.

Above about 200 - 250 V, the IGBTs start looking better. They are not as fast as the FETs, but newer ones are getting faster. In general, you can not parallel them without some small balancing resistance, you have to be VERY careful to never allow them to run in the linear region, even for 100ns or so. Short-circuit operation used to always end in catastrophe, but they have made them more robust. Still, when they are subjected to high currents, they can go into secondary breakdown, where they cannot be turend off. The worst thing is the positive temperature coefficient, so that if it ever gets into the linear region, either due to weak or slow gate drive or short circuit, then the center of the die gets hot, hogs all the current, and can fail to short in a couple microseconds.

And, you can't spend more to get lower saturation voltage, it is essentially a FET and BJT in a sort-of Darlington arrangement, so you can't get voltage drop much under 2 V. When you are switching 600+ V, that is less of a concern, and 600 V FETs don't have terribly low Rds either.

Jon

Reply to
Jon Elson

OK, will 'fess up. Linear, low current (0-500uA) high voltage (say 2KV) mode.

Reply to
Robert Baer

Thanks.

Reply to
Robert Baer

  • Thanks, i think. Had stupid idea to use one at 2KV, 0-500uA linear app.
Reply to
Robert Baer

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.