igbt vs. mosfet. why choose?

[JohnWondersWhy]

If you've read this post . . .



then you know I've had unfortunate success at killing mosfets. OK, that's over . . .

SO, for an h-bridge, at 20kHz, 10 amps, 200 volts . . . . there are so many igbt and mosfets availiable, I'm trying to understand which to choose. For my application, they seem interchangable.

What's the advantage to using a mosfet as compared to a igbt ???

 

[dknguyen]

Compared to IGBTs, MOSFETs switch faster, have lower losses at lower current levels, and don't suffer from latch-up. However, they cannot withstand the high voltages that IGBTs can and at extremely high current levels, the voltage drop across the MOSFET source-drain resistance can be higher than the gate-emitter diode voltage drop of the IGBT resulting in more losses at these extremely high current levels. But these current levels are so damn high in the first place that it's mostly theoretical. IGBTs can also suffer from thermal run away while MOSFET source-drain resistance increases with temperature reducing snowballing temperature-current increases. For the same reason, it is easier to parallel MOSFETs than IGBTs because the current is more apt to balancing itself out between devices (connecting like connecting resistors in parallel rather than connecting diodes in parallel).

It basically comes down to voltage ratings. MOSFETs can only withstand to about 600V while IGBTs can go up to about 1400V. SInce you are using higher voltages you also need less current for the same amount of power which means if you use an IGBT, you are probably working at very high voltages which means less current and the diode vs resistance loss advantage that the IGBT has over the MOSFET is probably something you will never encounter where the IGBT has less losses than the MOSFET.

20kHz is a little beyond the upper limit switching speed of an IGBT, and well under the max switching speed for a MOSFET. 200V is also well within the range of a MOSFET and 10A is so low current you'll get less losses across the MOSFET resistance than the IGBT diode.

 

[Speakerguy]

No reason for an IGBT here, there are plenty of MOSFET's that will work. Hitachi IIRC has IGBT's working in the 6kV range now, but I don't think they are on the market yet.

ETA: Have you scoped the gate-source voltage of the parts that are blowing? Is it the high side or low side ones, or both?

Also, I like the L6384 as a half bridge driver. You can increase the dead time between on/off switches with an external resistor.

 

[Styx]

yer this is def the relm of MOSFET's

One thing to note tho is MOSFET's have a triangular SOA which means when you are rating and choosing the FET for power switching you have got to be ever so careful about choosing a suitable part


IGBT's on the other hand have a all but square SOA (baiscally you can do full volts and full amps \o/)


also the latchup w.r.t. IGBT's are all but gone, 2nd Gen IGBT's (ie quite a few years ago) made it so the intrinsic thyristor would not latch in any expected normal or fault condition (ie it will fuse before latchup... they can easily take 10X their rated current with no latchup, in fact I can't remember the last IGBT I had that latched)

IGBT's are chosen when you have high volts, high amps or... high power


Also check the gate-drive of your FET's you gotta switch the FET's hard (ie a nice NPN-PNP push-pull stage is always good todo the actual FET gate-drive)

 

[Hero999]

You could add a 15V zener across the gate and source to absorb any high voltage spikes.

 

[causalitist]

try the IRFP series mosfets... check em out.. such as
IRFP4668
IRFP4768

and just for kicks since its only 75v:
IRFP4368 .. 1.8mOhm !!! 350amp 520watt

 

[dknguyen]

THis thread is more than a year old.

 

[tcmtech]

Still does have relivancey though!

The new gnerations of IGBT's on the market are fast catching the speed of the bigger mosfets. Many on the inverter based welder and plasma cutter power supplies I work on are running full H-bridges with Parallel IGBT's at 30-60 khz. And many use a pair of dual half bridges rated at 1200 volt 150 amps and running at 40 khz or more.

Still it amazes me to see them do those power, voltage and frequency levels efficiently as they do.
I have found International Rectifiers ultra fast series can be used as a direct drop in for many of the older mosfets they once used in inverter power supplies.

I have often rebuilt older model plasma cutter power supplies with the devices from the International Rectifier ultra fast IGBT line and found that the heat sinks run way cooler. Plus the old mosfets were usualy the IRF450 model with a 450 volt 15 amp rating while the IGBTS were 600 or 1200 volt and 23 or 28 amp rated. Depending upon which model I chose. Plus the IGBT's have internal reverse diode protection and higher disipation wattage capibility in the same physical package and pin configuration.
And now many are getting to be actualy much cheaper in cost even with the higher ratings too!

Here is an actual spec sheet on a 1200 volt 600 amp IGBT.

 

[Sceadwian]

Mosfets have issues at higher voltage that transistors don't. Thats' why IGBT's exist in the first place. I imagine with a proper driver at higher voltages (over 100 or so volts) transistors are better. Becaues the gate voltage requirements for Mosfets can become problematic. You start losiing a lot with a mosfet being driven in it's linear range from parasitics, which don't exists in the same way on a transistor.

 

[opeleg]

More Mosfet vs IGBT Questions

Bringing this old thread back to life a bit.

What about low voltage and high current.
I want to build an inverter with a 75V DC Bus and 1000A. Reliability is more important than efficiency.

It seems that mosfets would be optimal at this voltage, but again this is a high current to find suitable mosfets for. Its much easier to find IGBT's at this power rating.

Thanks in advance for any possible help.
Omer

 

[dknguyen]

Low voltage high current is MOSFETs. However, 75VDC might be considered high voltage. I know I consider it as high voltage. But industry might classify it differently.

YOu might just also be looking in the wrong place since there are way more FETs out there to choose from than IGBTs. The fewer number of IGBTs definately have a narrower target market. Because most of the IGBTs that are out there seem to be made for industrial sized applications, while MOSFETs cover a much broader range and so that might be why it seems easier to find what you want in IGBTs. You need to look for MOSFETs in large screw terminal power packages like the SOT-227. And look from manufacturers like IXYS that actually make industrial sized MOSFETs.

**broken link removed**

**broken link removed**

 

[opeleg]

Even in the links you sent me, there are very few models that come close to the current ratings that I need. I would have to parallel them. At 75V would it be ok to use IGBT's if efficiency is not my utmost concern? I fear running into problems building an inverter with paralleld semiconductors.

Thanks,
Omer

 

[shortbus=]

Even in the links you sent me, there are very few models that come close to the current ratings that I need. I would have to parallel them. At 75V would it be ok to use IGBT's if efficiency is not my utmost concern? I fear running into problems building an inverter with paralleld semiconductors.

Thanks,
Omer

Hi, in my short time learning about mosfets I found out one thing -"DATA SHEETS LIE!" For get what they say at on the specs. the listed ratings are only for IDEAL conditions. They won't work in the real world.

Check out this; FET Current Ratings -- Chuck's Robotics Notebook Read it closely and do the simple math(?) and you'll see what I mean. And IGBT's suffer from the same basic short comings when it comes to the data sheets.

I haven't figured out if the manufacturers do it to discourage DIY's, or to sell more parts to us(me?) dummies

Cary

 

[dknguyen]

Even in the links you sent me, there are very few models that come close to the current ratings that I need. I would have to parallel them. At 75V would it be ok to use IGBT's if efficiency is not my utmost concern? I fear running into problems building an inverter with paralleld semiconductors.

Thanks,
Omer

It's jsut the target market. Like I said, most IGBTs are made for industrial sized applications while most MOSFETs are not even if MOSFETs might be theoretically better within a certain range. In the end, you're stuck with what you is available and if that's IGBTs, no point looking for a MOSFET. One thing should be said though...you can't parallel IGBTs nearly as easily as IGBTs...and like shortbus said- "datasheets lie!". So you need to sit down and do the heat and power dissipation calculations to find out how hot the IGBT will actually get and how much current it will actually be able to pass (which will be significantly less than the datasheet). ANd if you can't find such an IGBT then you're stuck paralleling and you definately want to parallel MOSFETs rather than IGBTs.

I never look at those datasheet current ratings. I sit down with the my switching frequency, load current, rise and fall times, ambient temperature, maximum temperature, and the MOSFET on resistance AFTER it has heated up and do the calculation to find out how much current it will actually be able to carry

 

[shortbus=]

One thing that surprised me is Rds on. Every one is always quoting that in their answers. But no one ever says to look at the graph showing the
Rds/junction temp. Rds increases by around tree times when you put it into real world!

Also the package constraints are never explained. Showing large amperage ratings for a device that is only capable of a MUCH smaller amperage due to the size of the leads is very deceiving.

It,s taking me a long time to learn this stuff, but the more you understand the more you learn!

Cary

 

[dknguyen]

One thing that surprised me is Rds on. Every one is always quoting that in their answers. But no one ever says to look at the graph showing the
Rds/junction temp. Rds increases by around tree times when you put it into real world!

I thought I did in my previous post.

 

[shortbus=]

I thought I did in my previous post.

I didn't mean in this thread! Every one always quotes the data sheet number, not the real world number when answering questions to help forums.

Meant no disrespect toward you! Cary

 

[Ane]

One thing that surprised me is Rds on. Every one is always quoting that in their answers. But no one ever says to look at the graph showing the
Rds/junction temp. Rds increases by around tree times when you put it into real world!

Also the package constraints are never explained. Showing large amperage ratings for a device that is only capable of a MUCH smaller amperage due to the size of the leads is very deceiving.

It,s taking me a long time to learn this stuff, but the more you understand the more you learn!

Cary

 

[Ane]

Good day all.

Am new to the Forum, and am very aware that this is an old thread, but am interested in what was being discussed.

Am in need of a simple IGBT driver for Inverter. 5 to 10kva.

Am waiting to hear from anyone willing to sort that out for me.

 

[tcmtech]

An IR2110 or such similar type of driver IC will do what you need.