Why IGBTs Can Handle More Power Than Mosfets and BJTs

[likephysics]

I was reading about IGBTs. They can handle much more power than Mosfets and BJTs.
I thought Mosfet could handle more power than BJTs, since RDSon was munch lower than BJTs.
Can someone explain why?

 

[vk6kro]

Insulated Gate Bipolar Transistors have very high voltage ratings, so they can be used to switch high voltages.

Their current handling capabilities may not be any better than normal BJTs, but this extra voltage capability means they can handle very large quantities of power, depending on the device.

Extra current can be handled by buying larger devices or putting smaller ones in parallel.

 

[likephysics]

But Mosfets can be paralleled to get similar power capability.

 

[vk6kro]

But Mosfets can be paralleled to get similar power capability.

Of course, but presumably we are talking about individual devices of similar size.

You might have a Mosfet that can pass 20 amps and not be destroyed by 300 volts.
Or, you could have a IGBT that could only handle 10 amps but not be destroyed by 5000 volts.

Obviously the IGBT could control more power. 50000 watts vs 600 watts.

 

[Phrak]

Of course, but presumably we are talking about individual devices of similar size.

You might have a Mosfet that can pass 20 amps and not be destroyed by 300 volts.
Or, you could have a IGBT that could only handle 10 amps but not be destroyed by 5000 volts.

Obviously the IGBT could control more power. 50000 watts vs 600 watts.

vk6kro, you seem quit familiar with IGBTs. There should be a crossover region at lower voltages where power FETs become start to become advantageous over IGBTs. In other words, given the same package, and heat sinking, in the lower voltage regimes the FET will supply more power throughput (to resistive loads to keep things simple). I'm familiar with FETs but not IGBTs so I don't know where this crossover area resides.

 

[likephysics]

vk6kro, you seem quit familiar with IGBTs. There should be a crossover region at lower voltages where power FETs become start to become advantageous over IGBTs. In other words, given the same package, and heat sinking, in the lower voltage regimes the FET will supply more power throughput (to resistive loads to keep things simple). I'm familiar with FETs but not IGBTs so I don't know where this crossover area resides.

Textbooks usually mention Automotive application as an example, where FETs are more preferred than IGBTs. Automotive is probably 24V. But a lot of SMPS use FETs. I guess the cross over area is not just voltage, it's probably voltage and power.

One more question:
BJTs handle more power than Mosfets?
How can I compare. I can find out the Rds ON resistance of Mosfet, but how do I compare it with saturation resistance of a BJT?
Is the saturation resistance of BJT, less than Mosfet?

 

[uart]

Historically BJT’s were the first of the three to be available as high power switching devices with Mosfets following and IGBT’s being more recent. All semiconductor devices have (internal) design tradeoffs that are required to allow high voltage operation. With both BJT’s and the Mosfet these can really limit the devices usefulness.

For BJT’s the most pronounced limitation (for higher voltage devices) is lower current gain while for Mosfets it is higher resistivity. The OP mentions the lower resistance of the Mosfet but this is not actually true for higher voltage (200+ V) devices. Here BJTs typically have lower resistance than Mosfets (or to say more correctly, have higher current density than Mosfets at the same conduction voltage).

One of the limitations in all semiconductor devices as we go to higher voltages is the need to use lower doping density in one or more of the layers. This is because higher doping levels lead to shorter depletion regions with higher voltage gradients and hence a lower breakdown voltage. This design requirement can greatly increase the resistivity in high voltage devices.

Mosfets are majority carrier devices whereas BJT’s are minority carrier devices, and this difference is very important when considering the impact of the lower dopant levels (carrier concentrations). In minority carrier devices the downside of using lower doping density can be mitigated by minority carrier injection during on-state conduction. This is why high voltage BJT’s actually have better current handling capability than high voltage mosfets. The downside for BJT is that the base width has to be increased to accommodate high blocking voltage and this can drastically lower the current gain. For example, while low power BJTs can easily have current gain greater than 100, high voltage high power BJTs often have current gain as low as 10 or less.

 

[Phrak]

Textbooks usually mention Automotive application as an example, where FETs are more preferred than IGBTs. Automotive is probably 24V. But a lot of SMPS use FETs. I guess the cross over area is not just voltage, it's probably voltage and power.

I think you misunderstand. With the same package, power loss is the same, or nearly so before the silicon degrades or the device doesn't function as intended. If you prefer, the two technologies can be compared with power throughput. The breakovers are somewhere in the 80-300 volt region. Specifically where that is, I don't know.

 

[likephysics]

uart, thanks for the nice explanation.

 

[vk6kro]

vk6kro, you seem quit familiar with IGBTs. There should be a crossover region at lower voltages where power FETs become start to become advantageous over IGBTs. In other words, given the same package, and heat sinking, in the lower voltage regimes the FET will supply more power throughput (to resistive loads to keep things simple). I'm familiar with FETs but not IGBTs so I don't know where this crossover area resides.

I wish that was true. :)
I have been watching IGBTs for a while, but they have been mainly expensive industrial devices capable of thousands of volts at hundreds of amps. Costing hundreds of dollars, too.

I notice that Digikey have a $4.13 device the IRG4BC20MD-S which is capable of 600 Volts and 18 amps operation. Not at the same time though.
They have a Gate- Emitter turn on voltage of about 5 volts and a lot of input capacitance.
They also have about 2.1 volts across them when conducting 11 amps. So, 22 watts dissipation on a heatsink.
None of this makes them very attractive at low voltage.

However, the price is OK and cheap enough to get one or two to play with.

I checked the list of Mosfets in LTSpice and there is one that works at 800 volts, so that upsets the voltage range where you might use each device.
Normally, Mosfets operate below, say, 200 volts, so you would consider IGBTs above that voltage.

Rooftop solar panels can generate 300 volts or so, so a switch mode device to convert this to AC could probably use IGBTs.

 

[Phrak]

I wish that was true. :)
I have been watching IGBTs for a while, but they have been mainly expensive industrial devices capable of thousands of volts at hundreds of amps. Costing hundreds of dollars, too.

I notice that Digikey have a $4.13 device the IRG4BC20MD-S which is capable of 600 Volts and 18 amps operation. Not at the same time though.
They have a Gate- Emitter turn on voltage of about 5 volts and a lot of input capacitance.
They also have about 2.1 volts across them when conducting 11 amps. So, 22 watts dissipation on a heatsink.
None of this makes them very attractive at low voltage.

However, the price is OK and cheap enough to get one or two to play with.

I checked the list of Mosfets in LTSpice and there is one that works at 800 volts, so that upsets the voltage range where you might use each device.
Normally, Mosfets operate below, say, 200 volts, so you would consider IGBTs above that voltage.

Rooftop solar panels can generate 300 volts or so, so a switch mode device to convert this to AC could probably use IGBTs.

I've not yet had a design application requiring looking into IGBTs. However I installed an induction stovetop in the kitchen and one of the danged IGBTs already blew out. Arg.

But my baseline for FETS is International Rectifier. IR bought HexFet, where someone at HexFet was smart enough to notice that he could make a better power mosfet by adding together a lot of tiny FETS instead of using one big one with bad characteristics. Because transistors come with thee nodes, a 3-axis geometry was used; therefore "HexFet". They designed a waffer with a lot of hexagonally shaped, low power mosfets and wired them together.

Now, if you only look at page one of IR's data sheets you will be mislead. Their page one data is for a die at ambient temperature; like that's going to happen. This is their highest rated device which they quote at 300 volts. See https://ec.irf.com/v6/en/US/adirect/ir?cmd=catSearchFrame&domSendTo=byID&domProductQueryName=IRFP4242" . Click on 'datasheet'.

What's your favorited reference for IGBTs?

 

[cabraham]

There are 3 parameters to consider. At low voltages, FET is a better choice vs. IGBT. Rdson is low for low voltage FETs. FETs have a linear characteristic, Vds is a linear function od Id, since Rdson is linear. IGBT is non-linear, less than a 1st power relationship. For IGBT, twice the current, Ic, results in less than twice the voltage Vce.

At higher voltages, IGBT is better regarding conduction losses. The crossover point depends on current. At high currents, the voltage where IGBTs have lower conduction loss than FETs gets lower.

Finally we must consider switching loss. At 20 kHz, used in PWM motor drives, a FET has lower switching loss. At several hundred Hz, motor commutation frequency, switching losses are so low, the device, FET or IGBT, should be selected based on lowest conduction loss. At several hundred kHz, i.e. SMPS, the FET will almost always win, since switching losses at that freq for IGBTs is very large.

You must compute all losses carefully. Which device, FET or IGBT, is better is application dependent. Parallelling FETs is also a good option because each FET carries half the current. The conduction loss is Id^2*Rdson, so that each FET carries 1/4 the dissipation as 1 device alone would carry. Having 2 in parallel results in half the conduction loss vs. 1 device. Did this help?

Claude