I'm very familiar with the use of MOSFETs, I use BJTs on occasion, and I understand that the IGBT is just a combination of the two, but does anyone know when and why a particular device would or should be used?

For everything I do and can think of, I can use a MOSFET instead of a BJT or an IGBT with very few exceptions.

MOSFETs can switch faster and have a low on resistance (Rdson) so they make better switches than BJTs and IGBT which have a non-zero collector-emitter voltage (Vce). This also means that for high power they can typically consume less power. If I were driving 1A through a MOSFET with an Rdson of 250mOhms, it needs to dissipate 250mW. While a BJT/IGBT typically has a Vce of 0.5V and would dissipate 500mW. If the current requirement gets higher (say 10A), then that MOSFET definitely loses and dissipates more heat (25W vs 5W), however I can just choose a MOSFET with a lower Rdson (the smallest I've seen is 0.4mOhms) but I'd be hard pressed to find a Vce much lower with a different BJT/IGBT.

MOSFETs, BJTs, and IGBTs all can be setup in configurations that allow for amplification. MOSFETs can be a little trickier than BJTs and I assume IGBTs, but it is not impossible.

Controlling a MOSFET or IGBT is very power inexpensive as the control voltage uses almost no power, while a BJT constantly requires current for biasing, control, and amplifying. So in a power limited system a MOSFET/IGBT is the clear winner.

In low voltage systems, or voltage critical systems, you can deliver more voltage to the load with a MOSFET (again because of Vce). For instance if you have Blue LED (3.1V+) being powered in a system running at 3.3V, then a loss of 0.5V from the Vce of a BJT/IGBT is unacceptable, while the MOSFET will do it without any issues.

The MOSFET and I assume the IGBT however are susceptible to ESD on the gate, while the BJT is very robust. This is actually the main reason I still use BJTs.

The gate-source voltage (Vgs) is typically very low (20V typical, but I've seen as high as 30V) on MOSFETs (and I assume IGBTs), which is a disadvantage that a BJT does not suffer from. However, placing a voltage divider in front of the gate can resolve this issue most of the time.

MOSFETs allow current to pass both ways when on, and only one way (body diode) when off. This can be useful for some applications, but the body diode can be frustrating in others. The BJT (and I assume IGBT) are unidirectional and for the most part are fully off, when they are off. However the body diode is great for controlling something with inductance (motors, solenoids, relays, etc) as the back EMF generated when it is turned off can use it as a free wheeling diode, which reduces complexity and part count.

MOSFETs come in two main flavours (P-Type and N-Type). The N-Types are easy to control from low voltage sources as they are referenced to ground and can be used with a standard microprocessor pin connected to it without any issues; They are typically used on the low side of a load. P-types are referenced against the power source, which means you can't directly drive it from a microprocessor if the voltage source is higher than your micros. This can be resolved by using an additional N-type FET; but that adds cost and complexity. The P-types will typically control from the high side of the load.

So is there something I'm missing? Some limitations that I'm not aware of? Are some better at doing something than the others? Is it just that people are more familiar with the older technologies and not looking into other products? I'm genuinely curious and want it to be more nuanced then I have stated. I want to be a better designer, and I feel there must be something here that I'm missing.

 

See PDF.
Max.

 

but does anyone know when and why a particular device would or should be used?

BJTs are good for small and very large current applications, also good at high voltage. The downside as well as the beauty of BJTs is that their voltage drop at saturation is independent of current. This makes them undesirable at very small current and large current switching, as their saturation losses can be far bigger than the ohmic loss from a MOSFET. At very large current, however, their constant and current-independent saturation voltage drop is a savior.

MOSFETs are good with large current applications, mostly switching. At very large current, because MOSFETs are ohmic devices (behaving like a resistor), the conduction losses (I^2*R) becomes significant. and BJTs win over.

IGBT is just a BJT driven by a MOSFET.

 

BJTs are good for small and very large current applications, also good at high voltage. The downside as well as the beauty of BJTs is that their voltage drop at saturation is independent of current. This makes them undesirable at very small current and large current switching, as their saturation losses can be far bigger than the ohmic loss from a MOSFET. At very large current, however, their constant and current-independent saturation voltage drop is a savior.

MOSFETs are good with large current applications, mostly switching. At very large current, because MOSFETs are ohmic devices (behaving like a resistor), the conduction losses (I^2*R) becomes significant. and BJTs win over.

IGBT is just a BJT driven by a MOSFET.

I totally agree about power loss and how for a given BJT it can be really good at scaling to higher currents since the power dissipation is Vce * Ic (Linear) while the FET is Id^2 * Rdson (exponetial). The linear vs exponential growth definitely favours the BJT (and I assume IGBT as well). However, I can easily choose a FET with a lower Rdson. Instead of 250mOhms, I might choose one with 25mOhms, so instead of dissipating 25W at 10 amps, I'm now dissipating 2.5W, which is still lower by half of what a BJT is. I've even seen a FET that goes as low as 0.4 mOhms which would dissipate 40mW.

 

instead of dissipating 25W at 10 amps, I

Do the same calculation at 100amp, or 1000amp. Does it still favor the MOSFET?

That's the point.

 

See PDF.
Max.

The gist of what is in this PDF covers the differences between a MOSFET and an IGBT and can be summed up in these two pictures:

and


This is helpful, and it does add clearly defined boundaries where one product would be clearly better than the other. The arguments for why each device is better seems to be well thought out, which I greatly appreciated. It does not cover where the BJT fits into the design picture though.

 

Do the same calculation at 100amp, or 1000amp. Does it still favor the MOSFET?

That's the point.

I will concede that point. There does come a point where you can't escape the exponential relationship, and you have to switch over to BJTs and/or IGBTs. You can theoretically go really high with a MOSFET, but I bet the dollar cost would make it prohibitive before you hit that crossover point. So you would have to decide on cost versus heat dissipation well before there was a clear winner.

 

Hello I caught this thread late, but I agree with MaxHeadRoom. Also, when you can find Mosfets that can block greater than 600V - 900V the Rds_on tends to be sacrificed (worse conduction loss than the minority carrier IGBT). Also in applications like motor drives the IGBTs actually have a short circuit rating (for a short period of time) where Mosfets would be damaged under the same event. But in many applications I may have 900A pulses in my PWM, but the carrier is only 6kHz with an IGBT.

 

BJT vs MOSFET vs IGBT ==>
BJT vs MOSFET vs IGBT vs (Normally-OFF SiC Junction Transistor)
for example GA50JT17
vds=1700V id(T=25)=100A id(T>100)=50A Rds=20mOhm
It works like a bipolar transistor. Current amplification factor of about 100.
There is spice model. is fast enough (250kHz) td_on=17ns td_off=39ns

 

HDS_65, that makes sense when explained like that.

 

MOSFET used for solar micro inverters where IGBT used for high power solar/wind inverters.