What's the main difference of the desat overcurrent protection for MOSFETs and IGBTs?

[hndxwzq]

Hi, all, I am totally confused on the overcurrent protection for IGBTs and MOSFETs.

1. Some gate drive chips have desat function, but it seems they can only be used for IGBT, am I right?
2. According to their datasheet, the protection threshold voltage is set around 7V for IGBT, but how about MOSFETs?
3. I know the fact that IGBTs and MOSFETs present different output characteristics, i.e. IGBTs have flat active region while MOSFETs usually don't, but how does the impact of this difference on the design of desat protection circuit


Thanks a lot

 

[E-design]

IGBTs when switching too much current, will start to desaturate. We detect this condition by setting a trip point where we need to switch off the device. For MOSFET's we normally just sense the peak current for our protection scheme.

 

[hndxwzq]

IGBTs when switching too much current, will start to desaturate. We detect this condition by setting a trip point where we need to switch off the device. For MOSFET's we normally just sense the peak current for our protection scheme.

But why we don't use the desat-protection for MOSFETs?

 

[FvM]

Desaturation overcurrent protection can work for high voltage MOSFETs. Their output characteristic isn't very different from IGBT in the high current range and a threshold of e.g. 7V would protect it quite well. Low voltage MOSFETs have Vds < 1 or even 0.5V at rated maximum pulse current. They would be destroyed by overcurrent before a protection circuit ever triggers.

 

[hndxwzq]

Desaturation overcurrent protection can work for high voltage MOSFETs. Their output characteristic isn't very different from IGBT in the high current range and a threshold of e.g. 7V would protect it quite well. Low voltage MOSFETs have Vds < 1 or even 0.5V at rated maximum pulse current. They would be destroyed by overcurrent before a protection circuit ever triggers.

Very good points. But I still cannot understand your statement "Low voltage MOSFETs have Vds < 1 or even 0.5V at rated maximum pulse current".
I know the fact that the low voltage MOSFETs have very small Rds(on), e.g. several milli-Ohms, but under overcurrent condition, MOSFETs will go to the constant current region (active region) and the Rds(on) would become very large. So the Vds should be high enough to trigger a desaturation protection circuit. I fail to find a commercial low voltage MOSFETs have Vds<1 V at rated maximum pulse current, and Could you give me a example if possible?

 

[steveelliott]

https://www.st.com/web/en/resource/technical/document/datasheet/CD00001526.pdf
STS12NF30L has Rds(on) of 8 mohm and Pd(max) of 2.5W.
Figure 4 shows Id of 30A @ 1V Vds for normal gate drive levels
1V*30A = 30 W.
toast!

 

[hndxwzq]

https://www.st.com/web/en/resource/technical/document/datasheet/CD00001526.pdf
STS12NF30L has Rds(on) of 8 mohm and Pd(max) of 2.5W.
Figure 4 shows Id of 30A @ 1V Vds for normal gate drive levels
1V*30A = 30 W.
toast!

Good comments. But I think Pd(max) can only be used to estimate the maximum current and voltage through the device during the steady-state operation.
I mean, during short-circuit transient, even though the instantaneous power is higher than Pd(max), if the time duration is very short (due to the trip of a protection circuit), the device should be OK. So I think the total energy Ed(max) instead of Pd(max) should be used to determine whether a device would be damaged or not under overcurrent condition.

 

[steveelliott]

"but under overcurrent condition, MOSFETs will go to the constant current region (active region)"

I think my point was that a there is a large range of Id that will cause device destruction long before Vds rises (see figure 4).

"I fail to find a commercial low voltage MOSFETs have Vds<1 V at rated maximum pulse current, and Could you give me a example if possible?"

However to answer your point, the STS12NF30L has a pulse current rating of 48A; at that current (again according to Figure 4), Vds is about 1 V for Vgs > 4.5V. I guess technically it's not less than 1V, but it's pretty close.

 

[Mr.Cool]

as usual i agree with FvM. i use de-sat for IGBT at any votlage and Mosfets (greater than 150V rated) . only thing that changes is additional series diode for the mosfet so as to build the votlage (Ids * RdsON is not big enough voltage to easily detect so it needs some help). you can do this detection using an integrated circuit gate driver whihc has the function built in which is easy or using discrete components but you need a blanking period (i.e. you only want to detect de-sat when mosfet commanded ON).

 

[hndxwzq]

"but under overcurrent condition, MOSFETs will go to the constant current region (active region)"

I think my point was that a there is a large range of Id that will cause device destruction long before Vds rises (see figure 4).

"I fail to find a commercial low voltage MOSFETs have Vds<1 V at rated maximum pulse current, and Could you give me a example if possible?"

However to answer your point, the STS12NF30L has a pulse current rating of 48A; at that current (again according to Figure 4), Vds is about 1 V for Vgs > 4.5V. I guess technically it's not less than 1V, but it's pretty close.

Thank you and I strongly agree with your explanation. But I am still confused.
For IGBTs, we usually set the knee point (e.g.7V) as the protection threshold. The corresponding current level at the knee point is generally larger than the maximum pulse current. So there is also should a large range (from the normal operating point to the knee point) that would cause device destruction before Vce rises. But they actually are not destructed according to our experience. Does this difference mean that usually IGBT is more rugged (i.e. better short-circuit capability) compared to MOSFETs?

 

[FvM]

I'm under the impression that the relevant differences between MOSFET and IGBT have been already explained. IGBT desaturation detection works in the constant current region of the characteristic. Generally speaking, you have similar behaviour in the "saturated" region of FET output characteristic (don't get confused by the terminology differences). The point is, that at least for low voltage MOSFETs, the respective region lies far beyond rated peak currents. This means, the FET will b destroyed before the protection triggers.

What you can do is to measure the MOSFET current by rdson voltage drop in "linear" region. This is done by some switched mode cotrollers. But it's much more sensitive to type and temperature variations. And the threshold voltage is rather low, e.g. a few 100 mV. So you can't use industry standard protected IGBT drivers.

In other words, if you read the datasheet details, you can answer the question yourself.

 

[mtwieg]

In principle there's no reason you couldn't effectively use desat protection for any MOSFET, though for low resistance devices you would likely have to set the threshold well below the actual desat point. Once you do that you're really using the FET more as a current sense resistor, since it's still operating in its linear mode. One issue with this is that the threshold voltage will be pretty low, meaning the sensing circuitry could be very sensitive to interference, especially in a hard switched power circuit. You would need a fair amount of dead time in the protection circuitry to stop false triggers, and by that point the protection might lose a lot of its effectiveness. However if you use low inductance packaging and good layout it's probably doable.

 

[Mr.Cool]

using high voltage diode which has 250uA for forward bias and a zener diode to set the voltage. should work. remember, the desat is for trigger between 125% and 200% FLA so there is a big range of trip which is required with mosfets and their small voltage.

put it in PSPice and see what comes up.