[SOLVED] Individually protect parallel MOSFETs

[red_alert]

I'm looking for a method to individually protect (overcurrent) multiple parallel connected MOSFETs.

It's a H-bridge design and I want to use a separate fuse for every single MOSFET. The easiest method could have been putting the fuse on the drain circuit but I'm using a single non insulated aluminium heatsink for all parallel MOSFETs thus all drains are already shunted.

If I put the fuse in source circuit, I'm affraid that when the fuse will blow (to protect the MOSFET) the source terminal will float so the gate might catch a parasite overvoltage or something.

If I put the the gate-source resistor (1k) before the fuse, could I avoid this scenario?

Is there any other method to individually protect those MOSFETs?

 

[D.A.(Tony)Stewart]

the current imbalance is affected by the following parameters:
- Driver resistors for turn on and turn off
- Length of wires
- Parasitic inductance
- Distance of resistors from the gate pad

Disabling one switch places extra burden on the other switches. They should be balanced by design and use asymmetic gate resistance with diodes or similiar such that they gate turns off faster than turn on..
Adding small conductive ESR to each RdsOn reduces the overall variance and possibility for thermal runaway. The track resistance ought to be the standard deviation of the RdsOn but much less than the RdsOn.

 

[red_alert]

I'm not worried about the current imbalance durring the normal operation (or at design time).

But a possible fault of a single MOSFET from a paralleled group could lead to two situations:

1. The affected MOSFET has drain-source shortcircuited; that would affect (shortcircuit) the other leg (group of MOSFETs) of that half bridge.

2. The affected MOSFET has open circuit between drain and source; that will put a higher current load on the rest of MOSFETs from the same group.

That's why I want to individually protect them (I'm going to use 6-8 paralleled MOSFETs in each group and I don't want to blow them all together). Does a source fuse affect the normal operation?

 

[dick_freebird]

I don't think you need to worry about protecting individual
FETs, because

- FETs tend to share nicely to begin with

- a damaged FET that stops listening to gate control
probably won't listen to a gate control based protection
either

- you won't like the cost or performance impact of a drain
based protection (source, even worse).

I think you're better off selecting components that will
survive whatever causes D-S shorts in power FETs,
than trying to use a cheap one to protect a cheap one.
Presuming that your plan does not involve expensive
plus expensive....

 

[red_alert]

I need to switch 200A (with 600A peaks) and I'm going to use 100A rated MOSFETs (6-8 of them) to decrease RdsOn.

I really don't have any other option but to put them in parallel (there's no single chip to match those specifications). Beside, it helps to distribute the heating across the entire heatsink.

Anyway, I wasn't thinking of complex Drain/Source protection circuits (nor Gate either) but to use a simple fuse in series with Source terminal.

Is there a problem if Source is floating (if the fuse is blown) and there's still PWM voltage applied to the Gate?

There will be a 1K resistor between Gate an Source though. Also, I could put a resistor across the fuse to (somehow) keep the Source to ground.

 

[FvM]

As already explained by dick_freebird, protecting parallel MOSFETs with fuses is a bad idea and will most likely end up in catastrophic failure if a fuse ever trips.

Paralleling MOSFETs is pretty standard in power electronics and usually works with just individual gate resistors to supress parasitic oscillations. A problem of unequal current share due to threshold voltage differences will only arise in linear operation, if the gate voltage rise/fall time is very slow or the drive voltage too low. Otherwise source inductance will enforce uniform current share during edges and the positive rdson t.c. static current equilibration.

It's neither clear against which kind of overload you want to protect the MOSFETs nor if a fuse can actually help.

 

[red_alert]

It's neither clear against which kind of overload you want to protect the MOSFETs nor if a fuse can actually help.

It's not overload but abnormal (faulty) behaviour of a single MOSFET during normal operation. It might be oversensitive to transient or any other parasitic voltage/current. It is not a commercial design; it's for my personal use and I just want to minimize any collateral damages.

But could you explain why a fuse won't help (or might become a source of a catastrophic failure)?

How else to avoid a half-bridge switch shoot-through when the other switch get damaged (shortcircuited) or in case of a driver failure (simultaneous ON commands for both switches)?

 

[schmitt trigger]

In the past I've used very succesfully SmartFets from STMicroelectronics, that incorporate internal protection circuitry.

OnSemi, Infineon, International Rectifier and others also make similar products.

 

[dick_freebird]

A fuse in the source leg makes the FET unable to hold
Vgd to safe levels and will likely result in a punched out
gate oxide (and possible drain-to-driver leakage or even
high current depending on how the blowout goes).
You would trade an overcurrent fault for an overvoltage
fault further upstream.

A drain-leg fuse would at least cut it apart at the high
voltage point and fail more safely.

 

[red_alert]

I want to make some tests with a big toroidal transformer (5 kW) for a sine wave inverter.

Till I find the right startup SPWM software procedure to avoid that powerful inrush current I want to protect the MOSFETs against any failure (they weren't a cheap ones). I could not make any tests with fewer MOSFETs (for current rating reason) so I have to play safe.

I might not use any MOSFET protections in the final design but, once again, I want to get there "alive" (from MOSFETs point of view).

There are any other solutions (instead of using fuses)?

 

[schmitt trigger]

Based on many years of power electronics experience, with my fair share of high-power blowups, this is what I would recommend:

Make a low voltage, low current prototype first, on which you can actually validate your theories and test your software first.
If anything blows up, it is less dangerous and a lot cheaper to replace a 1 amp, 50 volt mosfet than several 100 amp, 100 volt mosfets.

This sentence alone should be enough for you to click the "helped me" button.

 

[chuckey]

I guess you are using PWM, use a HF current transformer* on each FET drain (or source lead), sum the transformer voltages each via a diode, use a comparator operating on the voltage to close the drive down. it will beat a fuse and you can adjust it
Frank
* or a field sensitive chip if its 50 HZ

 

[red_alert]

It is a good idea but 4 x 6 current transformers is just too much.

Maybe I'm going to use a single fuse on the (common) drain to at least protect one group of MOSFETs from shoot-through failure condition.

It seems like I cannot protect individual MOSFETs (using simple/compact/cheap circuits) so I have to live with that. Maybe after one more week of learning about "toroidal transformer inrush current" I would be more confident in my tests.

Thanks everyone for those useful advices.

 

[D.A.(Tony)Stewart]

Over Current protection might look like this in SmartFets

 

[red_alert]

Maybe I could try that durring tests only. Thanks for sharing!