why do mosfets burn when they are protected by fuses

Thread Starter


Joined Jul 18, 2012
I have bought broken DC to AC inverters to repair and I notice the internal fuses never blow, but the DC side mosfets always short and burn out.
For example an IRF 3205 has a 110 amp rating continuous. pulsed 390, at 100*c 80 amps.

It is protected by a group of 30 amp fuses.
Why do mosfets self destruct without the fuse blowing first?
If the fuse wont do a thing, why bother with internal fuses at all?

Here is a pic of a board I am replacing mosfets on.
So you can see the board layout. there are 2 input boards.
Six 30 amp fuses, so three 30 amp fuses on each board side.
There are four input mosfets on each side, total eight per board.
4 input mosfets share 90 possible amps until some fuse supposedly blows, which they dont.
I think these mosfets are turned on in pairs, each board has two drive chips, on one side of the board, one driver gate line runs to two mosfets, the other driver gate line runs to the other 2 mosfets. So they are alternatively turned on and off by the mosfet driver.
So that tells me each pair of mosfets would be exposed to 90 amps which is well within half their max rating.

All I can think, is fuses blow too slowly to do any good.
The mosfets see a huge current spike, and they short source to drain becoming low ohm resistors which limits the DC current so the fuses stay intact, while the mosfets burn.

Yes, the inverter supposedly have internal over current protection. Which apparently does not work reliably?

This inverter board burnt out when attempting to turn on a 16000 btu heat pump.
I turned on the heat pump switch and heard some sizzling and every input mosfet on both input boards burned. The AC output mosfets are ok.

So what I will do is put a relay on the heat pump, so that if inverter is powered on, the heat pump can not run.
A heat pump like this is a massive inductive load, does that somehow affect the overload protection built into the inverter circuit? Or is somehow that circuit defective? I have made no modifications to any such circuit in this inverter.

And would a 20 amp push button thermal breaker on the inverter's AC output blow before the internal mosfets blow? This inverter has a 3000 watt AC rating.


Joined Jan 15, 2015
Another consideration is will the heat pump run on the modified sine wave the inverter is outputting unless the inverter outputs a true sine wave. All of that if the inverter is capable of supplying the total load placed on it. As to the fuses? Possibly the fuses never reached their rated current. If this is a MSW inverter and it was driving a high load designed for a TSW input the MOSFETS could have failed due to other reasons. Things like an inductive spike from an inductive load? Just a guess.



Joined Mar 14, 2008
I was told when I first learned about semiconductors that "the transistor is always faster than the fuse".
This is because the thermal mass of the critical semiconductor junction is much smaller than the fuse element.
Thus, under overload, the transistor will overheat to the point of failure before the fuse heats up sufficiently to melt and open.
Fast blow fuses can help but the only way to reliably protect a transistor from overcurrent is to use a fast semiconductor current limit circuit in series with the transistor.

It could be that voltage spikes, not overcurrent is zapping the MOSFETs.

Thread Starter


Joined Jul 18, 2012
This MSW inverter, when it works, is able to power other loads like the stove top and oven, they are only resistive loads but still can draw some decent current.
It can easily run the 1600 watt microwave oven, which is an inductive load, the fridge, fluorescent lights, vacuums, circular saws, fans.

It runs everything I try but when it sees that Cruisair heat pump compressor, that is the end of it.
It has actually burnt twice when I tried to run the heat pump.
The first time it failed years ago, I thought maybe it had interconnected with my running Onan generator, but after I fixed it recently, I decided to test it again against the heat pump, and it failed again, so maybe it just cant run it without self destructing? I know that a heat pump compressor can draw significant startup current. When it is running this heat pump only draws 10 amps, which is a little less than the microwave.
It must be the massive surge that kills it? Would the mosfets be seeing a voltage spike coming from the heat pump motor? I did not even spin the motor that I could tell.

I know that class T fuse (fast fuse high amps) is what is recommended when installing an inverter. They are really expensive. Do you think class T fuses can actually save mosfets?

For example here is a 250 amp one


Joined Jul 18, 2013
Why do mosfets self destruct without the fuse blowing first?
If the fuse wont do a thing, why bother with internal fuses at all?

All I can think, is fuses blow too slowly to do any good.
You can get fuses know as Rectifier Fuses, they are constructed for very fast sweep through and fuse time.
Especially intended for semi conductor protection.
Last edited:


Joined Dec 13, 2013
motor starting loads are different from ovens and other resistive loads. a compressor load like that heat pump is rven greater, starting current depoends on how much pressure remains against the compressor motor.

Thread Starter


Joined Jul 18, 2012
motor starting loads are different from ovens and other resistive loads. a compressor load like that heat pump is rven greater, starting current depoends on how much pressure remains against the compressor motor.
Hi, yes they can demand a significant surge, but I thought the inverter OCP device built into these things would prevent this from shorting and burning mosfets.

The mosfets I am replacing with, this time I am using a different supplier. Cost a couple dollars more, but I just dont know if, it could be poorly made fake chinese mosfets, a defective OCP circuit in the inverter, a normal expected result from trying to start a heat pump compressor, or something else.

What I seem to recall years ago when I tried to run the heat pump that the inverter simply shut down on overload, which makes me wonder if the circuit board has an issue.
There are many current sense resistors on the output board and an IC chip with a few little transistors. So I wonder, how do these things sense current overloads and then shut down?

Is there a potentially non destructive way to test an overload current protection device on these inverters?

Looking at this board, what would be determining the overcurrent sensing? I think the transistors are driving the output mosfets which are on the back, which leaves that IC chip in the middle.
Or is it something on the main boost board?

The little ceramic disc capacitor is thoroughly cemented on top the IC, so I can't check its number easily to see what it is. Maybe it drives the mosfet drive transistors? Or maybe that is on the main boost board.

I watched a youtube video, and supposedly removing current sense resistors lowers the shut down point. Adding in resistors increases the shut down overload point.
Last edited:

Thread Starter


Joined Jul 18, 2012

Thread Starter


Joined Jul 18, 2012
Rectifier or semiconductor fuses are the fastest you will probably get.
Thanks, I see some have DC ratings.
And also can you get a fast acting fuse that will fit the existing fuse holders on the board?
That would be 12 times 30 amp fuses.

These IRF 3205 says can take pulse drain current of 390 amps, so if the fuse blew within the time frame of that pulse, maybe then the fuse would protect the mosfet?

Reading some more this herehttp://www.cooperindustries.com/content/dam/public/bussmann/Electrical/Resources/solution-center/technical_library/BUS_Ele_Tech_Lib_High_Speed_Fuses.pdf

talks of the datasheet listing saying
semiconductor data sheet specifies a maximum I2 t withstand for a semiconductor device. To offer short circuit protection to the semiconductor device, the fuse selected should have an I2 t let-through less than the I2 t withstand rating of the semiconductor device. High speed fuses have excellent current-limiting ability, as indicated by their low I2 t let-through and peak current let-through. High speed fuses are
However the IRF datasheet does not mention that parameter 'maximum I2 t withstand'
Anyone know more about this?

I think it is worth getting one of these fast fuses.
Yet reading some more, not sure a fast fuse can stop a semiconductor device from damage
This was specific to switching circuit DC power inverters. If I get one of these fuses, I certainly wont be testing if it does protect the inverter. Although if the mosfet is rated for 390 and this fuse blows at 350, then it should. I was thinking to get a LITTLEFUSE L15S 350AMP .

Does a fast fuse blow at its rated number? Or does it overshoot for a short time. If it allows an overcurrent beyond it's rated value, then it is no good for protecting semiconductors
Bi-polar Power Transistors and Darlingtons
It is difficult to protect power transistors with fuses. The power transistor
usually operates extremely close to its power limits of current and voltage.
Only a short excursion beyond the safe operating area will damage the
functional aspect of the transistor and even high speed fuses will not react
fast enough to protect the device. However, like IGBTs, when the function of
the transistor is lost the current is only limited by the low resistance of the
damaged silicon and very high currents result. These will melt any connecting
wires and will, in the case of press pack configuration, eventually melt the
silicon. The resultant arcs will cause the packaging to fail with catastrophic
results. Even though device protection cannot be offered by fuses, it is still
essential to use fuses to prevent case rupture and provide circuit isolation.
Last edited:


Joined Feb 24, 2006


Joined Sep 9, 2010
Fuses are too damn slow.
So true. It was an eye-opening moment when I looked up the specs of a typical (Littlefuse) "fast-acting" glass fuse. I had always figured a fast-acting fuse would quickly open at any current over its rating. Not even close! It might take half an hour at even double the rated amperage. And that's "fast"!

Screen Shot 2015-04-28 at 6.05.25 PM.png

Thread Starter


Joined Jul 18, 2012
Well, I called and emailed Littlelfuse technical and asked if a 350 amp semiconductor fuse opens at 350 amps and how long it takes.
I already decided to lock out the heat pump from being able to run if the inverter is on, so at least it wont inadvertently be blown up a third time by the heat pump. That is what the Onan generator is for and it starts effortlessly with the 6500 watt gen. What is nice to have is to run the microwave, vacuums, lights, the anchor winch, TV, electric grill off the inverter without needing the gen to run. See you can run those items with the main engines charging the batteries.

It should have an inline fuse in the DC supply line anyway. A fast fuse should blow fast if there was a catastrophic DC short. The inverter is powered off 5 feet (10 feet round trip) of 2/0 copper cable.

Ok, They sent me a PDF of that fuse, can someone look at it. It seems to say it will blow after 3 seconds but at what amperage, over a thousand amps? it seems way high?
This is a semiconductor fuse?


Last edited:


Joined May 10, 2009
Well, I called and emailed Littlelfuse technical and asked if a 350 amp semiconductor fuse opens at 350 amps and how long it takes.
You need to understand \(i^2t\) ratings. The datasheet gives these for your fuse (if I have the right one)

in general with devices in series, even with the same fusing rating, the one with the lower \(i^2t\) rating will blow first.

Just google 'i2t rating' and you will get lots of info, how long it takes to blow depends on the current pulse shape as well.

I had a quick look at your MOSFET datasheet, but I'm no expert in this area. Using the \(i^2t\) rating for your fuse you should be able to estimate how long it takes the fuse to blow and you can try to determine whether the device can withstand it. If not, look for a fuse with a lower \(i^2t\) rating.
...the fridge, fluorescent lights, vacuums, circular saws, fans.
You may wish to reconsider prolonged operation of induction motors and transformers (cip refrigeration compressors, fan motors and magnetic ballasts) on MSW waveforms --- excessive heating is a virtual certainty...

Best regards