Hi

has anyone observed the following effect?

In a circuit a gate driver (FAN73901) drives a halfbridge. Rgate are 15R. The high-side Mosfet reads 0R from gate to source. The resistance measured from driver output to the source reads 15R. Conclusion: The Mosfets gate is shorted to the source (internally).

After removing the Mosfet (with a lot of heat involved because it's mounted on an aluminum PCB) the Mosfet seems to be working fine. No short circuit on the driver output either. Put a new Mosfet in and the whole thing works, so it's like a temporary gate to source short circuit.

Obs: Deadtime had been adjusted too short so there was some 100ns cross-conduction when that happened. 180A peak. Anyway this was a 180A continuous current Mosfet.

 

A MOSFET that big has a decent gate capacitance. Your meter might take a while to charge it up, looking like a near short in the meanwhile. Wild guess.

 

If it's not gate capacitance, check the rest of your circuit...I've been fooled by shorted parallel components (parallel FETs, protection diodes, driver devices, etc) etc. in MOSFET and power BJT circuits before.

 

It's unlikely that the MOSFET itself experienced a temporary short, so I would look elsewhere, as suggested.

 

Depending on the internal structure, some of these power MOSFETs can latch (which is recoverable).

I had this phenomena with a TL494 dc/dc converter when working it at the upper voltage limit, changed to a large BJT which equally worked, and of course does not latch.

 

The meter read from gate to source (in-circuit) immediately 0 Ohm and stayed there. Measuring directly at the output of the driver I measured the gate resistance. Replacing the Mosfet made it work again. Only that the short at the Mosfet disappeared.

As for latching, I thought this would be a temporary effect on "energized" Mosfets, as it happens on the internal gate driver output FETs. THIS is an effect I observed once but the latched output (always ON) was reset when removing the drivers power supply. (not in this circuit)

I just checked the suspected MOSFET with a power supply and it doesn't seem to be defective.

Remains a mystery to me.

Btw, it was this one: http://www.mouser.com/ds/2/196/IPB017N06N3_Rev2.2-81637.pdf

 

I had a bunch of mosfets on the bench lately, after changing out a bunch, and checked them with my fluke87, they checked shorted s to d, it turns out that I checked them g to s first and turned them on. they stay on for hours unless you short the g to s or reverse the charge. probably wouldnt do that in circuit tho.

 

fried a couple of irf9540's lately S/C g s and d. the fluke 77 never lies

 

I had a bunch of mosfets on the bench lately, after changing out a bunch, and checked them with my fluke87, they checked shorted s to d, it turns out that I checked them g to s first and turned them on. they stay on for hours unless you short the g to s or reverse the charge. probably wouldnt do that in circuit tho.

Mine was reading a short from gate to source.

 

You can't trust an in circuit reading.

In TV repair we got into the habit of desoldering the component lead before testing, provided it was a single sided PCB and the lead did not touch the copper.

On double sided PCB we would cut the PCB track with a scriber, then test the part.

You could tell a TV PCB that had been through fault finding procedure by an oldtimer, it would have tracks cut and rejoined in numerous key places.

 

You can't trust an in circuit reading.

In general I agree, especially if you don't know the circuit. In this case it was pretty obvious.

A completely discharged/deenergized circuit measuring 0Ohm from gate to source with no other component in parallel. It was the Mosfet proven by the fact that after replacing it it works now... Unless the insulating layer on the aluminum PCB was damaged, but I don't believe that at these low voltages.

We will see if after the deadtime adjustment if the same will happen again. The customer had observed the same effetct on a different unit , a temporarily disabled High-side switch. I couldn't ask them to actually measure inside the unit though.

 

I have seen faulty power semis that were intermittant, ie; shorted sometimes and not other times. Usually affected by heat.

Looks liek you ruled out the other possibilities now, so that might be the case?

 

If this is a thing, I want to know more about it, because it would explain some truly weird MOSFET behavior I've observed from time to time...anyone know anything about the physics involved in an intermittent transistor short??

 

Probably a tiny fused (melted) area of silicon somewhere on the chip.

In most cases the fused short would be permanent, but there can be a possibility that the fused blob is not fused to both + and - layers of the silicon.

It's just fused to one layer and touching the other, so thermal or electrical effects can make the short come and go.