any guesses what is going on?
for those who don't want to watch the video
IPB107N20N3 as a low side switch for a 12v lamp, 10k gate pull-down, a dual lab supply for the 12V and the gate
after the fet has been ON it no longer turns OFF by dropping the gate voltage to zero, it stay ON at a few hundred mA and cooks itself
think it is unlikely it is a fake, four parts including one from mouser behaves the same way
Are you sure you are draining the charge from the gate when you turn it off. The lab power supply may drop to a Volt or so, and then just sit there.
You generally want a gate driver that both drives the gate hart to turn it on, and also will drive it hard to remove the charge when you want it to turn off.
Also, all those long wires could be causing oscillation which could overheat and damage the FET.
Jon
I wasn't me doing it I just saw the video, I had the same thought that the output from the supply didn't go to zero, but there is s pull-down and a voltmeter shows ~zero volt
it could be oscillations, but what triggered the test was that he had tried it in an amp and it failed in the same way, a different part number FET with similar specs worked
+1
Power supplies are 1-quadrant devices, some more than others.
Hang a voltmeter on the gate and see what's really going on. Or replace the 10k with 500 ohms and see if that works.
If it's not that, the thing's probably oscillating.
Cheers
Phil Hobbs
Den tirsdag den 26. september 2017 kl. 01.27.11 UTC+2 skrev snipped-for-privacy@gmail.com:
he had a voltmeter on the gate, zero volt
could be, but it failed the same way in a circuit were a similar spec'd fet works
Gate is spec'd at +/-20v, I don't think it's a logic level fet.
Cheers
If you look at SOA DC curve on page 4 , he probably went past it when he turned it on with that knob "winding it up slowly." He needs to current limit that supply to 10A.
On a sunny day (Mon, 25 Sep 2017 15:38:15 -0700 (PDT)) it happened Lasse Langwadt Christensen wrote in :
I did not wath the video, but long ago had problems with oscillating HV MOSFETS driving a lamp in linear range. Adding a few nF between drain and source stopped the oscillations.
ual lab supply for the 12V and the gate
ltage
inear
At Cambridge Instruments all MOSFETs got a 10R resistor in series with the gate.
If the MOSFET did oscillate, 10R might not be enough to stop it, but it lef t a space on the printed circuit board where you could fit a resistor big e nough to stop the oscillation. We had enough trouble with oscillating MOSFE Ts (sometimes when purchasing changed the supplier without telling engineer ing) that this became something hard to distinguish from a company standard .
I'm guessing some oscillation. They do make 'scopes for diagnosing this type of thing. :^)
George H.
they do? ;)
it isn't my problem, I'm just following the youtube saga of a fet that fails in circuit where other similar work and on the bench it act more like a triac than a fet
Is the fet now blown out?
Den tirsdag den 26. september 2017 kl. 17.53.57 UTC+2 skrev John Larkin:
maybe but it would have to very fragile.
straight from the esd bag from mouser, turn on with 8V gate voltage, load is a 12V/21W lamp all is a cool 25'C, turning the gate voltage down to zero verified with a meter doesn't turn off the FET the current just drops to 200mA and the fet start cooking
It's possible that the slowish turn-on, into the cold filament, excedded SOAR and popped the fet faster than that silly floppy thermocouple could have sensed.
Sloppy engineering!
Den tirsdag den 26. september 2017 kl. 18.49.23 UTC+2 skrev John Larkin:
the DC SOAR at 12V is something like 5A, he was using a 2A supply anyway he finally got the scope out and it was screaming at a couple of MHz
47R stopped it, the 6.8R that was in the original circuit was apparently not enoughdon't know if he is an engineer, he was repairing some kind of class-d amplifier
OK, one failure mechanism is partial conduction. Much bigger problem in IGBTs, but it can happen in FETs optimized for switching performance only. If you have a load well within the devices specs, but just barely turn it on, only part of the die may conduct. This causes localized heating, which can destroy the transistor in a cascading failure. Not turning off could be another way of saying failed shorted.
Jon
YUP, EXACTLY! Ramping the gate voltage SLOWLY to zero while under load will almost guarantee frying the FET. Most power FETs have a big slope in their safe operating area when run in the linear mode. Some are optimized for switching performance, and they do not guarantee even current distribution over the whole die near the gate threshold. (I have some experience with this, I have a ramp-up circuit that charges big caps in a servo amplifier, if the amp ever faults and is drawing current when the ramp-up transistor is shut off, it usually cooks the ramp-up transistor.)
Also, putting the gate at the threshold will put it in the region where oscillation is likely.
Read the manufacturer's data sheets, they often suggest NOT using most FETs in the linear area. If you DO, then be sure to derate it a lot.
The 200 mA load current could be from oscillation, or it could be the cascading thermal damage in the transistor.
Jon
..........
I would have posted earlier supporting the oscillation theory suggested by others but couldn't because my newsserver albasani has been acting up quite a bit lately.
The clue to me was that, although it was stated that the gate voltage was turned off again, the VM reading did not quite drop back to the previous value of essentially 0. The erratic reading indicates that the meter was trying to read AC at the gate on a DC range. IOW, oscillation.
On a sunny day (Wed, 27 Sep 2017 17:31:10 +0530) it happened Pimpom wrote in :
What is interesting, is the WHY it oscillates, apart from 'long wire theory'. If you look in those 12V bulbs, you see a little coil. Just a few turns. Get diameter, length, apply formula, get Cds abd Cgc and Cdg and see if frequency matches. In my case it was about 20 MHz... Just increasing the gate resistance without knowing 'why' is a bit of a lottery. Interesting is also the Q. As the little hot wire coil cools when the FET switches off, the wire resistance drops by a factor 10 or so.. Q goes up... Oscillation starts.... Anybody done slimulations on this? One could probably measure the bulb inductance, maybe later.... Or wait Car headlight .38uH cold
As to albasani: I left, they do not even reply to web forms. nntp.ioe.org is free and shows this group. news.datamas.de also no longer is usable.
The guys also blank out each group with things in it they do not like. Will only get worse.
WTF does he expect with those long leads making all the connections. This demo is unrepresentative of the MOSFET properly mounted.
Have something to add? Share your thoughts — no account required.
Ask the community — no account required