MOSFET doesn't stay conducting

Thread Starter

evgen422

Joined Oct 14, 2015
3
Hey guys, so i have a 2KS2267 MOSFET and when i test it by applying positive voltage to GATE it does open, but after a few seconds resistance between source and drain starts increasing and it becomes turned off.
How long mosfet should stay conducting?
 

Hypatia's Protege

Joined Mar 1, 2015
3,228
Hey guys, so i have a 2KS2267 MOSFET and when i test it by applying positive voltage to GATE it does open, but after a few seconds resistance between source and drain starts increasing and it becomes turned off.
How long mosfet should stay conducting?
It should remain 'on' for (at least) as long as bias is applied --- following removal of (external) EMF, the gate will be controlled by residual charge -- many factors (including, but not limited to, ionizing radiation and humidity) greatly effect this... -- Succinctly: If the device is saturated during application of appropriate gate EMF sans measurable gate current -- it's working! :cool:

Best regards
HP
 

Sensacell

Joined Jun 19, 2012
3,432
Think of the gate as a capacitor - with some leakage.

When you touch the gate lead (with some voltage) to charge it up, the FET turns on, then the charge leaks away slowly once the wire is removed.

Touch gate and source leads at the same time with your fingers - the charge will be drained quickly.
 

alfacliff

Joined Dec 13, 2013
2,458
I ahave had some mosfets stay on for a day or two. if the mosfet is in circuit, the circuit will affect the g-s resistance. if out of circuit, clean the base of the mosfet to remove all flux, dirt, dead animls and such. a few seconds seems a bit short, could be a leaky mosfet.
 

Hypatia's Protege

Joined Mar 1, 2015
3,228
I ahave had some mosfets stay on for a day or two. if the mosfet is in circuit, the circuit will affect the g-s resistance. if out of circuit, clean the base of the mosfet to remove all flux, dirt, dead animls and such. a few seconds seems a bit short, could be a leaky mosfet.
Dead animals!?:confused: I take it this is an automotive application???:D

With marked curiosity...
HP:cool:
 

ian field

Joined Oct 27, 2012
6,536
You have to continuously apply the voltage to the gate.
The gate impedance of a MOSFET should be for all practical purposes, infinite - I'd expect the gate capacitance to remain charged for at least an appreciable length of time.

As someone else mentioned; humidity could leak the charge away, but other than that I'd regard the part as potentially suspect.

If the application uses an off the shelf gate driver that drives the gate fully on and fully off, the device may well work OK. However, the gate insulation layer is fragile and vulnerable to ESD, minor damage may not disable the device immediately, but can cause "in the field" failures.

The only acid test is; obtain a fresh part to compare it with. If a new part holds a charge on its gate, the one that doesn't is probably best discarded.
 

ian field

Joined Oct 27, 2012
6,536
I just checked a 2sk1940, it holds the gate charge for at least 1/2 hour sofar. I pulled it out of a small smps to check it.
Once or twice I've found MOSFETs with leaky gates, the drain current would switch on and off with applied gate voltage, but wouldn't hold as the gate charge leaked away.

There's probably less risk with anything working at less than 50V, but I wouldn't chance it on the chopper MOSFET in a mains powered SMPSU.

Typically MOSFET gates are rated somewhere around 18 - 20V, but I've seen a published article that suggests anything over about 15V incurs a kind of derating factor. Maybe half the life expectancy or something like that - but you're still going to have other components fail before the MOSFET.
 
I just checked a 2sk1940, it holds the gate charge for at least 1/2 hour sofar. I pulled it out of a small smps to check it.
IMHO such cannot be considered diagnostic - again, ionizing radiation (including low energy UV) -- and even high RH, smoke or other aerosols may produce rapid discharge - to say nothing of electrolyte residue contamination of the case --- If you suspect gate leakage, more satisfactory examination may be achieved via application of bias through a large (Ca. 47MΩ) resistor -- under such conditions Ig leakage of even 426 nA will drop fully 20V across the gate resistor guaranteeing zero bias...

Best regards and good luck!:)
HP
 

alfacliff

Joined Dec 13, 2013
2,458
i use a fluke dmm, no voltages high enough to damage the gate. just a quick check method of finding whether the thing is grossly defective, just marginaly worse than looking for craters in the epoxy. but it has been working for me. most power mosfets, unless they are blown up, tend to punch the gate and leak.
 

ian field

Joined Oct 27, 2012
6,536
IMHO such cannot be considered diagnostic - again, ionizing radiation (including low energy UV) -- and even high RH, smoke or other aerosols may produce rapid discharge - to say nothing of electrolyte residue contamination of the case --- If you suspect gate leakage, more satisfactory examination may be achieved via application of bias through a large (Ca. 47MΩ) resistor -- under such conditions Ig leakage of even 426 nA will drop fully 20V across the gate resistor guaranteeing zero bias...

Best regards and good luck!:)
HP
There's nothing to stop the TS from making sure the device package is clean and dry. Ionising radiation sufficient to discharge a MOSFET gate is probably something that needs to be investigated for safety reasons.

A leaky gate certainly hints at ESD damage to the gate oxide layer - this type of fault is known to worsen with further use.

Its also possible to compare it to a new example of the same type - it might have some peculiarity like a built in leak resistor.

But off the cuff, I'd advise against using it in any equipment where reliability is important.
 
Ionising radiation sufficient to discharge a MOSFET gate is probably something that needs to be investigated for safety reasons.
FWIW: Even so feeble a source as U238-bearing glaze (as applied, for instance, to original 'red' Fiestaware) will dissipate the charge on an unconnected MOSFET gate within milliseconds at a distance of 10cm - as will extremely low PE EMR (e.g. Ca. 4eV Woods lamp spectra, or, indeed, unfiltered sunlight) -- while safety and caution are never inappropriate, I feel the public hysteria evoked by the mere mention of ionizing radiation is both unwarranted and unfortunate... (Note: above comment not directed toward anyone in particular:))

A leaky gate certainly hints at ESD damage to the gate oxide layer - this type of fault is known to worsen with further use.
Agreed! Assuming the described rapid discharge is indeed owed to 'internal causes'...

Its also possible to compare it to a new example of the same type - it might have some peculiarity like a built in leak resistor.
Good idea! -- especially as ESD protection is often poorly (or non) documented...

But off the cuff, I'd advise against using it in any equipment where reliability is important.
Aye! When in doubt don't 'bet the farm' on a $5 part!:eek::D

Best regards
HP
 
Last edited:

ian field

Joined Oct 27, 2012
6,536
FWIW: Even so feeble a source as U238-bearing glaze (as applied, for instance, to original 'red' Fiestaware) will dissipate the charge on an unconnected MOSFET gate within milliseconds at a distance of 10cm - as will extremely low PE EMR (e.g. Ca. 4eV Woods lamp spectra, or, indeed, unfiltered sunlight) -- while safety and caution are never inappropriate, I feel the public hysteria evoked by the mere mention of ionizing radiation is both unwarranted and unfortunate... (Note: above comment not directed toward anyone in particular:))


Agreed! Assuming the described rapid discharge is indeed owed to 'internal causes'...


Good idea! -- especially as ESD protection is often poorly (or non) documented...
Actually - IME: protection devices generally are documented. A good example is the ongoing generations of dual gate MOSFETs used in TV tuners, the original family of devices that included the 40673 didn't have any, later generations like the BF981 or 991 etc usually have an inverse series pair of zeners per gate. Most data sheets I've seen described it in adequate detail.

I vaguely remember having seen the data sheet for a power MOSFET that had an input resistor, but it had various other protection features that set it apart from the usual types.
 
Actually - IME: protection devices generally are documented. A good example is the ongoing generations of dual gate MOSFETs used in TV tuners, the original family of devices that included the 40673 didn't have any, later generations like the BF981 or 991 etc usually have an inverse series pair of zeners per gate. Most data sheets I've seen described it in adequate detail...
Indeed - proper characterization of small signal devices, will, of course, require something approaching 'full disclosure'

Howbeit examination of large signal discretes with a component analyzer not uncommonly reveals undocumented passive components/semiconductor junctions (in addition to documented protection features) --- when asked (read: pressed) a Vishay Tech Rep stated that said features are typically not documented inasmuch as their effect upon device performance is not inconsistent with Spec -- fair enough! -- Though I'm bound to say it 'argues my point' - To wit: brief 'floating gate' discharge times are not necessarily indicative of faulty or failing insulated gate devices... That said, protracted discharge times are always a good sign!:)

Best regards
HP
 
Last edited:

ian field

Joined Oct 27, 2012
6,536
Indeed - proper characterization of small signal devices, will, of course, require something approaching 'full disclosure'

Howbeit examination of large signal discretes with a component analyzer not uncommonly reveals undocumented passive components/semiconductor junctions (in addition to documented protection features) --- when asked (read: pressed) a Vishay Tech Rep stated that said features are typically not documented inasmuch as their effect upon device performance is not inconsistent with Spec -- fair enough! -- Though I'm bound to say it 'argues my point' - To wit: brief 'floating gate' discharge times are not necessarily indicative of faulty or failing insulated gate devices... That said, protracted discharge times are always a good sign!:)

Best regards
HP
I'm sort of guessing that a built in leak resistor would show a steady reading on a DMM high Ohms range, and a damaged gate layer would probably fluctuate a bit.
 
I'm sort of guessing that a built in leak resistor would show a steady reading on a DMM high Ohms range, and a damaged gate layer would probably fluctuate a bit.
Agreed -- Though, in my experience, when present, intentional (undocumented) G-S shunts tend to range in the 'low hundreds' of MΩ - which being beyond the range of most general purpose VOMs/DMMs...

Its also possible to compare it to a new example of the same type
I begin to think that's the best approach from a practical standpoint:)

Best regards
HP
 

Aleph(0)

Joined Mar 14, 2015
597
I agree that test another with same part # is good idea if worried but sometimes differences in lots and runs too:rolleyes: It is the day for mosfets! Yours won't stay on and other guy's won't shut off:D!
 
Top