MOSFET Failure Problem

[Patrick Johnson]

I am attempting to drive 8 logic-level MOSFETs (IRLR2905) with an 8-bit bus
switch (FST3244). I have a 300 Ohm resistor in series with the gate of each
MOSFET. Fast switching time isn't a necessity. I also have a 30V Zener
diode connected across the drain-source to handle reverse transients. The
MOSFETs are being used as low-side switches for 12V automotive fuel
injectors (solenoids). I have 4 of these particular circuits built and am
having trouble with all of the MOSFETs failing short-circuit. The first
circuit worked for a little while, the second has yet to fail, and the third
and fourth failed immediately after power-up.

The MOSFETs have a maximum Vds rating of 55V (I am only using 12V) and a Id

20A (I am only using 1.2A). The bus switch has a logic-level output (5V).

I can't see how I am stressing these parts. Any ideas of what may be
happening?

 

[Nermal]

One possible problem: you are switching the MOSFETs too slow, the is,
they are staying too long in the linear region where the power
dissipation is too high.

 

[petrus bitbyter]

I am attempting to drive 8 logic-level MOSFETs (IRLR2905) with an 8-bit bus
switch (FST3244). I have a 300 Ohm resistor in series with the gate of
each MOSFET. Fast switching time isn't a necessity. I also have a 30V
Zener diode connected across the drain-source to handle reverse transients.
The MOSFETs are being used as low-side switches for 12V automotive fuel
injectors (solenoids). I have 4 of these particular circuits built and am
having trouble with all of the MOSFETs failing short-circuit. The first
circuit worked for a little while, the second has yet to fail, and the
third and fourth failed immediately after power-up.

The MOSFETs have a maximum Vds rating of 55V (I am only using 12V) and a
Id
I can't see how I am stressing these parts. Any ideas of what may be
happening?

Seems you zeners are not fast enough to handle the reverse transients you
mentioned. Why didn't you use (fast) diodes across the solenoids? Are your
zeners not damaged? Also make sure the gates of the MOSFETs are never
floating.

petrus bitbyter

 

[petrus bitbyter]

One possible problem: you are switching the MOSFETs too slow, the is, they
are staying too long in the linear region where the power dissipation is
too high.

Yet another possibility. Especially if the MOSFETs have no heatsink. (You
can try with your finger. When you burn it, it's too hot and too late.) Most
of the times the driving circuit is rather too fast then too slow but both
are possible.

petrus bitbyter

 

[Robert Baer]

One possible problem: you are switching the MOSFETs too slow, the is,
they are staying too long in the linear region where the power
dissipation is too high.

1) Reduce that gate resistor to 10 ohms; that will help speed up the
turn-on and turn-off times (RC with 300 ohms is 0.5uSec).
2) Add "snubbers"; the flyback voltage from the inductance of those
solenoids is the major cause of the failing FETs.
Use a diode with series resistor no less than the solenoid coil
resistance, and no more than 10 times the solenoid coil resistance.

 

[Patrick Johnson]

Thanks for the replies...

What exactly happens when the MOSFET gates are floating when Vds is applied?
I think this may have been the problem. I was using a test circuit when the
MOSFETs failed. The test circuit applies a purely resistive load on the
MOSFETs so I'm thinking flyback voltage may not have been a problem (at
least for now). It is possible that the gates were floating when Vds was
applied.

My thought with the zener diode was that by allowing a higher flyback
voltage before clamping would allow for a faster closing of the fuel
injector. Would a snubber circuit consisting of a fast diode and 2 Ohm
resistor (the fuel injectors are ~2 Ohms) in series be better? How would
this affect the injector closing time...or would it?

Thanks!!
Patrick

 

[Robert Baer]

Thanks for the replies...

What exactly happens when the MOSFET gates are floating when Vds is applied?
I think this may have been the problem. I was using a test circuit when the
MOSFETs failed. The test circuit applies a purely resistive load on the
MOSFETs so I'm thinking flyback voltage may not have been a problem (at
least for now). It is possible that the gates were floating when Vds was
applied.

My thought with the zener diode was that by allowing a higher flyback
voltage before clamping would allow for a faster closing of the fuel
injector. Would a snubber circuit consisting of a fast diode and 2 Ohm
resistor (the fuel injectors are ~2 Ohms) in series be better? How would
this affect the injector closing time...or would it?

Thanks!!
Patrick

There is a tradeoff when using any kind of flyback voltage snubbing.
At FET ON time, the current rises to maximum "set" by the supply
voltage and the total circuit resistance (mostly the solenoid resistance).
At turnoff, that current will continue due to the collapsing field.
No snubbing will result in a large voltage induced across the
solenoid inductance, which will add to the 12V and cause the FET to
avalanche (guarantee on that point).
But that means the solenoid is energized only during ON time.
A simple diode across the solenoid allows that current to flow thru
it and limits the induced voltage to near 1 volt peak...but...that means
the magnetic field will last longer due to that "discharge" current thru
the diode (about double the FET ON time).
A fast diode does not help; what is needed is to dump that energy in
the form of heat and thus remove it fom the inductance as fast as
possible...but keep the current low enough so the solenoid cannot
continue to be activated.
So, on the bench, activate a solenoid with a typical mechanical load,
then slowly reduce the voltage until it "drops out"; measure that
(probably in the region of 0.4 times rating).
Also measure the solenoid current at full voltage and at dropout point.
Figure diode series resistance from full voltage divided by dropout
current and add 10 to 20 percent.
That way, the maximum diode current is less than that to keep the
solenoid on, and one would have minimal flyback voltage.
My guess would be 20 ohms.

The tradeoff: flyback voltage VS solenoid energize time.

 

[neon]

di/dt or dv/dt is getting into it . forget zeners use snubbers to kill transient try a 100ohms and .22 mfd in series acroos each device with a scope i could tell. This way i am guessing. the gate 300 ohms is optional but smal cap to gnd is benefitional.