bipolar better than mosfet?

[Hank]

So yesterday I wired up a simple circuit to use an N-channel mosfet as a switch
to turn on a mechanical bell consisting of a coil of wire and a contact
interupter thingy. The bell draws about 5 amps at 24 volts at roughly 50% duty
cycle and I am guessing it pulses at about 15-20hz (haven't bothered to measure
this). Anyway, I drove the gate of an IRF2807 through an npn darlington
optoisolator whose collectors were connected to 5V regulated from the 24 volt
supply (the opto was from my junkbox, a hp 6n139 and only rated to 18V). I
connected the emitter to the gate of the mosfet and had a 1k resistor to ground.
So the gate of the mosfet would charge up to ~4 volts when the opto was on. I
also had a 1n4004 diode across the bell leads - not a very fast diode I know,
but the mosfet has an inherent DS diode feature anyway.

Ok, so, powering up, energizing opto, bell comes on for about 2 seconds then the
IRF2807 fails open and bell turns off.

???

I replaced the IRF2807 with a 2N6078 pulled off some old circuit
board...basically a TO66 medium power npn. Wired it's base to the opto emitter,
it's emitter to ground, and collector to the negative side of the bell. Works
like a champ. Hasn't failed in 45 minutes of off/on operation.

Sooo...

1) Maybe the IRF2807 was partially damaged by static electricity from hanging
out in my junk drawer for a few years?

2) 4 volts on the gate was not enough to drive the n-channel device into
saturation and thus Rds was significant and caused an internal meltdown in 2
seconds?

3) The IRF2807's inherent diode protection and the additional 1n4004 diode were
not sufficient to protect against the nasty voltages coming from the mechanical
bell solinoid/interupter thingy?

4) In noisy situations like this, stick to bipolar parts?

Thanks for any tips

Hank

 

[Spehro Pefhany]

On Thu, 27 May 2004 16:11:32 GMT, the renowned "Hank"
<nospam@nospam.com> wrote:

So yesterday I wired up a simple circuit to use an N-channel mosfet as a switch
to turn on a mechanical bell consisting of a coil of wire and a contact
interupter thingy. The bell draws about 5 amps at 24 volts at roughly 50% duty
cycle and I am guessing it pulses at about 15-20hz (haven't bothered to measure
this). Anyway, I drove the gate of an IRF2807 through an npn darlington
optoisolator whose collectors were connected to 5V regulated from the 24 volt
supply (the opto was from my junkbox, a hp 6n139 and only rated to 18V). I
connected the emitter to the gate of the mosfet and had a 1k resistor to ground.
So the gate of the mosfet would charge up to ~4 volts when the opto was on. I
also had a 1n4004 diode across the bell leads - not a very fast diode I know,
but the mosfet has an inherent DS diode feature anyway.

Ok, so, powering up, energizing opto, bell comes on for about 2 seconds then the
IRF2807 fails open and bell turns off.

Failed *open* ? That's odd. You were NOT putting enough gate drive on
it, but I would not expect it to die quite that fast, and I would
expect it to fail shorted. You may have a gate-source short due to
layout issues in your setup.

???

I replaced the IRF2807 with a 2N6078 pulled off some old circuit
board...basically a TO66 medium power npn. Wired it's base to the opto emitter,
it's emitter to ground, and collector to the negative side of the bell. Works
like a champ. Hasn't failed in 45 minutes of off/on operation.

Sooo...

1) Maybe the IRF2807 was partially damaged by static electricity from hanging
out in my junk drawer for a few years?

2) 4 volts on the gate was not enough to drive the n-channel device into
saturation and thus Rds was significant and caused an internal meltdown in 2
seconds?

3) The IRF2807's inherent diode protection and the additional 1n4004 diode were
not sufficient to protect against the nasty voltages coming from the mechanical
bell solinoid/interupter thingy?

4) In noisy situations like this, stick to bipolar parts?

Thanks for any tips

Hank

Best regards,
Spehro Pefhany
--
"it's the network..." "The Journey is the reward"
speff@interlog.com Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog Info for designers: http://www.speff.com

 

[John Larkin]

On Thu, 27 May 2004 16:11:32 GMT, "Hank" <nospam@nospam.com> wrote:

2) 4 volts on the gate was not enough to drive the n-channel device into
saturation and thus Rds was significant and caused an internal meltdown in 2
seconds?

Yes.

3) The IRF2807's inherent diode protection and the additional 1n4004 diode were
not sufficient to protect against the nasty voltages coming from the mechanical
bell solinoid/interupter thingy?

Maybe that, too.

4) In noisy situations like this, stick to bipolar parts?

Not really. You need a lot less gate drive power for the fet, and they
should work fine if properly protected.

Bells draw a lot of peak current! Try to measure that (or just DC coil
resistance, and calculate), and check the fet saturation curves to
make sure it can stay hard on at the peak current.

John

 

[Joerg]

Hi Hank,

Adding to what John said, I'd hook up a scope to the gate and see if the gate
voltage dips back down when the drain pulls down. That works best in single-shot
mode. If the drive impedance isn't low enough that can happen and if long enough can
cause a FET to be destroyed. That is a common failure mode in switch mode supply
prototypes.

Regards, Joerg

http://www.analogconsultants.com

 

[Jan Panteltje]

On a sunny day (Thu, 27 May 2004 16:11:32 GMT) it happened "Hank"
<nospam@nospam.com> wrote in <Uqotc.7558$nJ6.6945@nwrddc02.gnilink.net>:

2) 4 volts on the gate was not enough to drive the n-channel device into
saturation and thus Rds was significant and caused an internal meltdown in 2
seconds?
Possible

5A is a lot, use voltmeter or scope to see if voltage is low if on.
Anyways, use a higher drive voltage (10V would be nice).
JP