FET Killer

[kinarfi]

Sorry folks, my apologies.
Kinarfi

 

[kinarfi]

try limiting Vgs to 10V, using a voltage divider. with the 10k resistor you already have, a 4.7 k resistor going from the gate to the switch should suffice. the 4.7k resistor and the gate capacitance will also serve to debounce the switch (to debounce it better add a 1 nF or 10nF cap across the 10k resistor). there are noise spikes every time you operate the switch, which the FET amplifies, and the burst of noise could be reflected by the load back into the FET. what is your load? is it a motor or other highly inductive device? if so, you might get an inductive spike that could kill the FET. yes there is a body diode in the FET, but that doesn't automatically protect the gate from being sent outside the +/-20Volt Vgs limit. when you check the D-S "junction" of a MOSFET, the gate should always be connected to the source during the test. also you said the device was heatsinked, which in the photo, it's not...

Thank you, in the end, I will be driving a motor with an H bridge and the bridge is protected by a 6 amp bridge rectifier, but I lost 2 IRF3205 FETs with more load than a 120 ma test light. I'll do what you suggest and also investigate the problem later today. The photo was taken after I removed the dead FET and used an IRF2805 for demo purposes, out of 3205s so I'm going to 2805s, the motor I'll be driving runs at 10 amps and stalls with over 70 amps. I really like your suggestion,
Thanks again,
Kinarfi

 

[ke5frf]

ke5frf, you appear to be trying to help, but I wonder if you know any more about FETs than I do.

Excuse me?

Why wouldn't I?
Judging by this thread, I wonder if you even know Ohm's law but I didn't say anything because I thought you were being mistreated, but I guess I was in error.

 

[kinarfi]

Sorry ke5frf My apologies.
Kinarfi

 

[kinarfi]

I did just find something interesting, this is from the data sheet on IRF2805
ID @ TC = 25°C Continuous Drain Current, VGS @ 10V (Silicon Limited) 180 amp
ID @ TC = 100°C Continuous Drain Current, VGS @ 10V 120 amp
ID @ TC = 25°C Continuous Drain Current, VGS @ 10V (Package Limited) 75 amp
I always did think that the TO-220 package had it's physical current limits in the amount of current that the leads can handle and the heat that can be transferred out, looks like it's 75 amps,
Kinarfi

 

[ke5frf]

ME TOO, and if I'm a genius, you are also, we both know something about FETs, maybe we're equal, It wasn't a put down!!!! and you have been respectful and supportive, Thank you, I didn't mean to offend.
Kinarfi

I have no idea what you are talking about, nor do I assume your knowledge about anything. I actually can't find a comment on the thread where I said anything about FETs, or your knowledge of them....only that people should be giving a novice trying to learn a break, and I stand by that opinion. That is a GENERAL comment, btw. I hadn't even looked at who the posters on the thread were, only the comments.

Please calm down and work on saving those poor FETs and drawing more accurate, standardized schematics.

 

[audioguru]

How are you going to cool the case to 25 degrees C? With liquid nitrogen?
Even if you use a huge heatsink with a fan, the case will be much warmer than 25 degrees C. Then the on-resistance will be higher than if the case was 25 degrees C.

 

[kinarfi]

unclejed613,
Can I ask you for some suggestions on other parts of my circuit involving the use of FETs?
**broken link removed**
**broken link removed**
In the bottom left corner is an N FET drawing similar to what I had when I Killed my FETs, minus the cap. I figured there was something weird going on and the noise answer sounds correct, I was suspecting that the gate was charging backwards (blue circle) when off and when the switch was closed, it discharged and recharged 'violently' punching a hole in the gate insulator and killing the FET.
Thanks for your help on the forum, I took you advice and added a resistor and a cap and then hooked the motor to the output and ran it several times and even let the motor stall for a few seconds while watching current via a .04 ohm resistor in line, running voltage was in the .5v area and stall was in the 3.1v area or 12.5 amps and 75 amps, the FET and the resistors were all mounted on a 3/16" X 3" X 4" piece of aluminum, which did get warm, but not hot.
On the other part of the drawing: This is from the data sheet on a LM339, at the OUT, I have two lines that are for analysis purposes only, one at a time. When the out is on, high, output transistor is off and it needs a pull up resistor which is by the FETs. When it goes low, output transistor on, This appears to be perfect of the P channel FET, but the N channel FET has no gate to source resistor to turn it off. Would you suggest adding any resistors or caps to this design? Or maybe even a transistor or another P FET to turn the N FET off?
Thanks
Kinarfi

 

[Hayato]

I think that the biggest problem is that you are driving inductive loads.

 

[Roff]

With an LM339 driving an NMOS, the pullup resistor turns it on. The saturating NPN output stage turns it off.

 

[kinarfi]

I think that the biggest problem is that you are driving inductive loads.

You may be right, but that is my goal, to drive a motor at 12 to 50 amps, do you have any suggestion? I know the FETs can handle the load as long as I don't turn them both on at the same time.
Thanks

 

[audioguru]

Are you controlling the speed of the motor with PWM?
The LM339 has an output current that is much too low to drive the high gate capacitance of Mosfets quickly so the Mosfet spends a lot of time being linear and getting hot.

Are you quickly switching the H-bridge so that the motor must immediately reverse without stopping? Then its current is extremely high, much more than 75A.

 

[Hayato]

And you get voltage spikes as high as 250V.

 

[kinarfi]

Yes the motor is PWM at aprox 100 HZ and maybe the input can be reversed very quickly, but it goes from forward to zero to reverse and at zero it has dynamic braking, there is a 6 amp full wave bridge rectifier across the motor to handle spikes.

 

[kinarfi]

With an LM339 driving an NMOS, the pullup resistor turns it on. The saturating NPN output stage turns it off.

Do you think the saturated NPN is sufficient? It's probably more so than the 5K resistor of the PMOS, correct?

 

[Roff]

Do you think the saturated NPN is sufficient? It's probably more so than the 5K resistor of the PMOS, correct?

The saturating NPN will turn off the NMOS relatively quickly, compared to the turn on speed provided by a 5k resistor. The opposite is true for a PMOS - relatively fast turn on, slower turn off. Neither case is probably fast enough to drive the big MOSFETs you need for driving your motor, as Audioguru pointed out.

 

[kinarfi]

The saturating NPN will turn off the NMOS relatively quickly, compared to the turn on speed provided by a 5k resistor. The opposite is true for a PMOS - relatively fast turn on, slower turn off. Neither case is probably fast enough to drive the big MOSFETs you need for driving your motor, as Audioguru pointed out.

Thanks,
Do you have any suggestions of how to speed things up?
Kinarfi

 

[Roff]

Thanks,
Do you have any suggestions of how to speed things up?
Kinarfi

Not specifically. Others here might have experience with specific parts. You can learn a lot by searching "mosfet drivers".

 

[Hayato]

Use BJTs in totem poles to drive the mosfets gates.

 

[kinarfi]

Does any know what the optimum PWM frequency for driving a motor is?