MOSFET fried driving solenoid

[edrean]

So I'm busy with an automotive project whereby a PIC12F675 monitors the vehicle battery voltage. When the voltage increases to above 13.2V (indicating that the vehicle has been started), it delays for 10 seconds and then has to switch on a solenoid. The solenoid in question is a Cole Hersee 85A (http://www.colehersee.com/home/item/cat/174/24059/). Its coil seems to draw around 700 mA when powered. When the battery voltage drops below 13.2V it disengages the solenoid. I am using an IRF9Z34N MOSFET to drive the solenoid coil.

Everything worked great until I fried the MOSFET. At first I could successfully switch the solenoid on and off for a number of times. At that stage I was only testing my circuit, so only the solenoid coil was connected while the solenoid contacts were not connected to anything. When that worked, I then connected the solenoid contacts (one end to the vehicle battery and the other end to a load). On the first test the solenoid switched on and wouldn't switch off again. The MOSFET now gets hot and won't switch off. I'm not sure if it was the switching on or the switching off that killed it...

I have attached a schematic showing my connections from the PIC to the MOSFET. I have also attached a graph of the rise time of Vgs on the MOSFET (2V and 50us per div). I was worried that the rise-time was too slow and that consequently the MOSFET didn't turn on fully quickly enough, but that is evidently not the case. This measurement was done WITHOUT any load connected to the MOSFET. Don't know if that will make a difference.

One assumption I made is that because this MOSFET has a built-in diode, I would not be needing an additional diode across the solenoid coil. So I did not connect a diode across the solenoid coil. Was my assumption wrong here?

Any ideas as to why I am breaking my MOSFET? I only have one left so I would prefer not destroying another one...

 

[Mithrandir2008]

Well I really can't guess why it fried and I have fried quite a few(admittedly N-channel 500V mosfet's) till date

But I do have one piece of advice for the circuit. It is always better practice to keep some sort of isolation between the PIC and gate driver, just in case something goes wrong and your controller gets screwed too. A simple optocoupler IC(6N137) will do nicely. It may not matter since you are using only a 13V source, but still.

And the rise time shouldn't matter since, as you mentioned, the body diode should be able to take the current transfer path. It may slightly increase your losses but not really enough to fry it. I don't suspect anything is wrong with your MOSFET/gate circuit, it maybe an issue with the solenoid and I have no clue about it.

Hope that helped

 

[edrean]

Thanks for your advice re the optocouplers. I will strongly consider using optocouplers for my outputs as protection.

So I am correct that the MOSFET's own built-in diode is enough flyback protection from the collapsing solenoid coil?

So what killed it?
Too much current? The coil only draws 700mA and the MOSFET is capable of much more.
The coil's magnetic field collapsing? There is diode protection for that, right?
Overvoltage? I'm on max 15V.
Overheating? Rdson is 0.1 Ohm. At -13V the MOSFET should be turned on fully if I'm not mistaken. Even if the Rds is 0.5 Ohm it gives a power dissipation figure of 0.245W. At 62 degrees/W that give a temp of 15 odd degrees. So no problem there, right?

So what am I missing?

If I were to sacrifice my remaining MOSFET, what should I monitor so that I could at least get closer to the cause before it also dies? I'm thinking of monitoring the coil voltage to look for any voltage spikes that may occur...

 

[ronv]

I would add the diode.

 

[Diver300]

You should have a freewheel diode. Without it, the energy stored in the solenoid ends up in the MOSFET each time the circuit is turned off.

There are other suppressors that you can use.

 

[edrean]

I would add the diode.

You mean add a flyback diode across the solenoid coil? I have a 1N4007. That would do, right?

You should have a freewheel diode. Without it, the energy stored in the solenoid ends up in the MOSFET each time the circuit is turned off.

So you're saying that the built-in diode in the IRF9Z34N is not enough?

 

[ronv]

The 4007 should do it. In therory the FET should be ok but strange things happen when the flyback pulse starts flying around.

 

[OlPhart]

The internal diode is a byproduct of the transistor structure, it helps, but it's best to have another (Schottky) in parallel. However, this only protects drain->source.

You still need protection from common-mode negative pulses on turnoff (gate->ground). At the solenoid put a Schottky diode in reverse bias (conducts negative pulse to ground) and (I like) a zener diode (~14V) across as well to shunt high volt positive spikes to ground. This keeps the gate between ~-.7V <-> ~+15V.

When I burn a MOSFET, it's usually I didn't protect the gate well enough and it punches a hole in the source-drain junction. Good Hunting...<<<)))

 

[edrean]

At the solenoid put a Schottky diode in reverse bias (conducts negative pulse to ground) and (I like) a zener diode (~14V) across as well to shunt high volt positive spikes to ground. This keeps the gate between ~-.7V <-> ~+15V.

I have a 1N5822 in my bin which I can test with and if it works I will then probably settle for the 1N5817.

Good Hunting...<<<)))

Thanks! I hope it's a short hunt.

 

[edrean]

You still need protection from common-mode negative pulses on turnoff (gate->ground). At the solenoid put a Schottky diode in reverse bias (conducts negative pulse to ground) and (I like) a zener diode (~14V) across as well to shunt high volt positive spikes to ground. This keeps the gate between ~-.7V <-> ~+15V.

On second thought. I think I'm confused. I have drawn a diagram showing how I understand you. But I don't think I did understand you correctly, because I cannot see how this would protect the gate of the MOSFET. If my diagram is wrong, could you please explain in more detail?

Thanks!

 

[audioguru]

Without having a diode across the coil, the drain voltage of the Mosfet suddenly goes to -50V or more when the inductance produces a spike when it turns off. Then the drain-gate capacitance causes an over-voltage at the gate.

With an ordinary 1A diode across the coil then the drain voltage rises to only about -0.7V and then the gate does not have an over-voltage spike.

You do not need another diode across the Mosfet. If the switching is not very fast then an ordinary rectifier diode (1N4001) can be used across the coil.

 

[edrean]

Without having a diode across the coil, the drain voltage of the Mosfet suddenly goes to -50V or more when the inductance produces a spike when it turns off. Then the drain-gate capacitance causes an over-voltage at the gate.

With an ordinary 1A diode across the coil then the drain voltage rises to only about -0.7V and then the gate does not have an over-voltage spike.

I see...

You do not need another diode across the Mosfet. If the switching is not very fast then an ordinary rectifier diode (1N4001) can be used across the coil.

The MOSFET will switch on 10 seconds after the PIC senses that the vehicle has been started. Then when the PIC senses the vehicle's engine was stopped it will turn the MOSFET off. So the switching frequency will be very low. Not even once per second.

I was wondering about the reasons for destroying the MOSFET....what are the chances that when the solenoid contacts opened (coil turns off) it momentarily drew an arc on the contacts and this somehow got into my circuit and destroyed the MOSFET? I'm just wondering. Everything went rather well until I connected power and a load to the contacts of the solenoid. Will the flyback diode help with that, or is that where the Zener diode comes in to clamp a voltage spike that may be induced on the coil side? How do I protect against that, if something like that is even possible?

 

[audioguru]

When the drain suddenly goes to -50V or more then the capacitance in the Mosfet between the drain and the gate will drive the gate voltage too high which damages the Mosfet.
The diode across the coil dumps the turn-off current from the inductance into the power supply so the drain voltage spike is very low.

Maybe the contacts are also driving an inductance that caused a high voltage arc to jump to the Mosfet. A wire feeding a load that is a few feet long has inductance.

 

[edrean]

Maybe the contacts are also driving an inductance that caused a high voltage arc to jump to the Mosfet. A wire feeding a load that is a few feet long has inductance.

What should I do to protect from that?

 

[audioguru]

Insulate (use spacing) the contacts from the coil.

 

[OlPhart]

Howdy, I'm not about to differ with audioguru, I simply prefer to be redundant in my protection. High side aspect is just to keep gate spikes in rein. G.H <<<)))

 

[edrean]

Thanks to every one of you for your contribution to help solve my problem.

I used the 1N5822 that I had lying in my bin as a flyback diode across the solenoid coil terminals. I added load to the solenoid contacts and retested a couple of times and I haven't damaged the MOSFET yet. So I'm seeing significant improvement. It works as it should now.

I actually feel kinda stupid for not seeing that the internal MOSFET diode would not provide adequate "freewheeling" for the solenoid coil. I am certain that is what caused my issue. I'll also add some Zener diodes for extra robustness in the final build.

Thanks again!