MOSFET Switch - 24VDC at 10A

[Mike Warren]

I need to switch 24VDC at a maximum of 10A using a MOSFET and am wondering what I need to do to protect it. The load will vary and may be inductive. It would typically be switched no more than once every few seconds or minutes, but certainly no faster than 10 times per second under worst case abuse.

I also need it to be able to be wired to switch high side or low side and would prefer to use an N-channel.  I do have 40V available for the gate switching.

The attached picture is the beginning of what I have in mind. The idea is that it can be wired drain to 24V and use the source as the high side output, or ground the source and use the drain as the low side switch.

The 9.1V zener is to make sure the G-S breakdown voltage is not exceeded, but there will no doubt need to be some better protection, perhaps a MOV and/or a R/C snubber across the D-S?

Any ideas?

[BiOzZ]

what i would do personally at a high current inductive load is put a larger diode across the fet and a zener clamp (that your doing) ... short of setting up a gate drive system that's as good as your going to get without knowing the exact characteristics of any reverse voltage or voltage spikes ... if you are still having problems stick it on your scope and try and get the peak and frequency of such and you can go threw the process of making a low pass filter but i highly doubt that will really be necessary

but if its a simple switch switching just DC there is nothing wrong with using a simple fet and 2 resistors! ... but you might not pass UL doing that XP

[Psi]

You could also add a TVS diode across the fet to absorb any positive spikes that might exceed the mosfet voltage. It would also block negative voltage from an inductive load, however that might be asking a bit much from it. Maybe use a bidirectional TVS and a regular diode to cap the inductive spike at -0.5V before the TVS conducts.


I would also use a TVS to protect the gate instead of a zener.
A large spike may kill a zener but a TVS diode can handle hundreds of watts.

[BiOzZ]

Yeah TVS diodes do work a treat ... a bit of an overkill for the purpose if your getting them in the 100W range they can run a cent each in 1k quantities but if your just making 1 than they can work a treat and you can scatter them around

but i think for this a zener would work fine as if you fry the zener you probably need to rethink your design

[dodad_guy]

Source to the supply, drain to the load.  Make sure the pulse width is short enough to prevent transformer saturation or use some kind of load limiter.  The MOSFET will heat up if the pulse width to a transformer is too long.

[dannyf]

Any ideas?

You may want to do some calculations.

To reduce conduction losses, you have to use a switch with low Rds. Those switches tend to have very high gate capacitance. Since you have used very high gate resistance (44k total), the switching losses will be significant. ie. you can get into a situation where the average power dissipation is low, the switch can be burned out when it is opening or closing.

May suggestion would be to look carefully into the specs for the switch and the zener.

[Mike Warren]

but if its a simple switch switching just DC there is nothing wrong with using a simple fet and 2 resistors! ... but you might not pass UL doing that XP

I don't need to pass UL, but why would that be a problem?

Thanks for the replies, everyone. That's given me something to think about. I am concerned about switching time.

[BiOzZ]

I don't need to pass UL, but why would that be a problem?

Thanks for the replies, everyone. That's given me something to think about. I am concerned about switching time.

if anything breaks you can be shoving the input voltage threw the gate ... it does happen so i still suggest a larger diode across the fet but but if your not triggering it with anything vital it works ... another workaround is to have it triggered with a opto coupler

but if you have a zener or a TVS just shove one of those in there

if its triggering slowly and timing is not key you can also put in a cap or an LC filter in there to take any spikes ... inductive loads can act as a boost inverter and shove as much as 1kv in to your design if the inductor is large enough so be careful ... a suppression diode like i said would take most of that away

[jmcdonald]

Could you use a automotive high side switch instead?
http://www.irf.com/whats-new/nr100304.html

[Mike Warren]

Could you use a automotive high side switch instead?
http://www.irf.com/whats-new/nr100304.html

That looks nice, but unfortunately I need to have this capable of high side or low side switching.

[Mike Warren]

Yesterday afternoon I built a test circuit using a NTD5806NT4G because I happen to have a bunch of them from another project.

Something weird is happening.

For the test I'm just running into an electronic load set to 10A CC and only using a 15V supply because that's all I have that can supply 10A.

The MOSFET is failing intermittently when I switch off and on the 15V supply. I've gone through 6 MOSFETs so far and still haven't been able to determine exactly why it's failing. Triggering the oscilloscope on the power supply I never see the voltage on the gate exceed 10V above the supply voltage so I'm obviously missing something.  I ran out of time yesterday so will get back to it on Monday.

The MOSFETs are always failing short (approx 1 Ohm) drain to source and about 100 Ohms to the gate.

My first thought was that the turn on/off time was too long, so I tried feeding a 10Hz signal into the control FET. I ran it for over a minute and the MOSFET was slightly warm, but nothing I'd be concerned about.  It immediately failed when I cycled the 15V supply.

Now I'm going to be wondering about this mystery all weekend.

[dannyf]

Now I'm going to be wondering about this mystery all weekend.

When you design your circuit to do everything, it does exactly nothing.

[Mike Warren]

When you design your circuit to do everything, it does exactly nothing.

I guess I need this to work like a solid state relay (although I don't need isolation). A Google search only showed me usage rather than how they are designed internally.

[RossK]

First thing, put a PTC in the high side. This will save your FET's from many common development mishaps - although i'm not sure if it will help with this case as its a bit of a at the moment. What kind of load are you driving? are you just shorting the power supply? if so add a current limiting resistor, your PS might be losing constant current control as power switches off (just a guess). On that note do you blow fets if you disconnect the power supply instead of switch it off? Are you driving a DC motor or other inductor? if so be sure to have a diode across it.

It'll be interesting to learn what is causing this  O0

[envisionelec]

Reduce your gate resistor to 10-22 ohms.
Check that the 15V isn't doing something weird at switch-on.
Drive the gate with a totem NPN/PNP or just a shunt wired PNP so it quickly turns off.
Page 14 -- http://www.ti.com/lit/ml/slup169/slup169.pdf
Hang a UF diode across D-S. The body diode is only so quick.

[Mike Warren]

Reduce your gate resistor to 10-22 ohms.
Drive the gate with a totem NPN/PNP or just a shunt wired PNP so it quickly turns off.

I don't think my problem relates to normal switch on or switch off. I can run a 10Hz square wave into the gate of the driver MOSFET for any length of time I want and the switch MOSFET only raises about 10C above ambient (finger test only).

Hang a UF diode across D-S. The body diode is only so quick.

Would a TVS be better?

Check that the 15V isn't doing something weird at switch-on.

This is where I expect the problem lies. It's a cheap 3-15V @30A supply and I'm switching it on and off via the mains switch. But I do need my circuit to be able to survive this sort of thing.

I've started some more testing and will report back later.

[Mike Warren]

Well, I didn't manage to blow up a MOSFET this morning, but I did see some nasty spikes coming from the power supply when I started. Unfortunately. I didn't think to save the waveform, and later, when I did have a USB drive connected, there were fewer spikes and they were much smaller.

The first attached waveform shows one of the turn-on spikes. The yellow trace is the 15V power supply and the blue trace is the source of the MOSFET. The switch is being turned on about 50mS after power-up by a uC.

I also tested the circuit driving a solenoid that will be one of the loads these need to run. At a 10hz cycle time the MOSFET did start to get quite warm. Not that they will ever be switched more than every few seconds at most in real life.

The third and fourth attached waveforms show the switch-on and switch-off times using the values shown in the schematic in my first post.

The reason I want to keep the gate resistance quite high is that there is the possibility that these could be wired incorrectly with the source connected to 24V and I'd like the device to have a good chance of surviving until the fuse blows.

Anyway, I'll order some better MOSFETs and some TVSs and then I'll build a new test circuit that will hopefully fix the problems I have seen this morning.