Experiences, over voltage on FET drain source

It probably varies somewhat from manufacturer to manufacturer, but in my experience a 40V FET has about a 45V breakdown voltage. (If you build a motor drive, pay attention to the voltage the motor can _generate_ when being spun by external forces at faster than the design speed).

Depends heavily on the circuit configuration.

Maybe turn the FET _on_ to snuff the spike? ...Jim Thompson

Is there an inductive load? Clamped? Current? Fet part number?

Post the circuit!

Hi Klaus. You may be fine, with no problem at all.

Virtually all power MOSFETs are completely happy doing the avalanche thing. At some voltage above the rated value, say 35 to 40V for a 30V parts, the MOSFET will avalanche. That is, it will act like a zener diode, with the usual zener sharp breakdown edge. [Note, what makes a transorb, or similar TVS device, unique over a zener, is added thermal mass to absorb energy. MOSFETs also have added mass; the die is mounted to a tab, which mounts to your heat removal system.]

The only question will be, how much energy will be dissipated during breakdown.

Where does your breakdown energy come from, flyback-transformer leakage inductance? The datasheet has Transient Thermal Response plot, showing how much energy can be absorbed for how long. If you run the numbers, you'll see this is simply derived from the thermal mass and the energy required to raise the junction to 150 degrees C. Prolonged operation above 150 to 175 C can damage the MOSFET.

One last note, when the junction temperature goes up dramatically like that, the breakdown voltage also rises (e.g., 25mV/deg) about 10%,

On a sunny day (Sat, 6 Feb 2016 09:09:06 -0800 (PST)) it happened snipped-for-privacy@gmail.com wrote in :

My only experience with MOSFETS breakdown was C-drain_gate causing the gate isolation to be pierced. So first make sure the gate is clamped or driven from a sufficiently low impedance. The power MOSFETs I use have a reverse substrate diode that has zener capabilities, for example IRLZ34, and that diode is fast.

If you re not talking MOSFETs but FETs use a clamp diode to supply??

Or use a MOSFET with a higher Vds AND a clamp diode?

I am monitoring the 24V line, and using a brake FET to clamp the energy if it runs into generator mode

Cheers

Klaus

I could do that, but then it will fight against the capacitance on the 24V line also, and then I would need a monitoring circuit to let the FET go into linear mode, and right now the FET can only be controlled either fully on or fully off

Cheers

Klaus

FET number:

Really nice device, very low price

Cannot post the circuit, but it's a FET in a stepper motor drive.

Cheers

Klaus

The energy comes from a lightning surge, so the standard 8/20us surge tests apply

I am thinking about adding a separate capacitor to clamp the pulse. It woul d no be connected to the 24V line, but connected to the line through a MOSF ET. So when I detect a surge pulse, the FET bridges the connection to the c apacitor allowing the capacitor to clamp the energy (must have parallel res istance to discharge between pulses)

The capacitor is sitting at 0V normally, so it has a lot more clamping ener gy capacity, instead of a capacitor riding on the energized 24V line

Good point, I may just go ahead and make a test. It's on the drawing board now, so trying to do it all without needing to go to the lab

Uh, didn't know that. So the avalance voltage of a FET has a general temper ature dependency of 25mV/K?

Cheers

Klaus

I would really not like to go to 40V. Something strange about that. Going from 40V to 30V rating suddenly reveals a swarm of devices with low RDSon and low prices. Guess other people than me are using 30V devices at 24V!

Cheers

Klaus

I read in the Art of Electronics that if you run an FET at very low drain s ource voltages, say less than 0.5 V it acts as more or less a pure variable resistor. I tried that for a regenerative receiver where you need gain to drop of with increasing signal amplitude (or at least the same). I couldn 't get enough gain out of it at such low voltages in the quick tests I did. Maybe there was a wire up error because you don't need much gain for a re gen. Could you expect the same behavior from a MOSFET at low drain source voltages?

It's because they're using them at 12V: computer CPU and GPU converters.

Even 40V is kind of marginal here. 60V might get you some economy from automotive applications, but that's far enough that Rds(on) will simply be higher for the same die size, so probably not as good capacity/$ as 40V.

Tim

Oh for heaven's sake. Would you kindly be more specific in future?

This energy has to be clamped and redirected before it ever hits the electronics. Residual energy should be low enough for the semiconductors affected to absorb the single pulse.

Fets are fairly immune to single pulse avalanche energy.

RL

On a sunny day (Sat, 6 Feb 2016 14:34:21 -0800 (PST)) it happened snipped-for-privacy@gmail.com wrote in :

I ment FET of course, see context.

Seems u use a MOSFET. Like I said, if it is a MOSFET it should be OK, it will zener.

Well, in my first post I described that I use a 26V tranzorb. The tranzorb is used to clamp the energy from the lightning surge. But, the problem is that clamp voltage is way over 30V (40V at 19A AFAIR)

Cheers

Klaus

On a sunny day (Sun, 7 Feb 2016 06:49:34 -0800 (PST)) it happened snipped-for-privacy@gmail.com wrote in :

I think at this point with lighning etc a circuit diagram would be in order.

Something like this:

M1 and M2 are 30V MOSFETs V1 is a 24V source D1 is 26V transzorb

Cable is beteen V1 and D1, so this is the potential point where the surge can couple into

"Monitor block" checks the 24V rail. If a surge is detected M3 is turned on to dump energy into Rclamp

Cheers

Klaus

On a sunny day (Sun, 7 Feb 2016 09:03:28 -0800 (PST)) it happened snipped-for-privacy@gmail.com wrote in :

Ah, you are trying to protect the supply line! How about some series transistor after the transzorb, one with a high Vce, either as an NPN emittor follower (with a zener to ground on the base), It would drop about .7 V, but OK. or else a PNP forward conducting by default with some driver ciruit as voltage stablizer (less forward drop)? That would allow the transzorb to go way higher, The other thing I personanly would design in is a small inductor in front of the transzorb, to absorb any fast spikes?

Yes, that may be a good idea. I just need some mass to absorb the energy

I will have that in a later version, since I do not think the design will pass conducted emission tests without a CM inductor. I just want that inductor to be small so if I have a good clamp I do not need that function elsewhere

Another consideration is to make a special driver for the halfbridge transistors. Say we turn them on for 1us (or lower), so they clamp some energy without damaging them. The problem with that route is that it will only work if the unit is up and running

Cheers

Klaus