MOSFET body diode as 'zener'

[jan.didden]

Hi,

I noticed that in some MOSFETs, the body diode is shown as a zener diode. For instance, in the IRFR9310, a hv-p-channel device.
In my design it would be very good if it actually was a zener because I could use it as overvoltage protection.
But I'm not sure it really functions as a zener and what the max zener current is.

Anybody knows this and/or has experience with this?

jan didden

[SemperFi]

Yeah it acts like a low voltage zener, (or can we say high voltage diode?), if Drain is more positive than Source by a voltage that leads the diode into conduction.

Those parasitic diodes in a P-MOS usually have a voltage drop 2-4V. Is that enough? N.MOS' have lower voltage drops on the diodes, around 1V. (The P-channel devices have higher contact resistance between metal and P-Si and therefore higher voltage drop for the diode compared to N-devices. ref irf 'Power MOSFET Basics')

So, if 3-4V is what you need, go fer it, tho the diode's forward voltage values probably vary a bit from device to device.

[Elvee]

Originally Posted by janneman

But I'm not sure it really functions as a zener and what the max zener current is.

Most modern MOSFets are avalanche-rated, and avalanche is the nearest thing to zener for voltages >5.5V, so yes, that diode has a zener-like behavior.
But in general, it is not specified the way usual zener/avalanche are: they have an energy rating.
I guess it would be unwise to try to reach the max Pd of the transistor by pure static dissipation in that diode, but several watts is probably safe enough.

[Mooly]

Originally Posted by janneman

Hi,

I noticed that in some MOSFETs, the body diode is shown as a zener diode. For instance, in the IRFR9310, a hv-p-channel device.
In my design it would be very good if it actually was a zener because I could use it as overvoltage protection.
But I'm not sure it really functions as a zener and what the max zener current is.

Anybody knows this and/or has experience with this?

jan didden

The devices max Vds has to be observed at all times. By definition the diode will be non conducting up to that point. If it conducts at say 420 volts (20 volts over max rating) then you have exeeded the devices limits.

It would be interesting to try one though, on a limited HV PSU with the FET fully off and see at what point breakdown occurs. How repeatable is it device to device.

Whatever the outcome, ultimately I would say if the diode does enter the breakdown region then the device has already been overvolted. That would apply to transient as well as steady state conditions.

I would look elsewhere for a more well defined protection scheme.

[Elvee]

I just tested the concept on a IRF620:

At 100mA, the breakdown voltage is a stable 255V.

It does not seem to pose problems. I stopped the experiment after ~2 minutes, because the heatsink was too small and was becoming uncomfortably hot, but I think that with a suitable cooling, it could work reliably as a power zener.

[darkfenriz]

Avalanche is different physical phenomenon than zener breakdown, but it may work for some circuits.
The difference might be a small amount of hysteresis i.e. the threshold of starting to conduct is higher than ceasing to conduct threshold. This makes (modern) Mosfet's diode very efficient at absorbing overvoltage transients or spikes in an SMPS, but I wouldn't build a series regulator around it. Theoretically one could try and build a relaxation oscillator around this diode I guess
For the record, for modern 'trench' devices the real life avalanche voltage is 15-25% higher than voltage ratings (at 25 deg.C) and goes up with temperature by some 0.1% to 0.15% per deg.C

[Elvee]

Originally Posted by darkfenriz

Avalanche is different physical phenomenon than zener breakdown, but it may work for some circuits.
The difference might be a small amount of hysteresis i.e. the threshold of starting to conduct is higher than ceasing to conduct threshold.

In the case of the IRF620, there is no hint of negative resistance, unlike bipolars.

[darkfenriz]

OK.
Are you able to measure the noise and its spectrum?

[SemperFi]

hmmm...I've completely misunderstood the OP...or something else is wierd...The body diode starts to conduct when the Vds goes opposite polarity from normal, and for a N device it starts around 1V while for P-device it starts a little higher 3-4volts.
How come you guys are talking hundred of volts here? What diode are you guys talking about?

[fpitas]

Originally Posted by SemperFi

hmmm...I've completely misunderstood the OP...or something else is wierd...The body diode starts to conduct when the Vds goes opposite polarity from normal, and for a N device it starts around 1V while for P-device it starts a little higher 3-4volts.
How come you guys are talking hundred of volts here? What diode are you guys talking about?

Some MOSFETs have a body diode that's avalanche rated. Generally the datasheet has a spec for the maximum pulse power etc. The avalanche diode is good insurance when the transistor is used in high-power switching circuits, especially if there's an inductance involved.