5
\$\begingroup\$

Some MOSFETs have a built-in ESD protection in the form of back-to-back Zener diodes between source and gate.

An example can be seen here.

DMG2302UK internal schematic

This particular MOSFET has a Vgs(max) spec of +/-12V. Unfortunately, no spec is given for the Zener diodes so I don't know how they're rated.

I am wondering if it's possible to apply a higher Vgs than the maximum amount specified, given that a series resistor is used between voltage source and gate.

I have a sample of this MOSFET, so I wired it with a 10 kΩ series resistor on the gate, applied a bench PSU between resistor and source, and connected a multimeter between gate and source.

Plotting Vin vs Vgs:

Vin Vgs
5.00V 5.00V
10.00V 10.00V
12.00 11.97V
14.00V 13.93V
16.00V 14.97V
18.00V 15.36V
20.00V 15.59V
30.00V 16.21V
40.00V 16.57V
50.00V 16.82V
60.00V 17.01V

I did the same with a 910 kΩ resistor in series instead of 10 kΩ (which is closer to the value the real application would use):

Vin Vgs
5.00V 4.55V
10.00V 8.88V
12.00 10.43V
14.00V 11.77V
16.00V 12.84V
18.00V 13.54V
20.00V 13.87V
30.00V 14.30V
40.00V 14.46V
50.00V 14.59V
60.00V 14.67V

I've run the setup with the 10 kΩ series resistor and 60 V input overnight, and the MOSFET seems to still work (Rds still changes accordingly to Vgs), though I am clearly going above Vgs(max).

Does anybody have any insight on this being a good practice?

EDIT: I am aware that external zener is the proper way to go, but it has a cost and requires a space. I am asking for previous experience from someone that possibly have been in the same situation before.

\$\endgroup\$
6
  • \$\begingroup\$ I would guess it's okay (particularly if the overvoltage is occasional), but that and $3 will get you a coffee. There is really no way to tell for sure unless the manufacturer indicates it's acceptable practice. \$\endgroup\$ Sep 14, 2022 at 9:14
  • \$\begingroup\$ Certainly not. You will wear out the GS insulation and have a breakdown. \$\endgroup\$
    – winny
    Sep 14, 2022 at 9:19
  • 1
    \$\begingroup\$ Anecdotally, I just repaired a cheap LiFePO4 battery management PCB. It had 40V going through 10 kohm resistor to three parallel N-MOSFETs (NCE7580), rated 25V absmax Vgs. The MOSFETs clamped Vgs to 28 V (I assume internal zeners but datasheet does not specify). It lasted 3 years, then one of the FETs shorted out. \$\endgroup\$
    – jpa
    Sep 16, 2022 at 18:56
  • \$\begingroup\$ @jpa thank you! that's the type of knowledge/information I was looking for. \$\endgroup\$
    – Qwerty8787
    Sep 22, 2022 at 13:59
  • \$\begingroup\$ @jpa NCE7580 doesn't seem to have built-in gate-source zeners though, that's odd. Any other external components? \$\endgroup\$
    – Qwerty8787
    Sep 22, 2022 at 15:07

3 Answers 3

3
\$\begingroup\$

I am wondering if it's possible to apply a higher Vgs than the maximum amount specified, given that a series resistor is used between voltage source and gate.

Make sure you don't exceed the maximum: -

enter image description here

It's non-debatable if you want peace of mind. So, if you want to use a higher drive voltage in series with a resistor then add an external Zener diode of (say) 10 volts or so.

Why "10 volts" you might ask: -

enter image description here

Note that the above characteristic graph works with gate-source voltages up to 4.5 volts and, there's no great reason to exceed that hence, limiting with an external 10 volt Zener diode is fine if you are worrying that the drive level could exceed 12 volts.

\$\endgroup\$
7
  • \$\begingroup\$ As mentioned in response to another comment, external zener is not a trivial solution because additional cost and space. The way I'm using these mosfets means source is "floating" (or better, not tied to a low impedance net) so it could see almost any voltage between +/-15V. Previously I've been using MOSFETs with Vgs max +/-30V but in this particular iteration MOSFET output capacitance (Coss) is an issue and needs minimising. All MOSFETs I found so far with small capacitance have +/-20V Vgs max at best. \$\endgroup\$
    – Qwerty8787
    Sep 14, 2022 at 9:51
  • \$\begingroup\$ @Qwerty8787 nobody with any credibility is going to say you can exceed the maximum. You have to face this fact if you want peace of mind. That is why the limit is specified in the data sheet. That is why it's called a maximum rating. If that doesn't work with your circuit then you have to find a different MOSFET. \$\endgroup\$
    – Andy aka
    Sep 14, 2022 at 10:05
  • \$\begingroup\$ It's not about credibility. Possibly someone have been in the same situation as me before and can share experience on a particular mosfet where things work out just fine (i.e. zener voltage safely below Vgs max). It still puzzles me why they'd embed a zener with voltage above Vgs max. \$\endgroup\$
    – Qwerty8787
    Sep 14, 2022 at 10:22
  • \$\begingroup\$ @Qwerty8787 marketing people dream up schemes to try and get business. If adding some form of over-voltage protection into a MOSFET gets 0.1% more business then it might be deemed a successful idea. I also have no idea where it might be useful to use it especially since they give no clue about the continuous current into the device. But, then again, they only claim it can handle ESD and not continual cyclical over-voltages. The phrase "ESD" appears twice on pg1 with no indication what type of ESD or model. \$\endgroup\$
    – Andy aka
    Sep 14, 2022 at 10:48
  • 1
    \$\begingroup\$ @Qwerty8787 Correct operation is not guaranteed outside of the absolute maximum ratings - but you could easily see why they might want a better non-guarantee than the competition \$\endgroup\$
    – user253751
    Sep 14, 2022 at 19:44
3
\$\begingroup\$

Not answering directly, but some comments on ratings generally.

  1. You'll generally choose a zener/TVS so that its worst-case clamping voltage is still within ratings of the thing it's protecting. We might assume the part in question has a high enough Vgs(max) that biasing the zener is okay.

  2. Pin ratings are always given with a bit of margin, even when injected current is allowable. Example: logic inputs are often rated VSS - 0.3V to VDD + 0.3V. Depending, they may be rated for some current injection (functional operation at say 0 to 10mA, latchup-free up to 100mA, etc.), which will surely drop more than 0.3V beyond the respective supply -- but what's really being said is as long as the current is within bounds, you don't care about voltage. Whereas if you're applying a fixed, zero-ohm ideal source voltage, you sure as hell want to avoid injecting current into those ESD clamp diodes because current is exponential with voltage, and hence pushing 0.5, 0.6V beyond supply is likely to draw indeterminate and excessive currents; but at 0.3V, it's safely low, even at max. temp.

  3. Gate breakdown occurs slowly at voltages near Vgs(max), corresponding to a small leakage current (some nA to µA). Some of which (the electrons in the leakage current) remain lodged in the gate oxide, so this can cause a shift in Vgs(th), and if enough accumulate, eventually (catastrophic) breakdown occurs and a pinhole is blown in the gate oxide -- complete device failure. (Well, complete in the sense that it's probably no good anymore. It may very well still transist, just that there's this minimum resistance between gate and/or drain and source, due to current being shunted through that pinhole blown through the poor thing.)

    So, the datasheet rating is chosen so that wear occurs slowly enough to be practical, at process corners (e.g. minimum oxide thickness) and maximum temperature. (Or maybe it's at room temp. Ask the manufacturer to be sure!), and so that catastrophic breakdown does not occur.

  4. They put the zener in there, really just for ESD protection, and really at that, more just to bleed off a little excess charge, since these low-voltage and low-current devices really are quite sensitive even to ambient induced fields, let alone direct contact. A very small zener is used, to avoid increasing Ciss and Igss, and as your measurements show, this leaves quite a lot of internal resistance, which will drop a lot of voltage under transient (direct ESD strike) conditions. It may be that it's dimensioned to still withstand a few hundred volts (unfortunately, they do not specify an ESD level and method -- again something you'll need to ask them about to be sure).

    Which, because it's only for transient purposes, we might rightfully wonder if #1 has indeed been followed, or if the ordinary (gate, by itself) rating really is the 12V shown, and there's actually no safety margin at all (indeed negative margin).

Personally, I would not count on it. One more piece of supporting information: low voltage MOSFETs always have lower Vgs ratings. A comparable part without clamping, would still have a 12V or so rating, I'm pretty sure.

If you don't need the full current rating of this part, consider a "prebiased" or "digital" transistor instead. These are BJTs with bias resistors internally, so they can be directly wired to logic levels. A wide range of resistances are available, making them suitable even at quite high logic levels (like 24V).

If you do need it, consider a SOT-323 or even smaller part, and a chip or DFN style zener. You're only spending one more BOM item, and the footprint can indeed be smaller than of the full size SOT-23.

And if you need multiple, consider SOT-363 or etc. Pairs can save quite a bit of space, or at least parts counts. Ditto for zener/TVS arrays, resistor arrays, etc. (The main downside with arrays is they're somewhat more awkward to route, and can't be changed individually, if during prototyping, you find you need different resistor values or whatever.) (Quad or more transistor arrays are surprisingly uneconomical; stick to pairs there. Quads are quite common and attractive for R, C and D though.)

\$\endgroup\$
1
\$\begingroup\$

What a spongey Zener. I never run Vgs at the zener volts. These internal back to back zeners are good at dealing to occasional short term spikes. If you apply a DC current then the zeners will limit VGS and heat up in a very small part of the semiconductor die. Zener short circuit fail would make the device useless. If you want to feed the gate from say 60 VDC and say 10k then you can play safe with a lower voltage zener across GS. For your application a 10 volt zener would be good.

\$\endgroup\$
1
  • 1
    \$\begingroup\$ I know external zener would be the safest option, but it's not free and takes up space. I was hoping to get away with interal zeners but I understand it's very much uncharted (or undocumented) territory. My use case is in a +/-15V rail circuit, so absolute worst case is +/-30 Vgs max. \$\endgroup\$
    – Qwerty8787
    Sep 14, 2022 at 9:36

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.