how to protect a MOSFET

[newxito]

Some professional clock kits use 3 x 2n7000 MOSFETs to drive 6 rgb leds. The gates are connected directly to the microcontroller pins. I tried that and it works fine. My problem is that the controller and the MOSFETs are not on the same PCB.

If I power up the two boards without the wires between microcontroller pins and MOSFET gates, there is a good chance to destroy the MOSFETs, I killed a lot of them :-)

Unfortunately, I’m not an electronic engineer, so my question is how can I protect the MOSFETs. I have tried several things on the breadboard. Connecting a 1M resistor between gate and ground and a diode between connector and gate seams to do the job, but I’m not sure if that is the correct solution.

[newxito]

I forgot to draw the led resistor...

[Jon]

Is there a particular reason for using a MOSFET in this application rather than a jellybean NPN transistor? Small signal MOSFETs are fragile things anyway, and exposing them to the set up you have with separate boards just seems like an inherently troublesome approach.

Jon.

[newxito]

I used these MOSFETs because I found them in several schematics dimming RGB LEDs using PWM.

I’m trying now with a BC548 with a 4.7K resistor between the base and the pwm controller pin. It works great. So, bye bye 2n7000, welcome BC548 :-)

Thanks a lot for your input!

[gregebert]

I've used many MOSFETS in my clocks, with zero problems. In my opinion, the main nixie-ish advantage of a MOS device versus bipolar ( NPN or PNP)  is that the MOS device requires essentially zero power to keep the device on, whereas bipolar devices require a small amount of base-current, typically 100uA or less to drive nixie tubes. Also, various driver chips like the HV5530 are NMOS devices.

There are 2 precautions you must observe

• Overall circuit-design: Make sure none of the datasheet parameters are exceeded. 
• Electrostatic discharge (ESD) protection.

 I dont know what the diode is for; circuit-wise and ESD-wise I dont see any functional purpose for it. Having it there does not cause any problems that I can see.

Is the 1meg resistor across the gate-source on the same board as the MOSFET ? If not, then that's most likely the problem because there isn't any path to prevent parasitic charge from accumulating on the MOSFET and destroying it.

If the resistor *is* on the same board as the MOSFET, then next thing to consider is how you connect the 2 boards together. It's best to have the grounds connect together first. If you are already doing that, is there anything to protect against excess voltage at the drain, such as a zener-diode or a bleeder-resistor ? I often use 10meg resistors at the drain terminal of high-voltage MOSFET circuits to bleed-off any charge that could try to accumulate during board assembly/handling.

What about the power supplies ? Is the LED's power supply the same supply that runs the PWM controller ? If not, that's another potential problem.

I doubt you are exceeding the maximum-power dissipation of the device, especially if the LED supply voltage is 12V or less AND the max current is 20mA or less. If either of these values are exceeded, you will need to calculate the power dissipation, and from there, the temperature-rise of the device.

------------

Now regarding ESD....

My workbench has an ESD-mat, I wear an ESD wrist-strap, use an ESD-safe soldering iron, store all of my ESD-sensitive components in conductive bags, store my PC boards in conductive bags once they have sensitive components mounted, and I also have an ESD jack on the PCB that I connect to ground before I remove the board from it's bag. I dont wear a conductive smock (I should), but at least I only wear cotton clothing when handling ESD-sensitive devices. It may sound paranoid, but I've never zapped a device.

I realize not everyone can work in these conditions, and if you cant, then at least work on a concrete floor and frequently touch the circuit-ground of your project with your fingers to equalize the potential between yourself and the board. I also suggest touching the circuit ground at the same time you pick-up a conductive ESD bag of parts. I've done this at a minimum for 40 years now and have never zapped a part, even when ESD was not well-understood and many devices had little-or-no internal protection.

[newxito]

Maybe I should explain my design. For now, I have the following boards:

Power supply board, delivers 5V and 170V

ATmega controller board with connections to drivers, RGB leds, PIR and GPS/NTP

Two types of driver boards, one using 74XXX and transistors and the other using HV chips

Four types of “display” boards, IN-8, IN-12, IN-18 and Z5660M, all of them with the MOSFETSs and RGB leds

I like the idea of a modular system. If I want to switch from an ATmega to an ESP32 or a Raspi I just have to change the controller board. Or if someday I want to make a 12-digit clock I just can cascade two driver boards. Even if the clock has a nonstandard design I can use the drivers with direct wiring to the nixies. Well, that’s why I have separate boards.

Back to the 2n7000 problem…

In my case, the MOSFETs only die if the gates are not connected to the controller. When I started to test the display boards, I killed all the MOSFETs on all the boards because I tested without the RGB connection. The grounds are connected and the power supply is the same for all boards. Adding the 1M resistor and the diode near the MOSFETs, stops the massacre :-) I don’t know why the diode helps but that works for me. But as Jon pointed out, with my design it’s probably better to switch to transistors.

[gregebert]

The modular approach has worked well for me, and I agree with all your reasons for using it.

I suspect an ESD event is causing the damage; we just need to find out where it's coming from and mitigate it.

The first place to look is power supplies. In order to minimize the impact of an ESD event, all of the connections on the power supply need to be at the same potential when the power-supply is connected to your board. Also, all of the power connections on the board need to be at the same potential before you connect the supply. 

This is easily done with large resistors between each power-supply voltage, and your circuit ground; typically I use 1-10Megs for higher voltages like the anode supply. For lower voltages, 10-100K is fine. These resistors eventually discharge any power-supply capacitors; you can calculate the time constant T=5*R*C, where T is seconds, R is ohms, C is farads. The smaller R is, the shorter the time is to discharge, but also there is more wasted power. If you have multiple boards, etc, each one needs it's own set of bleeder resistors.

I dont have the time to go into all of the nuances about ESD design practices here, but if you have all supplies tied together, most of the other components will protect eachother thru conductive paths; this is especially true for MOS IC's because they have built-in ESD protection circuitry that essentially ties all pins together. Even though a discrete MOS device is unlikely to have ESD protection built-in, if it's connected to another device, such as a microcontroller, that has ESD-protection, both will be protected in-circuit.

If you really want to be as careful as possible, touch the ground of the power supply with one finger, and the ground of the board with another finger before connecting the power cable. This will equalize the 2 components to the same voltage thru your body resistance within milliseconds. When you connect them, there is NO ESD event.

Finally, make sure you are connecting power-supplies first, then signal connectors. That way, there will not be an ESD event when the signal connector is inserted because the power cable has already equalized the 2 devices. Be careful, though, because signal connectors that tie grounds together can create ground loops in your system. If you have noise-sensitive circuits, like clock signals, you may need to use differential signalling, schmitt-triggers, or opto-isolators.

[newxito]

I agree 100%, it must be an ESD problem, these things are really fragile, even on a breadboard circuit just touching 2 or 3 times the gate kills the 2n7000. I think I will not be able to solve the problem in a reasonable amount of time, so for now, I will patch the boards with transistors and continue with the tests. Just finished to solder one of each type of driver boards…

[gregebert]

Agreed. Move-on with a more robust solution.  From your hand-drawn schematic, replace 'gate' with base, 'drain' with collector, and 'source' with emitter. Since you will be using NPN transistors, you will need to limit the base-current to about 1/50 of the maximum LED current with a series resistor (where you show a diode). If your LED is 50mA or less, then a 2.2K base resistor is about right. (Let me know if you need help calculating exact value if you are trying to minimize power consumption). The resistor from gate-to-source (base to emitter) is not needed. Just about any NPN transistor will work; 2n3904 is perfect and dirt-cheap. Be sure to check the pinout; I've seen variations that may not be pinout-compatible with the MOSFET.

[GastonP]

This circuit is the one I always use and I have yet to see a zapped MOSFET ;).
Diodes can be 1N4148 or almost any fast signal diode (never a rectifier such as 1N400x) and a resistor of 1K-10K is enough.
IIRC SMD BAV99 has this exact configuration for the purpose of being used as input protection devices.

[gregebert]

That's basically the same method used on ICs for input devices. I typically use a zener diode.

As long as there is another device driving the MOSFET, such as an IC, you dont need to add that form of protection because the driving IC will already have that onchip.

[Dekatron42]

What are those IC's on the board with the 2N7000?

/Martin

[newxito]

There are no IC’s on the board with the 2n7000, just some resistors, leds, nixies and the connectors to driver, controller (led pwm) and psu. I replaced the 2n7000 with BC548 and 4.7K resistor (patch picture). One of the driver boards on the picture has two HV5622 and the other one 8 x 75HC595 and 64 x MPSA42 transistors.

[newxito]

sorry, 74HC595

[Dekatron42]

Aha, I thought they were on the board with the 74HC595s in the photo.

/Martin

[newxito]

Just tested your circuit on the breadboard... the mosfet is still alive :-) 

This forum is great, thank you all for the support.

[GastonP]

Yes, that's right but as this circuit does not have a constant connection, there are cables and connectors which can be open, have false contacts, and even can have high impedance ground connections, it's better to integrate the protection into it. And I didn't tell I invented it :) ; this circuit is repeated ad nauseam in every CMOS semiconductor manual of the 70's-80's, that's why I love it.
Zeners are a good alternative when one does not have a VCC reference or when the input voltage can be higher than the highest available power source (which can happen too with an un-energized circuit whose input must not be shorted in that situation). I don't use them much because they are more expensive and are slower to start conducting, besides having a higher capacity. But for this kind of hobby applications most of the times is just a matter of taste :).