# Mosfet troubles Updated

I currently have the following circuit shown below on a prototype PCB. The circuits intended purpose is to do the following:

1. Receive two 5V PWM signals from a microcontroller.
2. Boost the current a little using the FDS6912A N-channel MOSFET
3. The signal then passes through a 4N25 optocoupler. Each optocoupler has its own independent battery, which isolated the microcontroller power from right hand side.
4. The output of the 4N25 optocoupler is passed through to an ESC (Electronic Speed controller). To me the speed controller is a "black box", but it is used to drive a motor.

The problem I'm facing is that the FDS6912A N-channel MOSFET fails (for a reason I can't understand). The PWM signal at the input of the PWM_CH1 and PWM_CH2 is fine when I scope it. The output of both optocouplers remain at 0 V.

What I've found is that after replacing the FDS6912A mosfet the PWM tends to work fine again. My observations lead me to believe that the mosfet is blowing when I initially connect the batteries but I can't work out why. This is my theory but I can't be sure.

Because the circuit is already on a PCB I am severely restricted in what I can do to try and fix it. If I can put in a better replacement MOSFET that suits an SOIC-8 package or a suggestion can be made as to the cause it will greatly assist.

I've attached datasheets of the mosfet and Optocoupler in the links below and any help would be appreciated.

Update:
I Re-Examined the Gate voltage after another mosfet Failed and yes.. The PWM duty dropped from 5V to approximately half a volt (0.5v) after it failed. If I disconnect the mosfet it returns to 5V. Can anyone suggest a more resilient mosfet... Or will i have to resort to somehow modding the PCB to try and include a series gate resistor?

FDS6912A datasheet

4n25 Optocoupler datsheet

• What voltage are the gate signals? – Olin Lathrop Jun 27 '13 at 12:31
• 5V Peaks, with a frequency of 100Hz. The output of the microcontroller appears to be fine. It's after that stage where the problem appears. – Peter H Jun 27 '13 at 12:46
• If the Mosfets are blowing it means that either a spike voltage is exceeding the rating (positive or negative) or they are exceeding their current limit. Clearly not the latter as R3 and R4 would probably have exploded. I would try adding a diode (say a 1N4007) between the drains and ground for each Mosfet. I know they have internal ones but there's no harm in beefing up the protection. – JIm Dearden Jun 27 '13 at 12:57
• It looks like gate damage to me - try putting 1k in series with the drives from the controller - this shouldn't affect the performance at 100Hz - do you mean "Hz" or "kHz"? The 5V supply to the opto diodes - is this returned (0V) at GND (symbol)? – Andy aka Jun 27 '13 at 13:13
• 100Hz... The right hand side of the circuit is driving an Electronic Speed Controller which is connected to brushless motors via the PWM. This is the reason for the isolation in the PWM signal. – Peter H Jun 27 '13 at 13:14

I understand that you have a PCB ready and don't want to modify it but why didn't you use this circuit: -

It doesn't need the FETs and it doesn't need an intermediary 5V supply. The LED input is very robust and all you have to ensure is that you don't reverse bias the LED in the opto. A reverse connected diode across the input terminals would achieve this.

The schematic looks sound to me (through, with all the components upside-down or oriented weirdly, it's hard to say) so I would suspect some non-ideal behavior as the cause.

Exceeding the maximum gate-source voltage even for an instant is a sure way to destroy a MOSFET. If you are driving a motor with PWM you will have high $\frac{di}{dt}$ somewhere, and this could possibly be coupling into the gate through unintended mutual inductance. Or, ringing caused by the inductance of the gate and source traces and the capacitance of the gate might result in transient voltages enough to cause damage.

Things you might try:

• keep the loop area of the gate trace, and its return, from the source back to the MCU, as small as possible to minimize its inductance
• keep the loop area of your high-current traces likewise small, and far away, and not parallel, to the extent possible
• lower the impedance of the gate (make R1 and R2 smaller)
• add a zener with a breakdown well below the maximum $V_{GS}$ to clamp any transients at a safe level
• modify your motor drive for a lower $\frac{di}{dt}$
• use BJTs instead of MOSFETs
• add resistance in series with the gate to dampen ringing
• Yes..sorry about the orientation... I know its not so pleasant to look at that Optocoupler upside down. Some great tips though Phil. What value would you suggest for R1 and R2? – Peter H Jun 28 '13 at 14:01
• @PeterH Well normally, 10k would be fine, but for the purposes of troubleshooting, I'd make them as small as you can without exceeding the current sourcing ability of your MCU. Of course if you make them too small, then you might as well be driving the optos directly, since you will be sourcing just as much current. – Phil Frost Jun 28 '13 at 16:39

Whilst I was soldering the SOIC-8 mosfet, I was applying flux to help the solder flow to the mosfet and avoid bridges between the pins. I then used both a heat gun and soldering iron to help the solder flow. As it turns out I was applying to much flux. After un-mounting the mosfet there was still an extensive amount of flux on the underside/beneath the chip and the PCB. Not all of the flux evaporated.
I discovered this completely by chance and was skeptical because the voltage at R3 would vary at different times, when I was testing. Another give-away which I didn't detect earlier enough was that when measuring the resistors R1 and R2 at times no resistance was present. I.E A short circuit. This was again due to the flux that was short circuiting the mosfet pins. Sometimes it would work fine and at others it wouldn't. I guess the flux fluid was moving causing all sort of weird stuff to happen depending on the pins it was short circuiting and the angle of the board.

After cleaning the PCB and applying a new MOSFET, this time with VERY minimal flux it turns out that things appear to be ok. I can only hope that it stays that way. So far all tests have been positive.

I appreciate all the feedback and tips. I will certainly incorporate them into a future design.