0
\$\begingroup\$

The following is a schematic for a solenoid that failed after ~200k cycles. I found that the MOSFET drain to source failed short. I am getting~ 26 ohms resistance.

The way this circuit works is as follow: SW1 is a microswitch that gets cycled mechanically (one cycle is when the switch moves from NC to NO then to NC) when that happens, the solenoid turns off after ~45 ms delay it then turns on again.

I have removed the dead MOSFET and simulated a ground removal (MOSFET turning off). The scope screen shot is captured below. You could see a very fast pulse of almost 84 V with a 54 ns rise time.

The MOSFET drain-source max rating is 100 V.

What could have caused the MOSFET failure specially after so many cycles and how I could make this circuit more robust?

Mosfet datasheet link: https://www.diodes.com/assets/Datasheets/DMT10H009LK3.pdf Looking forward to your feedback. Thank you in advance.
TVS1 Datasheet

TVS2 Datasheet link:

enter image description here

Updating the MOSFET Turn off time scope shot

It takes about 7ms to turn off the mosfet. enter image description here

SOA The Solenoid initial draw is ~ 1.5A enter image description here

\$\endgroup\$
1
  • \$\begingroup\$ Comments are not for extended discussion; this conversation has been moved to chat. \$\endgroup\$
    – Voltage Spike
    Mar 5, 2021 at 23:03

2 Answers 2

2
\$\begingroup\$

What I think killed the MOSFET:

  1. slow on/off switching resulting in too much power dissipation in the MOSFET, you're using it outside the MOSFET's SOA region. Learn about SOA here.

  2. voltage spikes because you used a TVS (in parallel with the solenoid), I would recommend using a Schottky diode instead. I agree that maybe a TVS is good enough, however, it is uncommon to use a TVS in this situation. I would go for the tried-and-tested fast Schottky diode. I'd select one that can handle at least 2 A of forward current and at least 50 V reverse voltage.

To solve 1) I would try this:

schematic

simulate this circuit – Schematic created using CircuitLab

I added another N-channel MOSFET (I suggest a 2N7000 or similar) to act as a "poor man's comparator" to switch the large MOSFET on/off quickly.

R1, R2 and C1 with the switch form a delay circuit. It's "upside down" from what you had since I added M2, M1 is on when M2 is off so I flipped delay part. Tweak the values of R1, R2 and C1 to suit your needs. If you're familiar with a circuit simulator, simulate this thing!

Oops, I forgot that the 2N7000 has a Vgs,max of 20 V so I added R5 to limit the M2's Vgs to 14 V.

To fix 2) speaks for itself :-)

\$\endgroup\$
10
  • 1
    \$\begingroup\$ Personally I believe the TVS will protect the MOSFET drain. Also, I am sure using a TVS causes the MOSFET to turn off faster. But if you used a flyback diode, then you could use a lower voltage MOSFET instead of 100V. I think switching the MOSFET faster will lead to good results. \$\endgroup\$
    – mkeith
    Mar 5, 2021 at 21:51
  • 1
    \$\begingroup\$ if M2 turns off slowly would that make M1 turn on slowly? Yes/no, ;-) The idea is that even when M2 switches on/off slowly, that still makes the voltage at the gate of M1 go up/down much more quickly (compared to the change at the gate of M1). If the switching of M1 still isn't abrupt enough, you can increase the gain (amplification) of M2 by increasing the values of R3 and R4. There is a limit though as the gate capacitance of M1 will then get more influence. \$\endgroup\$ Mar 6, 2021 at 12:35
  • 1
    \$\begingroup\$ Then it is still possible that the SOA isn't the issue and that it is the TVS instead of a diode. Fact is, you're hitting the limits somewhere and you want to stay away from those limits. Yes, you can dig deeper and if you spend enough time you will find the root cause. However, it is quicker to just address both possible issues and make your design more robust against both issues, at least, that is what I would do. \$\endgroup\$ Mar 6, 2021 at 16:51
  • 1
    \$\begingroup\$ ESD is usually only relevant when a person or object can touch the actual conductor. So a long (but isolated) wire isn't going to cause an ESD event. I don't think a TVS across the MOSFET is needed but if you still think ESD can be an issue, I so no harm. What you can do is place the TVS, you can always remove it if it causes an issue. \$\endgroup\$ Mar 6, 2021 at 21:21
  • 1
    \$\begingroup\$ To the OP: you can add a comparator to the circuit without using a regulated voltage. There are plenty of comparators that can operate from a single 28V supply. This would allow you to completely separate the time delay from the driving element. \$\endgroup\$
    – mkeith
    Mar 6, 2021 at 22:35
3
\$\begingroup\$

Is there any reason you can't replace the bidirectional TVS diode with a simple silicon rectifier diode with the anode at the Q1 drain and the cathode connected to the 28V rail? Like so:

schematic

simulate this circuit – Schematic created using CircuitLab

Since the current only ever flows in one direction through the solenoid coil (from LC+ to LC-), the inductive kickback will always be LC- spiking high relative to LC+. With a regular silicon diode, this will get clamped to the forward voltage of the diode (0.7 volts or thereabouts) rather than the 58 V required to activate your TVS. It should also be a little bit cheaper, too.

I should also note that if you're worried about the silicon diode being too slow, you can use a Schottky, which has a faster switching time.

\$\endgroup\$
3
  • \$\begingroup\$ Thank you for your feedback. The reason for TVS2 instead of a shotcky diode is to allow the solenoid to release quicker by not restricting the voltage to a very low voltage. \$\endgroup\$
    – Rocky79
    Mar 5, 2021 at 21:10
  • \$\begingroup\$ @peter what rocky79 is saying is that V=Ldi/dt. So after shutoff if V is 0.6V, then di/dt in the solenoid will be small. It will freewheel for a long time. But if V is 50V, then di/dt will be large, meaning the current through the solenoid will drop off very rapidly. I hope you see what rocky is saying. It is perfectly legitimate, assuming the coil current must be reduced rapidly. \$\endgroup\$
    – mkeith
    Mar 6, 2021 at 22:39
  • \$\begingroup\$ @mkeith Exactly. Thank you for clarifying it. \$\endgroup\$
    – Rocky79
    Mar 7, 2021 at 2:58

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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