0
\$\begingroup\$

I am using 4 IRF840 N-channel mosfet,converting 325VDC to 230VAC with SPWM and H-bridge.I have no problem with resistive load,but when I connect inductive load or a device with capacitor power supply,mosfets of one side burn.here is m circuit: M question is:Would placing 350V TVS diode on source-drain of each 4 mosfet,solve this problem? enter image description here

I mean placing diodes like this:

enter image description here

\$\endgroup\$
  • 1
    \$\begingroup\$ Before you just throw TVS diodes at the problem you should really root-cause the problem. If it's overvoltage on the FETs choosing different FETs or maybe IGBTs would be the right solution. If you want to use TVS diodes to protect the FETs you need to know how much energy you have to dissipate in the diodes otherwise you may end up just popping the diodes AND the FETs. \$\endgroup\$ – John D Aug 8 '17 at 16:48
  • \$\begingroup\$ What is (cap) ESR/RdsOn and (inductive) DCR/RdsOn ratio and expected current? What is deadtime? and then what is L/R ratio? Also what is SRF of L? Where are your current sensors and waveform? choose 75mV drop Rs. If FET RdsOn/DCR ratio is too high, it will burn out. \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Aug 8 '17 at 16:51
  • \$\begingroup\$ Thank you for quick reply,I don't think the problem is result of SPWM factors or driving components because as said above, I don't have any problem with resistive load.I used the different low pass filter shown in second image.Does this cause this problem? \$\endgroup\$ – Tom Hanks Aug 8 '17 at 17:11
1
\$\begingroup\$

Check your voltage with a scope on the FET side of your filter. (use a x100 probe and be careful). You will likely be seeing excursions above the FET breakdown voltage of 500 volts, and your FET is doing what you want your TVS diodes to do.

Your real issue is that with reactive loads (capacitors or inductors), the current and voltage are 90 degrees out of phase. This means that during part of the sinusoidal waveform, your current will be flowing in the opposite direction to that experienced with a resistive load.

enter image description here

During this period, the current is flowing "backwards" into your power supply. This can cause your problem. For example, with an inductive load, if your DC supply cannot absorb and dissipate this energy, your input voltage will rise until the current conducts somewhere (in your case the FET breakdown voltage or reverse VGS). These are not small currents. Although the TVS diode will work in theory, it is not your best solution because it is inefficient; you are storing energy in your inductive load, the pulling it back to heat up your TVS diodes.

If you want to run reactive loads, you will have to do some power factor correction.

| improve this answer | |
\$\endgroup\$
  • 1
    \$\begingroup\$ I placed P6KE350A(600W 350V TVS) for D1 to D4.It solved my problem.Works fine with inductive or capacitive load. You are right,maybe it's not best solution,but probably easiest one.Thanks. \$\endgroup\$ – Tom Hanks Aug 13 '17 at 17:55
0
\$\begingroup\$

These FETS already have TVS like Zeners included. That's not your problem or a solution.

Body Diode Voltage TJ = 25 °C, IS = 8 A, VGS = 0  Vsd = 2.0 V 2% pulse
Body Diode Reverse Recovery Time  TJ = 25 °C, If=8 A, dI/dt = 100 A/μs trr=460typ 970max ns
Body Diode Reverse Recovery Charge Qrr - 4.2 typ 8.9 max μC

as I said in comments....

  • What is (cap) ESR/RdsOn and (inductive) DCR/RdsOn ratio and expected current?

  • What is deadtime? 1us?

  • what is L/R ratio? Also what is SRF of L?
  • Where are your current sensors and waveform? choose 75mV drop Rs.

  • If FET RdsOn/DCR ratio is too high, it will burn out. If SRF of L is too low then load impedance drops FETS may burn out during hard chargeup phase without soft current limit.

Solution?

Add soft start and current sense for protection and surge regulation.

| improve this answer | |
\$\endgroup\$
  • 1
    \$\begingroup\$ I choose a 600W TVS with lower breakdown voltage,So it can clamp overvoltage before it can damage mosfet.I don't think that mosfet body diode has clamping action or maybe it clamps voltage upper than 1000V as many other diodes .Even,It is not fast diode.Deadtime is 400us.Thank you so much. \$\endgroup\$ – Tom Hanks Aug 13 '17 at 18:10

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.