When I was testing the inverter circuit as shown below (it is only one phase of the inverter with three MOSFETs in parallel), there was a phenomenon that I couldn't understand. enter image description here

Initially, I didn't connect the motor terminals, the inverter outputs were floating and the motor was at standstill. The DC power supply that provides power to the whole system was configured to have current limit at three times of the actual motor operating current.

Then for test purpose, I applied six-step gate driver signals based on standstill hall sensor reading (HIGH side PWM and LOW side full-on). Since the motor was not connected, the drive signal could only be one out of the six combinations. However, the DC power supply generated over-current protection.

After I connected the motor terminals and drove the motor in six-step operation, everything was fine and no over-current protection was generated.

My question is, why was there over-current (short-circuit?) when the motor terminals were not connected?

[update] The PWM duty cycle used in both cases was the same.

[update 2]

The negative side of the inverter was connected to shunt resistors shown below and then connected to ground. Sorry for any confusion.

enter image description here

  • \$\begingroup\$ Presumably the missing ground connection on lower FET sources exists in reality. | Just musings along the lines Spehro suggested: C3 D4 provides high side gate drive via bootstrapping. It should not need load to achieve bootstrap action as low FETs on charges C3 and upper FETs on drive VB > VHV+. Seems unlikely, but IF the lower ground IS missing and the system works due to 3 phase couplings then bootstrap fails with no load. \$\endgroup\$
    – Russell McMahon
    Sep 1 '16 at 10:58
  • \$\begingroup\$ @RussellMcMahon, I am sorry for the confusion, but the lower FET sources are connected to shunt resistors and then connected to ground, as I updated in the original post. \$\endgroup\$ Sep 2 '16 at 2:07

Since there was no motor connected, you know it had to be from 'shoot through' (high and low MOSFETs on simultaneously).

A bit of a guess- the driver chip you are using is a bootstrap type that cannot drive the top MOSFET on at 100%- switching is required to charge the power supply. While there is UVLO protection on the high-side driver, it's possible there is some sequence of operations that can leave the gates powered long enough to trip the overcurrent protection.

You can calculate whether the 10K will discharge the gate charge of the three particular MOSFETs you are using fast enough for the overcurrent trip to not operate. This may or may not be a problem in the application depending on the tolerance of the MOSFETs to such abuse.

  • \$\begingroup\$ The duty cycle I set when the motor was not connected was the same as when the motor was connected. So if it was the shoot through, wouldn't it happen too when motor was connected? \$\endgroup\$ Sep 1 '16 at 4:34
  • \$\begingroup\$ Did it happen on power-up? If so, how many times was the start-up repeated do each configuration ? \$\endgroup\$ Sep 1 '16 at 8:40
  • \$\begingroup\$ No, it didn't happen on power-up. It only happened when the switching signals started. \$\endgroup\$ Sep 1 '16 at 9:16
  • \$\begingroup\$ What state are the MOSFET gates in just before the switching signals start? And how repeatable is the issue? \$\endgroup\$ Sep 1 '16 at 9:49
  • \$\begingroup\$ All MOSFET gate drive are LOW before the switching, and this issue happens every time and on all five pieces of inverter board I have. \$\endgroup\$ Sep 2 '16 at 2:33

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.