This question is related with the one I asked sometime ago about the heat generated when driving a BLDC motor BLDC drive inverter circuit problem. The problem is, the MOSFET chip got heated up obviously even the RMS current is about 200mA. The motor to be driven is 24V 90W BLDC motor with a phase resistance of 0.15 ohms and phase-to-phase inductance of 200uH. The MOSFET and its gate driver I chose was SI4564DY (P+N) and FAN3278. enter image description here

@Bruce Abbott gave some excellent explanation why the chip got heated up in my last question, but after some digging, I still have some more questions regarding how to select the proper MOSFET.

Correct me if I am wrong, but according to datasheet, SI4564DY is supposed to handle up to 9A at 24V VDS, which I though is more than enough for the bldc motor used, and the "Continuous Source-Drain Diode Current" is 2.3A, which I assumed to be RMS value rather than instant value. So I didn't expect it to become so hot when the RMS drive current is about 200mA, although as @Bruce Abbott pointed out and verified by experiment that current spikes at 15% PWM duty cycle can as high as 2.3A. Even TI has a design using similar MOSFET chip to handle 24V/5A bldc motor without using heat sink. If I drive the motor at full load, I guess the MOSFET chip will burn out. So my question is

  1. Did I select the suitable MOSFET to drive the 24V/90W motor (phase resistance 0.15 ohms and phase-to-phase inductance of 200uH)
  2. Normally, when is heat sink needed for the MOSFET chip? I though SI4564DY is more than enough and heat sink should not be needed, as TI's example shows.
  3. If SI4564DY is actually not enough, what kind of current capacity should I look at when selecting another chip?
  4. To drive a 24V bldc motor, normally at what wattage should I consider using heat sink?

I am still new at handling this kind of bldc motor control and hopefully these questions don't appear to be silly to you gurus.


Here are some related info. The size of the PCB is 15mm*5mm with mainly the MCU, gate driver, MOSFET and some sensing circuit on it. The PWM freq used is 20Khz, and I did try 10KHz, but no obvious temperature drop. Currently I am using PWM on lower arm N-channel MOSFET.

  • \$\begingroup\$ The MOSFETs have Rdsson in the 20 to 30 milliohm range spo at 200 mA mean current power dissipation when hard on is about W = I^2 x R = 0.2^2 x 0.03 = 0.0012W = 1.2 mW SO you are doing something "wrong".The FET should take 500 mW- ! W dissipation depending how it is cooled. At 30 milliohms 3A gives 0.27W so your heatsinking would need to be very poor for that to bother it. Note that a BLDCM driven with 24 V MAY produce ~=2 x Vin on another phase so yuou may be "overvolting" your MOSFETS. \$\endgroup\$
    – Russell McMahon
    Commented Aug 27, 2015 at 15:12
  • \$\begingroup\$ 90 Watts???? 24V x ~4A ~= 90W. Without going into current per phase average etc 4A >> 200 mA to the extent that summat aglae. Where does the 200 mA figure come from? \$\endgroup\$
    – Russell McMahon
    Commented Aug 27, 2015 at 15:14
  • \$\begingroup\$ @RussellMcMahon would you please explain more about bldc driven with 24V may overvolt the MOSFET? \$\endgroup\$ Commented Aug 27, 2015 at 22:02
  • \$\begingroup\$ @RussellMcMahon when I control the PWM duty cycle to be 15%, the motor would run at about 1/5 rated speed and the current draw from a DC power supply shows 230mA. Actually in my last question, Bruce Abbott explained and I verified in experiment that the current spikes are much higher than 230mA, which could cause loss in body diode. \$\endgroup\$ Commented Aug 27, 2015 at 22:02
  • \$\begingroup\$ I'd need to look at how the motor and drive waveforms are implemented, but a motor can be and often is a "transformer" with windings coupled together magnetically. If you apply a voltage across one or two windings you are liable to get a voltage relative to the driven windings on the undriven windings. If this or these are open ciruit that wilol adopt a voltage based on trandformer action. The phase angle of the windings may affect the overall voltage seen between the driven and open circuit This may well not be a problem. But may. .... \$\endgroup\$
    – Russell McMahon
    Commented Aug 28, 2015 at 12:10

2 Answers 2


Friend I have seen your schematic , I found that there is no separate switching paths for your MOSFET..since your gate resistor helps in regulating load current but it also comes in series to drain to gate capacitance which leads to slow turn off of MOSFET. Try to keep a diode across resistor of gate which provides quick turn off,leads to less heating of MOSFET. I hope it may work.

  • \$\begingroup\$ The diode is not a must-have in many designs, not even in TI design referred in my question. \$\endgroup\$ Commented Aug 27, 2015 at 22:05

Check you aren't getting shoot through, which occurs when high side and low side fets are on at same time.

Introduce a small dead time between turning one fet off and the other on. Dead time needs to be longer than turn off time of fet which will be gprimarily set by gate charge and gate resistor.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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