This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

UCC24612: Heating problem

Part Number: UCC24612
Other Parts Discussed in Thread: TIDA-010015,

Hello,

I am developing 220-48V PSU, using TIDA-010015 as an example. Essentially schematic is the same, with some small changes in feedback circuit and with bigger Q4 (that is temporary solution). In place of Q5, Q7 are IPP110N20NA. I am having some heating issues with MOSFET drivers U7-U8. When PSU is loaded for 300 W (maximum power available should be 500 W), PCB in places, where drivers are placed, is heating up to 105-110 degrees Celsius. The more the load - the more of the heating. This is without any cooling, and this is by far the hottest place on board. I can confirm that it is, indeed, drivers that are heating up, because I see it quite clearly with thermal imaging camera.

I have tried increasing VDD and REG capacitances, lowering input voltage at VDD pin with no luck. Waveforms at VG also seem normal, without any overshoot.

What could be the reason for heating of those drivers?

Regards,
Aleksey

  • Hello,

    Could you tell me what the VDD voltage is set at?  You have a series pass regulator with Q4 should be regulating the input to 15V less a base to emitter voltage. If that is running higher than expected that could be causing the UCC24612 gate drivers to run hot.

    Based on the thermal information that you gave me I estimate the UCC24612 is dissipating 0.6W if the PCB teampature was 25 C (T1). 

    That seems like a lot of power maybe your FET's gate charge is too large.  You could select a different FET with a lower Qg.

    The thermal impedance  for the device is section 6.4. 

    Rjc = 97.6 W/Deg C., thermal impendence junction to case

    Rjb 44.2 w/Deg C., thermal impendence junction to board.

    so if the temperature of the case of the UCC24612 is 110 C (T1) and the teampature of the  PCB is 25 C (T2).

    T1 = Pd*(Rjc+Rj1) + T2

    (T1-T2)/(Rjc+Rjb) = Pd = ((110-25)C)*W/(141.8C) = 0.6W,

    You can calculate the power dissipation (PVDD) based on gate charge (Qg) and converter switching frequency (fsw).

    The VDD current to drive the FET Ids1 = Qg*fsw

    The current to discharge the FET is Ids2=Qg*fsw 

    PVDD = VDD*(IVDDrun + Ids1) + Vreg*Ids2. 

    >IVDDrun = 0.92mA, Vreg = 9V which should be the peak gate driver voltage.

    To recap VDD maybe running to high causing the UCC24612 to dissipate too much heat or the gate charge of the FET is so large it is causing excessive heating in the UCC24612.

    1. Reducing VDD will help reduce heating.
    2. Selecting FETs with lower Qg will help reduce heating.

    Regards,

     

  • Hello,

    1. I measured VDD voltage on one of the drivers and got 14.7 volts, which seem to be right. Voltage is stable, without overshoots or anything. I also tried before changing VDD power circuit so that Zener at the base of NPN transistor would be 11 V, not 15 V. Results were as if nothing changed - at the same load drivers would heat up to somewhat the same temperature.

    2. For the sake of experiment I placed IPP60R170CFD7 instead of original MOSFETs. They are not really good for this purpose, but have way lower Qg, though their Rds is way bigger. With those, it took drivers 1 minute to heat up to 70 degrees Celsius at 300 W load, after that I had to turn off PSU, because MOSFETS were heating up too. It is the usual heating rate of the drivers with initial MOSFETs.

    So none of this methods helped me. Is there any unorthodox things, which I might try to change? I have a feeling that I am missing something very simple, like PCB tracing error, though I checked that one a few times already. Drivers pinouts are not shorted together and everything seems fine from the outside, though problem persists.

    Thank you for advice.

    Regards, Aleksey.

  • Hello Aleksey,

    Did you go through calculations for calculating the power dissipation of the UCC24612 based on the Qg and switching frequency?

    Based on your ambient teampature and power dissipation you should be able to estimate how hot the surface of the UCC2412 should get.

    Originally when I read this post I thought the UCC24612 was reaching 105 to 110 degree C.  However, are the FETs what are heating up the board?

    The following application note discusses how to calculate FET losses in your design.

    If it is the FET causing the heating you could select a FET with a lower Rdson and reduce switching frequency to reduce switching losses.

    In some designs you may require heat sinking.  AVVID has a nice software tool for sizing heat sinks that you can find at the following link.

    Regards,

  • Hello,

    Yes, the problem was with the FET's. Apparently, thermal imaging camera doesn't show right temperature of TO-220 metallic case not under the right angle. So on camera I saw cold FET's, while area at their feet (where drivers are) was heating up fast. So it were the FET's heating up the board while I blamed drivers for that.

    Thank you very much for help.

    Regards,
    Aleksey.