• State
  • Locked Locked
  • Replies 5 replies
  • Subscribers 30 subscribers
  • Views 1281 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

DRV8840 Faults with LC output filters

Patrick Marcus
Prodigy 30 points
Other Parts Discussed in Thread: DRV8840, DRV8840EVM

  • We are using a DRV8840 in a somewhat non standard fashion.
  • We are driving 3 8.4V Peltier coolers in series (TECs) through an LC filter for smoothing.  I've attached the schematic for reference.
  • Near 90-95% duty cycle we consistently get faults on the DRV8840.  The output current should only be around 3 to 3.5A at full output.
  • We have tried a variety of PWM ENABLE frequencies between 10KHz and 100KHz (which is our desired frequency of operation).
  • We have also tried both braking modes and they don't seem to have an impact.  
  • We do not believe we are hitting the temp limit as the fault occurs at approximately the same duty cycle over a variety of warm up times.
  • We are also avoiding LC overshoot issues by ramping the PWM duty cycle slowly.
There is a possibility that our external PWM control is stepping on the internal PWM control comparator function set by the current sense resistors.  We are about to test that by shorting the sense resistors out.  It isn't clear though, why the fault pin would be set in any case
.
  • 0
    TI__Guru** 114905 points
    Hi Patrick,
    When the fault asserts, does the device stop driving until nRESET is applied or VM is toggled? If so, this is an overcurrent event.
    If the device begins driving again after some time, overtemperature is probably the cause. If overtemp, the thermal connection to the board may be suspect. Can you check your thermal connection?
  • 0 in reply to Rick Duncan
    Prodigy 30 points
    Rick,

    Thanks for your response. My answer is mixed.

    It appears that when it is on the edge of tripping the the fault line toggles, but after pushing the PWM a little higher the fault latches. I am fairly certain we have a proper thermal connection (thermal pad has 11 12x24mil thermal vias to a solid ground plane on an inner layer and 3sqcm on the bottom layer all 1oz copper. It does get warm but I don't think we've seen it above 80 or 85C on the package surface at 100KHz.
  • 0 in reply to Patrick Marcus
    TI__Guru** 114905 points
    Hi Patrick,
    So it sounds like you may have both issues. Let's first concentrate on the toggling nFAULTs. 3 to 3.5A current about where overtemp can occur.
    Are the thermal vias 24mils in diameter? This is twice the recommended diameter from the datasheet. The vias can allow the solder to wick out the bottom, leaving little thermal connection. This can then cause the device to cycle internally.
    There is a DRV8840EVM available for evaluation. Also all schematics and layouts for the board are available on line.
    Now to the possible overcurrent. Do you have a current probe? If you do, please look for current spikes of greater than 6A for more than 3us. These could cause the overcurrent event and cause the device to shutdown. I suggest triggering with the nFAULT and zooming in on the 10 to 20us prior to nFAULT falling.
  • 0 in reply to Rick Duncan
    Prodigy 30 points
    Thanks Rick,

    I'll keep the via size in mind. I know we are out of recommendation but we have those sizes due to rapid PCB turn around spec limits.

    As for the current we do have a probe and I don't think we ever saw any significant spikes.

    One of my engineers just came in with an update, he reduced the current sense resistor from 0.2 ohms to 0.1 ohms and that resolved the problem. I don't quite understand why unless the internal PWM pulse mode control was fighting with our external PWM when the current was near the comparator trip threshold.
  • 0 in reply to Patrick Marcus
    TI__Guru** 114905 points
    Hi Patrick,I am glad to hear the problem has been resolved.
    Assuming no changes were made to the VREF voltage, reducing the sense resistor should have doubled the current. I would think this would remove any current chopping and allow the input to control the output directly.
    It is hard to explain why this would remove the faults. There is the possibility that the absolute maximum was being violated on ISEN when using a .2 Ohm sense resistor. Lower the sense resistor to 0.1 would reduce the voltage on ISEN.