• State Resolved
  • Locked Locked
  • Replies 6 replies
  • Subscribers 29 subscribers
  • Views 354 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.

DRV8906-Q1: Short Protection

Martin Burgmair
Prodigy 20 points
Part Number: DRV8906-Q1

Dear Team,

I have a prototype board using the TI DRV8906-Q1 grouping 2 x 3 channels in parallel. While the datasheet says "The over-current protection (OCP) ensures the device protection in any short scenarios like the phase short, phase to ground short and phase to supply short conditions", I am facing blown-up ICs (see image below) when accidently shorting PWM groups that drive 0% and 100% duty-cycle respectively. 

I am using the following overcurrent protection settings, since I have a capacitive load to drive:

  • PL_MODE_EN = 01b = Parallel mode OCP slow turn-off slew is enabled
  • OCP_DEG = 100b = OCP deglitch time is 60 µs

Since I obviously don't want to destroy more prototypes, I would like to know if there could be any other register configuration in which the built-in OCP is not performing as expected?

Thanks in advance, Martin

Damaged DRV8906-Q1:

Damaged DRV8906-Q1

  • 0
    TI__Mastermind 37683 points

    Hi Martin,

    Can you answer the following questions:

    1. What is the load value?
    2. At what voltage are you driving the load?
    3. The OCP deglitch time is very long. Is there a reason for the long deglitch time? capacitive loads can have large current increase the OCP fault won't trigger until the current is above the threshold for longer than the deglitch time. If the current is very large, it can cause damage before OCP is triggered. I think this is what cause the damage.
    4. What is the expected current through the load?
  • 0 in reply to Pablo Armet
    Prodigy 20 points

    Hi Pablo, thanks for your reply.

    1. Fault occurs with no load connected, but shorting between PWM groups.

    2. I'm driving at 24V.

    3. There is a reason: I am driving electromagnetic valves. These come with a very basic electronic to regulate the current through them. (Via MOSFET, that is PWM-opened and closed.) This electronic has to be fed from the same source which supplies switching and holding currents for the electromagnetic valves, i.e. the DRV8906's high side.  While the acual current through the electromechanic valve is no problem, the electronic requires to have loaded its capacitive circuity first. If i use shorter deglitch times, the DRV8906 shuts off before this circuity is loaded and I dont get into a steady operating state. When the circuit is already loaded - there is no problem with OCP shutdowns even at shorter deglitch times.

    4. The fault occurs with no load connected, but shorting between the PWM groups, so calculation for max current is only based on assumptions, since it depends on the on resistances of the FETs in their respective states.

    I would love to set the deglitch time shorter, but need to load the capacities first. Maybe you can recommend a way to go for the descripted problem?

    Thanks alot!

  • 0 in reply to Martin Burgmair
    TI__Mastermind 37683 points

    Hi Martin,

    Thank you for the detailed information.

    Can you clarify what you mean by shorting PWM groups? Maybe drawing a diagram may be helpful for me to understand?

    The combination of the longer deglitch time and current needed to load the capacitive circuit may be causing high currents for long times which may be causing damage to the IC. Have you measured the current required for the capacitive load to be charged? Is this current higher than the OCP threshold?

  • 0 in reply to Pablo Armet
    Prodigy 20 points

    Hi Pablo,

    thanks for your effort, I really appreciate it! 

    See the screenshot from my schematic: Note that I split the DRV8906 up in a power and logic part, depicted is only the logic part. Also, Net Ties (NT*) are just copper traces in the PCB. The PWM groups are configured according to this circuit, which matches the IC's design. (=Group0: Out1,2,5; Group1: Out3,4,6) The error case (short) is drawn by hand.

    Exactly, it's the combination of high current and long deglitch times. (Yes, inrush current > 1.3A/Output, that's why the OCP flag is risen with shorter deglitch times. Which in turn is why I chose the deglitch times that long.) But I just assumed, the device would not offer any OCP settings capable of destroying it without any outer circuity - especially since it claims to be short-safe (rf. to my initial comment). 

    At this point, I am thinking about additional current limiting circuity, to limit the inrush current for the initial capacitive load and then switches off. See e.g. this snippet:

    But the problem is, then I won't see the open circuit condition anymore, which is one of the reasons I chose this IC in the first place. But maybe I am not the first one with this problem? :)

    Thanks again!

  • +1 in reply to Martin Burgmair
    TI__Mastermind 37683 points

    Hi Martin,

    Thank for providing the information.

    This App-note has some information on how the OCP circuit of this motor driver functions. In short summary, there is a analog current limit circuit which clamps the current to a level which is typically much higher than the OCP limit. After the deglitch time expires, an OCP fault is triggered. Under ideal conditions, this higher clamping current before OCP is triggered should be low enough to prevent damage to the IC. however, the current may overshoot initially before it is limited. Depending on many factors like voltage, temperature, load, this overshoot may be very high. It can be high enough to cause damage. The other possible cause for damage is the long OCP deglitch time. If you can limit the inrush current to stay below the OCP threshold, the OCP deglitch time can be reduced. The device should still be able to protect against shorts but may not be damaged due to the lower deglitch time.

  • 0 in reply to Pablo Armet
    Prodigy 20 points

    Hi Pablo,
    That's what I am going to do. Maybe some additional information for next readers: If one does not need the degree of freedom to choose the rising time of the newly limited inrush current, one can calculate the current limiter's resistor network in such a way that it does not exceed at least the "normal" open circuit test current of the DRV8906. At this point, I am waiting for HW to verify these assumptions.
    Thanks!