• State
  • Locked Locked
  • Replies 2 replies
  • Subscribers 29 subscribers
  • Views 667 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

DRV8432 Source Lead Current Sense Temperature Drift

Chip Slinger
Prodigy 110 points
Other Parts Discussed in Thread: DRV8412

We are using the DRV8432 as shown in datasheet figure 12, with a GND sense resistor, but with two separate motor channels operated in bridge mode. We are driving a very low inertia motor in a haptics application, with low friction. We differentially sense the current from the two bridge GND legs. When we command zero current, there is inevitably some amplifier offset due to sense resistor tolerance, other tolerances, etc... We zero this out. However, we have noticed that the zero point drifts as the DRV8432 warms up.

We have instrumented the system in several ways and have pretty much ruled out that this drift is coming from either the current sense resistors or the sense amplifier. It is real! What seems to be happening is that there is additional current from other circuit elements that is not equal for the two bridge legs, and that this current drifts over time/temperature. This probably wouldn't show up in most motor control applications b/c there is enough friction and inertia that the small offset current change wouldn't be observable. In our low friction haptics application, though, we can observe this, and the motors move slightly.

We are interested in learning:

  • whether this problem has been observed by TI or by other TI customers.
  • if this problem exists on all chip versions (sometimes parts are improved after their initial release).
  • what output circuitry, if any, is connected to the power GND pins and, in particular, whether the output MOSFET drive circuit current also returns through these pins. This is mainly important in corroborating our observations and hypotheses, but also would be helpful in evaluating other approaches should we need to implement the current sense circuit differently. For example, would there be any reason to expect a similar issue to appear were we to sense the MOSFET high legs? (e.g., do the output driver return currents appear through this leg).

Finally, we have one question regarding the DRV8432 that is apart from this issue...We have a power supply "watchdog" function that cuts off the PVDD power in the event of a processor upset. We'd like to confirm that no chip damage can occur from operating the chip with this supply removed.

Thanks in advance for your help.

  • 0
    TI__Guru 55605 points

    Chip,

     

    Not exactly sure what is going on here as we have not seen this before.  

     

    We are interested in learning:

    • whether this problem has been observed by TI or by other TI customers.

    Not that I am aware of.  

    • if this problem exists on all chip versions (sometimes parts are improved after their initial release).

    No die change has occurred since initial release.  

    • what output circuitry, if any, is connected to the power GND pins and, in particular, whether the output MOSFET drive circuit current also returns through these pins. This is mainly important in corroborating our observations and hypotheses, but also would be helpful in evaluating other approaches should we need to implement the current sense circuit differently. For example, would there be any reason to expect a similar issue to appear were we to sense the MOSFET high legs? (e.g., do the output driver return currents appear through this leg).

    The output MOSFET current absolutely returns through the GND_x pins that are associated with each output.  I assume you have inserted sense resistors in this path and are measuring the current through a voltage drop across this resistor.  What size is your sense resistor?  We only recommend dropping 100mV maximum across this resistor as stated in the datasheet.  

     

    Also, with a low inductive load (assuming here), then you may also need to insert some series inductance between the outputs and haptic elements to "smooth" the current measurements and boost your circuit efficiency.  

     

    I have attached attached a DRV8412 eval board schematic.  The DRV8412 is a low power variant of the DRV8432, but the circuit is the same.  You can use this method of current sense as a reference point.  

     

    Regards,


    Ryan

  • 0 in reply to Ryan Kehr
    Prodigy 110 points

    I understand that the MOSFET current returns through the legs, but is any other circuitry connected through these legs?

    This is how we are measuring the drift: we set the bridge to 50% output so that there should be no current going through the legs.I watch the current amplifier output level and it drifts over time. If I short circuit across the current sense resistor, the effect mostly goes away, which means our amplifier is fine.

    I have tried a few different sense resistors, and we observe the same behavior with any of them. We do have a standard LC pi filter at the output, but this is to reduce motor ripple current and motor heating rather than to provide sufficient inductance in the output. I have reviewed the documentation on all the other chips in the family and the evaluation board schematics.

    It might be helpful to add that I've built this kind of amplifier quite a few times before, using the Intersil HIP408x series drivers and with other much less integrated approaches as well. I'm familiar with the requisite design and layout constraints. This is the first time I have ever observed anything resembling a temperature drift in the current sense circuit. It is significant, too, something like 1.3% of full scale!

    Is it possible to have a more direct conversation, to send schematics, charts, etc? I've been looking at this issue for 2 weeks and we're precipitously near product release. I need to understand the problem or implement the current sense in the motor leads, which means making significant design and layout changes.

    Thank you in advance.