• Resolved

DRV8305: BEMF bias when using 6-step trapezoidal sensorless control

Part Number: DRV8305

Dear all,

we are intending to use DRV8305 for sensorless 6-step trapezoidal control. We are using evaluation board DRV8305-Q1EVM, connected to an external micro-controller (signals EN_GATE, INHA, INLA, INHB, INLB) in 1-PWM mode with active free-wheeling.

What we observe is a really bad estimation of the zero-crossing events. Minor modifications on the commutation time (around 50 us) cause the zero-crossing time to move with an unexpected 10x factor (around 500 us).

This effect made us think that the BEMF voltage measurement is disturbed by some effect we do not understand. While investigating this, we have observed  current flow in the un-powered phase, which might confirm the existence of an undesired voltage component which is masking the real BEMF.

We have configured the device with maximum MOSFET currents, and different dead-time values (that do not have an impact on the observed effect).

Is there any reason why we cannot get good BEMF measurements using the out-of-the-box evaluation board? This effect is currently blocking our project since we cannot use the BEMF to estimate motor position/speed...

- Plot of current in one-phase, along with the 3 GPIOs used to implement 1-PWM mode, the zoom shows the unexpected current in the un-powered phase:

D021.BMP

- Plot of phase voltage and current, with zoom in the current when the phase is not powered:

Best regards,

  • Hi Diego,
    An engineer has been assigned and we will respond shortly.
  • Hi Diego -
    1) We are unable to view D021.bmp. Could you please re-attach this file.
    2) Are you able to get your motor spinning with the controller on the EVM.
    3) Have you disconnected the onboard MCU completely or are you programming the same MCU for the 1-PWM mode ?
  • In reply to Anuj Narain:

    Hi Anuj,

    I am enclosing the picture again and answering your questions below:

    1) We are unable to view D021.bmp. Could you please re-attach this file. 


    2) Are you able to get your motor spinning with the controller on the EVM.

    We are able to spin the motor in open loop (fixed phase switch time manually controlled from debugger), but not to use BEMF to infer motor speed and close the loop accordingly.


    3) Have you disconnected the onboard MCU completely or are you programming the same MCU for the 1-PWM mode ?

    The setup is:

    - Micro-controller in EVM not used at all, and 0 Ohm resistors removed to completely disconnect MCU from pre-driver.

    - Signals for phase control INHx and INHLx plus SPI connected to our micro-controller. You can see the 3 phase control signals we provide to the pre-driver (INLA, INHB, and INLB) in the previous figure.

    Just to mention it, we got the same results originally in a custom hardware developed by us. We switched to use the EVM just to confirm the observed effect with your reference HW (i.e. dscard possible error in our schematic/layout design). Both platforms behave exactly the same.

    The main point here is that we cannot get proper Zero-Crossing detection due to the fact that BEMF voltage is somehow masked by some unknown effect. This is mostly confirmed by the fact that current in the un-powered branch is not measred to be 0 as we would expect.

    Best Regards

  • In reply to Diego Alegre Oliva:

    I forgot to mention:

    2) Are you able to get your motor spinning with the controller on the EVM.

    We are not using the EVM micro and software as it is base on FOC and we are using trapezoidal control. We can´t investigate the FOC control as we are very limited in time.

    Please, help us on this topic.

    Regards
    Diego
  • In reply to Diego Alegre Oliva:

    Hi,

    I am  with a couple of pictures of the current setup, I think this can clarifies the way we are connecting the external microcontroller to the DRV8305-Q1EVM.

  • In reply to Diego Alegre Oliva:

    Diego,

    None of your pictures came through if you could please resend them.

    Regards,

    -Adam
  • In reply to Adam Sidelsky:

    Hi Diego,

    We received the setup pictures in the earlier email, so there's no need to re-upload to e2e.

    I'm going to mark this thread as closed, since we're handling this issue directly off-line now.

    Thanks,
    Garrett
  • In reply to Garrett Walker:

    Thanks for you help on this, Garrett,

    No problem, let´s work on it off-line and communicate as per our yesterday´s meeting.

    Regards.

    Diego