DRV8353: DRV8353S: two devices damaged due to unidentified reason

Johann,

Thanks for posting on the MD forum!

Which IDRIVE setting are you using?

Typically we do not recommend driving the PWM pins without the driver powered but I can't say for sure if and what damage this would cause.

Are you spinning the motor with VM/VDRAIN unpowered?

If you remove a damaged DRV from the PCB and measure resistance of it's pins to GND, how are these values different when compared to a fresh unit before placement on a PCB?

Regards,

-Adam

Hello Adam,

Which IDRIVE setting are you using?

Here are the IDRIVE settings we use:

• IDRIVEP:
  • HS = 50mA
  • LS = 50mA
• IDRIVEN
  • HS = 100mA
  • LS = 100mA

Please find below the configuration we use for the driver:

static drv8353_cfg_t m_motDrvCfg =
{
  .lock       = DRV8353_LOCK_UnlockAllReg,

  /**
   * Over-current protection settings.
   */
  .ocpAct     = DRV8353_OCP_ACT_AllHalfBridgeShutdown,
  .ocpMode    = DRV8353_OCP_MODE_LatchedFault,
  // TRETRY time not used in OCP_MODE Latched Fault.
  .tRetry     = DRV8353_TRETRY_8ms,
  .ocpDeg     = DRV8353_OCP_DEG_4us,
  // Over-current detection threshold:
  // SEN_LVL / RSENSE = 0.25V / 0.004 Ohm = 62.5A
  .senLvl     = DRV8353_SEN_LVL_Ocp0_25V,
  .disSen     = DRV8353_DIS_SEN_OvcFaultSenseEnabled,
  .cbc        = DRV8353_CBC_FaultClearedAtNewPwm,
  // Value based on max. diode forward voltage of MOSFET (see datasheet).
  .vdsLvl     = DRV8353_VDS_LVL_1V,
  .lsRef      = DRV8353_LS_REF_OcpMeasAcrossShxToSpx,

  /**
   * MOSFETs gate drive control settings.
   */
  // IDRIVEP > Qgd / tr
  // tr range (arbitrary): 100 to 300ns (based on datasheet example)
  // IDRIVEP1 > 8.7nC / 100ns = 87mA
  // IDRIVEP2 > 8.7nC / 300ns = 29mA
  // --> nearest value = 50mA
  .iDrivePHs  = DRV8353_IDRIVEP_50mA_1,
  .iDrivePLs  = DRV8353_IDRIVEP_50mA_1,
  // IDRIVEN > Qgd / tf
  // tf range (arbitrary): 50 to 150ns (based on datasheet example)
  // IDRIVEP1 > 8.7nC / 50ns = 174mA
  // IDRIVEP2 > 8.7nC / 150ns = 58mA
  // --> nearest value = 100mA
  .iDriveNHs  = DRV8353_IDRIVEN_100mA_1,
  .iDriveNLs  = DRV8353_IDRIVEN_100mA_1,
  .tDrive     = DRV8353_TDRIVE_4000ns,
  .deadTime   = DRV8353_DEAD_TIME_100ns,

  /**
   * Current sense amplifiers settings.
   */
  .csaGain    = DRV8353_CSA_GAIN_20V_V,
  .vrefDiv    = DRV8353_VREF_DIV_VrefDiv2,
  .csaFet     = DRV8353_CSA_FET_SenseSpx,
  .csaCalA    = DRV8353_CSA_CAL_NormalOp,
  .csaCalB    = DRV8353_CSA_CAL_NormalOp,
  .csaCalC    = DRV8353_CSA_CAL_NormalOp,
  .calMode    = DRV8353_CAL_MODE_Manual,

  /**
   * Driver control settings.
   */
  .pwmMode    = DRV8353_PWM_MODE_6x,
  // Default value (not used in 6x PWM mode).
  .pwmCom     = DRV8353_PWM_COM_SynchRectification,
  // Default value (not used in 6x PWM mode).
  .pwmDir     = 0U,

  /**
   * Fault report settings.
   */
  .disGduv    = DRV8353_DIS_GDUV_UvLockoutFaultEnabled,
  .disGdf     = DRV8353_DIS_GDF_GateDriveFaultEnabled,
  .otwRep     = DRV8353_OTW_REP_OtwReportedOnNFault,

  /**
   * Commands.
   */
  .coast      = DRV8353_COAST_MosNotInHiZ,
  .brake      = DRV8353_BRAKE_BrakeOff
};

Typically we do not recommend driving the PWM pins without the driver powered but I can't say for sure if and what damage this would cause.

We can add a firmware protection that turn off the PWMs as soon as the voltage on VM drops below a certain threshold.

Are you spinning the motor with VM/VDRAIN unpowered?

Do you mean spun by hand? The motor in our case is powered with the same voltage as VM which means that when the driver is off, the motor is no more powered.

If you remove a damaged DRV from the PCB and measure resistance of it's pins to GND, how are these values different when compared to a fresh unit before placement on a PCB?

We compared 2 faulty ICs with 1 new IC out of the PCB and the measurements between pins and GND are the same for faulty ICs and good IC.

We are currently running out of ideas. Have you checked our schematics? Is there anything that could be the origin of this issue?

Thank you.

Regards,
Johann

Johann,

Thanks for the information.

The schematic looks good, no issues there.

The IDRIVE seems appropriate for the Qgd of your FETs.

The firmware change is a good idea.

Yes when asking about spinning the motor when unpowered, I am asking about backdrive in the application or any way that the motor would turn with the power off.

Did you check all pins? I find it strange that we would not see any differences here if the chip is damaged.

Regards,

-Adam

Hello Adam,

Yes when asking about spinning the motor when unpowered, I am asking about backdrive in the application or any way that the motor would turn with the power off.

No, there is no mean to drive the motor by any other way.

Did you check all pins? I find it strange that we would not see any differences here if the chip is damaged.

Yes, we checked all pins (see document attached).117401_DRV8353_Dead-ICs_Meas.xlsx

We suspect that this is due to the following scenario that happened:

1. The motor was moving (FOC controller active).
2. The motor was then power off (as well as the DRV) but the FOC controller was still regulating (= sending PWM signals).
3. We then powered on again the motor while the FOC controller was still trying to regulate.

Is there anything that is mandatory during DRV startup to avoid such permanent damages?

Or do you have any other ideas?

Thank you.

Regards,
Johann

Johann,

I don't see values for the BAD units, only "idem".

FOC is a complicated algorithm and I don't think it would work well if the feedback system or DRV is not powered.

Likely the FOC controller went into a strange state and tried to drive the DRV at very high duty cycle at powerup or became un-synced and the DRV was damaged.

Hopefully if the FOC controller is shut down while the DRV/feedback isn't powered this will not occur again.

Regards,

-Adam

Hello Adam,

Sorry, this is a French term. "Idem" means that the value is the same.

Of course, I agree with you, the FOC will be disabled while the DRV is not powered. We were doing tests during development without any safety feature. We thought that the DRV would be protected in such use cases.

So, if I summarize:

• The schematics are OK.
• The configuration of the DRV is OK.
• The solution to avoid DRV damaging is to disable the FOC controller while the DRV is shutdown.

Regards,
Johann