• State Resolved
  • Locked Locked
  • Replies 3 replies
  • Subscribers 82 subscribers
  • Views 908 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.

Earth Ground Fault Detection

RLetizia
Intellectual 990 points

Hi, 

Hoping to get your expertise on detecting U,V,W to Earth ground faults.

From what I understand, one of the most common faults when driving high voltage 3Ø AC motors are phase (U,V,W) shorts to Earth ground (PE).    When using low side shunt resistors for phase current measurements this fault can’t be detected since the fault current is not returned through the shut resistors.

I can think of three hardware solutions for this:

1) An external equipment protection GFCI (RCD) would catch this fault but probably not in time to protect the 3Ø driver IGBTs, plus the drive wouldn’t be inherently protected since an external GFCI device is needed.

2)  Current transforms on the U,V,W lines that trip if the net current is not zero.   This seems like a costly solution but probably the best to quickly shut down the drive stage without damage.

3) High side current sensing of the 320 VDC bus.   High bus current but low U,V,W would indicate the fault to Earth.    This requires high side isolated current measurement and trip circuitry.

Are there other approaches that you motor guys use or could recommend trying?

Is there a software solution measuring U,V,W low side shunt currents?  Has anyone tried using the three ADC measurements to detect this fault?

Thanks.

  • 0
    TI__Mastermind 27240 points
    Solution #1 will not be quick enough to protect the inverter.
    Solution #2 seem to be the most cost effective one but will not detect the exact phase with GND problem.
    Solution #3:
    I could not fully get your point here. But I would imagine that if you have high side and low side sensing, then you not only identify GND fault but also the phase with this problem. But it is more expensive with so many sensors and adc channels.

    Low side shunt measurement should help in implementing #2. We don't have any code, but believe it should work. If you verify it, let us know.

    regards,
    ramesh
  • 0 in reply to Ramesh Ramamoorthy
    Intellectual 990 points
    Hi Ramesh,

    I don's see how solution #2 is helped with low sided shunt measurements. The U,V,W phase currents should net to zero, but you can't get this information from the low side shunts only. I think solution #2 would require direct phase current measurements using current transformers or similar.

    Solution #3 monitors the HV bus current and and trips before the driver is damaged. If this over-current trips but not the over-current provided by the individual shunt resistors than it would be presumed the short is to earth ground. It wouldn't be able to indicate which phase is shorted to earth ground, but troubleshooting instruction could indicate to remove the motor connections from the drive and use a megger to determine if any of the motor phase were shorted to ground in the motor or motor cabling.

    Have you seen implementations to detect phase shorting to earth ground?

    Thanks.
  • 0 in reply to RLetizia
    TI__Mastermind 27240 points
    Letizia,

    If you need to continuously track the outputs for GND fault with NO latency, then you do need a series current measurement hardware.

    The solution 3 from my prev post will have time latency and so will not apply for continuous tracking.

    Just an additional thought to what you mentioned reg solution 3, if you monitor just one polarity, you may not be able to detect the GND fault of the HV bus itself. In that case, you may need to sense both + and - currents and trip if they are not equal. If the wire length of the dc bus is insignificant, then it may not apply.

    I have not implemented/ verified GND fault detection logics yet.

    rgds,
    Ramesh