• State Resolved
  • Replies 3 replies
  • Subscribers 30 subscribers
  • Views 133 views
  • Users 0 members are here
Support feedback
Options
Options
Related

CSD88584Q5DC: CSD88584 failing after everal weeks of operation

MIKE TITTA
Prodigy 90 points
Part Number: CSD88584Q5DC
Other Parts Discussed in Thread: DRV8306, CSD, CSD88599Q5DC, , DRV8363, DRV8353, DRV8323

We are currently designing a motor driver brd to control a 36Volt BLDC---The motor only draws 3 to 5 Amps---We are using a DRV8306 to drive the motor CSD88584 FETs---as mentioned we are seeing failures after several weeks of running the motors---We pretty much copied the eval brd design which was optimised for 24VDC operation ---We are susupecting that because we are running the FET near its max VDD(36V) some ringing might be pushing it over its max VDD. So we are going to go to the CSD 88599 FET with a Max Vin of 60Volts . My question is can the ringing on the VM line possibly get into the VM input on the DRV8306 and if so can a RC filter feeding the 8306 be an option . Additionally I noticed in the data sheet for the motor FET a note stating a snubber circuit shoud be use either instead of or additionally to the .01uF bypass caps on all the FET VINs. As mentioned earlier we are suspecting ringing to be the issue in the design.The enginner involved on the project has did some preliminary measuring but has been pulled away so he has not gone thru the brd in detail with a scope to look for ringing . Would appreciate your thoughts ----Thanks, Mike Titta  R&D   

  • 0
    TI__Mastermind 39270 points

    Hello Mike,

    Thanks for the inquiry. The CSD88584Q5DC has an abs max Vin = 40V and operating with 36Vin does not leave much margin for transients and voltage spikes. It is quite possible that the FETs are being repetitively driven into avalanche when VDS > BVDSS which can cause cumulative damage to the FETs resulting in failure. I would definitely go to the CSD88599Q5DC power block with the higher voltage (60V) FETs. These are high performance power blocks and 3A to 5A should not cause any damage due to excessive power loss. I would recommend the 0.01μF on the Vin pins for each power block whether you're using a R-C snubber or not. It's a good design practice. The R-C snubber will dissipate additional power but can help to reduce the switch node spikes. I'm the applications engineer for the FET product line and I cannot answer your questions on the DRV8306. I will reassign this thread to the motor drives applications team.

    Best Regards,

    John Wallace

    TI FET Applications

  • +1
    TI__Expert 5635 points

    Hi Mike, 

    The DRV8306 was designed for 24V systems. This being said, your operating voltage of 36V is extremely close to the abs max of 38V so the damage here is not surprising, especially during your long term operation tests. 

    We typically spec the abs max of our DRV parts to be ~1.5x to 2x the nominal operating voltage. This is because transient spikes are common in motor systems and can be a frequent source of damage. This is why the abs max is usually higher than the recommended operating condition.

    If you are able to operate at a lower voltage (~24-26V) this would be much safer for the DRV device and I presume you would be much less likely to see damage. If this is not a possibility, I would probably recommend a higher voltage device like DRV8353 or DRV8363 in order to meet the 36V operating voltage of your system.

    RC snubber may be able to suppress transients, which would probably help in this case, but you would also have to worry about negative transients manipulating the behavior of the FET. (unintended turn on/ turn off).

    Thanks,

    Joseph

  • +1
    Prodigy 90 points

    John--Thanks for the help we are going to go with the higher voltage parts---we just got in an eval brd using the DRV8323 in hardware mode with the CSD88599Q5DC FETs---Just powered it up looking good so far---I have been looking on how to use DRV8323 which uses 3 separate r sense resistor and associated op amps to a single r sense circuit similar to the driver we were using ---As far as I can see TI  recommends connecting 2 of the FET sources to gnd and just use one FET connected to a sense resistor and associated op amp which will be outputted to our main control MCU ---A second scheme TI recommends is connecting all three FET sources together and connect them to a single sense resistor and op amp --- Like to know what you think John --I know your not the controller guy but if you could pass I would appreciate it-  Thanks Mike Titta R&D