Hello,
We have a custom brushless motor controller design that uses the DRV8303 device. The design is used in a low volume battery operated (36V lead acid battery set) remotely operated vehicle - one for each of the drive wheels. The design is in production for about 1 year but at low volumes (less than 50 PA).
In the last few months we have had a lot of random failures on these drive controllers. We recently started an analysis of these failures and we see the following. Failure rate is currently running at approx. 20% so this needs to be resolved quickly.
We analysed 10 of the failed units. The only device that is damaged on the failed units is the DRV8303 driver IC. On these failing units the charge pump power supply GVDD only reaches 1V dc when the chip is enabled. Also on these failing units it looks as if the protection circuitry in the chip (Vds voltage mon etc.) 8303 is still functional in that 8303 is enabled the fault line (pin 2) goes high. Presumably this is because there isn't sufficient gate drive voltage to turn on the power mosfets (GVDD is too low) and the Vds sense protection circuitry kicks in a fault.
When the devices are changed out for new devices the drives function normally. No other repairs were required.
Our main suspect was an overvoltage of some kind so extensive testing on a bench set has been done to try to determine what the problem might be. These tests have concentrated on high speed accel/deccel tests (reversal) using a motor with an inertia wheel. These tests produce a higher regeneration voltage on the DC bus than on the actual machine itself (gearbox in the machine reduces the reflected inertia) but the maximum voltage measured was still only approx. 39V DC from a nominal dc bus voltage of 37V dc. These low regen levels are expected because of the good quality batteries and cabling.
Drive controller has run for several hours on these accel deccel tests without problems (this would indicate that the fundamental design looks OK) but random failures are still occurring both in factory testing for the machines and also in the field.
I do have 8 failing DRV8303 devices that were changed out. Is there any possibility that these could be analysed to try to determine the root cause of the failure? Has anything like this been reported and investigated before? (I have seen some similar posts on the forum but they all fizzeled out before any definitive conclusions were made).
We are really desperate to try to get a resolution for this problem. In the meantime we are going to fit transil devices across the 8303 power supply but this is just a stopgap measure in reality as we do not know the rootcause of the failures.
We would really appreciate any help from anyone with this problem.
Many thanks in advance.
.
