Battery Backup Causing Motor Drivers to Short

[Topher]

Hi guys. On our printer racks, we have Model S batteries in parallel with the output of the power supplies for every 2 racks (24 printers) powering all Duet3 minis. The batteries are "ideal" diode isolated for min working voltage drop while preventing surge charging. We just encountered our first major power outage where we lost power for about 2 hours during full operation. The good news - the majority of the printers survived the event. Bad news is 2 printers heater faulted which is to be expected as the battery voltage became too low to heat the hotend to temperature but more interestingly, 3 other printers had a motor driver short fault once the battery hit around 20 volts. The fault was persistent even when the power was restored bringing the rail voltage back up to the normal 25 volts. Hitting the reset button on the printers did not help but power cycling did. None of the drivers were actually damaged. I suspect this simple inexpensive approach for battery backup of this scale may need a pricy investment in a custom power supply on the battery to give me a stable CV output. Cost aside, I would like to not go that route as adding any additional switching would cut battery runtime even with high efficiencies of 90% or so. Curious of your thoughts on this @dc42
image url)

[dc42]

@leckietech were you running the drivers in stealthChop mode? stealthChop is a voltage controlled PWM mode, so it seems likely that reduced voltage would affect stealthChop operation. In the rev. 1.03 datasheet, Trinamic gives this warning on page 42:

A motor stall or a sudden change in the motor velocity may lead to the driver detecting a short circuit or to a state of automatic current regulation, from which it cannot recover. Clear the error flags and restart the motor from zero velocity to recover from this situation.

And:

With low velocities, where the back EMF is just a fraction of the supply voltage, there is no danger of triggering the short detection.

The implication is that when the back EMF approaches the supply voltage, the short detection may be triggered. This will obviously happen at lower speeds when the supply voltage is reduced.

[Topher]

@dc42 thanks for the quick response. The board is using 2209 drivers so I believe the default mode is stealth chop according to Duet documentation as I didnt declare it with M569. I am running 256 steps, what mode would you suggest I run the drivers in to overcome this issue in the future? Spread Cycle, maybe?

[dc42]

@leckietech the default mode in 3.3 and earlier is stealthChop at very low speeds switching to spreadCycle at higher speeds. in RRF 3.4 the default mode is spreadCycle.

If the spurious short-to-ground reports are a side-effect of running in stealthChop mode, then using spreadCycle instead should avoid them.