DRV10970: DRV10970 control issue

Hello Annie,

I'd like to apologize for the delay as a result of the holidays.

It sounds like the customer is applying a fixed and constant PWM duty cycle to the PWM pin of the DRV10970 and they expect a constant speed;or electrical,  commutation frequency; of about 200Hz. However, they see some variance in the motor speed of 188-202Hz, is this correct?

I can think of three possible reasons: Hall sensor placement, nonuniform load, and a weakness in the sinusoidal algorithm (which is still related to hall sensor or motor problems). 

Hall Sensor Placement:

For hall sensor placement, generally speaking, the DRV10970 is only as smart as the inputs of the sensors. If the hall sensors are misaligned by fractions of degrees, the device will be given incorrect information and then commutate incorrectly enough to cause a fluctuating but not bad enough to cause the rotor to stop.

This can be checked by disconnecting the device from the motor, powering the hall sensors, and spinning the rotor externally (I like attaching the bare rotor to the end of a drill) and seeing if each hall signal is triggering with the same frequency.

This is the most likely.

Non-uniform Load:

The load is the combination of the rotor shaft and whatever is interacting with the rotor's motion such as air, liquid, ball bearing friction, etc. You can usually tell this by trying to manually spinning the rotor at a constant speed (again, I use a drill) and looking at the phase's BEMF to make sure it is uniform and of a constant frequency for a full revolution.

This can happen but is usually unlikely. 

Weakness in Device's Algorithm or Control:

In this case we start looking at how the algorithm starting interacting with the motor. At the basic level, variations in the comparators, internal FETs, digital delay, etc will add some error in commutation.

These errors are usually on the microseconds or microvolts type levels of errors or they are constant which shouldn't affect the electrical speed, especially fluctuations of Hz. This is very unlikely.

In addition, the device algorithm sometimes plays a role, where sinusoidal commutation might make errors based on motor parameters (low inductance for example). These require troubleshooting but usually result in the device not spinning the motor at all, as opposed to <1% in speed. Try spinning the device in the other settings (e.g. sine -> trap) and see if the fluctuation gets better or worse.

I will say this is likely.

Best,

-Cole