I'm actually doing some pre-qualification tests on the DRV8711 and two other drivers from competitors. One of the testpoints is to compare the microstep accuracy. I used an Oriental Motor PK268-E2.0B with coils in parallel (2.8A nominal current), DRV8711EVAL at 24V power supply and an Heidenhain Encoder with 1024 lines with 1Vss output, processed by a Beckhoff EL5201 with 1024-times interpolation (leading to a resolution around 2,912 increments / degree).
General settings: All default exept the following: GAIN=5, TORQUE=61 (leading to 4.0A peak?)
I run the motor in 1/4 step mode with 16 single steps forward and backward and recorded the actual position with TwinCAT scope. The values were transfered to an excel sheet were I calculated the position error as well as the relative inaccuracy (in % of a quarter step). I started with DECMODE=001, as I found this gives better high speed performance than the default setting. But with this setting, the accuracy was very poor. In some positions the relative error reached 120% and even more. Going back to DECMODE=100 the situation got better. In the fullstep positions the error is below 12%, while the maximum error is still 45%. Other drivers showed maximum errors of 16% and 26,7%. To me the DRV seems also a bit more noisy than the other devices. But in contrast to that, the stand still noise on the position signal is below 0.002°, which is only 3-4 increments.
I also monitored the coil currents (with LEM hall sensors) to make sure that the motor currents are comparable. The current measurements show no obvious problems, but I recognizes the the waveform slightly varies from one electrical revolution to another?!
