unexpected motion pattern while measuring backlash

[mayoff 0]

I clamped a dial indicator to my bed to measure my Ultimaker's backlash. While working on that, I noticed something interesting about the behavior of the needle on the indicator. Please watch this video to see it:

 

I told the head to move 2.54 mm (0.1 inches) slowly along the Y axis, to make the needle revolve once. The interesting thing (to me) is that the needle didn't move smoothly all the way around. It moved smoothly for about 0.008 inches, then paused, then moved smoothly another ~0.008 inches, then paused, and so on.

I can think of several things that could cause this pattern. One is that the indicator itself is causing it. If so, that's a benign problem in the sense that it doesn't affect prints. Here is the certificate of inspection that came with my indicator:

I don't know if there's enough resolution in the certificate's error graph to decide whether the pattern is caused by the indicator.

Another potential reason is interference between the firmware's clock frequency and the wave frequency required to drive the stepper at the desired speed. I don't think this is the problem because if I run the test at a higher (but still fairly slow) speed, I still see the pattern, with the same distances (but shorter time intervals) between pauses. If it were a clock interference problem, I would expect the time interval between pauses to remain constant.

Another possible reason is that the stepper itself has irregularities.

The last reason that occurs to me is that the timing belt/pulley interface is irregular. Perhaps due to the way the teeth mesh, the pulley's response to timing belt motion is nonlinear.

[mayoff 0]

I decided that it's definitely not the indicator, because if I put my finger on the print head and run a move command, I can feel the pauses even when the head's not touching the indicator.

It also occurred to me that it could be some flaw in the linear bearings cause friction periodically, but I get the pattern on both axes so that seems unlikely.

[joergen 2]

I would point the finger towards the stepper, because your measured "error" coincides with the micro stepping distance:

the UM has 78.7402 micro steps/mm (1/16th steps), or 0.0127mm/step, or (if you insist on dirty units) 0.0005in/step.

0.008in/0.0005in=16

I think what you are seeing is the stepper "nudging" to a full step position, before it continues with the micro stepping, which can possibly explained with the energizing of the coils.

[Daid 306]

You lost me at "inches". But if your power levels of the motors are not properly configured then there is a slight offset in the motor positions with microstepping.

[mayoff 0]

Thank you for the explanation. I only used inches because that is how my indicator is marked.

Should I try to adjust the motor power, as described here: http://wiki.ultimaker.com/Electronics_build_guide#Tuning_the_stepper_motor_drivers I do have a multimeter but I'm not sure how I would use it for this.

[msurunner 2]

Thank you for the explanation. I only used inches because that is how my indicator is marked.

Should I try to adjust the motor power, as described here: http://wiki.ultimaker.com/Electronics_build_guide#Tuning_the_stepper_motor_drivers I do have a multimeter but I'm not sure how I would use it for this.

How are your prints looking? If they are looking good, you are probably getting the measurement associated with the slight tightening and loosening of the belts as the eccentric point of the pulley on the axis comes around. If that's the case, then I wouldn't touch the drivers. They can be a bit finicky and are easy to fry. Snowygrouch looked into some high tech pulleys that would maintain concentrically correctness but found they cost far too much.