Circuit Mill Development Log: Z Slip

Shane Oberloier's picture

I've noticed that the circuit mill in development is having issues with small adjustments (0.3 mm). During small movements, the axis will do one of two things - it will either not move at all, or it will move too far. For example in the picture below, the left pattern was done keeping the Z axis at a steady height. The right pattern was done with the Z axis adjusting to keep the cut level. As you can see, the mill appears to be "slipping" over time, and the bit cuts further and further into the board.

No Z movement (left) vs Level compensation (right)
No Z movement (left) vs Level compensation (right)

So I decided to gather some data. Using a dial indicator, I took measurements at varying increments. For the 0.1" and 0.01" inch tests, I would always get movement on the first attempt. On the 0.001" tests, it would take multiple attempts to even get the axis to move (the #of attempts column). I was careful to compensate for backlash so it would not effect the results. Imperial units were used because that is what my dial indicator indicated with. I also made sure the belts were tensioned first.

The test setup
The test setup

Here is the data gathered from the test.

  Positive Positive Positive  
Trial 0.1” 0.01” 0.001” # of attempts
1 0.096 0.008 0.005 3
2 0.096 0.008 0.004 6
3 0.096 0.008 0.003 3
4 0.096 0.008 0.007 6
5 0.096 0.008 0.001 3
6 0.096 0.008 0.006 3
7 0.096 0.008 0.004 6
8 0.096 0.008 0.003 3
9 0.096 0.008 0.001 4
10 0.096 0.008 0.005 3

 

  Negative Negative Negative  
Trial 0.1” 0.01” 0.001” # of attempts
1 0.1 0.01 0.001 3
2 0.1 0.01 0.002 3
3 0.1 0.009 0.006 3
4 0.1 0.01 0.001 3
5 0.1 0.01 0.003 3
6 0.1 0.01 0.006 3
7 0.1 0.01 0.005 6
8 0.1 0.009 0.001 3
9 0.1 0.009 0.005 3
10 0.1 0.01 0.004 6

For big movements (0.1" and 0.01"), the machine works accurately. There is more accuracy moving down (dropping the spindle towards the ground) than moving up (picking the spindle up). The small movements are extremely inconsistent and varied and cannot be compensated for in software. They also happen fairly consistent in each direction. The RAMPs board that is driving the motors has also been setup for full micro-stepping. There are two Z motors that share a driver. It may be wise to consider using larger motors - or enabling them to use individual drivers.