CNC10 (Linux) Z axis dropping over run time

I am a hobbyist and use this mill for hobby uses so it seldom gets ran for long periods of time - I made 9 parts Sunday that take about 30 minutes each to run. At the end, I noticed that the last part i ran was about .020" thinner than the first - I face off the back so I can make these work, but after checking each part, it looks like the Z axis is dropping a few thousandths of an inch with each part. I rechecked the Z axis part 0 and it confirmed that it had dropped by about .020" according to my reference tool (an electronic edge finder). Belt is tight, no pulleys slipping. It seems pretty consistent to be encoder counts being dropped - any other ideas? Could this be a bad encoder? Any thoughts on this? The machine was ran from start to finish on the same session - now new homing or new part zero setup during the several hours I ran it - just making the same part over and over.

if you are using steppers, it sounds like missing steps on Z axis up travel when it is fighting against gravity. Try reducing velocity and acceleration

Control is Centroid Servo3IO DC running SEMS MTS30U4-57 brush type DC servo with brake on the Z axis (factory setup - this is an Ajax DM45 NC mill). I do have some backlash in the Z axis - was wondering if that might affect things as well. Checked pulleys - everything is tight on the keyed shafts, timing belt drive is in good shape - so seems to me like it has to be the controller or the encoder in the servo missing counts.

On the PID Configuration screen (F1/Setup -> F3/Config -> F4/PID) see the "Abs Pos" column for the affected axis. Also note the asterisk ('*') next to the axis label in the left-most column.

The asterisk shows when the encoder index pulse is present. It should only appear in a very narrow band (1-3 encoder counts wide) once per turn of the motor.

If you have homed to the index pulse, and you check the display on the PID configuration screen immediately after homing, you should see the index pulse at an Abs Position of 0 counts, and again every 8000 counts away from there (assuming you have the typical 2000-line encoders).

Now run the machine until it has clearly lost position.

Then, without powering down or rehoming, go back to the PID Configuration screen and locate the index pulse again (e.g. by slow jogging back and forth with successively lower feedrate override settings). If the Abs Pos value is no longer at a multiple of 8000 counts when the index pulse comes around, then you are losing encoder counts. This could be due to faulty cabling or connections, but is more likely due to a faulty encoder.

If the index pulse still appears only at multiples of 8000 counts, then something has slipped mechanically. That could be the encoder hub itself, slipping on the motor back shaft.

I may be running it again this afternoon (if I can get to my shop! I has been raining all night and all morning - may have to swim!). I ran it yesterday for a couple hours and could see about a .005 difference in the z axis height. Any idea what a replacement encoder would cost? Pretty sure you are correct on the counts. Thanks for the help - this is exactly the kind of information I was looking for!

Dan

I, or Centroid, or most any other dealer can provide a replacement encoder for $240.

You would need to verify the diameter of the motor stub shaft that it clamps on to.

Ok - tried to check this out and had some luck. I will make sure I have a printed copy with me next time and see if I can get better results, but so far I think the encoder may be flakey - want to see if I can duplicate the results before tearing into it. I could not hit the homing mark with the jog switches - will try that again next time. I could not even see it flash after running parts - I could on the x and y no problem - the Z seems maybe a little jumpy - of course it does have the weight of the head hanging on it all the time as well. I did finally see the homing mark flash after dropping the head an inch or two from the home position. The home position is just maybe 1/16 to an 1/8" from the top of travel. I do see it look for the mark during homing and it seems to find it so I'm pretty sure the Z output is working but it is very difficult to hit manually (so far).