Probe Calibration Renishaw MP12
CrashCNC wrote: I'm not having success integrating this code with probe screen as suggested.
I believe the probe screen python code is using linuxcnc call stat.probe_position() to identify where the probe was triggered after each move.
Since the code supplied pulls variables #621-#628 (offsets) through a Gcode subroutine and adjusts the measured probe value within the subroutine into variable 5001 and 5002 probe screen is missing the offset.
Anyone else have this issue? Are variables 5001 and 5002 special? Maybe i'm missing something.
**EDIT - I think i understand . This code is attempting to change the absolute position on the mill before probe screen has a chance to read it. Isn't that dangerous? Would it not throw off the homed location? Probe screen still would never get the result because it is determining the probe trip point from the previous G38.2 calls which do not include the offset table variables.
The probing calibration routine worked great. I agree with what was calculated. Thank you plopes9000 for your efforts!
Checkout the original post, I've just updated it - I believe the integration with probe screen is now taking into account the calibration values.
Data was gathered with 10 touches each at 10 degree increments. The 10 touches were averaged to remove contact variation and also used to determine the variation. The pre-travel plot is the probe error when using the physical probe ball diameter and not the corrected diameter. The residual error is after the probe pre-travel has been removed by using the electrical probe ball diameter instead. The shading in the plots indicates the 95% confidence interval. All plots have probe center alignment calibration applied.
As my machine has 1um resolution on the encoders I consider anything around 5um to be good enough. Both the renishaw and wako probes were able to be calibrated to around 5um with a x-y offset and probe ball diameter correction as seen in the residual plots. The 95% confidence variation on both probes was about 1.2um so they are likely repeating better than I can measure on my machine. Both probes are good to 15um with only x-y calibration.
For the Imsrv TP-100 the story is a bit different. This probe is a low cost unit priced at about $250 that I picked up many years ago when I started out with a Taig CNC. I think it would be fairly representative of hobbyist level probes. It is a simple switch with no resistive sensing as is used in the renishaw to improve sensitivity. This is apparent in the pre-travel plot where the travel to trip is around 120um. The variation is also much higher with a average 95% confidence variation of 4um and a maximum of 10um. Probe diameter calibration is required for this probe to be accurate on just about any machine. After probe diameter calibration there is still about 20um of error. It would be possible to compensate this using either a curve fitting or multi point approach. I was thinking of implementing a multi point calibration but I am not sure how many would benefit from it. It would only apply when using a low cost probe on an high resolution machine.
PS Yes I know they should be polar plots but I haven't figured out plotting in python yet and excel doesn't have polar plots