Master Probe

You reset the master probe to THEO and calibrate it with "yes, the sphere has moved"; the THEO offsets of the master tip are used to calculate the position of the sphere. You then calibrate all other probes. An unbroken chain of relations.

Now, if you calibrate the master tip answering "no", some deviation (machine uncertainty, different temperature, vibrations from a truck, etc.) will give the master probe MEAS offsets that differ from THEO (more or less), which will affect the relation between master and everything else until everything else is calibrated. It is a broken calibration chain, although with (mostly) very small deviations. But larger errors may lurk in the background... Read on!

Next time you calibrate the master with "yes, the sphere has moved" these deviations will be in use when calculating the position of the sphere, consequently included as an error in the calculated position of the sphere. This error will be included in the offsets of every probe that is now calibrated.

If you use different probes when you answer "Yes, the sphere has moved", each time you add a small offset to the sphere position - statistically these will cancel out, and the sphere will still be almost correctly located, but it is (relatively) easy to devise a sequence of calibrations that will consistently move the sphere more and more away from its actual location, adding more and more offset to the calibrated probes, until you finally can't calibrate a probe which is reset to THEO because it misses the sphere (or collides with it).

- OT -

For the historically interested, the physical procedure on a Johansson CMM with our older softwares was different. The sphere locating process always began with measuring a calibrated hole in the table to 'locate/calibrate the probe' (this is echoed in PC-DMIS by resetting the probe to THEO), and then locating the sphere.

You reset the master probe to THEO and calibrate it with "yes, the sphere has moved"; the THEO offsets of the master tip are used to calculate the position of the sphere. You then calibrate all other probes. An unbroken chain of relations.

Now, if you calibrate the master tip answering "no", some deviation (machine uncertainty, different temperature, vibrations from a truck, etc.) will give the master probe MEAS offsets that differ from THEO (more or less), which will affect the relation between master and everything else until everything else is calibrated. It is a broken calibration chain, although with (mostly) very small deviations. But larger errors may lurk in the background... Read on!

Next time you calibrate the master with "yes, the sphere has moved" these deviations will be in use when calculating the position of the sphere, consequently included as an error in the calculated position of the sphere. This error will be included in the offsets of every probe that is now calibrated.

If you use different probes when you answer "Yes, the sphere has moved", each time you add a small offset to the sphere position - statistically these will cancel out, and the sphere will still be almost correctly located, but it is (relatively) easy to devise a sequence of calibrations that will consistently move the sphere more and more away from its actual location, adding more and more offset to the calibrated probes, until you finally can't calibrate a probe which is reset to THEO because it misses the sphere (or collides with it).

- OT -

For the historically interested, the physical procedure on a Johansson CMM with our older softwares was different. The sphere locating process always began with measuring a calibrated hole in the table to 'locate/calibrate the probe' (this is echoed in PC-DMIS by resetting the probe to THEO), and then locating the sphere.