Different measurement results using 2 different angles.

[Turbodan]

I have a part that has a dimension of 3.95mm (straight distance of 2 flat surfaces with same vectors). The part is mounted to a holding fixture. In order to measure this dimension with the part mounted on the fixture I have to take a hit with the probe angle of A90B-90 then turn my probe to A105B90 and come under the part to take a hit on the other surface. I have to use a 2X40 ruby probe to get under this part. My hits are at the same Y and Z axis. So the hits are parallel of each other. If I flip the part over and measure these exact same points with my probe at A0B0 for both hits I am seeing about a .05mm difference in measurement from the measurement taken with change in probe angles. Using micrometers my measurement where I don't change probe angles is much closer to what the actual is. I know longer probes will affect deviation a little. What are your thoughts? All probe angles have been calibrated. I have used 2 different medium force TP20's and got the same results.

[Schlag]

Do you use a master probe in your calibration routine ? If you don't you need to start. Look up other posts referring to master probe calibration as a start.

[JEFMAN]

First, you measure a distance between two planes by only two hits, so I believe the distance can't be accurate.
I would check also if it's really the ball that touch the part in A105B90.
What is the stem vector of your calibration sphere ?
Calibrating A105 angle needs to use a sphere with a angled stem.
Did you calibrate the tips at the same time ?
What are the results (diameter and std dev ) ?
Does the part is rigid enough ?
Could it be deformed by the holding fixture ?

Sorry, I don't answer, I only ask some questions...

[NinjaBadger]

First, you measure a distance between two planes by only two hits, so I believe the distance can't be accurate.
I would check also if it's really the ball that touch the part in A105B90.
What is the stem vector of your calibration sphere ?
Calibrating A105 angle needs to use a sphere with a angled stem.
Did you calibrate the tips at the same time ?
What are the results (diameter and std dev ) ?
Does the part is rigid enough ?
Could it be deformed by the holding fixture ?

Sorry, I don't answer, I only ask some questions...

Calibrating A105 angle needs to use a sphere with a angled stem. - Are you sure on this one Jeffman? I'm pretty sure I can do this with a 0,0,1 Tool vector.

[Douglas]

I also calibrate with wrist at A-115 with sphere at 0,0,1 it has tons of clearance... if anything the ball needs to be higher from the table I find

[JEFMAN]

Calibrating A105 angle needs to use a sphere with a angled stem. - Are you sure on this one Jeffman? I'm pretty sure I can do this with a 0,0,1 Tool vector.

I'm pretty sure that the accuracy is better in this case ! (depends on the number of hits... If the calibration is done with 5 hits, the difference shouldn't be visible - Maybe someone here remember what I think about calibrating with 5 hits !)

[UKCMM]

As well as the points raised it is also possible that this difference is being caused by the features used in the alignments and the method of dimensioning, a bit in the dark unless more information is given.

The attached sketch gives an idea of how alignment could cause this error and if using distance with points relies on the features being perfect parallel to each other.