formula and point export weirdness

[[Be...]]

We have a feature that we want to measure with a symmetry point to make sure it is well centered on a fairly thin wall when measuring a line on that edge.  So there are two symmetry points, one for each point in the feature.

The feature is connected to its own alignment, which is just a 90° rotation about the X-axis so that what was +Z becomes +Y.

The symmetry point is in the base alignment.

We also wanted to do a test to see if the symmetry point was significant and this is where we saw some weird stuff.

The edge we are measuring is at 45° to the Y axis and perpendicular to the Z axis when in the base alignment.

We created the measurement and put a formula in for the measured point's Y location (remember the feature is in its own alignment).  the formula reads something like:

GetActual("symmetry point 1").z

This formula is in the definition of the point inside the measurement strategy and not on the nominal of the feature itself.

 

Then we duplicated the feature and put the same formula in except we put a 0.015" offset so it reads:

GetActual("symmetry point 1").z - 0.015

 

It looks pretty good so far.  It says that the measurement point in the one is 0.015" lower than the other prior to any measurements being taken.

 

Then we test this six separate times, export all the nominals and measured points to text files, import them all into a common excel sheet and start to analyze the results.

There are a few troubling observations.

1) the difference in the nominals of each point is not exactly 0.015".  It ranges from around 0.012 to 0.016, but is not exactly 0.015.  Be aware that both points are using the same symmetry point.

2) all nominal and measured Z-values match.  Between the six tests there are 12 points for the regular and 12 points with the offset. The Z-value for each point was unique, but the nominal and measured values for the point were a perfect match.  

Why would either 1 or 2 happen?  

Before you ask, I cannot share any pictures or files.

 

[[Ma...]]

So mainly those deviations can relate to vectors of touch.

You base alignment will not be perfect so in every run you can touch the part in another locations, so there is first variable - another spot - another touch vector.
Then you should use in formula "GetNominal" instead of "GetActual" - this can lead also to this deviation ( i did conversion and your difference is 0,1mm in Z axis )