Scanning Path and Stylus Orientation

[[Do...]]

Hi everyone,
I wanted to see if anyone had information pertaining to stylus orientation when scanning. We currently scan a large part with the stylus in the -Z orientation and the part is oriented perpendicular (normal) to the probe. My question is, is this the best approach and what are the pros and cons of scanning from a different angle and direction? My initial assumption was that scanning normal to the face would inherently introduce more noise in the scan via the friction introduced, however, even if I were to change the angle, wouldn't the measuring force be the same as well as the friction? My hypothesis is that scanning at an angle (anything but normal to the surface) and following a trailing path would lead to optimal results whereas scanning towards the surface or normal to it would yield a less accurate measurement. Any feedback would be greatly appreciated!

Scan Normal.pngScan Away.pngScan Towards.png

[[Da...]]

I would compare it to writing with a pen. This seems the most natural to me:
[img]download/file.php?id=13627[/img]

[[Je...]]

Please sign in to view this quote.

Oh my. This topic has been nerded-out ad nauseum. There are so many variables to consider that fortunately don't have to be considered if your task involves reasonable tolerances.

Oversimplification: if your stylus weight and length meet the requirements for your CMM's sensor, and if scanning speed and point density are scaled appropriately, the measurement will reach an acceptable level of plausibility. Also, technologies like Zeiss's Navigator help CMMs optimize settings to achieve an agreement of cycle time and accuracy.

Active sensors like Zeiss's VAST adapt to larger stylus weight and lengths and abnormal probing angles better than passive systems like the XXT you have pictured. Yet, even the XXT performs well scanning perp. to a surface's normal due its internal spring mechanism that senses deflection and compensates accordingly.

It boils down to how much form error you need to plausibly evaluate. What size of tolerances are you working with, and how important is cycle time? Are these for pass/fail decisions on high-volume production parts, or do you need to capture a digital twin for more tedious projects like reverse engineering?