I was wondering if anyone has any sort of rule-of-thumb about how many points to take on diameters. I know that varies for size and/or maybe what kind of tolerance on the feature. I mean, I know how many hits I usually take on certain features and what works for me, but is there any sort of stuff out there that talks about recommended numbers of hits for features? Just wanted peoples thoughts on this...
There are so many variables - type of probing system, uncertainty of the entire system (probe + CMM), tolerance you are trying to achieve, specific customer requirements etc...
For example, if you need to apply a UPR filter, you need to scan the complete circle (no partial arcs) and have at least 7 hits per undulation for the calculation to work correctly so, for a 50 UPR filter
MinimumPointsPerMM = 7 / ((diameter*pi) / 50)
I've also come across several OEM's that have their own guidance on number of hits - some provide a table where they might list the number of hits they expect you to take for certain ranges of sizes (up to Ø5, over Ø5 and up to Ø10, etc). I've seen others apply a formula - number of hits = 10% of the circumference of the circle, rounded to the nearest whole number for example.
I think kingsld1's idea is probably the most scientific and the only real "rule" you could follow.
There are so many variables - type of probing system, uncertainty of the entire system (probe + CMM), tolerance you are trying to achieve, specific customer requirements etc...
For example, if you need to apply a UPR filter, you need to scan the complete circle (no partial arcs) and have at least 7 hits per undulation for the calculation to work correctly so, for a 50 UPR filter
MinimumPointsPerMM = 7 / ((diameter*pi) / 50)
I've also come across several OEM's that have their own guidance on number of hits - some provide a table where they might list the number of hits they expect you to take for certain ranges of sizes (up to Ø5, over Ø5 and up to Ø10, etc). I've seen others apply a formula - number of hits = 10% of the circumference of the circle, rounded to the nearest whole number for example.
I think kingsld1's idea is probably the most scientific and the only real "rule" you could follow.
Neil, I was not aware that you cant use a UPR filter on a partial arc. If I'm scanning a circle that I must scan as two separate halves how would I apply a UPR in this case? Do I need to construct a circle from the two halves and apply the filter then?
