Correct reading of a 90° of radius

[[Ma...]]

Hi experts,

i have a task, where i have to measure a part of machined a cylinder.
It's about max 95° angle and radius is 15mm with tolerance of +/- 0,05mm.

I got measurement protocols from a customer with numbers and i was trying to get same values.

After tweaking strategy ( clockwise/counter clockwise; speeds; angle coverage ) i got close to +/- 0,01mm to customers numbers ( while tweaking strategy i got +/-0,10mm different numbers )

Customer stated only speed: 8mm/s - which is a little fast i think ( active head ) and radius measurement for cylinder ( 3 circular paths ) and a radius measurement for 3x circle.

Now question: should i use Radius callout or Radius measurement?

One note - using profile i got +/- 0,015mm deviation, only one part ( first 5mm of path ) is showing -0,07mm deviation from model - which is skipped by 0,5s from start of a scan.

 

Thanks for inputs. I am out of ideas - i am not willing to sign for this report - i would like to tell customer about problematic evaluation ( different direction of scan will dramaticaly change results )

Edited Thursday at 01:30 PM

[[Ch...]]

for size I prefer radius callout and constrain location .. all based on plausibility constraining does not all produce the most accurate results.

My opinion: radius measurement is tied to an alignment or location, where radius can be evaluated independently

[[Cl...]]

In addition to that. I would use single points instead of scanning.

[[Ma...]]

Thanks for ideas

Please sign in to view this username.

 

Please sign in to view this username.

I would also choose points rather than scan - at least 5 or 7 points is my choice.
Here is a sketch so i wouldn't restrict location. There is no plane/point where i would lock it to.
This is only half of a part and there is no dimension telling me where to start and end ( except for width )

Edited Thursday at 03:06 PM

[[Ke...]]

Y'all are going to call me crazy, but... I scan pretty much everything, including partial circles.

I don't support production at all, so I have no issue scanning at slower speeds: 7mm/s is my typical "max" for "smaller" diameters... for this size, I would probably be in the 3-5mm/s range. 3mm/s may seem slow, but slower speeds usually = less filtering being done at the start of the scan... for shorter segments.

I might start the scan before the edge of the radius, so that the probe rides up the lip of the raduis and I'm including as much of the segment as possible (If you are getting significantly different results depending on your scan direction, I wonder if that may be reated to the form of the radius, at each end of the segment?)... a couple of other scanning options: 1)Using curve to begin and end the scan, going from top plane to side plane, and then recalling those points into circles/cylinders 2) starting the scan in the middle of the segment, and scanning in both directions to to cover as much of the segment as possible, without having data at the edges of the segment being filtered out (I have never done option 2, but I remember someone suggesting it, and it seemed like a sound idea)

I have been able to correlate the results of my scanning method to results from traces on a MarSurf, on plastic molded parts with a 60 degree segment... I even had to go as far as checking them visually with radius gages to prove that my numbers were right, because a supplier was checking the same parts/location with a 6 point method (which I think can be fine, if the form is "good enough"; but when the parts are deformed (in this case,by plastic shrink) 6 points is not enough.

For correlation with the customer, I would ask for *exact* settings, and scan heights,lengths,start points, etc.

[[Ma...]]

Please sign in to view this username.

 Thanks.

Because of customer stated scanning speed, then i tried that - then i reduced it to 3mm/s ( first 1.5mm is masked ) - that was looking right speed for that.

I've played with point density ( started with 0.05mm - then finished with 0.1mm )

What i saw with profile callout - a few mm from right edge was about 0.07mm lower then rest of model, so with starting from there with high speed it will eliminate that part.
I agree with scanning and deformed surfaces.
I often do a profile check to know it's form and then decide how to approach with programming.

Here customer stated angle coverage. speed and location ( not stating start/end angle and direction ).

All i wanted to confirm is usage of "Radius" ( reported value from feature ) or "Radius measurement" which reported me for 189 points that it's within tolerance.
Real values was: Radius: 15.2048 / Radius measurement: 15.13.. - 15.15.. So i was a little confused.