Parallelism and Flatness of 3 surfaces

[[Ja...]]

I have a part that has 3 pads (not real large) on opposite sides of each other. One set of pads is established as Datum A and has a Flatness callout of .0002" (three surfaces) the other is a callout of Parallelism of .0002" to Datum A (three surfaces). I have scanned all pads individually then recalled them into one plane each. I then call out my Flatness to the one plane and my Parallelism using both planes as stated per print. the problem is doing an indicator check on a surface plate vs. what we are doing on the CMM we are getting completely different results. Why?????

Capture.PNG

[[Ow...]]

A few questions:
Between indicator vs CMM, which characteristic are are you seeing the biggest difference, Parallelism or Flatness?
What do the plots look like, does it show any odd ball spikes that may not be filtered out adequately?
I assume you're holding the part parallel to the Z axis and scanning it with two different probes, how confident are you that the part is being secured enough to not deflect or move?

[[Ri...]]

0.0002" is extremely tight of a tolerance. You would need a machine that was accurate within 0.0005µm.

There are so many questions that could be asked:

How accurate is the indicator?
Are the parts clean?
How stable is the environment?
How many points are you scanning?
Speed?
Probe size?
Filter?
Outlier Elimination?
What type of CMM is this?
What sensor are you using?

[[Ja...]]

I have the Filters and outliers turned on. The Flatness is always checking between .00008-.0001 and it is the Parallelism that is off usually two of the three pads are off a little but not always the same ones either. Part does not move at all. Form looks great on all 6 pads. Probes are clean. Same speed on all pads. Datum A pads are helping establish the base alignment in Z.

[[Ja...]]

Richard-
Indicator used .00005"
300 points per pad
.3 Speed
Filter on
Outlier on
1mm probe
Accura
Vast XTGold
Parts always cleaned before run
Stable environment 68°

[[Ch...]]

What about the evaluation method on each feature (Tangent vs LSQ)?

[[Ja...]]

LSQ

[[Da...]]

So, is the part standing up on edge so you can access both sides ? OR are you putting datum A on the surface plate of the cmm?
How big (rough diameter) is the part, and how thick is it ? Material ?

[[Je...]]

.


It's second shift's fault.

.

[[Er...]]

What are the readings from each method? I would assume the CMM passes what the granite fails?

Have you tried measuring on a different piece of granite? Ours is worn in different spots, but still passes certification. And it has chips. 🤣

The granite should touch the most extreme points on Datum A. The CMM would have to hit those exact points to get the exact same reading. Datum A being a smaller diameter would magnify the differences to the pads on the edge.

[[Dy...]]

Have you tried making 1 plane and take points on all 3 pads in the same strategy instead of recalling them? I've just personally seen some issues where recalling multiple planes into one skews my results.