Measuring a pocket location

[WolfMan]

Out quality department is measuring a location of a cast pocket, basically taking two opposing points and checking the center of the midpoint. The pocket depth is tapered and the form of the pocket is far from ideal. The depth varies from .015-.030. They are Using a .3mm probe. I am trying to explain that when the pocket is that shallow you cant really trust the results. They are convinced that since both sides of the pocket may have a probing issue it should be canceled out and should not effect the location of the pocket. How do I explain that it is not recommend using a CMM to measure pockets this shallow especially when the surface finish and the form may be uneven.

[Douglas]

with just 2 points one side might hit a high spot of the form and roughness while the other hits a low spot, it would shift the position by the amount of the total variation. Location certainty could never be better than the surface and form deviation combined? You'd need more hits on the profile pf the pocket to improve location, what shape is it a square pocket or round pocket or oddball?

[WolfMan]

Most pockets are somewhat the same, occasionally they not. My concern is related to the the depth.

[Douglas]

you mean it is probing 2 points on the bottom to determine depth?.I thought you meant a point on each side to determine location. such a small ruby diameter will find every bit of variation too

either way each point might have error and when in opposite directions cancel each other out, but when error is in the same direction for each point it will not cancel out, the error will be reported

[louisd]

if the pocket depth is 0.015 to 0.030, although you have casting roughness, radii/break edges, draft, and whatever casting defects there...
you can technically shove a 0.3mm probe down to that pocket and hit SOMETHING.
You have just 0.003" from the bottom surface for clearance along prehit/retract distance when it's only 0.015". Better make prehit/retract pretty small, lol.

I would think sampling a few more hits than just one on each edge, to generate a line for each edge, then construct a midline from both sides would be more repeatable.

[louisd]

Probe a hit at bottom of pocket, directly in front of where you take a point. origin to that point, then measure your hit, exactly 0.015" away from your origin. this will give you 0.003" clearance from bottom plane, and hopefully still snag something there, which captures your pocket...