Cylinder Alignment Problem

[hwilly]

We have a job on now, it's a housing with an internal cylinder as the only datum on the drawing. I need to check the perpendicularity of all the machined faces to this cylinder, also all the other bores for concentricity.

I measured the cylinder using 8 points, I measured two rows of four with the second row twisted 45 degrees for an even spread. I levelled it in Z, then set it as the origin in X and Y. I measured a plane further up on the part and set this as the origin in Z.

When I check for concentricity, I get completely random dimensions, ranging from 0.35 to 1.45! Also the perpendiculrity od the planes to the cylinder is way out, I have checked it on a height gauge.

Am i doing something wrong here? Is there a better way of measuring the cylinder and setting it as the datum? We took a part over to the customer and they checked it on their zeiss machine, and embarrassingly for me, it came out consistently correct...

[vpt.se]

Try three levels (3 circles) for the cylinder? Make sure you don't get any shank hits (shanked out?) when measuring.

[RCW52]

1/ while you were there , did they set it up/check it in the same way as you did ?
2/you ought to post a drawing of what your on about, if possible.
3/ was the same version of software used ?

[zerotolerance]

3 rows, and throw in some sample hits if possible.

[vpt.se]

3/ was the same version of software used ?

Unless the customer were running PC-DMIS on their Zeiss, it wouldn't matter?

[John R]

Align the z axis around the x axis then align the z axis around the y axis then origin the cylinder for x and y

[hwilly]

Their Zeiss has a scanning probe, so I'mnot too bothered how they checked it.

I'll try three rows of hits for the cylinder, then try and align it as JohnR advises, see how we get on...

[RCW52]

Unless the customer were running PC-DMIS on their Zeiss, it wouldn't matter?

Oh..................really ...............thanks for that.

[dwade]

Unless the customer were running PC-DMIS on their Zeiss, it wouldn't matter?

Differences in software will not give you much of a difference in results providing the alignment, measurement techniques and algorithms are the same.

[Gregl]

along with 3 rows of hits, you should move up to 5 or 7 or 9 hit per row, a 4 hit circle is one of the worst to use. The software like a odd number of hits in its calculations

[Roberto]

Align the z axis around the x axis then align the z axis around the y axis then origin the cylinder for x and y

I don't understand this. Too early I need more coffe

[Matthew D. Hoedeman]

An 8-hit cylinder in an invitation to disaster. A 6-hit would be better. If you are a good and careful operator, there will be an "unlimited" number of possible cylinders that can be made from an 8-hit, 2-level cylinder with the second level '45 degrees off', especially if the distance between the levels is close to or the same as the diameter of the cylinder. 3 levels and use a different number of hits per adjacent levels. I will generally do 5-3-4. Never had a bad cylinder from that yet.

[John Riggins]

Concentricity is going to give you problems.
Are you sure that is the design intent and not coaxiality?