Best practice for correlations between cmm

[[Go...]]

Hello everyone,
I have this doubt on the subject of how to do the best practice or way to correlate the dimensional results between different cmm but same software (Calypso 5.4), I am commonly asked for a maximum variation of 10% depending on the tolerance of the characteristic, which explanation it must be given or justified when that 10% variation is not fulfilled based on the same measurement program between cmm.
Thanks for your comments.

[[sh...]]

What features are you questioning? Tapped holes? Runout? T.P, Ø ?


Are both CMM's on the shop floor? Is one in an temperature controlled environment and the other is not?
Has the part been given 24hrs to adjust to the climate controlled environment in QA?
I'm sure the list of questions could go on and on.

[[Go...]]

Please sign in to view this quote.

The characteristic with more variation are the line profiles, the diameters, heights, flatness, roundness are more stable

[[Fa...]]

hello if you are talking about very small differences (a few microns) the measurement strategies should be adapted according to the type of sensor. The scanning speed of a XXT TL3 is probably different from a VAST XT, if you correlate several machines each small difference affects the result, as mentioned in the previous message also the temperature of the measured piece must be taken into account.
In the laboratory I have 5 different Zeiss and to check if they are all aligned I have made a program that measures the calibration sphere in order to quantify the differences. In this way you evaluate if the problem is given by a single stylus system or if the whole machine must be calibrated.

[[Go...]]

Please sign in to view this quote.

Hello, in your program that you use to measure the sphere that you evaluate?

[[sh...]]

We've also come across worn ruby tips. we have a microscope with 200X Mag with a light
May be possible whether your features are not correlating in a certain axis, it may be worn.

If you have a ruby tip pointing in Z- and you receive different results for flatness.
but good results for roundness, the ruby tip may be worn at the end of the tip but not "around" the ruby tip.

[[Ri...]]

These are correlation practices I've used for about 17 years now.

Controlled atmospheric conditions.
A shop based CMM may not correlate well to environmentally controlled CMM.
Test CMM to CMM sharing the same environment.

This may cost you more, I use new styli for correlation tests. Eliminates wear variance issues.
Setup on one CMM, when done move the same tools to the other CMM. Of course pre-qualify all styli.
If this test is completed once every 6 months you should get by with the new styli for several years with out wear variance concerns.

Your correlation tests do not have to be complicated geometry which requires specialty styli. Not much different than a typical CMM Calibration by an OEM, they only use a few styli and single points. About as simple as it gets.

One of my tests was utilizing a Master Sphere and a Sphere feature.
You create a grid of your CMM volumetric area. From that grid you place the Master Sphere (MS) in a grid section, Reference the Master SPhere and run a prgram to perform a 3D True Position of the Sphere using predefined styli. Reporting Size, form and location, Sphere diameter used was recalled the Master Sphere calibration certificate. We performed one test per month for 5 years changing grid locations (All corners, High Midway At the table etc.). Using this method you get a good idea of how the CMM is measuring an object in the greatest portion of its volumetric area.

Depending on your environment, that will control your tolerances. Typically I use a tolerance of 0.0003 (Inch) for the 3D True Position result, 0.0001 total for the diameter and 0.0001 for form (all Inch mode) for our shop conditions.
On articulating heads I used what I deemed the five standard articulations.
Z-, X+, Y+, X-, and Y-.
Same axis's were used on fixed heads with Star probes.
The above has been utilized on trigger and scanning heads providing some very good data about the CMMs tested thru out.

Data was recorded in Excel, Most CMM software outputs upwards of 8 decimal places, great for SPC.