Inconsistency on profile measurement

[[Ry...]]

Hello all,

I am in the middle of performing a 25pc type 1 MSA on an injection molded plastic part. We have created a fixture with spring loaded clamps to hold the part perpendicular to the CMM measuring space. There is a profile characteristic that is a critical to function, meaning we need to perform GRR, capability etc. I am unable to obtain consistency on the same part. The spring loaded clamps are applying pressure to an area that is adjacent to the surface that has a profile callout on it, so I am suspicious that may have something to do with my issue, but I would also think that the springs are allowing for repeatable pressure on the clamp. The profile callout only has DAT A in the FCF and the datum is a plane - so the only real deviation from nominal we should see is in the Y axis. 

I have the profile being evaluated with 3 separate methods. I have a scanned plane, a plane composed of 5 points (non-scan) and some individual space points that I assigned a perpendicular feature to from DAT A, reporting out the depth. On each of the perpendicular depths I have a result element with a formula which obtains the deviation and multiplies it by 2.

With each evaluation method I am obtaining a range of about .020mm, which I am almost certain will not provide passing results for the MSA. The profile tolerance is .12mm.

 

I was wondering if anyone had any suggestions or maybe I am missing something? I feel like the fixturing that we have for this part is very robust, but it could well be the cause of my issue. 

[[Ch...]]

MSA studies on surface profile are tricky and sometimes not recommended.  Here's my thoughts:

You need to understand exactly what is different between your profile plots from run to run.  Are the differences in the overall shape?  Or does it look like spikes in different places.   Your decision tree looks like this:  If the overall shape looks different from run to run, especially in places where you would expect the clamp to deform the part, then it is probably clamping that is causing your issue.  But if the differences look to be caused by spikes, then you are looking at a measurement strategy issue such as filters or point density.

MSA studies were designed around evaluating simple, one-dimensional variable measurements.   They do not contain a prescribed procedure for conducting an MSA on complex GD&T characteristics like surface profile.  They call this the silence of the standards.  It is up to us as practitioners to develop a method that works best for our organization.  

Due to the complex nature of profile measurements, it is sometimes recommended to do MSA on a select number of points instead of the entire profile.  For example, pick out a point and get your point nominals from the model.  Do an MSA study on just this one point.  What you typically see is that the MSA on those individual points is good.  

Edited Thursday at 02:32 PM