Profile Measurment - I have a SC call out for a profile of an assembly - Attribute?

[Rob Ewald]

Profile Measurment

I was wondering if someone could help me. I have a SC call out for a profile of an assembly of 8.9 mm in relation to datum A, B, and C. I am taking my measurements from nominal. In other words I have a tolerance of ±4.45mm. I am taking 3 measurements. How can I report capability using one numerical value (to represent the profile) when the profile requires a minimum of three measurements to create a plane? To me just taking the maximum measurement (one of the three measurements) and allowing this value to represent the “plane” is just plain wrong. We would never be able to show capability on the assembly. Any help that would shed some light on my problem would be greatly appreciated.

 

[DICKIE]

You might consider treating this characteristic as an attribute instead of a variable. Pass\fail verses your dilemma of defining the surface numerically. If you could use a cmm to probe the surface most packages will do profile also.

 

[Jim Biz]

Hmmmm ??? Not sure this will help and be way off base because I've not actually had to deal with this before but here's what comes to mind.

I would agree that taking one max measurment to represent the plane appears plain wrong.

JUST FOR THOUGHT:
Why could you not measure 3 to 5 parts to datum A - record average - Datum B average - Datum C average then do a comparison of the resulting averages to the ideal plane call out?

If you find the information to be well within spec you may not need to prove it by another method

Regards
Jim

 

[AJLenarz]

Special Characteristic, Profile, Capabilities… … … this has all the makings of a plot for the next Soprano’s episode.

You have a very interesting question here. Let me see if I can take a stab at it.

A profile callout is generally assigned to a surface with curve and form (ex. bumper, body panel, dash boards, pistons… … …). These types of surfaces are generally hard to define mathematically. The definition is done usually (but not always) via a CAD model. It is not uncommon to assign a profile callout to ensure that the entire surface area fits within a defined window. That is a easy way for the designer to dimension this type of surface on paper, but very difficult for the individuals down stream measure and report on it (is this beginning to sound familiar?).

Now profile measurements are defined by a single value, which represents deviation from nominal. Also defining the profile measurement is the datum. The datum defines the position and orientation of the object being measured. Usually (not always) the datum will be based off the way an assembly will interface the mating components.

Now, to answer your question “How can I report capability using one numerical value…”. I will use the automotive bumper as an example. Let’s say for example that a bumper has a profile tolerance of 1” to datum A, B, C. The datum will define how the bumper is to be fixtured, held, and orientated. Now the entire surface must not deviate more than +/- 0.5” than defined. But it is not always feasible to measure the entire surface area. And in your case, it is not easy to summarize the entire surface profile into one capability value. So to over come this obstacle, pre defined coordinates are selected and measured. It is these predefined spots that are used to monitor and collect the deviation from nominal values. And from this you can begin to calculate a CPK for this one particular spot. Multiple positions are selected across the entire surface area that is being monitored. And from this you could calculate CPK’s at multiple points of the object you are measuring.

By having predefined spots or positions, this will help improve correlation between customer / supplier. Instead of discussing one value for the entire surface, you can discuss specific points and begin to work out disagreements.

I hope this provides you some guidance. For further clarification of profile callouts, see ANSE Y 14.5:1994.

Furthermore, in the years I have been in quality, I have never seen a SC on a profile callout. You might wish to bring this issue to the customer. Make sure that this characteristic is truly important enough to put a SC designation on and not something that a design engineer has done for his own amusement or ignorance.

Right or wrong…. that is my opinion and I am sticking to it.

 

[Paul F. Jackson]

Process Capability of profile measurements is not any different than the process capability predictions of other feature tolerances that report the contain(ability) Cp, Pp and contain(ment) Cpk, Ppk of the distribution of the consumed tolerance and its associated mean location relative to the span of the specification (Cp, Pp) and its boundary (Cpk, Ppk).

It is important to recognize that every inspection performed is an abbreviation. It is an abbreviation because there is no way to examine every point on a feature and there are no perfect gages to examine it with.

The number of points you choose to gather on the surface is your decision and the integrity of your analysis rests in your ability to capture the variability of the points on the surface. That is somewhere between (one point) and (enough to not to expect significant change) It may indeed be only one point if the surface is small or you trust that one point is sufficient because you know something else about feature. You would monitor the variability of those points in your process control procedures and develop upper and lower control limits. After control been established a prediction of capability can commence.

Profile has a target value of zero and may either be constrained (controlling location or orientation and form) or unconstrained (controling form only). One can simply compare the distribution of consumed tolerance to the specified tolerance available to compute the Pp, Cp and compare the mean of the measured data to the Upper Limit to compute the Ppk, Cpk.

Profile tolerances that are not targeted nominally such as a profile that has a tolerance zone specified unequal bilateral or unilateral are a little trickier. To compare the measured data to the specification tolerance for process Capability analysis one must portray the size of the tolerance zone required to contain the worst deviation. Consider a profile tolerance that is depicted unequally bilateral -0.2 ~ +0.6 from its basic profile and is declared a total value of 0.8 in the feature control frame. If the worst deviation (the one nearest or beyond the given limits) was -0.1, one must determine what size tolerance zone could contain that point as it relates to the total tolerance 0.8. I assert that the consumed tolerance must be portrayed as 0.6, which is twice the deviation from the mean of the available tolerance. Likewise if the tolerance was depicted unilaterally 0.0 ~ +0.8 and the worst deviation was -0.2 the deviation would be reported as 1.2. Pp, Cp and Ppk, Cpk assessments for these consumed tolerances would follow the standard models.

 

[Rob Ewald]

First of all I would like to thank those involved in their responses. While varied all had some good points to make. This is what was decided upon by our company GDT teacher. Take three points measured from nominal which the fixture is now designed to do. Take the LEAST capable point ( the one agreed to that will give the most trouble) and allow this to represent your variable profile point. Include a separate attribute check to make sure all points are within tolerance. Have the customer agree to this method in writing. Enter this into the lessons learned manual and try to get right the next time.

Once again thanks for everyone’s input. It was helpful to demonstrate the variance of good answers.