Best Practices for Choosing In-process Inspection Dimensions & SPC Dimensions

Jszymkowski

Registered
We manufacture small and complex injection molded plastic parts with commonly applied tolerances of ±0.1 mm. As a design engineer, I apply all necessary functional tolerances to the drawing and ask Quality to identify in-process inspection dimensions and SPC dimensions. Quite often, I question Quality's choices in this matter. The in-process inspection dimensions chosen by Quality are prone to subjective results because the available measurement methods on the shop floor are limited (calipers, height gauges, etc.) or the point of inspection may vary because the inspector isn't limited to a single location across a feature of size. Often, I find the features chosen for inspection are not inherently subject to change due to process variation. My argument to Quality is: If we choose dimensionally stable features to measure, what's the point in inspecting these, especially for SPC dimensions?

My question to the reader is: What are the best practices for choosing in-process inspection dimensions and SPC dimensions? Are there any publications that cover this?
 

Emmyd

Involved In Discussions
I'm a Quality Manager for an injection molder and can provide a little feedback based on my experience.
#1 - Are functional gages available for in-process inspections?
#2 - are parts adequately cooled and shrinkage rate stabilized when in-process checks are done?
#3 - are the point chosen by Quality mating points? Ideally points where parts mate with others should be stable and not show as much variation (for example hole sizes)
#4 - Is there enough discrimination in the measurement resolution to detect variation within the specification tolerance?
#5 - can you provide some specific examples of points of disagreement between you and Quality?
#6 - Has a DFMEA been provided to Quality showing which areas are subject to concern if out of spec?
 

Jszymkowski

Registered
Thanks Emmyd. See my answers in red
#1 - Are functional gages available for in-process inspections? Yes, go-no-go could be made available. Are there other gages that could be used?
#2 - are parts adequately cooled and shrinkage rate stabilized when in-process checks are done? Yes
#3 - are the point chosen by Quality mating points? Sometimes, but often this is driven by input from Engineering. Ideally points where parts mate with others should be stable and not show as much variation (for example hole sizes)
#4 - Is there enough discrimination in the measurement resolution to detect variation within the specification tolerance? This is something I frequently ask. Digital calipers are frequently used to take measurements, and I think between their precision, and operator variables, there is too much noise to discriminate well.
#5 - can you provide some specific examples of points of disagreement between you and Quality? One example: a 1.75 mm ±0.05 mm dimension (we use ISO standards) was designated as an in-process inspection dimension. This is a solid feature of size (~1.5x typ. material thickness). In my experience there is little risk of this feature varying significantly in size unless there is a VERY BIG change in the process, maybe not even then. It really can't be affected by warp as it is a very short feature. Maybe I'm missing something?
#6 - Has a DFMEA been provided to Quality showing which areas are subject to concern if out of spec? Usually, Engineering meets with Quality to review the selections, and at that time we can communicate which features are associated with the highest risks of functional failure.

In the end, Quality owns the designation of in-process and SPC dimensions, but I wish I could point them toward some Quality standard that described the minimum requirements for choosing dimensional tolerances to be used for statistical evaluation or alerting Production that a process change has occurred. I'd rather not rehash the same discussions repeatedly. I've searched and Googled, but haven't found such a standard. My not having a Quality background probably isn't helping.
 
Top Bottom