Dingy2, you're right that 100% inspection is impractical if not impossible on large runs. The reality is, however, that large body stampings aren't 100% gaged either. Using statistical process controls, automotive companies have been moving successfully away from 100% verification for some time now. They select pieces for CMM inspection based on a statistical sampling algorithm that meets their quality requirements.
The CMM mechanism itself does indicate the center location of the probe at any given time. The software and algorithms therein determine the surface at the contact point. This extrapolation provides a minor error which is generally accepted by most industries. Using a cylindrical probe in a helical tracing path ensures that you are getting the smallest cylindrical zone that contacts the MMC condition of the minor diameter, and that the MMC cylinder is perpendicular to the CMM bed, which is typically a datum simulator. Additional errors will accumulate based on the length of the probe (deflection from normal increases with length), the scan speed/sampling rate, the surface condition of the probe, cleanliness of the machines surface, ...).
There are several key differences between gaging and inspecting that are driving manufacturing away from gaging and toward dimensional metrology, primarily using CMMs.
- gage fixtures for small parts with gage-maker tolerances are typically very expensive and are limited to one product whereas the cost of metrology equipment or a CMM is spread over a range of products
- gage fixtures for large parts with gage-maker tolerances are awkward to handle, damage easily, are difficult to store safely, and the cost is typically prohibitive whereas an appropriately-sized CMM can handle all workpieces from small to large with nominally the same accuracy and repeatability assuming normal maintenance
- verification of a gage requires dimensional metrology equipment, whereas dimensional metrology equipment requires calibrated artifacts; this puts gages at least one step further away from the national standard unit than the metrology equipment which translates to greater error
- functional gages only check the inner boundary, and do not indicate anything about the outer boundary condition; you would have to do LMC size-checks on each hole as well as doing the gaging, to be complete
- functional gages oly indicate PASS/FAIL at the MMC, with no indication of where the failure originates or data to determine if it is correctable
- functional gages do not provide any data regarding your process, and therefore a dimensional metrology plan must still be established to randomly inspect pieces to obtain the data needed to feed back into the manufacturing process
- functional gages do not indicate when a process has started to vary, whereas dimensional metrology with statistical tracking makes any variance immediately recognizable
- functional gages are typically applicable to one plane only, which means repeated setups, whereas CMMs can inspect all but the bottom surface of a workpiece in most cases
- each pattern of features must have its own gage fixture, whereas all patterns can be inspected simultaneously by the CMM with each pattern processed separately by the software
- functional gaging requires 100% part sampling whereas statistical sampling can be done when using dimensional metrology, which reduces your overall verification and process control costs
Functional gages do have a valuable function in situations where the tolerances are adequate to allow wear on the gage, where the gages are small enough to be easily handled and engaged with the workpiece, and where the variance of the manufacturing process does not impact the workpiece functionality.
My involvement has been in situations where cost reductions, reduction of scrap and rework, and tight tolerances were the rule. A gage-based system had been used for 30+ years for verifying standard tapers and diameters, etc.; warranty and rework costs were high but accepted due to market conditions. Problems became more evident when new (CNC-only) machinists were hired. The gages were not used properly and were abused so that they were constantly being reworked and replaced; the cost of gaging rose sharply. Because the gages weren't being used properly, the critical features were out of spec in reality whereas they "gaged ok". Around that time, the manufacturing-engineering group started to look into statistical process control, and found that they needed dimensional data. They started inspecting the critical features and
new gages, and found that the workpieces were dimensionally more accurate than the
new gages. I was indirectly involved up to that point, but became intertwined with the project when I indicated that the form, position, etc. that GD&T controlled needed to be verified also. We checked the workpieces and the
new gages over time and found some very interesting trends on the manufacturing equipment and on the gages. All gages were scrapped as a result.
I used to think that the traditional contact CMM was the be-all and end-all but I was
re-educated as to the reality of their limitations and errors. Each new software, hardware, and probe release reduces the error and improves the functionality of contact CMMs, but there are other systems that make some things a whole lot easier. Vision, laser, ultrasound and other technologies provide dimensional data quicker and with greater resolution and repeatability than contact systems, but are often limited bydepth-of-field to the outer surface.
No system is perfect. In the end, the method of part verification (gage, bench dimensional metrology, automated-system dimensional metrology) that you use should address what your goals are overall, not just at that one point in the manufacturing process.
Jim Sykes, P.Eng, GDTP-S
Profile Services
CAD-Documentation-GD&T-Product Development
![[idea]](/data/assets/smilies/idea.gif "[idea] [idea]")
![[wavey]](/data/assets/smilies/wavey.gif "[wavey] [wavey]")
