Glenn, I am very impressed by your methodical approach to the problem. Many times someone will come to one with the culprit rather than with the problem. The fact that the robot could not speak for itself was probably the reason in this case. The QC manager should have just said the parts were "bent" and left it to someone like yourself to determine what happened.
I ran into something similar with a piece of networking software. It turned out that it was the application design that was the culprit. The customer had put a network analyzer on their system and tried to lead me in that direction. The only problem was that there were almost no parameters that could be adjusted in this particular software. Once they took my suggestions about how to use existing mechanisms to restructure the software, performance was greatly improved. I even gained a reputation as the one who had solved a big problem for another company (who was using the same software). Really, it was only a couple of hours of work.
I don't recall ever naming robots or other machines, although I have worked in shops where machines had been given names. And my father alays told me that machines are female, so no, she (the robot) didn't drink beer.
Great story, Glenn. Sounds like a die design problem. "Die spit" (molten metal shooting out of the die) shouldn't happen if the die is properly vented, and "dirt" shouldn't build up on the die surfaces if it has sufficient overflows. Diecasting is a complex process with a lot of variables, but it all starts with good tool design. (One question: shouldn't something building up on the die make the parts thinner, rather than thicker?)
Dave Palmer; The dirt build-up was on the flat mating surfaces, rather than inside the cavity which forms the part. Another problem with die-casting is die deformation from heat stresses. This die had deformed by a few thousandths of an inch, which was enough that there was not enough material on a critical face. The dirt on the mating surfaces spaced the two halves apart by a few thousandths of an inch, making the part a little thicker, which brought that face into specification.
We have a large Fanuc packing robot that we named Big Bird. We also have a work station that flashes a red andon light when it breaks down. It stops enough that the operators have nicknamed her Roxanne.
I once worked at a place that had a strict policy against naming robots. The idea was that if robots had names, people would attribute humanlike characteristics to them, and that this could result in people blaming a robot for bending parts (for example). If robots don't have names, it's easier to remember that they are machines that do what they're programmed to do. At least, that was the theory.
We have a pair of Fanuc palletizing robots on a gantry system, and the German company that installed them, labeled one Tom and the other Jerry. Apparently, somebody is a fan of cartoons.
Their names are actually labeled on the robots with large lettering, and the software on the HMI station refers to them by name. It is useful to identify them, since there are two robots next to each other on the same system.
This topic was an area of study for one young man's H.S. Science Fair entry, and proved interesting to the judges... just not quite interesting enough to go to ISEF. I watched a relatively low-end academic performer garner a TON of attention from the judging staff as they had NEVER considered what could happen when anthropomorphism creeps into society. Your post should inspire further study as the engineers in my workplace ALWAYS "name" their prototypes. One was named Daisy Duke and we had huge expectations....
Glen, not only is this a tale of good detective work, it is also about the only reasonable way the investigation could have been done. A close-up examination as the process was running would have been both difficult and dangerous. Even plastic injection molding, a process sort of similar to the metal high pressure casting process, has a few hazards to be aware of and avoid.
Doing all of the trimming by hand, rather gently, is the way to avoid use of the trim press, which would have been my first guess as the "bending" culprit. A trim press is able to cause damage if the parts are not placed in the dies exactly right.
Festo's BionicKangaroo combines pneumatic and electrical drive technology, plus very precise controls and condition monitoring. Like a real kangaroo, the BionicKangaroo robot harvests the kinetic energy of each takeoff and immediately uses it to power the next jump.
Design News and Digi-Key presents: Creating & Testing Your First RTOS Application Using MQX, a crash course that will look at defining a project, selecting a target processor, blocking code, defining tasks, completing code, and debugging.
Focus on Fundamentals consists of 45-minute on-line classes that cover a host of technologies. You learn without leaving the comfort of your desk. All classes are taught by subject-matter experts and all are archived. So if you can't attend live, attend at your convenience.