Our industry is fraught with tales of quality audits from hell and other less than desirable places, usually the result of standards written by folks who know a little but expand it to encompass a lot. Even well thought out standards can lose their effectiveness in the hands of auditors who are systems people rather than technical specialists. Conflicts arise when the systems people begin pontificating on technical matters they may not fully understand but despite this lack of knowledge, the deck is stacked in their favor in such situations.
The quality auditor can cite rulebook and law—some or all of which—may be irrelevant to the issue at hand but their advantage is that the subject of their audit has to prove the audit findings are wrong to maintain their quality registration. I use the term ‘law’ loosely since, in criminal matters, the reverse is the case—the accuser has to prove his or her case.
If the auditor is accusing you of doing something incorrectly but you know you’re not, many follow criminal law practices and take a plea bargain approach to make it all go away. It’s often a lot simpler and faster way to put it behind you but it can come back to haunt you when you get a different auditor that knows what they’re doing and you have to undo the ‘fix’ that got the last one off your back.
What brings all this to mind is a discussion one of my technical reps had with a customer who was in such a situation and asked for help sorting it out. The ‘finding’ cited by the auditor was that our customer was using an inadequate instrument to measure a metric screw thread. The device involved was a three-roll comparator and a Series 2 digital indicator with a resolution of .001mm. In the days before digital indicators became the norm, such instruments used mechanical indicators having .01mm resolution which made sense considering the mechanics involved in the device.
So, when my rep began relating the story to me I figured that if anything was amiss in this setup it would be due to the indicator being too precise for the application. But that was not the case and when it was explained more fully, I uttered the title of this column: “You’re kidding me.” The auditor determined that it was not precise enough!
By his reckoning, since the thread spec was noted to three decimal places, the indicator resolution on the comparator had to be one tenth of that i.e., .000lmm for a product tolerance of .19mm!
Our customer decided to take the easy way out and change the indicator. He began checking instrument catalogs and realized that they don’t make them with that level of resolution. He would have to go to an LVDT type probe with electronic display at considerably higher cost. He called my rep for help which was a good move on his part because if he had purchased a higher resolution display his problems would be increased—he would have digits dancing all over the place while repeatability went off the charts.
My rep suggested a different approach: round up the product dimensions to two decimal places which would mean the product tolerance would be the same. It would also mean a miniscule shift of the manufacturing tolerance but the auditor’s criteria would be met. There has been peace in the valley since this was put forward.
Several factors caused this situation, most of which I have ranted about in the past. First of all the design computer produced numbers to a level that was impractical—a symptom of the Digital Disease. The auditor succumbed to the same condition with his suggested way to fix what he considered was a problem. He misapplied the rule of 10:1 in his rationale and didn’t think it through. The comparator being used is set using a threaded master which, following the auditor’s rule, would have to be made to near gage block like tolerances. I think I’ve caused enough trouble, so I’ll leave measurement uncertainty out of this discussion.
I am not aware of any gage maker being able to produce a master to that level of precision. In fact, the master already in use by our customer was as precise as you can get on this planet as far as I know.