Tolerances, Ridiculous Tolerances and Getting the Job Done


We serve a terrific east coast manufacturer of OEM transducers and custom strain gauges and load cells. We provide them with machined parts, printed circuit boards, wiring harnesses, sheet metal enclosures and other specialized hardware.

Many of the machined parts display unusual geometries, require highly specialized surface preparation, electroplating and coatings and call out exotic alloys. As our customer is a job shop, their orders tend to arrive without any notice, are varied and range from a handful to no more then 1500 pieces.

We recently quoted a familiar looking part that we call “hydrants”. Cylindrical base, a slim blade or “neck” and a smaller circular “head and shoulders. It’s one of a family of related transducer bodies machined out of aluminum and steel. This particular part specified a +/-0.0005 inch parallelism tolerance on the neck.

We took exception to the tolerance, politely pointing out that all of the other related parts specified only a +/-0.005 inch (WITP = “walk in the park”) tolerance. We asked both why the super tight tolerance was needed as well if we could have some relief. Without relief, this was going to be a “no quote”.

The initial response was a simple “no, we need it as it’s specified on the drawing”. So, we pushed back and asked our customer again. They asked their customer. And finally that customer asked the end user. After more then a month of back and forth, we got:

“In order to avoid stopping the manufacturing process of the transducer bodies and open POs, please ask your machinist to manufacture the next batch as usual and collect data (at least thirty measurements) on the parallelism. This exercise will give us a better idea of your capability to meet
+/-0.0005”. Based on the results we both can make the call on revising the parallelism spec.”

Now, please allow me to offer this translation of their comments:
“Tom signed off on the drawing in 2013, but he doesn’t work here anymore. The part sure looks like all of the others. It was probably just a typo and the tight tolerance isn’t really necessary. Regardless, we need a paper trail and this exercise to cover our backsides.”
One other point worth mentioning: due to the part geometry, almost no metrology lab can measure parallelism to that level of acuity.
So, we manufactured the parts, collected the data, yielded a parallelism of +/-0.0007” and the customer declared victory. And the parts worked like all of the others and we all got the job done.
The moral to this story is that while some tolerances are “must have” many are “nice to have”. If your vendor pushes back, please try to meet in the middle with a “kinda must have” solution.


Jack Daniels

Leave a Reply

Your email address will not be published. Required fields are marked *