Never Assume Anything

This topic reminds me of an old adage I have heard from many teachers. Perhaps you have also told your kids or friends this one. It is especially true in engineering. Those who disregard this warning do so at their own peril, or worse, at someone else’s peril!

Once during a period I worked in industry, I was asked to look at an assembly process which was another engineer’s program, because the product yield was poor and nobody recognized a cause for this. The performance of these products varied all over the place, from one unit to the next, and those units which actually met the customer’s specs were beginning to look like accidents. First, other persons suspected that the assemblers were making mistakes.

It did not appear to me that the assemblers were doing anything odd or contrary to what had been provided to them as instructions. I checked their method and checked the tooling they were using. We made a few improvements to the assembly tooling. I tested a series of product specimens in the field, employing exactly the same methods used by the customer. The customer was right; many samples from the lot were performing way below specs. I drank quite a few cups of coffee and munched on lots of cookies while contemplating the strange, erratic performance of these devices. The application for these products was non-trivial, indeed it could be quite serious for the user; but, I’m not able to elaborate on that.

The program personnel assumed that all of the component parts for these products were acceptable and met specs, as they had been procured using the same normally high standards of our in-house procurement department. Our procurement personnel were good and had plenty of experience. The conclusion I came to was, well, that was a weak link in the chain; there must be a problem with some component(s). It had to be quite erratic to explain the wild scatter in the performance results we were observing. It also had to be a problem which was “invisible” to our “incoming – inspection” department where certain characteristics of these parts were inspected before they were released into stock for assembly. It also had to be invisible to our assemblers. I visually looked at all of the parts which were being assembled into the final product.

There was a very small rectangular flat mirror which was used as a beam folding mirror to fold the light beam from an objective to an imaging detector (let’s call it a CCD). That step shortened the physical length of the product to meet specs for “form and fit.” A light went off in my head. I took a small sample of these mirrors out of stock, a step which was met with some objections. These were taken to our optics figuring and polishing department where our chief optician (a genuine expert in his field) tested them on an interferometer to determine their flatness. You guessed it! They were anything but flat, in complete non-conformance with the specs with which they were procured. They were in fact, “potato chip” mirrors; so deviating from flats they were hard to test on that interferometer. They also varied from one specimen to the next.

Here was our problem. Of course a person could not see that visually! The incoming inspection personnel tested them for scratch and dig specs and for overall dimensions; nobody tested them for surface flatness! As used in our product not only did they add power to the light beam, they deformed the wavefront from the objective in random ways depending upon how each mirror was potato chip shaped. The results were random image deformations from unit to unit. The common procurement process for these was to look for a low bidder, and it is likely they found one! Evidently, that supplier paid no attention to the flatness specification.

This is only one example of what happens in industry and elsewhere when assumptions are made. The reader may be familiar with this common story: One person designs a product. Another person procures parts for the product. Another person (in Q.C.) examines the parts, maybe, . . . but misses a detail they cannot measure. Another person assembles those parts but has no knowledge of the performance of the individual parts. Another person tests the final product, maybe, . . . and may or may not find it to be out of spec. The manufacturer, at their corporate level, may or may not care!

At no time during this sequence is there any engineer who is assigned the role of “mother hen” for the entire project, to watch over all processes and to be looking for “assumptions.” No, not the project managers. Does that individual cost the company too much? As a result, sometimes things “break” or “crash” or “sink” and result in preventable disasters.

Any project that is “worth its salt” should have one or more mother hen personnel, probably experienced engineers, watching over it. Their number will depend upon its size and complexity. Everybody should be communicating. They should be people who based upon their work experience, will not assume that everything is going to work as planned, simply because that would be nice!

Do good work!

End