Fix the Problem X: Learning from Interview Problems

Interviews are where companies and candidates learn about each other to see if there is a fit between a candidate’s background and personality, and the company’s needs and culture.

One of the things candidates, in particular technical candidates, get asked on interviews is to solve problems.  Often these are company problems that have been already solved, in an attempt to see how the candidate approaches problems.  Sometimes these are questions about extant issues.

Plusses, Minuses, and Danger Zones

On the positive side, this is a chance for candidates to show their abilities – ideally referencing other accomplishments as well to build a case for their skills.  On the negative side, such questions are often about problems that took multiple people hundreds, if not thousands, of man hours to identify and resolve.  Thus they can be very difficult to actually solve, in particular as a candidate is not as familiar with the details of the product and the processes by which it is made.

And if the issue is one currently extant, there is a danger that the line of inquiry is being used to obtain free consulting.  I’m not accusing all companies of doing this, but the possibility does exist.  So in the process of answering a candidate does need to be on guard.

A Two-Way Street

These problems, however, are an opportunity for candidates to learn as well.  I will discuss three situations where I was presented with problems in interviews in the hopes that these experiences will be of use to my readers as well.

Fiber Optic Bubbles

Back in 2001 after being in a massive RIF from Visteon, I interviewed with a company that made fiber optic cables.  The process was fascinating!  They would take raw, uncoated optical fibers off reels and pass them through a coating bath, lining seven fibers next to each other; this bath material would coat the fibers and adhere them together side by side into a ribbon.  Then, two of these seven-fiber ribbons would be lined up edge to edge, passed through another coating bath, and stuck together.  These ribbons would then be stacked up, one upon another, to form a rectangular bundle, and run through a machine that extruded a multi-layer protective sheath around the whole thing.

The issue they were having was bubbles in the coating around the fiber; bubbles distorted the coating and exerted a force on the fiber which, apparently, would allow light leakage out of the fiber – degrading the signal.  They handed me one of the 14-fiber ribbons to look at.

I have discussed looking for patterns in prior essays, for example here and here; I immediately noticed two things.

First, the bubbles were only present at the centerline bond between the two smaller optic-fiber ribbons.  This was an important clue as to where in the process from raw fiber to end cable this defect was being introduced.

Second, in looking at the bubbles, the spacing between them was remarkably consistent.  These bubbles, then, were not randomly formed, but something in the process was doing something that created a periodic “hiccup” creating these bubbles.  Using a ruler I got the distance between the bubbles, IIRC an inch-and-change, and in asking for the speed of the ribbon I was able to calculate the time between bubble formation.

My guess was that something was oscillating in the system, but probably not smoothly; there was likely “something” sticking slightly, with the sticking creating jerking motions as that “something” stuck and unstuck.

Since the serpentine path the ribbons took was visible, my approach would have been to use a high-speed camera and/or strobe light to identify whatever parts of the system were oscillating at that identified frequency as the place to focus attention.  (E.g., if the frequency was 10 Hz, light the strobe off with a frequency of 20 Hz.  The “something” – my a priori guess was one of the slack take-up reels – would appear to be shifting back and forth between two positions after some experimentation to synchronize the strobe with the “something’s” cycle.)

They were excited by my idea.  Unfortunately, I never did hear anything further about whether the idea contributed to the solution of the problem (I asked a couple of times over the next year).  I suspect my idea contributed to improving someone’s performance review.

Lesson: Strobe lights and/or high speed photography can be enormously useful if a frequency-dependence of a problem can be identified.

Cavity Numbering

Another interview was with a company whose supplier – in China – was molding their components and shipping them to the US for assembly.  There were eight cavities in each of two molds; these two components would then be assembled to form the product.  Some of the pieces, A and B, would not fit together.  Measurements indicated that there were dimensional issues causing a problem with some cavities of part A not fitting into part B.

Here was the problem: they didn’t know which cavity was which – translating to an inability to determine which cavities worked interchangeably and which ones didn’t.  They already were pressuring their supplier to take the molds offline for a day or so to put in some kind of cavity marking.

Lesson: If you have multiple cavities in an injection mold, number them to aid in troubleshooting when confronted with situations like this.

Engineering Fundamentals Always Apply

In one very thorny problem presented to me, a glass-encased thermocouple was placed into a blind hole machined into an aluminum piece and potted in place with an elastomeric material.  These units were subject to thermal cycling as part of the unit’s operation.  Failure analysis showed the thermocouple’s glass casing was cracking.

This is where engineering fundamentals comes in.  What happens when things get hot?  They expand; each material has its own expansion coefficient.  Resins – from which the potting material was made – typically have a higher expansion rate than metals, in this case aluminum.  With the resin trying to expand faster than the material around it, its attempt to grow was constrained by the hole walls – introducing a compressive strain from the difference in expansion rates.

Stress – force per unit area – is strain times Young’s Modulus.  The expanding resin, being constrained from growing outward, squeezed inward on the glass and fractured it.  The solution was to change to a softer resin with a lower Young’s Modulus.  Even though the thermal expansion strains were the same, because the resin was softer, the forces were lowered and the glass didn’t crack.

Lesson: Sometimes you really need to get back to basics.  In this case, mechanics of materials.

Interview Problems: An Opportunity

Not all interviews work out (alas).  But the presentation of a problem to solve can be an opportunity – not just to prove your mettle, but to take advantage of the experiences of others to improve your own problem-solving toolbox by being walked through a real-life case study.

© 2014, David Hunt, PE