By Mike Seibel, Contributing Writer
Having run out of obvious things to check while troubleshooting a faulty test system, an engineer postulates a far-fetched theory and it turns out he’s right.
Many years ago I worked on automated test systems for RF integrated circuits. These were mixed-signal test systems, with DC, Digital, and RF measurements performed at a high rate of speed on packaged RFICs. The stimulus and measurement signals were delivered to the IC testing site via a 3 by 4 foot “test head” that contained the testing socket.
One of our customers was having an intermittent fallout due to a DC stimulus reporting current limiting during testing. (A DC Stimulus is just a programmable, direct-current, power supply; in this case many of these were used to test the IC under test.) Several of our engineers visited the customer site to evaluate this, and replaced wiring and DC stimulus hardware to attempt to fix this problem.
Despite our best efforts, the customer still had the same intermittent problem.
As the senior engineer, I was called in once this problem had festered into a full-fledged enigma and the customer was seriously miffed. Lucky me! I reviewed all the previous repair attempts and found nothing lacking in the process. Everything obvious that had a chance of being bad had already been replaced.
Based on the symptoms, I hypothesized that the one remaining fault could be some conductive contaminate lodged internal to the test head, which could conceivably be intermittently shorting the DC stimulus lines. I went so far as to point out to my co-workers that I expected to find a piece of small, conductive debris “right below here” pointing dramatically to a crack in the test head. In truth, it was a far-fetched and bold prediction since I had absolutely no other fallback theory.
When I arrived at the customer site, I was “greeted” by the seriously miffed engineering staff, who quickly escorted me over to the offending test system. Along with the help of one of their technicians, I methodically removed the huge cover of the test head to gain access to “the crack.”
The entire time I was frantically thinking to myself “If this is not the problem, I have no clue what else to try and these guys will lynch me.”
With the cover removed, the technician and I began our inspection. I told him to “look closely for any conductive debris - it might be very small.” After several tense seconds, which seemed like an eternity to me, he said with a straight face, “You mean like this allen wrench right here?”
Sure enough, a two-inch-long allen wrench had somehow fallen into the crack in the test head, lodging near some exposed connector terminals. Random vibrations would cause the wrench to move, such that it would sometimes short out the DC stimulus lines. But most of the time (like while we were troubleshooting), the wrench would lie in waiting , only to pounce once again after we left.
Fortunately, since it was the customer’s wrench, there was blame to share all around.
The customer was happy, I got out of there (in one piece, thankfully) and back to the airport. We redesigned the cover to eliminate the crack, and I brought my trophy back to share with my amazed co-workers.
That wrench was awarded a premium position at work on “my bookcase o’ interesting stuff,” between my patent plaques and my “thank you coach” little league trophies. Unfortunately it did not survive a career move to my present company. Most likely, it simply disappeared into another crack.
Contributing Writer Mike Seibel graduated from California Polytechnic State University, San Luis Obispo in the late 70s, and worked for over 20 years at HP/Agilent on RF and Microwave test equipment and systems. Eight years ago Mike joined Turin Networks, which is now Force 10 Networks, where he is the director of manufacturing test engineering. He lives in Santa Rosa, CA.