Oh happy, happy me
Friday, about 4:30pm (at work) saw the final resolution of a problem bit of kit that has taken over 6 months to fix.
The machine would fail to start every Monday morning until it had been swithced off and on several time and generally bashed and sworn at by the operators. After that it would be fine for the entire week.
This is just about the worst kind of intermittent fault. It took us several month tracking it down to one control module, after eliminating all the usual relays, switches etc. Also, the machine is an early 1980s build, which was very much a 'transition' period and means it has just about every kind of electrical and electonic system in it
Once narrowed down to this module, all the other engineers gleefully washed their hands of it. "Transistors and chips mate - that means you." was their happy cry. To make matters worse, the production manager wanted the machine back in operation (such as it was) at least 3 days a week, so no pressure then!
The thing was so critical, that even the act of unbolting it from it's mountings would clear the fault. From experience I knew that just
had to be a hairline fracture somewhere, so I'd spend a day pooring over it with a very bright light and high power magnifier, which is a hell of a strain on these old eyes.
The main circuit board was actually in remarkable good condtion for its age, and having resoldered a number of possibly suspect joints. I'd take it back, and wait for the inevitable "Nope. Still the same"..
Finally I had the bright idea of creating a sort of protective plastic bag cocoon in the machine with the offending unit mounted as usual but without any of the casing. Monday came and I was waiting for it. The operators were told not to touch it at all, and I was the one who switched it on, just once. The fault was there and I could extremely carefully poke around the board. After a couple of hours of this (in an extremely cramped place under the machine) I finally spotted a dead internal power line. Bridging across it sparked the machine into life, and, of course cleared the fault.
It wasn't till Thursday I could take the board off the machine to try to find the exact cause, and then, even knowing
what the problem was didn't tell me where it was! By the end of friday I was getting deperate. Every part of that line that I could see was in perfect condition. I'm ashamed to admit it took one of the office girls to nudge me in the right direction. "Maybe it's a bit you can't see".
Removing a couple of large components reavealed it. Indeed a hairline fracture caused by years of tiny micro-movements of one of the components that had been glued to the board in manufacture, only the glue missed its target and hit this bit of trackwork. instead.
The thing with hairlines is that, left switched off, a film of oxide builds up breaking the connection, but the most minute movement rubs this off, and a sudden electical charge can also perfom a temporary 'micro-weld'.
