That time when a component exploded... our stories

[D.8080]

[D.8080]

[Wasmachineman_NL]

[D.8080]

[Meterman]

1: At a prior job, we had one fixture that was used for final testing for one of our customers' HVAC temperature / velocity probes, and which was originally controlled by an old IBM PC (the actual brand - not a clone) that also had a 10MB hard drive. One day, I see a cheap clone tower sitting where the old IBM desktop used to be... and learned that one of the other techs was running a few assemblies through the fixture when the HD had a head crash. The guy said it was so loud he was momentarily convinced something metal had simply vibrated off one of the fixture's shelves and landed on the floor. Nope. He then realized the fixture had stopped working, and finally realized it was the desktop, and some checking proved the drive was done. The customer managed to scrape together enough used parts to make a functional tower to send out as a replacement, and gee, it sure looked funny sitting there half-finished (none of the spare bays had covers). Surprisingly cheap for who they were.

2: At my current employer, I work as an electronics repair tech, and among the products I repaired some years back was a motor speed controller that incidentally went into some equipment that was used for manufacturing ICs (if I understand right, the wafers were loaded into this unit for various chemical rinses as part of the overall process from a blank wafer to just before the wafers are split into individual dies - and yes, this customer's customers included a number of very familiar brands). Unfortunately, the original design of this motor controller used a trio of wimpy MOSFETs for the ramp-up / slowdown of the turntable, and therefore prone to blowing up in spectacular ways (I even have one where the MOSFET burned a hole through the back of the TO-220 package - the METAL part). So, I do a full tear-down and rebuild (or so I think), giving the unit a new pair of driver MOSFETs, 50-watt brake resistor, new fuse, etc. I then set up to do a 'dummy load' test with a light bulb instead of the actual motor (first test before actually bolting it into a process controller). Two seconds after throwing the tester's power switch, a 2' tall flame erupts from the brake resistor! I shut off the switch just like that and unplugged everything, and once the smoke cleared and my heartbeat returned to normal, I found I missed one small diode between two large caps. Of COURSE all the parts I'd just replaced were bad again, and of COURSE I'd just used the last brake resistor in stock at the moment. In due time, parts were sourced to do a second rebuild, but for a year I kept handing off the 'light test' part to a fellow tech until I regained the confidence to do them myself. (Thankfully, supplies of that MOSFET finally dried up, forcing the customer to redesign the unit with far more robust IGBTs).

[D.8080]

[Meterman]

D.8080: Oh yeah, it was!

This was also when I was still pretty much a newbie tech, so that's why I still remember it in such detail. In the years since, I've seen (or heard) some of my coworkers having to deal with a variety of motor controllers that simply decided to 'let go' during test. Most of the time, all they got was a flash and a sharp 'bang' (and a few where one of the prescription bottle-sized capacitors accidentally got installed backwards, resulting in a cloud of smoke that left the area smelling like rotting fish for a few hours).