开云体育

Tantalum Capacitors don't always fail right away when installed backwards...

I too experienced the exploding Capacitor effect early in my career. I used to repair boards in a Fan-out all day long for a few years. The boards that were failed in system test, which they just pulled, tagged and sent to the Fan-out area for debug. The Fan-out was just a system where all of the boards were fanned-out in front on someone who troubleshot the boards with schematics, Design Verification Routines, a DMM and an oscilloscope. We would debug to the component, have it reworked until it was working. Often a pair of dikes was used to cut IC pins on older RTL/DTL Wired-Or'd Logic busses and circuitry. Ah yes, loved those 16-bit busses with 10 agents. Anyway...

About 20 feet away was bed-of-nail testers for first time power-up and test of the boards. It was common to be disturbed (or awakened for some - grin), by a loud BANG from a reversed Tantalum Capacitor blowing up. In those days they were the Axial Package Sulfuric Acid Tantalum Capacitors. The operators all wore safety glasses and smocks to prevent injury, but they did spray the acid on the board which caused latent problems too. They were supposed to inspect them before test, but you know how some people do their jobs...

The WORST occasion though was with a Yellow T495 SMT Tantalum Capacitor on a Server card in development. It was later as an Engineer with a product in development in a lab. A new Server CPU card in development had some Tantalums on it and one was reversed. It had been in test for a couple of weeks (yes weeks), and was left overnight in Burn-in. During the night, the air conditioning was reduced to save money across all of the buildings. This elevated the temperature in the labs too - bonehead management decision.

One night, one Tantalum that was installed reversed finally failed. It was found the next day. It had burned a hole in the board 2 inches in diameter. It was funny to look at it where all of the FR4 was burned away leaving several copper layers and plane portions visible through the multilayer board. The High Current Server Power Supply had not protected the system as this had appeared as a high resistance short until the adjacent Planes finally shorted together killing the Power Supply Fuse. I guess there wasn't enough smoke to activate the Fire Control System - good thing we had enough trouble just trying to get the 5V 60MHz Frying-pan Pentium chips anyway. The lab was full of systems in development.

This is also why you specify 94V-0 flammability rating from all of your PCB vendors and 94V-2 flammability for all Assemblies.


Ross