PSU Repair: A Case Study

Words Of Caution

When servicing electronics, do not presume you know anything before double-checking everything, and do not assume the manufacturer knew what it was doing either. Otherwise, you may get surprises. And surprises when you are poking around live surfaces at 100V or more with a significant energy source or reservoir attached are usually unpleasant, potentially scary, potentially costly, potentially painful and potentially worse. Lots of potential.

Another thing to keep in mind while working on circuits that carry significant voltages: even if the occasional zap might not cause any actual harm, over-reacting to the zap may cause worse injuries.

Can you guess what sort of mistake lead to this? The power supply was supposedly turned off when I picked it up to put elsewhere (while still plugged in), the chassis wall bowed in and shorted against the heat-sink. No, the mistake was not that the switch was actually on. I double-checked before picking it up and checked again after the fact; it was still off.

The mistake was assuming that the power supply's IEC socket had been wired properly. Somehow, Antec's manufacturer (Channelwell in this case) got the live and neutral wires backwards, which means that in the “off” position, the neutral line gets opened and everything on the primary side becomes live instead of neutral. That includes the primary-side heat sink, which is tied to the low-side input cap. With the wiring issue, live got shorted to ground through the rectifier bridge leg tied to neutral and heat sink. At least now I know my ground is good, otherwise it would have been a shocking experience. On the downside, this busted one diode inside the rectifier bridge, along with the fuse.

If some of you were wondering why there are separate neutral and ground wires in electrical installations when the two get bonded together at the breaker box anyway, the above would be one justification. Ground is practically always ground when it's present. Live and neutral get wired incorrectly far more frequently.

Hacking Forward

Since I am only using this supply on 120V, only one half of the bridge gets used. Thankfully, the only shorted diode from that mishap was on the normally-unused half of the bridge, so I worked around it by simply hacking off the affected AC-side leg. For the fuse, I went cheap and dirty with a thin jumper wire. This power supply is not going to be used unattended until I fix it, so I am not going to bother with being prim and proper until I have a confirmed fix I am satisfied with.

For safety reasons, I also took a few minutes to fix the incorrectly-wired IEC socket so that “off” really means no power anywhere beyond the switch. Inserting a thick zip-tie between the heat sinks and wall will prevent future shorts while the unit is open-frame.