Design flaw on some power strip designs for AU/NZ UPSs & maybe others

[Anonymous user]

Hi all,

Came across a problem today on one of my old APC SOHO UPS units.  The unit I have, or rather "had", is model BF350-AZ.  This is the ES 350 setup for 230 VAC and comes with 3-pin AU/NZ plug and sockets.  The overall shape of the unit is like a large, grey plug strip, measuring about 3x4x20 inches and having 5 sockets on top.  This one had a battery with a make date of 04-2003, so it had a good run but finally died late last week.

I shut down the unit, pulled the power cord, pulled the battery out, got a replacement, dropped it in, plugged the power in, and turned the unit back on.  Of particular note, the unit was completely disconnected from any equipment, fortunately.  The "no load" condition should have been a fairly safe event, but not this time.  The unit powered up but had a fairly loud buzz and after a brief test sequence went immediately into overload.  I pulled the power cord and pulled the battery connector to force a power down.  I probably could have pressed the power button or something, but experience with UPS designs from various vendors indicates that not all UPSs power down as fast as needed in emergency conditions.  I rechecked the battery connections and all were correct.  So I tried again and the same thing happened.  I pondered what could be causing an overload condition on a UPS with nothing connected to it.  Having board level and  some component level diagnostic experience, and knowing that the unit is way out of warranty, I decided to open the "no user servicable parts" area of the unit.

What I found shocked and frightened me.  On the back side of the power outlets I could see where the bare phase/line bus had shorted to the bare neutral bus.  Technically, that's inaccurate, but it presents the idea.  Techically, what appears to have happened is slightly different.  The internal electrical contacts on the back side of the socket do not present rolled metal edges facing the front of the socket for the pins (tabs) on the plug to slide on to get against the internal contact.  The internal electrical contact present two cut metal edges that the plug pin has to slip between.

At some point in time a plug had been pushed into the socket and had been slightly mis-aligned enough to snag on the contact and bend it inward, away from the plug and face of the socket and back into contact with the other bus.I don't know why it only occured when I powered it back up, but I suspect that it was pushed back at some point in the past but held clear of the other bus by pressure from the plug.  Since the unit was powered down when I disconnected everything there was not problem.  When I powered it back up it came up with shorted output, somewhere between a load of infinite and "arc-weld".  The two bits of metal were not fused together, but obviously had arced and vaporized about half the width of the bus.  I also notice that 3 other outlets were bent in similar positions, with only 1 outlet that looked normal.  The other outlets don't appear to have shorted.

In addition, while examining the circuit board, I noticed burn or arc marks on the inside of the clear relay case that switches the battery backed outlets from line to inverter.  I unsoldered the relay and opened the relay case and saw that both of the NO ("normally open") contacts had fuzed ("welded"?) themselves together.  I was able to split the fuzed contacts with the blade of a box cutter and tested the relay for correct operation.  That is to say that I put a battery on the coil and the relay moved correctly, but I didn't check for resistance or continutity across the contacts, I just looked and saw that they snapped open and shut as expected.

I straightened out the outlet connectors and put the relay back in and hooked it all back up again.  When I test it this time, with the case open ("don't do this at home, boys and girls...") the unit didn't come up correctly.  The relay switched back and forth several times, but the unit stayed in "battery" mode and the FET got very warm and buzzed like a bumble bee on Jolt Cola.  So, I expect that the charge/maintain/line sense circuits are blown, at least in the FET and maybe the diodes and so forth.  Frankly, it's probably not worth the money to try to repair it.  I mean, it's 10 years old and can be replaced with a new one for 100 bucks or so.

But the point of this gigantic post, which I just signed up for tonight so I could post it, is that this probably could have been a way more exciting event with red flashing lights and gigantic sprinklers, and lots of people in long, heavy, brightly colored coats and hats talking on their radios.  I don't think I would have been able to convence my insurance company that I replaced the battery in my UPS but didn't actually cause it to burn my house down 20 minutes later.

Hopefully APC has changed (improved) their outlet design in the last 10 years.  If they haven't, it's sorely needed.  I've been a fan of APC for a lot of years, 20 or more I reckon, so I'm not trying to bash APC.  But it does concern me that a power products company would produce a product with this risk.  Maybe, giving the benefit of the doubt, there was a recall that I missed and this unit should have been taken out of service a long time ago.  I am bummed that I have to throw away this UPS and buy another one, and I just shelled out cash for a new battery that the dealer probably won't take back.

Pictures can be taken and uploaded if anybody wants to see what rambling on about.

Cheers folks, and thanks for reading.

Alan