Postscript to UPS issues -- MIDRANGE-L

The new packs finally arrived yesterday (from an outfit called Planet
Battery, who has much better prices than APC or Batteries.com does on
these things, and [unlike some discount houses] sells the packs assembled
and ready to install). By that time, I'd already designed and built three
kinds of monitoring widgets.

The first kind, for the two boxes that are actually monitoring their UPSs
(because they're up 24/7) inserts between the monitoring cable and the
computer, with a second cable going to a "wall wart" plugged into an
UNPROTECTED outlet, and has a relay that's continuously powered by the
wall wart, which blocks the signal on pin 7 unless there's an actual power
failure. This prevents failed self-tests from causing false alarms that do
"disaster" powerdowns when there's no power failure.

The second kind is much simpler, with no "input" socket, to allow our V4R2
development box to respond to power failures. It has pins 5 and 8 wired
together (simply telling the compuer that a UPS is present; I don't know
if this actually makes any difference), and pins 5 and 9 connected to a
relay (powered by the same "wall wart" as the first kind) that ties them
together if the power fails.

The third kind is exactly like the second kind, except that it has a DB15
instead of a DB9, wired for our V2R3 development box (whose internal
back-up battery died around 5 years ago, and was never replaced), with
pins 11 and 2 tied together to indicate the presence of a UPS, and pins 11
and 9 connected to the relay, to warn of power failures. Since these two
boxes aren't on 24/7, are no longer used for production, and can do normal
shutdowns in less than two minutes, there's no need for them to know about
low battery conditions, and it saves us the cost of installing extra
monitoring ports on our UPSs. All four widgets are powered by a common
"wall wart"; in all likelyhood, there's enough capacity on it to handle at
least another half-dozen relays.

If anybody's interested (and can't duplicate my widgets just from my
descriptions), I can supply schematics.

In running tests, I also discovered that the self-test doesn't
consistently raise the "on battery" signal. It only seems to do so when
there's a possibility of failing the test, and raising both the "on
battery" and "low battery" signals at the same time. Go figure.


Oh, and since nobody around here could remember which UPS got new packs
when (even though we now have three pairs of packs waiting for me to see
if anything's still good enough to keep around as a spare), this time I
took the time to date-stamp the new packs just before I put them in. Ain't
20-20 hindsight wonderful?

--
JHHL