Hi William,

(and "All" -- I'll try to address all of the posts I have
seen here instead of with individual replies)

William R. Walsh wrote:

I rarely have power outages here, and my most recent UPS battery
change involved a unit that did nothing more than "buffer" a phone
system until a generator would come on. Its battery lasted 10 years
(1998-2008). The UPS in question is an APC Back-UPS 600.

IME, when the batteries are toast, you won't get *any* up-time
from the UPS. Power goes... and so does the UPS! :-/

I attribute part of that long lifetime to the fact that the battery
could have degraded to where it had almost no runtime left--but that
would have been fine as the generator started and stabilized within a
few seconds.

Can anyone with firsthand knowledge shed some light on
this?

I can't say that I have firsthand knowledge. (Meaning: I've never
designed or built a UPS circuit from scratch...) However, the trouble
is likely to come from the float charging. I'm of the impression that

Agreed.

at least some UPS units have an awfully "hot" float charge and it
probably leads to the battery being overcharged. I've seen some Tripp-
Lite UPS units that would slowly drive the water out of their
batteries if left float charging all the time.

I've seen lots of UPS's discarded because they have killed
their battery packs. Bulging battery cases, cracks, etc.

I also think there are some that don't put enough of a float charge on
a battery to truly keep it up. I have an older APC Smart-UPS 1000 that
demonstrates this behavior--if the power hasn't been out in a while,
the battery will drop more quickly than it does when the unit has had
only a short time (a few days) between the completion of a battery
charge after use and a subsequent power outage.

You also have to consider that some UPS designs depend upon their
battery and inverter to deal with *every* little power line anomaly.
This too will shorten the battery's useful life. This seems to be true
of the inexpensive APC "plugstrip" UPS units that many people have.

Some UPS's actually power the load continuously and are
recharging the battery from the AC line (like telco's operate).
I would think this would be more stressful on the switching
power supply than the battery (as, essentially, the same current
is flowing into the battery as out)

Better designs have methods by which to stabilize, boost or trim the
power coming out of them without having to use the battery and
inverter.

I (currently) have:
1x Back-UPS Pro 500
2x Back-UPS ES 550
2x Back-UPS XS 1000
1x Back-UPS RS 1500
1x MX 3000
(all APCC)

I discarded a Belkin unit that appeared to be a repackaged
Pro 500 (actually had the nice feature of using a removable
power cord!)

I've also previously discarded a Back UPS 600 (?) as it used
a different physical size battery that was hard to come by
(since it didn't offer any feature that I didn't have on other
UPS's -- and, also had two pigtail power outlets on the
rear which always seemed like a kludge).

I discarded a large HP "UPS 1000" which was a decent box (I
think Astec internals) but totally unsupported. IIRC, it
ran off 48V and had another 48V external battery pack.

I managed to blow up another larger UPS (3KVA... the size of
a dishwasher) by forgetting that power was still available
from the battery pack even though the AC mains were
disconnected... shorting a 120V battery into damn near anything
destroys the "anything"! : )

All of my UPS's are connected to PC's or servers. So, all feed
reactive loads. Most are oversized for the loads they power.
Occasionally, I will run an extension cord from one of them to
power a CF light in another room if we have an outage at night.
(a "100 equivalent watt" CF runs for a LONG TIME on a small UPS!)

All of the APCC devices have "replace battery" indicators.
All of them do a loaded test on power up (for about 2-3 seconds).
I am not sure if they periodically run other tests as it is
hard to hear the characteristic "buzz" that accompanies the test
when a server is running (fan noise). I think all are tied in
to their respective computer/server (typically USB or EIA232)
though that should just help the machine shut down before power
is pulled.

I think all of the APCC designs use low voltage primaries. E.g.,
12 or 24V. I don't know if the battery abuse is consequential
to this (perhaps the charging circuits on the UPS's that I've
had that utilized 48V or 120V primaries were better designed
of necessity?)

I've taken to powering the UPS's off when their loads are not
in use. I.e., they can (over?)charge their batteries only while
I am using their loads. Thereafter, they act like power strips
and let me shut down all of the loads. This is intended to
see if the problem I -- and associates -- have been seeing is
related to an aggressive float. (it also has the advantage of
silencing the silly alarms that signal in the event of a power
outage -- if the UPS is on but its loads are NOT, then I really
don't want to be bothered by a chorus of half a dozen little
"chirpers"!)

I don't think it prudent to use regular lead acid batteries
("car batteries") in these applications unless the batteries
themselves could be located in a ventilated area -- I'd be
wary of outgassing. I had thought of replacing the gelled
electrolyte batteries in the UPS that I blew up with car
batteries (this would have required *10* such batteries)
and locating the UPS in the garage with the batteries on
the outside of the building for this very reason. Then,
starving the electrolyte in those batteries to improve
longevity.

In the short term, I see no other remedy for "fixing" the
battery mongers... :