[Baylisa] UPS battery modes of failure

Wed Jun 27 10:27:18 PDT 2012

About a dozen years ago, I had an opportunity to obtain a UPS that
had a number of attributes that I found rather attractive at the
time; among them:

* It was rated for well above the maximum load I would put on it.

* It didn't need  a special power circuit pulled -- it uses a regular
  NEMA 5-15P <=> IEC 60320-1 C13 power cord.  (Under my load, it draws
  < 3A @120VAC.)

* It is of the "online" (or "double conversion") type.  (See
  <http://en.wikipedia.org/wiki/Uninterruptible_power_supply#Technologies>.)

* It can accept an arbitrary number of daisy-chained external battery
  cabinets, each of which can hold up to 2 sets of batteries.  Each set
  is quantity 5 12V 5Ah Sealed Lead-Acid storage batteries.  The UPS
  cabinet itself holds (up to) one such set, as well as the UPS "guts."

I originally got the UPS; a few months later, I found one of the
external battery cabinets at Halted.  Within a year or so, it became
evident from physical inspection that I needed to replace the 15
batteries -- many were showing signs of stress (bulges where the plastic
case should be flat; some cracks in the cases), so I did.

A few years later, someone on craiglist was offering a couple more of
the external cabinets for $80 (IIRC), so I picked those up.  Naturally,
the batteries in them were beyond hope, and by that time, I figured I
should replace the 15 that I had before, so that was another 140 lbs. of
batteries (quantity 35 at 4 lbs. each).

About a year or so ago, the UPS started -- occasionally -- whining;
in particular, whining about the batteries.  A visual inspection
showed nothing unusual; we hadn't had any outages of significant
length.  And the battery voltages each registered >12VDC when I had
disconnected one lead.  And since these are "sealed" batteries,
using a hydrometer to test the specific gravity of the electrolyte
isn't feasible.  (Had a battery registered ~10VDC, that would
indicate a possibility of one cell being shorted or otherwise damaged
-- a 12V lead-acid battery has 6 cells, each of which has a potential
of around 2.1V.  A mode of failure that can produce this is the
formation of "dendites" in the battery.)

Through a bit of hacking (trial-and-error), I found that hitting
the "silence" button alone only stopped the whining briefly (perhaps
as long as a couple of minutes), but hitting it, then unplugging
the UPS, then plugging it back in seemed to stop the whining for
some time (usually, several days).

While this behavior was obnoxious, it was evident that the UPS was
actually working (as the load remained powered when I unplugged the UPS)
-- though I wasn't especially keen to determine the length of time that
condition would prevail empirically.

Now, the UPS alone should keep my load up & running for a bit over 30
minutes (with new batteries).  Each full (2 sets) external cabinet
(again, with new batteries) should add a little over 80 minutes of
runtime.

I was thus rather dismayed to find that when we had a service change
(replacing the main breaker panel for the house) that the load was
maintained for less than 15 minutes: this was not the desired behavior.

Weekend before last, I finally set aside some time to investigate.  I
figured I'd first unplug the UPS, then plug in a power strip where the
UPS had been and plug the UPS into the power strip -- load should stay
up for the few minutes (at most) that would take... yeah.  Well, it
didn't work out that way: less than 15 seconds after I unplugged the
UPS, it flatlined (and I was annoyed).

So I brought everything back up with no power protection (scary!) and
started poking around in the UPS -- vacuuming out the dust, checking
connections, connectors, and voltages.... and found that it wouldn't
even keep itself up & running (no external load) more than about 20
seconds with the external cabinets attached, and less than 5 seconds
without them.

I had already contacted the manufacturer some years ago; they weren't
especially interested in further support, as the model had been obsolete
for a few years by then -- and it's been my experience that the
reversibility of this phenomenon is sufficiently rare as to not be worth
pursuing.

On a whim, I tried swapping batteries between the UPS cabinet an one of
the external cabinets.  After that, the UPS was able to keep itself up &
running for over 2 minutes -- not a long time, but demonstrably longer
than 5 seconds!

This led me to suspect that:
* The (basic) cause of the failure I was seeing was likely an issue with
  the batteries, rather than the UPS itself.

* The UPS was quite correct to whine about the batteries (though it
  might have been nice to have more of a clue -- e.g., which battery (or
  batteries), and what the nature of the issue is/was.

After some further research I suspect that -- since each of the
batteries still shows >12VDC -- the likely issue is something called
"sulfation".

It seems that -- especially in the absence of discharge cycles --
lead-acid batteries develop coatings of lead sulfate (which is an
electrical insulator) on the plates; as this happens, the battery
effectively loses capacity (as that is directly proportional to the
amount of exposed surface area in each cell).

Around 2000, a fellow named Alastair Couper published an article
(<http://homepower.com/article/?file=HP77_pg84_Couper>) that describes
the construction and use of a circuit to "desulfate" the plates of
lead-acid batteries.  (Note: there is an erratum for that orignal
article, reading: "In my battery desulfator article on page 84 of HP77,
the value for C2 should have been 0.0022 µF, not 0.022 µF. My mistake.")

Unfortunately, it appears that Sealed Lead-Acid batteries are reported
(<http://leadacidbatterydesulfation.yuku.com/>) to be poor candidates
for this "pulse conditioning."

I'm thinking of buying a commercial device that purports to use this
technology to see if I can reverse the "sulfation" somewhat.
Unfortunately, just as the layer built up over time, removing it also
takes time (as overheating one of these batteries tends to make caustic
messes).  And I have a lot of batteries to treat.  (Further, there can
apparently be some "interesting" effects with respect to distortion of
the pulse if one isn't fairly careful about the connections, and I
suspect that trying to "gang-condition" 5 or 10 batteries arranged in
parallel might be prone to failure, and may well take just as long as
doing them each, one at a time.)

But it seems to me that *if* this works, it would make sense to
incorporate the approach in the UPS charging circuit in the first place.
Given the time I bought the UPS, it's certain that it was designed well
before Mr Couper published his article.  Though I don't recall seeing
anything in recent UPS specifications or features that appears to
indicate that the approach is actually being used.

And I need to figure ouot some sort of evasive action I can take to try
to prevent a recurrence of the problem -- perhaps keep a battery cabinet
out of service most of the time, letting it self-discharge, and rotating
which cabinet is out of service...?  (In my case, the external battery
cabinets are hot-pluggable.)

Peace,
david
-- 
David H. Wolfskill				david at catwhisker.org
Depriving a girl or boy of an opportunity for education is evil.

See http://www.catwhisker.org/~david/publickey.gpg for my public key.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 196 bytes
Desc: not available
URL: <http://www.baylisa.org/pipermail/baylisa/attachments/20120627/f198d55d/attachment.bin>