UPS battery replacement schedule

I just replaced the battery in my UPS, and notice that the old one was swollen and hard to remove. Searching this NG and elsewhere on the net, I see this means that the battery had reached the end of its life some time ago and should have been replaced sooner. Okay, replacing the battery when it reaches the end of its life cycle instead of waiting until it obviously fails is fine by me.

So how do I know when to replace a UPS battery? This particular UPS has no self test feature and no com port with which to communicate with the computer, and I don't want to toss perfectly working equipment into the landfill just because it lacks bells and whistles to make my life a little easier.

--
Warren Post
Santa Rosa de Copán, Honduras
 Click to see the full signature
Reply to
Warren Post
Loading thread data ...

A good test would be to pull the AC plug!

formatting link
"

Reply to
Jamie

I have several ways to test batteries. The easiest is to install one or two 100 watt incandescent lamps into the UPS output, unplug the UPS from AC power, and see how long it will run. I do this when the UPS is brand new, and again at erratic intervals. I also measure the charging current to the battery, which is sometimes higher than I would consider acceptable. These are also the ones that suffer from premature battery death from overcharging and/or overheating.

You can also test the batteries outside of the UPS by using a 12v automobile headlight or similar load. Again, compare operating time with a new battery.

I recently purchased a West Mountain Radio CBA-II battery tester:

Charge the battery and then run it through the tester. Compare with a new battery of the same type. The problem is that the basic unit does not work well with large batteries yielding odd results due to connector contact and cable losses. The voltage is measured at the tester end, not the battery.

I also plan to use the CBA-II to match UPS batteries. I tend to purchase a fair number of "used" lead acid UPS batteries. The savings is substantial but there are hazards. One is installing two mismatched batteries in series.

--
Jeff Liebermann     jeffl@cruzio.com
150 Felker St #D    http://www.LearnByDestroying.com
 Click to see the full signature
Reply to
Jeff Liebermann

Well, I'm seeking a way to identify when a battery needs to be replaced

*before* it fails the pull-the-plug test. By the time it reaches that point, the battery has badly overheated and swollen. Surely I need to have replaced the battery before that.
--
Warren Post
Santa Rosa de Copán, Honduras
 Click to see the full signature
Reply to
Warren Post

I have used an ESR meter to check the condition of lead-acid batteries, but haven't kept up with the practice. Was thinking an easy test would be to have test points mounted outside of the UPS that one could use to access each battery for an in-circuit ESR reading. Charting the results would probably tellone when to replace the battery(eis).

formatting link
(bottom of page)

Using the Bob Parker ESR (Blue - Anatek!) kit of course! Available from us and others...

John :-#)#

--
    (Please post followups or tech inquiries to the newsgroup)
  John\'s Jukes Ltd. 2343 Main St., Vancouver, BC, Canada V5T 3C9
 Click to see the full signature
Reply to
John Robertson

If nothing else, just use a two year schedule. A good UPS will constantly evaluate the batteries and advise you when they are near the end of their life. But that is only for the high end ones, not a garden variety 500 to 1000 watt version.

A life test (such as the one suggested by the other poster, two 100 watt lamps (for a 400-500 watt unit) will work. You can compute how long it should keep the lights on, and see what you get. Generally when batteries get to the failure point, they discharge very quickly (a couple of minutes max) instead of the expected 10 to 20 minutes a UPS should provide.

Reply to
PeterD

Warren Post wrote in news:4987c72e$0$21065$ snipped-for-privacy@reader.greatnowhere.com:

IMO,a weak battery will fail a pull-the-plug test(under load) well before the batteries swell or crack open.

It just will not have a long run time.

--
Jim Yanik
jyanik
 Click to see the full signature
Reply to
Jim Yanik

Check the data sheet? I just had to replace the battery in my home setup UPS and noticed that the datasheet for the replacement states "For main and standby power supplies. Expected trickle life: 3-5 years at 25C, Approx 5 years at 20C." (Panasonic LC-RA1212P)

Actual lifetime would depend on the quality of the trickle charge, how often and for how long it's called upon to supply power, as well as how hot it gets.

Absent a test circuit in the UPS, your best bet would be something like a semi-annual load test, plotting the time-to-dropout versus lifetime.

--
Rich Webb     Norfolk, VA
Reply to
Rich Webb

Ah, that's been my problem, then. I have never tested for longer than it takes my system to shut down, which is only 60 seconds or so. Now that I'm getting a new battery tomorrow, I'll measure the run time I get, and again periodically.

Is there some rough guide as to when to replace the battery, such as (for example) once run time has dropped to half?

For the benefit of lurkers reading this, you don't want to do the pull- the-plug test while writable media are mounted. (Not while running Windows, for example.) A Linux live CD (free for the download) would be a safe choice for this.

--
Warren Post
Santa Rosa de Copán, Honduras
 Click to see the full signature
Reply to
Warren Post

That's a better idea than my earlier suggestion of running the computer with a live CD. I'll go with your idea.

"LearnByDestroying.com" -- that's a great domain name!

--
Warren Post
Santa Rosa de Copán, Honduras
 Click to see the full signature
Reply to
Warren Post

I don't think you want to do the UPS battery life test with a PC. A PC does not present a consistent or even a known load. During the typical 2-3 life of the battery, you could easily change PC hardware, resulting in a different load. That would make your historical runtime numbers useless. Methinks it's best to use a known, controlled, and reproduceable load, such as a light blub. While a light bulb is not the best possible load, it is commonly available, fairly consistent from bulb to bulb, cheap, crash proof, and offers an obvious failure indication.

--
Jeff Liebermann     jeffl@cruzio.com
150 Felker St #D    http://www.LearnByDestroying.com
 Click to see the full signature
Reply to
Jeff Liebermann

But beware, if like me you have installed a pair of fat antiparallel diodes across the test terminals inside the meter (in an attempt to protect the meter against charged caps) then for goodness sake put a low ESR electrolytic in series with the test leads before you connect a lead acid battery, or there will be melted wire everywhere. I don't know what Bob's ESR meter thinks of DC if the diodes are not fitted - I have never tried it and I can't remember the schematic.

Chris

Reply to
Chris Jones

Warren Post wrote in news:498872f7$0$21065$ snipped-for-privacy@reader.greatnowhere.com:

half seems reasonable. you know you're on the downside of the battery's life at that point!

--
Jim Yanik
jyanik
 Click to see the full signature
Reply to
Jim Yanik

You, uh, weren't topping it off with a Viagra soulution, were you?

In all seriousness... You just wait until the battery won't take a charge, then replace it. This means you might have to be without the SPS for a few days.

Reply to
William Sommerwerck

Not necessarily. The battery might conceivably have been damaged by a problem with the SPS.

If you're worried about the same thing happening to the replacement, simply open up the unit once a month. Most SPSs let you open them without disconnecting anything. (The APCs do, anyway.)

Reply to
William Sommerwerck

As long as nothing is being written to the media, this isn't a problem. I can state as a fact that Windows 2000 will tolerate a cord yank without damaging the operating system. (This is not true of all OSs.)

Reply to
William Sommerwerck

Actually, filesystem corruption is a subject I can address with more authority than UPS batteries. I personally have seen and dealt with several instances of filesystem corruption on Windows 2000 (and using the more robust NTFS, not DOS) following a power failure when the box was sitting idle.

A few caveats are in order:

Your condition is important and bears emphasis: "As long as nothing is being written to the media". Good point. Recall, however, that all modern filesystems improve performance by caching data and writing to the media later during system idle times. This means that at any given moment a filesystem may have dirty data, i.e. data cached that needs to be written to the filesystem but hasn't been done yet.

To deal with (among other things) power failures that cause lost dirty data and corruption, a journaling filesystem should normally be used. Windows' default NTFS is journaling, and so it is indeed more robust than its predecessor DOS in dealing with power failures. But perfect it ain't, as I have repeatedly seen.

I haven't run any tests, but my unscientific observations lead me to consider NTFS to be better than DOS but less robust than any other journaled filesystem I am familiar with. So when I am setting up a Windows box, I'll use NTFS, but only because Microsoft doesn't offer anything better. If I'm setting up a Linux or other *nix box, I have a choice, and I usually choose something bulletproof like XFS or EXT3.

Back to load testing UPS batteries. Jeff Lieberman successfully convinced me elsewhere in this thread that incandescent lights make a far better load than any computer running any OS. If for some hypothetical reason I had to load test using a running box, it would be with no writable media mounted, and as far as I know (correct me if I am wrong) that is not possible with Windows. I will, however, cheerfully concede your point that with Windows on NTFS you can *usually* get away with it.

--
Warren Post
Santa Rosa de Copán, Honduras
Reply to
Warren Post

There's another point here... You should have the software configured so that the computer shuts down fairly quickly after power is lost (within two minutes or so). That will give you time to do it by hand if you're sitting at the computer, while not (likely) running out of battery power before the machine shuts itself down.

Reply to
William Sommerwerck

On 02/02/2009 05:34 PM, Warren Post sent:

Hello Warren:

In a previous life, I oversaw many systems with UPS protection. Some of the UPS systems helped with super mission critical applications. As is hinted in your post, your UPS(s) are not as fully featured as many are today. You might wish to consider slowly moving your current UPS system(s) to less critical applications while replacing them with higher tech models. Some manufacturers had been offering a trade-up program at one time.

Most manufacturers will offer a load chart to predict the rundown time of a new/quality battery under various load conditions. While maintaining some of my UPS systems, I obtained a short heavy-duty extension cord and I stripped off the outer insulation covering to reach the individual wires for ease of current measurements. I used various combinations of output adjustable space heaters and heavy-duty soldering irons to establish differing power testing loads. I found the predictive charts rather accurate.

Some UPS manufacturers do seem to overcharge their batteries as has been reported in other threads. This in combination with various repeated full discharge cycles will limit the overall battery life of anything. Where a UPS was lightly loaded and not driven to battery exhaustion more than a few times, I'd see batteries last for five years. Less favorable conditions gave us sometimes just three years. Worst case I saw was about two years battery life.

Where a situation warranted it, I've mounted muffin fans directly to UPS exhaust ports to improve air flow in rack mount configurations. Ugly but effective.

Your reported observation would lead me to believe that your UPS is overcharging its battery. I believe this is not easily dealt with and probably not correctable without modifications.

HTH

Pete

--
1PW  @?6A62?FEH9:DE=6o2@=]4@> [r4o7t]
Reply to
1PW

To prevent data corruption during a pull-the-AC-plug test, just interrupt the bootup process so that you're sitting at a pre-Windows screen. Or boot from a DOS floppy. All components will be consuming power (though read/writes) take a bit more power) and you can accurately gauge how much reserve time you've got. The problem with an incandescent bulb test is that it is not your computer. Yes, it is standardized, unlike the computer, but equating its results to computer backup time is difficult.

Warren Post wrote:

Reply to
Bennett Price

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.