The Too Expensive UPS Battery Project

[papaof2]

If you've followed my comments online, you'll know that I needed to replace the 12 volt, 4.3AH AGM battery in some small, discontinued UPS units.  The UPSs were closed out by newegg.com and officedepot.com several years ago for less than the price of the battery inside - and with free shipping.  These were designed for a maximum of 75 watts and I have several in place providing power for home network and cable TV equipment plus the security system.  I recently discovered during a power failure that one of the UPS units needed a new battery.  The manufacturer wants $50 plus tax and shipping; upsbatteries.com wants $27 plus tax.  For a battery with at best a 3-5 year life and often less in UPS service?  There had to be a better option and I found it at batteryhookup.com - they had new, 3.8AH LiFePO4 cells for under $3 each including tax and shipping.  3.8AH is just slightly smaller than 4.3AH and the LiFePO4 cells can be discharged 80% perhaps 2000 times where the AGM battery will only survive that level of discharge a few hundred times.  Add a BMS to each battery pack plus the plastic cell holders and the price is still under $20 - looks much better than $27 + tax & shipping.

I finished assembling the first battery pack last night and put it on a slow charge until I was happy that it was working correctly (I have a battery spot welder and the nickle strip from a previous project so no cost for this project).  When the pack reached 13.5 volts (3.37 volts/cell, not quite the rated 3.65 volts/cell charge limit but well into the charging area), I powered up the UPS with amp and volt meters on the battery pack and a watt hour meter on the load (15 watt incandescant bulb - remember it's designed to run small loads for a long time).  The UPS went into "Low Battery" alarm just over two hours later when the battery pack dropped to 10.7 volts (minimum safe discharge voltage would be 10.0 volts (2.5 volts/cell) so the UPS' low voltage check will prevent over-discharging the battery.  The pack, starting at an estimated 90% charge (that 13.5 volts) ran 2.1 hours with a 1.64 amp load for 3.44AH so it delivered 90.5% of the rated capacity.  I'm happy with that.

The battery pack is on charge in the UPS as of 8PM and I'll let it run until the charge current drops to 0.01 amp or whatever the charger's final charge rate is.  I'm guessing 14-16 hours for the highest possible charge the UPS can deliver so I'll check during the day tomorrow.  The cells can deliver more power (12 amps continuous, 28 amps for 30 seconds) than the BMS can handle (8 amps) but the UPS is only rated for 75 watts, and 12 volts * 8 amps is 96 watts so the BMS is adequate for the intended load if the UPS is even close to 80% efficient (typical efficiency).

I love it when an idea actually works out in the real world ;-)

[papaof2]

The original 4.3AH AGM (sealed lead-acid) battery would probably have been 80% discharged at the 10.7 volt low battery shutdown of the UPS so let's do a little math: 4.3 * 0.8 = 3.44AH. The new battery pack has the same useful capacity as the original battery but for fewer $$ and with a greater number of charge/discharge cycles. The manufacturer's discharge chart for the UPS has 15 watts for 138 minutes. The new battery pack lasted 126 minutes. I usually activate some type of backup power 30 minutes into an outage so this amout of backup is fine. The load this specific UPS will have is under 10 watts and the run time will be perhaps 20% longer than with a 15 watt load.

20 hours later the charger in the UPS has shut off and the battery pack is at 13.32 volts. This is 3.33 volts/cell which isn't a full charge for LiFePO4 but the cycle life of lithium chemistry cells increases if they are not pushed to their limits: charging to 75% capacity and discharging to 25% capacity can push the cycle life beyond 10,000 cycles.

If this battery pack works out over time, I have a couple of higher power BMS units (40 amps) and more LiFePO4 cells to build higher current battery packs for some larger UPS units. It would be nice to not be concerned if a UPS got left on during an outage until it shut down from "Low Battery" because the battery would not be damaged by that.

If I have any cells left when the solar lighting in the shed out back needs a new battery, a 12 volt LiFePO4 pack out there should work well and it will certainly fit in the existing "battery box" (an ammo can).

I might also use eight cells in a 4s2p configuration to replace the starting battery on the Honda walk-behind mower. The existing very samll and very short-lived battery is fused for 40 amps so they don't expect more than that for starting current - that's something to test with a clamp-on DC ammeter to know exactly how much power the starter needs. A 4s1p pack could deliver 28 amps for 30 seconds and a 4s2p pack could deliver 56 amps for 30 seconds so it's just a matter of how much power the starter needs and how many cells can be fitted in the existing battery compartment (or perhaps another ammo can). The LiFePO4 pack could be connected to a 20 watt solar panel and left on charge all year because the BMS would prevent overcharge. I have all those pieces and I'm very much in favor of things that "just keep working".

[9idrr]

Pretty soon you're gonna be sellin' power back to the co-op.

[papaof2]

Nah - they want too much equipment between them and me.

I'd be happy to be able to provide 30% of our daily power usage from solar - that would require a BIG solar array when we're having clouds/rain 3-4 days a week. Think minimum 2400 watts of solar panels ($1000 used) and a 12 volt, 1680AH LiFePO4 battery bank (about $3000) to replace what I have now. Plus another few hundred $$$ for wiring, circuit breakers, etc. plus mounts for those solar panels. That could power fridge, freezer and limited lighting year-round.

[9idrr]

I suppose there're grants, rebates, and other monetary reasons you could do all this but my best guess is that it's the thrill of the hunt and the idea of havin' power when those around you haven't that might be the drivin' factor.

[papaof2]

When power was out in December several years ago, I was out in the dark at 30F, shoveling 7" of snow to get the generator out of the shed. Then it wouldn't crank so I had to remove the air cleaner and squirt in starting fluid to get it going - doing all this by the light of my 3 AA head light. The gen ran fine once it started. That's the night I wasn't holding on to anything when I started down the ramp from the shed to get the gen and put it away - slipped on the snow, went down on my butt and back and into the the snow and slush. Then I got up and put the gen back in the shed - holding on to the doorframe that time. Back in the house for dry clothes, warm beverage and time in front of the fire with an ice pack on the shoulder which took much of the fall. Didn't break anything but I sure did move slow the next day.

That was all the incentive I needed to ensure I had "Wait until daylight" power in the future. I'd like to extend that time to multiple days but the guys at the Treasury Department complained when I did the Democrap thing of printing as much money as I needed. I can't see why - my colored paper looked just like their colored paper...

There are some rebates and there might still be some tax credits but those are just "getting your money back" so they may not reduce the initial outlay and most folks want their $$ up front.

Having dabbled with solar as long as I have, I take every solar generator, solar wall, solar <whatever> ad with a pound of salt because I know they can't deliver what their ads appear to offer. If four of the next seven days are clouds/rain/thunderstorms, it is NOT likely that there will be enough solar power from their system to be useful. When the 36 volts under load of the 200 watts of solar I have on one charge controller drops to TEN volts during an afternoon thunderstorm - below the voltage of the then not loaded solar battery bank - you know the solar system is NOT delivering ANY power.

[9idrr]

Yeah, there's always that fine print. Offer not good except durin' alternate blue moons that fall on fifth Tuesdays of months starting with X.

[papaof2]

Got 8 hours from that full charge with the normal load (cordless phone charger, security system base unit, talking caller ID). The UPS is now plugged in with no load; I'll let it charge 16 hours or so and then run the test again at that charge level (which will be something less than a full charge). The next test will be "real world" of the UPS charging the battery to whatever level it's set for and then doing an "until Low Battery" run for timing how long that charge lasts.

The notes I've been making will be useful when I build battery packs for other devices. If I get industrious, I might open the UPS case and see if there's a way to increase the charge level it delivers to the battery.

[papaof2]

After the next test completes, I should disassemble the UPS and see if there's a way to adjust the maximum charge voltage to my preferred level - I want to do whatever gets the longest life from the LiFePO4 cells.

You think I'm going to a lot of trouble for a $27 battery?  If things looked like 2018's "normal", I might agree with you.  Based on today's grocery shopping, with limited supplies of some items and empty places for others - lots of empty in the frozen foods section, such as no Fudgsicles, just empty shelves - replacing things which may not be available in the near future with something which I can build for less and which will last longer than the original makes good sense.  I can also use the 12 volt LiFePO4 battery packs for powering ham or CB radios for portable/mobile communicaions if needed.

[papaof2]

Twenty four hours later, the UPS was back supplying power to the usual load - it ran for 8 hours and 27 minutes. I'd read that LiFePO4 cells could be charged to near full capacity if left on charge long enough, even at less than the recommended charge voltage - that appears to be true. I certainly didn't expect to get more runtime than on the previous test but I won't complain ;-) It's also possible that the cells have a slightly higher capacity than the marked 3.8AH, as some manufacturers rate their cells at slightly less than their expected capacity so they never have returned cells which don't meet the advertised rating. Whatever the reason, I'm very pleased with the cells I got from batteryhookup.com. I'll let the batteries rest a few minutes and then put them back on charge for future use. Now I must decide which of the other UPS units get the better batteries - guess I should have bought that case of 100 cells when it was available...

It seems there's no need to adjust the UPS' charge voltage as the battery pack appears to have had a full charge. I love it when a plan comes together ;-)

Now if I could just afford enough of those 280AH LiFePO4 cells to make the new and much larger battery bank for the solar backup system so we'd always have 2+ days of silent backup...

[papaof2]

The LiFePO4 cells I used to build the less expensive UPS battery were part of a group of 48 cells that I ordered. All those cells were charged to 3.55 volts (typical value is 3.65 volts, maximum is 4.2 volts) as part of the testing when they arrived. The ones not used were put back in the box in July of 2021. When I checked those cells yesterday, they were at 3.32 to 3.35 volts, with more than half of them at 3.34 volts. The fully discharged voltage of the cells is 2.0 volts, so they had kept most of their charge for 9 months. I wired up another battery pack of 4 cells with a BMS and put it on charge (0.5 amp) before I went to bed last night. When I checked the battery pack today, it was fully charged.

Why am I building another pack? The lithium ion cells I put in the hand vacuum can no longer supply the startup current - remember those were originally cells to power a laptop, not a power tool. The LiFePO4 cells for the new pack are rated at 3.8AH, but they can deliver 12 amps continuous or 28 amps for 30 seconds. The cells have higher ratings than the BMS, which can handle 8 amps continuous or 20 amps peak. I think the LiFePO4 pack should easily start the hand vacuum and run it for a long time - the longest run time is using the powered brush to clean the carpet on the stairs to the upper floor and that will be the hand vac's "How well does it work?" test when the new battery pack is installed, maybe later today. If I don't get it done today, maybe Saturday or Sunday as I get the second steroid injection in my back mid-day Friday and I can expect to be mostly non-functional the rest of the day.

batteryhookup.com still has 16 packs of these LiFePO4 cells for $34.40 plus tax & shipping, so under $3/cell delivered.
batteryhookup.com/products/16-k2-26650-3800mah-lifepo4-cells
You will need holders that fit the cells or nickel strip and a cell welder to build a battery pack, plus a BMS for the volts and amps you need. The "battery spot welder" ($20-$25US delivered) and the nickel strips can be found on eBay, banggood.com and aliexpress.com Many of the battery spot welders need a 12 volt battery of adequate capacity to power them. Your battery packs will come together more neatly if you use the plastic battery spacerss - assuming you have space for a rectangular battery. If you're replacing the AGM battery in a UPS or other device, the equivalent LiFePO4 pack will be no bigger and might be much smaller. Battery holders and spacers for 26650 cells are avalable on eBay.

[papaof2]

It's been almost a year since the first post on this topic but I did find an affordable AC-operated process timer that reads out in hours, minutes and seconds to test how long a UPS runs at a given load.

Electric Analog Alarm Clock with White Case & Lighted Dial $16.99
www.ebay.com/itm/373808224462

It looks like the electric alarm clock that was in every house in the 60's/70's: white dial, black hands, red second hand, gold alarm set hand. The clock works just like that original clock with an AC motor and a neon bulb illuminating the dial - except that it's a quartz clock with a constantly moving second hand and the illumination is a blue LED (with a brightness control). However, the battery has been replaced with a wall wart and it works exactly as the original clock: plug it and the second hand moves. Unplug it and the second hand stops. The clock draws about 1/2 watt so it won't have an appreciable effect on the UPS' run time.

What's all this verbiage for?
So I can report that the new battery pack for the UPS actually lasts 8:21:27 with the UPS' usual load - what I had measured as 8:27 by a roundabout method originally but now have measured more accurately because I can get that time to the second ;-)

The UPS does need a few minute wait before plugging it back in if you run it to UPS' "low battery" shutdown of 10.7 volts - the LiFePO4 pack has to get back somewhere near 12 volts before the UPS recognizes it and starts charging it again. The LiFePO4 BMS has a low voltage shutdown of 10.2 volts so running the UPS to its 10.7 volt "low battery" shutdown repeatedly won't harm the LiFePO4 pack. That would have killed the original AGM battery in probably less than 100 cycles.

Why am I writing about this at 4:30AM? I took an Rx pain pill a couple hours ago and the pain hasn't gotten better :-( It occasionally happens that way, so I'd like to be able to test the pills in that bottle so see if half of them are placebos... I see the pain doc Tuesday for a couple of lumbar blocks to determine where would be the best place for an ablation procedure - using low level RF to deaden the nerve so it doesn't pass pain messages for 6 to maybe 12 months. Not looking forward to being stabbed in the back twice Tuesday but I am looking forward to the possible future treatments Tuesday's explorations might make possible. Otherwise, I'm looking at just getting higher doses of the Rx painkiller. I'd rather not be medicated when I have things I want to do around the house that require at least a stepladder, such as replacing the halogen bulb in a light on the back of the house and I'd like to be driving myself to get a haircut.

[papaof2]

UPS followup: We've had several short outages and the UPS has worked just fine. I've already tested it to shutdown (above) so I know it's good without further checks (the LiFePO4 cells I used in the UPS battery are good for 3000 cycles to 80% discharge - some cells are even better, such as the out-of-my-price-range 320AH cells that are rated 4000 cycles to 80% discharge).

3000 cycles is one cycle per day for more than 8 years. No more outages than we get, the UPS now has a "lifetime" battery in it ;-)

[feralferret]

My last employer manufactured and repaired automated test devices for checking the braking systems on railcars. They switched from AGM to LiFePO4 for powering these units. They discovered real fast that they need to change to a different charger for the batteries. The AGM charge controller works off of the voltage to regulate the charge rate. The LiFePO requires the charge controller to work off of the charge current for proper control of charging.

You may be able to get away with the original charging circuit if you don't push the cutoff voltage, but you may be risking problems if you try to up the cutoff voltage.

I use one of their castoff AGM batteries that I desulfated with my fancy charger for powering radio equipment that I work on to sell on fleabay. It also works well to power my radios during a power failure during severe weather.

[papaof2]

AGM batteries have had their place for a long time. The previous set on my small solar system was almost 9 years old when I replaced them (typical life of 3-5 years or 5-7 years depending on the vendor). I replaced those with AGM because I got a very good price on some used batteries that still tested as almost new. I do a 10% discharge test once or twice a month because in my research I came across a report that AGM's last longer if given an occasional discharge instead of always being at float voltage. Ask me in 9 years whether that has made a difference ;-)

I'd like to replace the 420AH of AGM batteries with close to 1000AH of LiFePO4 to have 2+ days of autonomy without sun or gasoline generator but that's not in this year's budget. Silent power when everyone else is in the dark is a good thing if all your lighting is either oil or battery.

[papaof2]

Yet another followup ;-)

When we had the most recent 8 hour outage (Jan 28-29), the second UPS with the LiFePO4 upgrade was in place for the old Dell laptop (from 2011-12) that monitors the solar system. Between the life of the LiFePO4 UPS battery and the life of the laptop's almost new battery, that laptop ran just fine for 8 hours ;-) Most of that time the laptop was in "screen off" mode and drawing minimal power but that's how I configured it - to last as long as possible when power is off. Of course, when the backup inverter is running, I can also plug the UPS in there and recharge its battery if needed. I should have used a 3-way tap to add the timer to that UPS and see how long it actually lasted.

I did find 920AH of LiFePO4 Grade A cells for about $2600 (with the new BMS units, wiring, fuses, circuit breakers, and so forth ;-) Now to find the $$$$ to get all those pieces. That size battery bank would provide about 45 hours of stand-alone winter power (central heat, fridge, freezer - on a timer to be on 2 hours out of 8, some LED lighting and charging phones and tablets). The freezer's manual says foods OK for 16 hours but I'm a pessimist about that so I'll run the freezer a little more often.
With 1100 watts of solar power - and full days of sun - the backup time would be extended past a week with just solar power. If there was no sun, about one gallon of gas a day should keep up with our usage and I have 15-20 gallons of treated gas in the equipment shed out back. If no sun, I would probably run the gen every other day - maybe while running a chainsaw on bits and pieces of downed limbs to provide some cover noise and/or some distraction? Or maybe just build that generator "quiet box" I have plans - and most of the parts - for: 3/4 plywood, insulation, cooling fan, remote temperature sensor(s).

I like the idea of more than a week on solar only ;-) I could put those panels on the roof of the equipment shed and run the power back to the house through some of the orange "fiber optic conduit" that I picked up free on Craig's List. That does presuppose that I can handle some type of trenching shovel for 30+ feet - or maybe just the center two tines in the little tiller. Or maybe use a sharpened pick and put some 3/4" irrigation pipe in a much smaller opening - no one notices unused lawn irrigation connections on a house...