DIY UPS capacity upgrade, success and failure

Mattastrophe

New Member
Feb 2, 2020
5
1
3
I’d like to start off by saying I’ve used STH as a reference for a long time, but this will be my first post. I hope someone can find this valuable, if only as something to laugh at, or as a good warning.

In all honesty, I will however say that while I won’t recommend that anyone do this because of the hazards involved, it did work for me. If you burn down your house as a result of reading this, please don’t say I didn’t warn you that it is an awful idea.

This was a very simple DIY project that does not warrant the amount of text that follows, but I thought I’d add a little bit of history here to tell about the long-term results, anyways. I’ll include a few pictures, but if anyone wants more detailed info, I will be happy to share.

upload_2020-2-8_19-42-46.png
Above: this is how the unit sits now. Note the batteries in marine boxes with the lids securely fastened

At the facility where I recycle batteries, back when the battery dropoff was a trailer that was not access controlled (now you have to drop off all batteries at a counter), people would dump UPS units that had failed batteries, rather than removing them first and taking those. I figured (correctly) that was why most of those units were there. When dropping off car batteries one day, I snagged 2 1400 VA UPSes: an APC and some other brand that I cannot recall. Bot had bad batteries, which I removed and returned to the collection facility later.

Since these were basically free, I decided I would test out a hypothesis that I had at the time: If the battery capacity is increased in a UPS, the charging time would increase, but so would the usable backup time. Both units used multiple sealed lead-acid batteries setup in a 24V configuration. In order to test this, I purchased some battery cables, some connectors, and 2 of the cheapest SLI (note SLI, not deep cycle) automotive batteries that I could find that had decent CCA ratings. My reason at the time was that with the SLI batteries, I would have more run-time than deep-cycle, and that my power outages were infrequent enough that the damage from a very low number of deep-cycles would still not exceed the normal wear and tear in their intended environment (starting cars in all sorts of weather). Where I lived at the time, while we did lose power for hours or more during the winter or summer storms, it was fairly common that power would go off for just a few minutes or less.

To actually use the batteries in the 24V configuration with the UPS, the internal cables needed to be made to connect to the car batteries, and the batteries needed to be connected to one another in series. I don’t have any pictures from the no-name UPS, but I do of the APC. A hole was cut in the side to run the cables to the external batteries, and the sharp edges were covered to prevent cutting of wires and any shorts that would result from that.

In order to allow easy connection to the battery bank, I used some pre-wired automotive battery connectors taken from a golf-cart or electric wheelchair or something of the sort (I forget which, now). If I was going to do this without what I had on hand, I would probably still use pretty much the same thing: https://www.aliexpress.com/item/32977993126.html. This is a battery connector rated at 50A, 600V, identical to the one built into the back of the unit for the battery disconnect (meant to be pulled out when servicing the unit).

apconside.JPG
Above: APC UPS on its side, showing battery connector used to replace original connectors for internal battery, and matching hole made to allow for external connection.

I setup and tested first with the off-brand unit. One thing that was interesting about it was that it had several 100A fuses in parallel in between its original batteries. I kept that when switching to the SLI automotive batteries, and kept that fuse holder in place when moving over to the APC, later on.

The off-brand unit lasted for about 1.5 years before it appeared that it would no longer convert the 24VDC to 120VAC, even though it would still charge the batteries. I assume this is because I was constantly running the unit right at its capacity, and probably over-taxed it. My first lesson from this experiment: try not to overload UPS’s.

The cables and batteries got switched over to the APC at this point. I was very impressed at how this functioned. With the 1400VA running at right about full load (okay, so maybe I didn’t learn my lesson with the other unit), I would get several hours of run-time, as opposed to the 45 or so minutes that I believe those things normally give out of the box. Of course, as the charging circuit is designed for a much lower capacity, after a long power-outage event, it took a VERY long time to recharge the batteries to full. The batteries that I used for this setup lasted for a little over 7 years, and for that entire time they delivered very well.


oldsetup.JPG
Above: old setup. Visible is connection coming from inside of case through hole cut in side vent.

Eventually, something went wrong. I noticed that everything in my rack that was powered by this UPS was shutting off during the unit’s self-test, and was immediately concerned. I looked over at the unit, and to my horror, the floor underneath and in front of the batteries was wet. Apparently, one of the batteries had exploded. I had seen this in cars that I had worked on before (I was a mechanic and auto-tech instructor at the time), but foolishly, it had not occurred to me that this could happen with my UPS. What a mess! Lessons #2 and #3: Don’t use SLI batteries for UPSes (use deep cycle only), and if you use automotive batteries for a UPS, USE A BATTERY BOX!

20170530_193930.jpg
Above: electrolyte spray and battery that blew its top off
20170530_193942.jpg
Above: closeup of exploded battery. The vent cap and bits of lid were blown very far away, and the computer directly above this battery had the paint on the underside entirely ruined by the splash of sulfuric acid from below.

IMG_20190306_201722.jpg
Above: stain on floor from battery acid that leaked onto it.


For those interested, here is what I believe happened, and why I believe that it did:

SLI batteries are designed to delivery as much current as possible for a short period of time, but are not designed to be deep-cycled. In order to maximize current delivery, SLI batteries have thinner lead plates that are jammed as close as possible, with PVC separators to prevent them from touching and shorting internally. By using thin plates, the number can be increased, which allows for more surface area for the chemical reactions occurring. These thinner plates are more prone to stress from deep-cycling, but also by being close together, they are more susceptible to crystallization that can form between plates, shorting them out. When these plates become shorted and load is placed on the battery, it generates a tremendous amount of heat, boils the electrolyte mixture, and can eventually ignite the hydrogen and oxygen mixture that is given off in the charging and discharging process and from the boiling of the electrolyte.

Deep-cycle batteries are designed to be drained more fully without taking as much damage, but also may be designed to reduce the likelihood of such internal shorts. Even if this is not the case, I can say with certainty that if I had my batteries in proper boxes, the worst damage that I would have received was a few drips of electrolyte on the wood supporting the batteries within my rack. Instead, the underside of one of my computers was sprayed with a solution that stripped the paint off of it, and the floor underneath the batteries was permanently stained. Also, cleaning up sulfuric acid and an exploded battery was no fun. Again, this would have been much simpler with a proper case, which I started using immediately after, along with deep-cycle batteries that I bought to replace the SLI ones.

TL;DR:
  • Yes, you can replace the tiny internal batteries in your UPS with much larger batteries to extend the run-time.
  • Make sure to use deep-cycle batteries
  • Make sure to use battery boxes, and I would still recommend placing these on top of a tray of some sort if you have nice flooring underneath.
 
Last edited:
  • Like
Reactions: Jiaxinxi

T_Minus

Build. Break. Fix. Repeat
Feb 15, 2015
7,029
1,578
113
CA
Have you considered items such as the TRIPP LITE PV1250FC and sealed batteries?

If keeping lead acid as @JSchuricht mentioned they should be vented\outside only, and I'd probably go with Trojan, or another manufacturer that makes them specific for this purpose (solar\off-grid).
 

Mattastrophe

New Member
Feb 2, 2020
5
1
3
While that is bad, I would be more worried about ventilation with that type of battery. You could end up with a much bigger issue than you have already had.
That is a very good point, and I apologize for not addressing or stressing that before, in the event that anyone else considers this ill-advised modification. I will say that I had taken that into consideration when setting this up, but forget to mention it. But yes, it is very important to consider the consequences of doing stupid things with large amounts of chemically stored energy, especially within your own home.

My rack is in a very large room with a lot of airflow, and is not enclosed in any sort of cabinet. Indeed, when using vented lead-acid batteries, airflow is very important, as lead-acid batteries using sulfuric acid in its electrolyte solution will give off hydrogen and oxygen gasses during charging and discharging. I would also like to point out that whenever servicing this setup (or any UPS), I make a point to wear safety glasses, and suggest that anyone else dealing with lead-acid batteries do the same (whether they are sealed or not, as sealed batteries can also fail and vent).

In my previous career as a mechanic, most or all of the eye injuries that I was made aware of were the result of battery explosions. One such incident (of which I have photographs of the car and person in question) was the result of attempting to jump start a car with a dead battery, and the person in question did entirely lose sight in one of their eyes. While this is getting off on a tangent a bit, since on this topic, I have to also stress the importance of safety during jump-starting, and that it is not a bad idea to wear safety glasses or goggles when doing so, and it is also a good idea to leave the hood open long enough to vent any gas that has potentially accumulated if someone had recently tried to start said vehicle, or if the battery was recently being charged. Even with these measures, a failed battery can still ignite internally (such as with what happened with mine), which is why even with precautions, I always wear safety glasses when jumping cars, and strongly suggest that others do, as well. I'd rather look silly when jumping a car than lose my eye(s) because I didn't want to be seen in stupid glasses.
 

Mattastrophe

New Member
Feb 2, 2020
5
1
3
Have you considered items such as the TRIPP LITE PV1250FC and sealed batteries?

If keeping lead acid as @JSchuricht mentioned they should be vented\outside only, and I'd probably go with Trojan, or another manufacturer that makes them specific for this purpose (solar\off-grid).
I will look into those more, thank you. If I continue to use this setup for much longer, I will likely eventually switch to one of those. I will admit that cost had a huge factor in my initial selection as well, as at the time, I was able to get very large discounts from several local auto parts suppliers, and so I had limited my selection to what was available there.
 

maes

Member
Nov 11, 2018
67
30
18
For those interested, here is what I believe happened, and why I believe that it did:
It's not just a question of SLI vs deep cycle; 'flooded' lead-acid batteries (aka 'car batteries'), gel cells and AGM batteries have different charging and float voltage requirements. UPSes are normally factory-calibrated for AGM batteries (in most cases, tho some might use gel cells and some newer units use lithium packs), but, and this is important, that calibration can drift as the unit ages and/or tries to adjust itself to the batteries themselves aging and degrading. There's drift in both the software calibration, and drift in the charger components aging, so sometimes the charge voltage the UPS microcontroller thinks it's giving the batteries (measured from its internal ADC) can be different from what it's actually giving the batteries, and both can be very different from what the preferred charge voltage for the specific battery type used.

That calibration also doesn't necessarily 'reset' itself when replacing batteries and, for some models, might have to be altered manually. It's likely you were feeding the wrong voltage to the batteries over an extended period, enough to cause them to offgas and pop.

Here is a good resource to get started.
 
Last edited:
  • Like
Reactions: Mattastrophe

Tom5051

Active Member
Jan 18, 2017
236
29
28
42
wet cell batteries are totally the wrong design for use with a UPS, deep cycling destroys the lead plates rapidly, charging voltage is higher for AGM batteries, wet cells release hydrogen gas during charging.
 

Attachments

Mattastrophe

New Member
Feb 2, 2020
5
1
3
It's not just a question of SLI vs deep cycle; 'flooded' lead-acid batteries (aka 'car batteries'), gel cells and AGM batteries have different charging and float voltage requirements. UPSes are normally factory-calibrated for AGM batteries (in most cases, tho some might use gel cells and some newer units use lithium packs), but, and this is important, that calibration can drift as the unit ages and/or tries to adjust itself to the batteries themselves aging and degrading. There's drift in both the software calibration, and drift in the charger components aging, so sometimes the charge voltage the UPS microcontroller thinks it's giving the batteries (measured from its internal ADC) can be different from what it's actually giving the batteries, and both can be very different from what the preferred charge voltage for the specific battery type used.

That calibration also doesn't necessarily 'reset' itself when replacing batteries and, for some models, might have to be altered manually. It's likely you were feeding the wrong voltage to the batteries over an extended period, enough to cause them to offgas and pop.

Here is a good resource to get started.
Awesome! Thanks for the info. Right now, I don't have this unit in service, but I'll definitely put this information to use if I use it again. As Tom5051 mentioned, I am aware that AGM charging is done at higher voltages than standard SLA or deep-cycle automotive/marine batteries, but even that is typically (although at the upper end) within the voltage range used by some of the bulk charging units and fast-chargers used in shops, if I recall correctly. However, if this is forced to drift higher due to the software and/or mis-calibration, then that could definitely be problematic.

I know that this setup is not kind to the batteries I used, but my initial goal was to see what capacity UPS I could get for next to free, and when all is said and done, the thing lasted and was heavily used for at least 7 years. But this would definitely not surprise me if what you had brought up contributed to or caused that failure, and was causing issues when I tried to resume service, later.

Sadly, I must admit that since my initial post, I have replaced this with newer units -- besides being an experiment, at the time I initially put this together, I was a student also working as a teacher, so a huge part of this was that I couldn't justify buying a nice UPS with the funds I had available. I have not discarded this unit though, and I am now very curious about the calibration you had mentioned, and will definitely check that out on this unit when I get a chance, both with a meter while running the unit, and with what was mentioned in your link. Even if I don't end up using it again, it sounds like a fun thing to dive into. If what I read from your link is correct, though, it sounds like the voltage might be able to be adjusted to play nicely with some flooded cell deep-cycle batteries.

Thanks so much!
 

maes

Member
Nov 11, 2018
67
30
18
If what I read from your link is correct, though, it sounds like the voltage might be able to be adjusted to play nicely with some flooded cell deep-cycle batteries.
The adjustment does have a surprisingly wide margin, but you will probably need to get one of the proprietary-pinout serial cables (because APC can't do anything normal) to get the recalibration done. I know I've seen and salvaged units that were off by multiple volts. In one of those, the old AGM battery had swollen so much I had to completely dismantle the unit and take a 5lb mallet to it to bash the battery out.
 

SRussell

Active Member
Oct 7, 2019
292
147
43
US
Have you thought about doing this with a Lithium Ion UPS and a DIY Tesla Powerwall?



 

Stephan

IT Professional
Apr 21, 2017
138
45
28
Bavaria, Germany
I am on the lookout for a silent UPS in the 3000VA range (no fan when mains is active; fan on while on battery would be acceptable) that has expandable batteries. Modding something would be okay, too. Doesn't exist, right?
 

Lost-Benji

Member
Jan 21, 2013
424
23
18
The arse end of the planet
The APC UPS's lend themselves to being very easy to work with but after looking at the images in the first few posts, you need to understand the dangers that you are messing with as clearly, you don't. Not trying to be rude but it needs to be said.

Batteries ALWAYS need to be contained, they can and do leak and they do vent. Electrical. DC side or the AC side, both need respect, lethal AC and very high-current flow potential on the DC side.

For the ease, neatness and safety, look at the APC SmartUPS XL (eXtra Long runtime) models that allow external battery operation via the anderson plugs. Use a Network management card (Cheap on eBay) and set the EBM number to roughly that of the same A/h capacity of the APC EBM's (28.8 - 36A/h @ 24V).

Do not even think of using Lithium batteries on the UPS's, they are not designed for those chemistries.
 

Diavuno

Active Member
I am on the lookout for a silent UPS in the 3000VA range (no fan when mains is active; fan on while on battery would be acceptable) that has expandable batteries. Modding something would be okay, too. Doesn't exist, right?
I have a APC smart UPS 2500 tower that is silent when on the utility power.... small fan when on batteries.
I also just picked up a used APC smart UPS 3000 2U (part of a cheap rack I picked up) it is nearly silent... slight hum when on utility power.
Both run on a simple 48V DC connection (same Anderson connector as above)

I think I'll modify that 3000 unit with some big batteries, Im looking for some big AGM 6V (golf cart) batteries.
 

Stephan

IT Professional
Apr 21, 2017
138
45
28
Bavaria, Germany
Meanwhile I found an UPS that fits the bill for me, it's an APC SU2200 (old beige model from around turn of millenium) with blue Anderson connector for up to 10 external battery packs. These now 20 year old models have some known flaws: Dry capacitors, two overheating resistors, worn out metal oxide varistors, drifting float charging voltage, wrongly set battery constant after some years, etc... all that I hope to be able to fix with a soldering iron. Especially float voltage needs to be fixed back to optimal by replacing two resistors (will swap one with a 25 turn precision tunable resistor), otherwise batteries are either cooked after 1-2 years, because voltage was too high, or they will never be fully charged. Or sulfurization on low voltage will even damage them way before their EOL (I am shooting for 10 years). Also, this is the last model with reasonable battery capacity that will turn off the fan when not on battery and when not charging. SMT SUT or whatever came later all never turn off the fan. I believe they will slow the fan though, but never fully turn it off. Nah, don't like that in an office environment. Also found a nice 24V Noctua fan, will replace the loud fan with that. Got all the other parts already too, hope this works out and I will not electrocute myself... ;-)