APC Back-UPS capacity upgrade (#P11F1)
Was that foreseeable? Not sure, but the design called for it…
Now that my modified APC UPS is running smoothly again (#P11), I still had two pairs of these nice XT60 connectors laying around. And as old, to-be-recycled lead acid batteries from UPSes pile up at work, I took one or two of them with me for testing at my electronic load. Like previous batteries, quite a lot of them are really toast, but some of them are in decent condition. Not as in “like new, ready for another year of service in a commercial fire alarm system”, but still holding a significant fraction of their original capacity and delivering a suitable amount of current. We’re going to reuse some of them with a DIY spot welder system at work soon, but I got to take the smallest batteries with me – small being relative here, “only” 12Ah, compared to 26Ah bricks that are basically half a car battery…and we got dozens of them. The original RBC17 of my UPS is 9Ah…
So the two puppies measured around 8Ah on 12A (1C) load, probably exceeding their specs as they are intended for long-term battery backups, not peak current. 2/3 of original capacity is no longer suitable for that task, but adding 16-ish amp hours is like a 175% expansion of battery capacity – not factoring in the lower current for each battery, likely increasing total usable capacity even more. These use the same flat faston connectors as the original system (larger ones use screw terminals), so adding those is exactly the same procedure
as last year as adding the first external battery pack as described before.
Conveniently, my file has the exact width required for the connectors, so those cutouts look better than before. Spaced equally, not colliding with the internal PCBs and wiring, but requiring additional cuts to the battery molding of the bottom of the case.
I’ll spare you of the soldering pic that I took – I just removed the insulation on two non-overlapping spots as shown, and added plenty of solder to it. I would have made a four-way connection out of it, but given the necessary currents that WILL be drawn in edge cases, I didn’t have any suitable connector system at hand. Wago 221 for example only allow for 32A and 4mm², which is exceeded by the non-interchangeable 10 AWG positive wire as well as my test load (shown below). So not chopping cables to bits, but hitting them with enough solder was my solution. They are insulated with hot snot and the cables cannot move anywhere, so even if the snot slowly liquefies and drips away, having no insulation shouldn’t be a problem. Not my proudest soldering joint ever, and I’m fully aware that this most likely isn’t supported or encouraged by APC in ANY way
I should also mention that connecting batteries in parallel without any protective circuitry involved is a risky thing. There’s the obvious one of having different states of charge and therefore potentially huge cross currents until all batteries are at the same voltage – which can be avoided by charging each battery individually with the same charger until full and then connecting them to each other. But having issues with one cell of one battery also means every connected battery will pump power into the system – as this is entirely made up of old, worn batteries, damage to a cell resulting in a short isn’t completely impossible. Those 6S would then dump their energy in the 5S one, probably not to their liking. A few of the larger batteries tested have this 5S characteristic, so there’s one dead cell inside that makes the entire stack high impedance…until more current is dumped into and the thing heats up. I’ve seen batteries that accept 100mA at the beginning and then gulp away 30 amps a minute later. Let’s just hope this never happens on my setup
Anyway, this is everything connected and (trickle) charging – some folks on APC forums note that a fully depleted battery larger than the factory one could trigger UPSes into battery fail mode by a simple timing circuit. As this consumer UPS is pretty dumb and has no way of knowing a serial number, UUID or something of the battery, this would be resettable by a cold start, if that ever happens. Also, as charging current is pretty low even if the battery is flat, I don’t think this is a thermal issue long term. APC says this unit needs up to 16 hours to fully charge, most likely to discourage commercial use where this would be pretty devastating in case of a second power failure.
This is the load test, running my desktop (FX-6300, 16GB DDR3-ECC, Radeon 380, one SSD, one 10GbE card, not much else) at peak load with Furmark. Display wasn’t attached to the UPS at that moment.
33A peak at I’d guess 11ish Volts would be 350ish Watts drawn from the batteries, which split current pretty evenly relative to their original capacity (12A/12A/9A). Massive improvement over drawing that from the poor single 9Ah cell! I’m aware that the extended capacity does not raise the thermal capacity of the UPS converter, even though the removal of the battery from the case could be slightly beneficial to that. But as I do not intend to run Furmark for the lulz when I’m not at home, I don’t see a problem with sub-50W loads over extended periods of time. “Das muss das Boot abkönnen” (the vessel’s gotta put up with that), as the famous quote of Das Boot suggests. Actually, just tomorrow will be a planned outage for 2 hours, so I’ll see how that system fares.
Batteries are charged up to a moderate 13.60V and show up as 100% on their Windows monitoring software. The apcupsd under Linux doesn’t really allow for checking battery state, but does other funky things like setting a battery installation date…again, this is a consumer UPS and is therefore artificially crippled compared to the much more expensive business counterparts. I don’t mind, it’s my first and simultaneously last APC unit…
[…] after a very quick disassembly of my UPS thanks to the external battery mod in #P11F1 it became clear that there’s no imminent fire risk due to the UPS being overloaded yet still […]
[…] base. So my solution was an old desktop UPS unit, actually the very same APC 700VA one that I own (and have modified) myself. Add extension leads to not wear out the internal ones that are also a bit short, and then […]