I have a hodge-podge of UPSes, many are loaded to 80%+ capacity, and many are > 2 years old. As such my runtime during power failure isn’t even enough to actually shut down any of the machines. Instead of buying some new ones, I decided to experiment with using a ‘car battery’. These UPSes all have relatively tiny batteries, new ones will have the same tiny battery and usually cost a lot to ship or buy locally. Since the other components of my UPS are still working okay (And configured), I thought I’d give this a spin.
I already have an emergency sump pump in my basement which is powered by a large deep cycle battery. The sump actually came with its own battery which even includes a probe that goes down into one of the cells to monitor the fluid level. It is a non-sealed battery and I add distilled water to it about every 9 months. It’s about 5 years old, has been used on battery a few dozen times, and the battery has not yet failed. Of course the battery is large, probably 80lbs, and was $140 or so 5 years ago. It is kept inside a plastic battery box, and I keep some baking soda nearby incase something goes awry.
I decided to start a bit smaller with the UPS. I got another battery box and the smallest ‘deep cycle’ battery I could find locally. I also picked up some 4g ‘starter motor wire’s about 32″ long. All this after tax and core charge was about $104 from Autozone. I cracked open a UPS with a totally flat battery that was sitting on the sidelines. I drilled a hole in the metal case, put in the 4g wires, cut and stripped the ends of the UPS connections, twisted them together with a lot of metal touching, taped the hell out of them, the insulated the while setup form each other and the sides of the case with some plastic strips. This is to be a temporary setup, if it works, I’ll solder and heat-shrink the connections. The starter motor wires were then attached to the battery. The Deep Cycle battery had normal posts as well as wing-nut bolt type posts, which meat that the starter motor wires could be dropped on easily.
I plugged the whole thing in and it seemed to work. I connected the UPS to the computer via its serial cable and fired up ‘nut’. The UPS seemed happy with the battery and was charging it, claiming it was only 37% charged. Still it had a no-load estimated runtime of 3200 mins. I let it charge the battery to 100%. During this time there was some minor outgassing from the battery which I monitored closely. The hydrogen is explosive and toxic, I was in a reasonably ventilated 950 sq/ft room, so I wasn’t too worried, but too much outgassing is not normal and it could blow the top off and expel some acid. Fortunately this didn’t happen, after 90 min or so the ups claimed the battery was at 100% charge, 13.5v, and had some 13000 mins run time. I connected a lamp and gave it a quick test, run time came down and everything seemed fine.
Next I’ll put a real load on it and hook it up through a switched outlet, so I can drop power but maintain ground, and I’ll let it run the battery down to 30%, and see what the real uptime is. The UPS is an old APC/conext CNB900 which only provides 450w max, so if this scheme works, I’ll need a few more batteries.
I should also note that I verified earlier that the unit was a 12v UPS as it had a single battery which was clearly labeled as providing 12v. Some UPSes require 24v, which would mean 2 deep-cycle batteries, though that would also mean even more run-time.