DIY power backup system

From The Aquarium Wiki
Jump to: navigation, search

This DIY power backup system project is designed to use a typical computer UPS with a dead battery and a new, larger battery to maintain water flow and heating during power outages.

This version is combined with my DIY timed filter shutdown project.

Parts list

  • One UPS with dead battery (can often be had for free) with sufficient wattage capacity for requirements
  • One big sealed lead-acid or gel mat type deep-cycle battery, such as the 55 aH SLA0176 by Interstate Batteries (about $140) which is designed as a replacement for traffic light backup systems made by Clary
  • GFCI outlet, since on battery power there is no "upstream" GFCI protection
  • Fuse holder for battery wire sufficient for maximum current (20-30 amps) and fuse
  • Binding posts or similar system to connect battery to UPS

Building it

Big battery

I obtained a dead 260 watt UPS from my brother for the cost of shipping (he deals in recycling e-waste). I perused the Interstate Battery website to pick an appropriate battery - starting with SLA models, I looked for the best aH/$ ratio in the 50-70 aH range. My neighbor is an Interstate dealer so ordering it was easy.


Once I had all the goodies, I dismantled the UPS to see what alterations I would have to make, taking notes as to where the wires went (although the PCB was clearly marked). I had to move the transformer about 1/2"/1 cm in order to provide clearance for the GFCI outlets, since they are deeper than the conventional ones that were in there.

Clearance issue

To install the GFCI I used a nibbler to cut the normal outlet holes into a rectangle.

Moved transformer

It turns out that the network/telephone surge protection feature is completely standalone, so removing it left some real estate on the rear panel to mount battery connections. I used binding posts I had on hand once I determined that they were roughly 8-10 AWG to handle the current required. I cut out some of the metal to make sure they wouldn't short out.

Installed filter time-out, jack in background

I mounted the DIY timed filter shutdown board where the battery used to be, and put a 1/8"/3.5 mm jack in the plastic front panel to plug the switch into. I had to use my Dremel to make the front panel thinner to accommodate the jack's threads.


Then I carefully reassembled the UPS, attaching the leads that used to go to the battery to the binding posts with crimped-on o-rings. The output power from the PCB goes to the "line" side of the GFCI and the "load" side of the GFCI goes to the other pair of outlets, in my case, after a trip through the relay on the filter shutdown board.

Set up for testing

With the cover still not installed, I carefully hooked up the battery without a fuse, installed the fuse, plugged in the UPS and prepared to test everything. I went through a few stages of minor panic - first, the "site wiring fault" light came on on the back of the UPS. I feared the worst forms of miswiring inside, and hoped I hadn't damaged anything. Turns out the strip I had plugged into had no ground! (Note to self: fix that) Then, because when powered up the UPS runs a battery test, the various clicks and buzzes and lights on the front panel disturbed me, until I realized they were "normal". Finally, the GFCI I bought has a red light indicating when it is "on", which I have never seen before, so I thought it was a warning light.

Moved from lab to tank, ready to set up

I plugged a desk lamp into the switched outlets and turned it on. Then I unplugged the UPS and the light stayed on. Plugged it back in and it went back to mains power. So far so good.

Quasi-dual banana plug safety

Then I got my kitchen timer and tested the filter shutdown system. It still worked, luckily, and the first test ran about 3:30. I turned the VR about a third of a turn one way and the next test ran about 2:15, so I turned it the other way and got 5:30, which is close enough for government work.

The invisible button

Then I unplugged everything, glued the battery wires into a gutted dual banana plug housing for a slight safety factor, and hauled it all out to the aquarium. While the UPS is lighter now, that damn battery weighs about 40 pounds (600 kg).

Where it all resides when installed

I had planned on putting both parts on the floor under the tank, but I realized that I had plenty of height and relatively cramped "area", so I put the UPS on top of the battery after sticking some rubber feet on the top of the battery to keep things level and stable. I drilled a hole in the stand for the timer switch, which I had soldered to a 6 foot (2m) cable with a molded plug on the other end. After rearranging all my mains cables, it is all happy and functional.


Now when I go to feed the fish, I push the button and the filters stop, confusing the heck out of the fish. But Pavlov will prevail, and after a few days they are beginning to understand what the sudden stillness means: food! Well, the cichlids have it figured out already, at least. The tetras are a little slower on the uptake. After a few minutes the filters restart with a bit of a grunt and a gurgle.

If I remember, next time the power goes out I'll report on the results.

Side issues

One issue I have not addressed yet is that my tank has a 250w heater, which is too much for the UPS combined with the two Emperor 400 filters (I was running it for a while on a borrowed little battery, and it complained when the heater came on), so I will be adding a 150w heater at the other end of the tank, set 1-3 degrees (F or C) lower, as a backup unit for power outages. There's no rush since it is now late spring, and temperature maintenance won't be an issue for several months.

An important, but difficult to quantify, factor is whether the UPS's charging circuit will be sufficient to replace the battery's self-discharging over time, since this loss will be greater with the larger battery than the one it was designed for. I did run some very crude numbers through my head, and I suspect this will not be a problem (keep in mind I have no hard data on either the battery's self-discharge rate or the UPS's charging capability). Only time will tell, I suppose.

See also