If you’ve got a so-called uninterruptible power supply (UPS) on your system, you’re probably painfully aware that the “uninterruptible” part has some pretty serious limits. Most consumer units are designed to provide power during a black out only long enough to gracefully shut down your system. But with a few hacks like these, you can stretch that time out and turn it into a long-endurance UPS.
As many good stories do, this one starts in the trash, where [MetaphysicalEngineer] spotted an APC home office-style UPS. It was clearly labeled “broken,” but that just turned out to be a dead battery. While he could have simply replaced it with a 12-volt sealed lead-acid battery, [Meta] knew that his computer setup would quickly deplete the standard battery. A little testing showed him that a car battery would extend the run time significantly, especially if he threw in some extra cooling for the onboard inverter. Bess Solar Storage
His final design uses a marine deep-cycle battery in a plastic battery box with the UPS mounted on top. The vacated battery compartment made a great place to add a cooling fan, along with a clever circuit to turn it on only when the beeper on the UPS sounds, with a bonus volume control for the annoying sound. He also added accessories to the battery box top, including a voltmeter, a USB charger, and a switched 12-volt power outlet. And kudos for the liberal use of fuses in the build; things could get spicy otherwise. The video below shows the entire build along with all the testing. [MetaphysicalEngineer] managed to triple the estimated runtime for the load he’s trying to power, so it seems like a win to us.
If your needs run more toward keeping your networking gear running through a blackout, you might want to check out this inverter-less DC UPS.
A couple of months ago I scored a *really nice* industrial UPS rack-mount system. Both batteries are dead, but I knew that.
Some of the people at the local makerspace work at a UPS company, and they claim that tapping off of the battery connections in the UPS is an appropriate way to get 12 volt continuous power from the unit – the batteries are recharged as needed from the AC lines and the system is not using the inverter circuit.
(Of course if your UPS has a 12V output you don’t need to do that…)
I want an experiment to operate continuously for more than a year, so powering it this way from a UPS is a good hacker way to do that. I estimate 12 hours (-ish) of backup time from the 2 gel-cell batteries in the unit.
IIRC, sealed led-acid batteries have a shelf life of 5 years, after which they should be replaced. Since the cost of the batteries is almost the cost of a new unit, people just get a new unit rather than repair the old one.
Sort of like the batteries for your drill gun.
Additionally, a junked UPS can be used as an inverter/regulator for solar panels. You still need the MPP circuit and battery, but the high-power inverter in the junked unit should still work.
Old Cyberpower I imagine has more problems than that. Plus those cheap APCs don’t last.
Can you be more specific?
Other than the batteries, what goes bad in an electronic circuit like that?
I’m looking for a cheap hacker-friendly way to power something continuously for a year (hopefully 5 years), with backup in case of short outages such as we get in the Winter in the Northeast.
It’s an APC UPS 750, clean as a whistle. Unit was never left out in the elements. Is this not a high-end UPS?
Is there a reason I shouldn’t rely on this unit?
In my experience as a long time sysadmin and IT manager, the APC SmartUPS family are pretty reliable. Not industrial grade, but solid. The lower end APC units, not so much, for two reasons. 1) I’ve seen way to many of them fail for other than battery. 2) The cheaper units from APC and others are “modified sine wave” output, which is frankly a lie. There’s nothing sine wave about them, they’re square wave outputs.
Square wave outputs strain most power supply designs and I don’t recommend them, especially for long term use. I’ve had some power supplies for laptops that wouldn’t run on them at all.
If you can find a SmartUPS UPS or other brand with a full sine wave output with a dead battery, you’re much better off with that. Be careful with TrippLite, as the models I have here won’t start unless the battery matches their design specs, so probably isn’t as useful for this type of hacking. Perhaps that’s only this model.
Good luck and note the line in the above article about careful use of fuses!
“I’ve had some power supplies for laptops that wouldn’t run on them at all.”
That’s probably unrelated to the stepped square wave UPS output as the first thing switching power supplies (e.g. laptop supplies) do is rectify the incoming voltage. They don’t give a crap what the incoming waveform looks like (have you looked at a typical AC line waveform?) so long as the diodes/bridge rectifier can turn it into DC.
Likely the real reason they wouldn’t work is they were junk (HP perchance?), or the voltage was too high or too low.
I once looked at the current waveform into a laptop power brick on AC mains. It looked like a square wave with a sine half wave cut out in the middle. Probably the brick drew the max input current until the line voltage was high enough for the power needed, then reduced the current accordingly. Rectifier -> step up -> smallest possible capacitor… The minimum voltage would then equal the mains voltage when the cut-out begins. I also observed that the brick would run on DC from a pfc circuit, but only with one orientation of the plug. I didn’t figure out the reason, with a rectifier as first component polarity should’nt matter, so no idea on that.
What are the power requirements? A 12volt battery and a trickle charger should do it, if your power budget is less than a trickle.. (But that’s pretty much the same as using an ups in the way you said, so… )
It should work. Dont trust the naysayers. Experiment and enjoy like the guy in the video.
Heh, one of them was an HP supply….
However, it still could be low voltage. The square wave output of these things peaks at a much lower voltage than the sine wave it replaces, I think the goal it to have the same energy in the cycle as the sine wave would have.
Hmm.. I just Google’d this and I see different descriptions of “modified sine wave”. The one time I put an O-scope on the output of one, I got a waveform like the “square wave” in the page below. Perhaps other models are better.
https://www.researchgate.net/figure/Square-Wave-Modified-Wave-and-Pure-Sine-Wave_fig1_350521718
Although I don’t like modified sine wave, I certainly agree with this.
Argh It put this under my own comment. I was agreeing with Infrared above.
Replying to all comments – thank you for the insight!
The unit is a rack-mount APC smart-ups model, so should be reasonably robust. I’ll be taking 12V directly from the internal batteries, so issues with simulated sine wave won’t directly apply. According to my friends at the makerspace, the UPS should be OK with this.
I could run off of a car battery and trickle charger, but the UPS has a serial and USB interface that will allow me to log events relevant to the experiment (science rule: record in your notebook anything that you can think of that might affect the results), as well as conditioning circuits that are probably cleaner and more robust than the trickle charger.
And I don’t know how much current/power is required right now because I just recently discovered (from a different HAD article) that GPS disciplined oscillators are available on eBay, which is a bunch simpler than the GPS receiver plus TXCO I was planning on using, and I don’t know how much power the eBay device draws yet.
Again, thanks for the expertise.
The rack mount APC units don’t run on 12 volts. Battery pack is a series parallel configuration. Better open it up and see.
Hey PWalsh, I am currently restoring an Rack mount APC Smart-UPS 1250RM. The batteries are in series, so 24v. I did a couple upgrades; added a V/A meter on the AC output and one on the batteries so I can monitor status better. I also added battery disconnect plugs to more easily connect/disconnect the battery pack. Now l am thing of adding a pannel-mount plug to add an external pack. The unit has big heat sinks and a fan, so l think it would support this mod, as-is.
This is probably a “Perfect vs. Good Enough” situation.
Years ago, I modified a cheap CyberPower UPS with a lawnmower battery and was able to power my PC for several hours without AC power (instead of the 20 minutes that was originally possible). If I did it again, I would use a deep-cycle battery instead of a lawnmower battery. I’d expect the results to be similar.
The electrolytic capacitors used are priced to the penny point and do not last. It is a common repair to re-cap the unit. I just did mine after it died, it would not come back up after a power failure- despite having a new battery. It was a 22uF cap dried out but I replaced all electrolytics at the 15 year mark. Nothing APC makes for consumers is “high end” and the power supply is running 24/7. Heat can also be an issue as the inverter is not designed to run for a long time, miniscule heatsinks. Would add a fan for when it’s running. Also, some UPS firmware does know the battery’s Ah capacity and gets mixed up when you plop in a huge car battery. Any strange cutouts are because its model assumes the battery should be dead and the unit is likely getting hot, due to the long run time. I can’t remember the make/model on youtube that did this.
Good point about the caps on rescued ewaste of this sort. Add to the thermals, most of the video is him measuring and building a solution to that.
You can get LiFePo drop in replacements for the original SLA’s. Much longer service life and runtime (for a few dollars more).
I’m making a concerted effort to get Lead-Acid out of my life – having to replace a battery that I literally use maybe 3 times in 5 years just drove me nuts.
Tell me more — I have a couple of applications (mostly UPSs, but also a mobility chair) that I would like a drop-in alternative replacement for SLAs.
I used a Dakota brand lifepo4 as drop in replacements in my tripplites. I also got some no name lifepo4’s with a good rating on Amazon for mother in laws mobility chair. It works but needs a lifepo4 specific charger.
The downside on the mobility chair is the capacity meter is set for lead acid. So it’s green and the suddenly dead ( because lithium holds a flat charge for the entire time instead of slow decline in voltage like lead acid)
oof.. the price difference is tough to swallow, though. And the datasheets I’m reading about not connecting in series eliminate the ability to use in my dad’s chair (which uses two 12220 batteries in series). And the whole capacity meter thing would just confuse my mom.
But, thanks for the info — I appreciate it.
They last far longer, though…especially because you can use their full capacity without damaging them anywhere near as much as lead acid batteries if they’re fully cycled.
LiFePO4 batteries will last nearly a thousand full cycles. And because they store much more energy for the size/weight, they don’t get as deeply cycled.
I am actually doing this. I havd an old APC that had 10Ah SLA cells in it that I’ve replaced with 14Ah LiFePo. Extended runtime and the unit doesnt charge the lithoums past about 85-90% so itll extend their shelf life as well :)
I’ve done this before and the results are quite mixed. On the surface this does work as intended, but with consumer models you will find that they are definitely not designed with even the maximum running time fully in mind. The two problems that will stare you in the face when you open up a really dead unit is that the circuit board is un-vented and that the heatsink for the MOSFETs is undersized. The fastest failure was after 2 hours on a 45 watt load. When I opened it up afterwards the MOSFETs were burnt crispy. Others lasted much longer, but ultimately met the same end.
Commercial models are better but still questionable.
I’d recommend actually watching the video, most of which is him dealing with this exact issue, by careful measurement, design, and building of a forced air cooling addition. He didn’t just yolo some bigger battery in there.
Failure modes can include dropping out the AC output at random intervals *without* any input power failure! I found consumer ups’s gave me lower reliability than the utility it was supposed to be protecting me from.
I had this exact problem recently. Once a week my rack would reset. Turns out the UPS does a weekly automatic test by cutting input power so that it can detect and throw a battery error before the battery completely dies. For some reason it never threw the error but instead would completely lose power the instant it cut input power. Maybe powering the test circuit off the battery being tested wasn’t the best design…….
Former UPS Design Engineer here. The thermal design of the UPS inverter section only takes into account the runtime available from the internal battery. Running at rated power for longer periods than the internal battery capacity would allow risks exceeding thermal limits and potential overtemp shutdown or damage. If you only require a fraction of the rated load the inverter can run indefinitely without issue, but intimate knowledge of the internal circuitry and thermal design would be required to know what that output power level would be.
Most of the video is him measuring this very issue, and building a forced air solution to keep the temps in line. He even takes into consideration the transformer temps. I’d recommend watching it, if you haven’t.
I did this years ago with an APC UPS and a deep cycle battery…after a day of being on mains to fully charge I decided to test it by unplugging it. I had a fuse in the battery line – 20 or 30 amps IIRC – and it immediately blew. Subsequent tomfoolery ended when two fuses soldered to the circuit board went poof.
I never figured out why so I never tried it again.
Dan, thanks for calling out the builder’s concern for safety via use of correctly sized fuses. Based on the comments, y’all needed to point out that the bulk of the video is dealing with a proper cooling solution. That or people need to actually watch the video before commenting. But I guess that would be against the hackaday tradition!
This is not a YOLO throw some parts together vid, this is a proper hack, with bench prototyping and measurement of potential issues (heat buildup and float voltage), then designing and building a one off solution. Very nice!
I have the same APC and I strapped a larger SLA with nylon zip ties. APC was purchased for $4.99 and I can get more. This Hack by metaphysicalengineer (https://www.youtube.com/@MetaphysicalEngineer) motivates me to add a fan because I have noticed increased dissipative switching heat during blackouts. I have two more APC under-desk UPS units and a few Marine Battery boxes and I plan to replicate the concept of long duration UPS. I used to rely on my Canadian Hydro Power supply until recent storms had knocked out municipal utilities for days at a time. I also store a Gas Generator in my Garage now.
I did this with an added solar control, and 100 ah lithium 12v. Worked great.
We do this with cyber power ups’s for remote tower sites. Just make sure your ups uses a real transformer. I caught one of the lower VA UPS’s on fire because the digital voltage circuitry was not designed to run as long as the bigger battery could.
A for effort, but for the time, effort, and expense put into this, a power station (that has a LiFePO4 battery) with a UPS function (with a reasonably fast switching time, some don’t switch fast enough) is likely the better way to go. As an added plus, almost all of them use a pure sine wave inverter.
If you want a DIY bang for your buck solution, an inverter with a UPS function (with an integrated battery charger using mains and/or a solar MPPT) and 48v LiFePO4 server rack batteries is likely your ticket.
Also, LiFePO4 batteries have a higher upfront cost than lead-acid, but more than make up for it by typically having a cycle of 3000 (or more) cycles before retaining 80% of their original capacity.
Please be kind and respectful to help make the comments section excellent. (Comment Policy)
This site uses Akismet to reduce spam. Learn how your comment data is processed.
Transformer 110 To 220 By using our website and services, you expressly agree to the placement of our performance, functionality and advertising cookies. Learn more