Why are UPS power supplies such junk?

[SiliconWizard]

For my workstation, I have a Eaton Ellipse which I've had for years and it works pretty well. I changed the battery only once (after several years, I'd say at least 4 or 5), and it was a quick job. The only downside (which seems pretty common among UPS, at least in this price range) is that there is no indication that the battery is going bad. No early warning. When it is, and there is a power cut, it will just... cut.

[coppice]

For my workstation, I have a Eaton Ellipse which I've had for years and it works pretty well. I changed the battery only once (after several years, I'd say at least 4 or 5), and it was a quick job. The only downside (which seems pretty common among UPS, at least in this price range) is that there is no indication that the battery is going bad. No early warning. When it is, and there is a power cut, it will just... cut.

Sadly, the only real test of a battery's ability to do work is to make it do work, and this does cause problems for traditional UPSes. If you used a solar battery and inverter type system, which can charge the battery or discharge it, all while connected with all attached appliances operating, a battery could be discharged regularly, doing real work, proving it can. Split the battery in two, and you could discharge one, while keeping the other fully charged, in case a supply failure occurs at the wrong moment. That would be hard to do at low cost in a small UPS, but would be a breeze in a larger system where you may have multiple parallel battery chains already.

4 or 5 years battery life is a better life than others are saying in this thread, but it still pretty poor for a battery that isn't really being worked very much. Car batteries with 5 year warranties are common, and they are being guaranteed to achieve that life with the abuse they get doing stop start in traffic, and especially starting from cold on a freezing day.

[AVGresponding]

the float voltage of the batteries also being set at the top of the range in order to extract maximum runtime at the expense of battery lifespan.

It has nothing to do with "maximizing runtime". Long-term, lead acid needs to stay at 100% state-of-charge, there is no leeway in either "increasing" or "decreasing" usable capacity. It's 100%, or it's dead in no time. Both overcharging and undercharging self-destroy the batteries, just in a different way.

Some UPS brands indeed are "known" to use floating voltage a tad on the high side, but then again, anecdotal evidence of reducing that voltage is not very good either (see reply #9 for example).

Good lead acid solution needs to keep the batteries at normal room temperature instead of some hot enclosure, plus add temperature sensing and floating voltage compensation.

Are you integrated into a UPS manufacturer's design team? I was, at one point, which is why I know how chargers are often designed and configured.

Claiming that reducing float voltage doesn't help is meaningless without the context of knowing the original float charge, and what you intend to reduce it to (granted, the corollary is also true), so to be clear , I favour 2.25-2.3V/cell at 21˚C, derating by maybe 0.01V/cell/2˚C above that. Bear in mind also that it's a practical impossibility to have exactly matched cell voltages across a string.

You are correct about temperature regulation and voltage compensation. This is rarely paid more than lip service, save in enterprise grade equipment. Serious attempts at cell balancing aren't something you see in consumer grade lead acid battery systems either.

A float voltage of 26V(2.166V/cell), if anything, seems on the low side even taking into account the current internal temperature of 32.2^oC.

APC do have a habit of "calibrating" the expected run time by switching to battery and powering the load until the batteries are nearly discharged - I suspect that reduces the battery life a fair bit.

That's a little on the low side, but not horrifically so. 32.2˚C is a bit high, the battery will be happier at <25˚C.

Cycling the battery like that does shorten its life. It's a trade-off, I guess. Lead acid batteries have a very non-linear discharge profile, especially in the usually (relatively) high current, (relatively) short duration profile you'd expect in a UPS, and it would be difficult to give an accurate prediction of runtime off just a short loading. Emergency lighting central batteries are normally much easier to predict the behaviour of.

[nfmax]

What I want from my UPS is a fairly long run time at a fairly low power draw - say 1 hour at 75VA load. This to keep up network router, switches, ONT, CCTV, automation controller, and VoIP phones. Short outages are common here, we had one yesterday for 51 seconds. Longer outages are less common, but we had two in 2024.

I use a grossly over-rated, true sine wave UPS from HP I bought as a refurb off eBay. A new set of batteries gave me 45 minutes. After just over two years, it’s down to 35 minutes. It has an internal fan, and it’s clear from its operation the UPS does a maintenance charge every month or so.

This followed unsatisfactory experience with a variety of consumer UPSs. These were stepped wave output. An APC model was OK, but others gave an output, at low loads with a fully charged battery, with a peak voltage exceeding safe limits. Even though the RMS voltage was in tolerance.

Nobody seems to make anything ’just right’ for this job. A DC UPS won’t cut it, as the router has an internal PSU and the switches and ONT use different DC input voltages. Yet, with the impending shutdown of the PSTN, the demand must be rising

[AVGresponding]

What I want from my UPS is a fairly long run time at a fairly low power draw - say 1 hour at 75VA load. This to keep up network router, switches, ONT, CCTV, automation controller, and VoIP phones. Short outages are common here, we had one yesterday for 51 seconds. Longer outages are less common, but we had two in 2024.

I use a grossly over-rated, true sine wave UPS from HP I bought as a refurb off eBay. A new set of batteries gave me 45 minutes. After just over two years, it’s down to 35 minutes. It has an internal fan, and it’s clear from its operation the UPS does a maintenance charge every month or so.

This followed unsatisfactory experience with a variety of consumer UPSs. These were stepped wave output. An APC model was OK, but others gave an output, at low loads with a fully charged battery, with a peak voltage exceeding safe limits. Even though the RMS voltage was in tolerance.

Nobody seems to make anything ’just right’ for this job. A DC UPS won’t cut it, as the router has an internal PSU and the switches and ONT use different DC input voltages. Yet, with the impending shutdown of the PSTN, the demand must be rising

I assume by "maintenance charge" you mean an equalisation charge? This is very much a blunt instrument approach, and a bit of a holdover from the days of wet Plante cell battery strings. SLA's really don't like that much, and doing it monthly seems a bit often, to me.

Consumer UPS's are pretty much universally crap; the small enterprise unit you're using is probably as good as you'll get without spending serious coin. You might improve battery life by outboarding them, it's quite common for units like that to have an Anderson connector for that purpose

[mikeselectricstuff]

Used commercial-quality ones are often available cheaply on ebay, just needing new batteries. These will typically have automatic  periodic battery testing to give advanced notice of faiures

[nfmax]

Yes. I paid about £75, IIRC, for a refurb unit with new batteries.

[nctnico]

4 or 5 years battery life is a better life than others are saying in this thread, but it still pretty poor for a battery that isn't really being worked very much. Car batteries with 5 year warranties are common, and they are being guaranteed to achieve that life with the abuse they get doing stop start in traffic, and especially starting from cold on a freezing day.

I don't have actual data but my feeling is that a car battery is grossly oversized for its purpose so it can deteriorate a lot before giving problems. I only changed a few batteries on cars we owned and in such cases I fit the biggest (highest capacity) one which will physically fit and after that I never need to change it again (which means >8 years of use). Other cars ran just fine for such periods without needing a new battery.

[Siwastaja]

I don't have actual data but my feeling is that a car battery is grossly oversized for its purpose so it can deteriorate a lot before giving problems.

That's exactly what it is; car starter battery is a perfect example of case where capacity is irrelevant and battery is chosen for other parameters, namely cold cranking current (basically, low ESR), and manufacturing low enough ESR without the capacity would be difficult, so you get the extra capacity "for free". Car starter battery basically cycles between 95% and 100% SoC, and in some rare cases can go below 80%, for example if you have issues with fuel supply and need to crank for longer.

Deterioration unfortunately not only reduces capacity (that would not matter) but also increase ESR (decrease current capability).

[madires]

You can get SLAs optimized for UPS usage which last a bit longer, but are also a bit more expensive. IIRC, Panasonic's UPS SLAs last about 7 years, while the standard ones are fine for up to 5 years.

[grumpydoc]

That's exactly what it is; car starter battery is a perfect example of case where capacity is irrelevant and battery is chosen for other parameters, namely cold cranking current (basically, low ESR), and manufacturing low enough ESR without the capacity would be difficult, so you get the extra capacity "for free". Car starter battery basically cycles between 95% and 100% SoC, and in some rare cases can go below 80%, for example if you have issues with fuel supply and need to crank for longer.

Deterioration unfortunately not only reduces capacity (that would not matter) but also increase ESR (decrease current capability).

Capacity is (somewhat) more relevant these days because the battery not only has to start the car but keep the electronics powered up when the vehicle is stationary - my car has cameras front & rear and maintains an internet connection when not moving. I'm not totally wild about the latter and would not miss its main utility of locking/unlocking the car with my phone but its not optional on modern vehicles.

No one wants to buy a car which will have a flat battery after sitting in an airport car park for a week.

[grumpydoc]

What I want from my UPS is a fairly long run time at a fairly low power draw - say 1 hour at 75VA load. This to keep up network router, switches, ONT, CCTV, automation controller, and VoIP phones. Short outages are common here, we had one yesterday for 51 seconds. Longer outages are less common, but we had two in 2024.

I use a grossly over-rated, true sine wave UPS from HP I bought as a refurb off eBay. A new set of batteries gave me 45 minutes. After just over two years, it’s down to 35 minutes. It has an internal fan, and it’s clear from its operation the UPS does a maintenance charge every month or so.

This followed unsatisfactory experience with a variety of consumer UPSs. These were stepped wave output. An APC model was OK, but others gave an output, at low loads with a fully charged battery, with a peak voltage exceeding safe limits. Even though the RMS voltage was in tolerance.

Nobody seems to make anything ’just right’ for this job. A DC UPS won’t cut it, as the router has an internal PSU and the switches and ONT use different DC input voltages. Yet, with the impending shutdown of the PSTN, the demand must be rising

Sounds like you live near me - unfortunately consumer grade and SOHO units are really only designed to allow graceful shutdown, running for an hour or more is not in their job spec.

[ejeffrey]

What I want from my UPS is a fairly long run time at a fairly low power draw - say 1 hour at 75VA load. This to keep up network router, switches, ONT, CCTV, automation controller, and VoIP phones. Short outages are common here, we had one yesterday for 51 seconds. Longer outages are less common, but we had two in 2024.

Nobody seems to make anything ’just right’ for this job. A DC UPS won’t cut it, as the router has an internal PSU and the switches and ONT use different DC input voltages. Yet, with the impending shutdown of the PSTN, the demand must be rising

We have had a lot of scheduled all day power outages as they are replacing a bunch of utility poles and burying other wires that feed our neighborhood.  So I have a DIY solution.

 I have a small APC UPS that lasts 10 minutes and I have an RV true sine wave inverter with a 100 Ah LiFePO4 battery.  It runs my fiber ONT, router, switch, and three PoE access points for about 24 hours.  When (or before) the power goes out I manually unplug the small UPS and plug it into the inverter.  After the power resumes I move the plug back to the wall and manually recharge the large battery.  The self discharge rate is so low that I just leave it disconnected after reaching full charge.

It's a bit inconvenient but it works and has gotten a fair bit of use.  My wife works remotely from home and our scheduled outages are usually during work hours so this lets her work as long as her laptop battery lasts.

[peter-h]

Nice

I deal with this by having a 3.5kW Briggs & Stratton generator (which gets started up once a year, and has a 15hr petrol tank), but I don't have an automatic seamless switchover facility Those exist, at serious money levels. OTOH, with a UPS, the several-minute changeover is OK.

[coppice]

4 or 5 years battery life is a better life than others are saying in this thread, but it still pretty poor for a battery that isn't really being worked very much. Car batteries with 5 year warranties are common, and they are being guaranteed to achieve that life with the abuse they get doing stop start in traffic, and especially starting from cold on a freezing day.

I don't have actual data but my feeling is that a car battery is grossly oversized for its purpose so it can deteriorate a lot before giving problems. I only changed a few batteries on cars we owned and in such cases I fit the biggest (highest capacity) one which will physically fit and after that I never need to change it again (which means >8 years of use). Other cars ran just fine for such periods without needing a new battery.

Most people with a UPS only drain the battery and recharge it a few times in the UPS's life. I don't see how oversizing would be relevant unless the application cycles the battery a lot. A car battery needs the grunt to turn the starter on a cold day, so its chosen around that. Modern cars with stop-start systems require special grades of battery to tolerate all those starting surges, even though most of those starts are with an already warm engine that will spin up easily.

[peter-h]

Also, IME, that auto start feature gets disabled if the battery is not pretty fully charged.

A car battery would be good for a UPS because they are big and cheap, but not ideal due to easy acid spillage, and almost no UPSs are 12V. That's why I linked to that 24V aircraft battery.

[nctnico]

Also, IME, that auto start feature gets disabled if the battery is not pretty fully charged.

A car battery would be good for a UPS because they are big and cheap, but not ideal due to easy acid spillage, and almost no UPSs are 12V. That's why I linked to that 24V aircraft battery.

Modern day car batteries are sealed as well. I don't think I have seen a car battery with plugs you can open to add water in the last 20 years. And you can put two 12V batteries in series to get 24 V (this is typically done in lorries). In the end a 24V battery is just more cells in series.

[coppice]

Also, IME, that auto start feature gets disabled if the battery is not pretty fully charged.

Stop start only really matters when driving around town. However, my experience is if you only drive around down the software doesn't accept that the car is in a fit state to go into stop start mode, so it is never used.

[pqass]

A car battery would be good for a UPS because they are big and cheap, but not ideal due to easy acid spillage, and almost no UPSs are 12V. That's why I linked to that 24V aircraft battery.

No, a car battery would NOT be good for UPS duty.  You'll kill it after a few cycles. 
However, look for "marine deep cycle" instead.  They come in wet or AGM and are made for high-draw accessories like trolling motors. 
And don't discharge below 20%.

[jc101]

I have a Riello Sentinel Pro UPS in my office/workshop. It's a full double conversion unit, but I run it in ECO mode so it doesn't turn the fan on. USB connection to the NAS for auto shutdown and a network card in it for everything else. Been in there coming up to 5 years now with no issues. It has batteries and some supercaps according to the blurb. I might update to  a higher power one at some point when I do a re-org.

I have an APC Smart UPS in the house, to keep the FTTP box and router alive. Both are 1500VA, but the Riello was much cheaper than the APC for a higher spec.

[paul@yahrprobert.com]

After reading this thread I decided to check my Cyberpower UPS with a 1 year old battery in it.  Its already completely dead!
So I went shopping and found the Goldenmate line of LiFePO4 UPS's. Amazon has them in the US,
a 600 W unit for about $170.  Ordered one, we'll see....

[m k]

APC Smart UPS 1000 from 2011, decommissioned from priority use around 2020.
Current load around 11%, no issues.

Autumn '15 batteries changed summer '16, double something APC.
Was practically no batteries since who knows.
(quite a priority)
'16 set still running, periodic test says good.
And that set stays in until failure.