back to article Rant launches Eric Raymond's next project: Open-source the UPS

In February, developer and open source software advocate Eric S Raymond ranted that the Uninterruptible Power Supply market was overdue for open source disruption, and touched so many nerves around the world that the rant has become a project. Last week, ESR opened up the work-in-progress on GitLab: the Upside project is …

  1. Charles 9

    It has been my impression that UPS's fail to "off" precisely because they fail to "pass-through", which in their typical actual use case, means they pass through nothing since they're supposed to operate when the mains is out.

    As for power sources, what's wrong with lead-acid? Why not allow us to rig ubiquitous car batteries for the job rather than those tiny "sealed" versions? A case where KISS simpler would be preferred.

    1. Mark 85

      Car batteries? Why not have an option for a truck (semi type) battery since they're bigger and have more capacity? So the electronics could once acquired, could be used with the user's choice of batteries including some the monster (and usually ungodly priced) UPS batteries sold for commercial use. The later I think, use a gel electrolyte as opposed to the "normal" acid. The unit itself shouldn't care what kind of battery it is as long as it knows the input voltage/current from the battery and what it's monitoring.

      1. Ole Juul

        "Car batteries? Why not have an option for a truck (semi type) battery since they're bigger and have more capacity? "

        We almost do already. For my main desktop system I took an ancient discarded APC 300W UPS. So old that I can't even find the manual on line any more. This can run my fairly powerful computer, a big monitor, a router, and a wireless radio. To make this workable, I soldered in some heavy wire to connect a deep cycle marine battery instead of the puny internal one that would have been original.

        As it turned out a few months later there was an emergency with power off for a full 4 hours. I had no time to turn anything off, but when I returned everything was still happily humming along. This is a fairly neat setup and the 110Ah battery was the only expense and cost well under $200.

        As for lead acid, one can argue that it is the best choice because it lasts a very long time and the recycling is very well honed from 150 years of development. Regarding a problem with size, (as someone mentioned above) anyone who does not have 3/4 square feet to spare for a slightly larger than the average car battery is probably homeless.

        1. Jason Bloomberg Silver badge
          Flame

          I took an ancient discarded APC 300W UPS ... soldered in some heavy wire to connect a deep cycle marine battery

          When heatsinks and other hardware have been designed only to support the limited backup time the recommended battery gives, extending the backup time with a higher capacity battery may cause problems with that hardware, overheating, fire, or other failures.

          The same too for charging circuits. If designed only to bring a smaller battery back to full charge they may overheat or fail when charging for a much longer time which a larger capacity battery will require.

          Make sure you know what you are doing and have done a risk assessment in case you need to talk to an insurance assessor later.

        2. Anonymous Coward
          Anonymous Coward

          > I soldered in some heavy wire to connect a deep cycle marine battery

          Absolutely agree with the views in this, but there is a practical issue, which is that deep cycle wet lead acid batteries really need to gas from time to time. We're off grid, where admittedly the batteries are worked harder than is expected with a UPS. and there are a lot more of them than in a simple 300VA UPS, but when the batteries are charging the whiff can be a rather strong clunk of the egg-and-baked-beans-with-6-pints nature. Theoretically not too good for you, so those who operate on the assumption that any risk is unacceptable would find the wet lead acid option good grounds to exert their power.

          1. handleoclast

            lead acid and gassing

            deep cycle wet lead acid batteries really need to gas from time to time

            The gas is from the electrolysis of water, giving hydrogen and oxygen, and usually happens in overcharging conditions (such as the continuous trickle charge in most UPSs when the batteries are already full). Hydrogen can lead to embrittlement in a variety of materials, which may cause unwanted problems.

            In the early days of submarines it was found that the filament lamps in the battery room failed far more frequently than expected. Hydrogen from the batteries diffused rather easily through the glass envelope and caused embrittlement of the tungsten filament, making it more prone to failure from vibration. That glass envelope is sufficient to retain an inert gas in modern filament lamps, or even maintain the vacuum of the early filament lamps, but it won't stop hydrogen.

            Of course, most filament lamps are illegal these days, so you probably won't have that problem. But it's probably a good idea to ensure any area with lead-acid batteries on charge is reasonably well ventilated. I don't know if anyone has done any research on the effects of hydrogen on LEDs, for example, but I suspect it won't do them any good in the long term.

            1. onefang

              Re: lead acid and gassing

              "The gas is from the electrolysis of water, giving hydrogen and oxygen, "

              Set fire to it, change it back to water, which tends to not go through glass.

          2. Piscivore

            H2S is evolved from charging failing lead acid batteries that have plate sulphation (US sulfation). A healthy battery will generate more H2 and O2 that are harmless unless they see a spark in which case the ratio of each is in perfect explosive proportion. As my crew discovered when they decided to spray the top of a charging 150Ah battery with the residue from an angle grinder cutting steel. BOOOM!

        3. Anonymous Coward
          Anonymous Coward

          > APC 300W UPS

          APC, nice UPS those, at least for consumer use. Never seen one fail.

          My ex has had one for nearly twenty years now, she's replaced the battery two or three times I think and that's it. It is a bog standard motorcycle battery too, which she replaces herself.

        4. Nobhobbor

          Deep cycle marine battery is clever, but do they have Electric rental bikes or scooters any where near by because wouldn’t you rather use one or more of those batteries? Don’t lead acid off gas sulfuric acid fumes when charging and produce acid powder at the cathode or anode? I know that after handling a car battery, my clothes developed holes the next time I washed them, I can’t imagine purposely choosing to have that situation sitting right next to my computer if there were better options for around the same price. LiON batteries are recyclable as I understand it

      2. Stoneshop

        Car batteries? Why not have an option for a truck (semi type) battery since they're bigger and have more capacity? So the electronics could once acquired, could be used with the user's choice of batteries

        At the very least you would want to adapt the charging circuitry to the battery capacity: lead-acid batteries want to be charged at 0.1C (their rated capacity in Ah), so it's not quite one size fits all.

    2. Duncan Macdonald

      The big problem with lead acid batteries is that they are BIG (and heavy). (A Yuasa HSB019 12 volt 100 Ah car battery weighs 23.4kg and is 353x175x190mm). Not many home users would consider such a large battery acceptable. For commercial use, large UPS systems normally use non-sealed lead acid batteries (as they are normally in places with maintenance staff), smaller UPS systems use sealed lead acid batteries to reduce the chance of problems such as spillage and acid vapor release.

      Lithium rechargeable cells have a higher power density - but cost much more - to equal the HSB019 battery mentioned above would take about 120 high capacity 18650 cells with a total price several times that of the lead acid battery.

      Given the low power density of lead acid and the high cost of Lithium-ion, most small UPS systems come with small lead acid batteries that are only good for 5 minutes or so at full load. (Big commercial UPS systems are usually sized to cover the period between a mains failure and the backup generator starting up and taking the load. They often do NOT supply the cooling systems but rely on thermal inertia and a fast backup generator startup so a very long UPS runtime is not a desired feature in such a configuration.)

      1. Adam 52 Silver badge

        "to equal the HSB019 battery mentioned above would take about 120 high capacity 18650 cells"

        You don't need the same capacity in a battery that can be deep discharged without suffering damage. Still much more expensive, I grant you, but not as much as it appears at first glance.

    3. Christian Berger

      Well I've actually seen UPSes fail to "off" even when the mains is perfectly normal.

      Those sealed versions are actually far easier to handle as you won't have to worry about acid spills.

      1. Anonymous Coward
        Anonymous Coward

        "Those sealed versions are actually far easier to handle as you won't have to worry about acid spills."

        But what would you be doing with your UPS to cause an acid spill?

    4. DougMac

      >> It has been my impression that UPS's fail to "off"

      The problem I've encountered is that for most of the small-ish UPSs, that when the batteries go past their useful life, the UPS starts cycling the power, even if the wall power has been steady on. APC is particularly nasty about this for certain models.

      Battery past useful life is almost always only a light on the front, I can't tell you the last time I've looked at the front of my home UPS.

      >> Why not allow us to rig ubiquitous car batteries..

      It was covered later to be mostly size. Your UPS would be quite large and heavy. As it is, most small-ish UPSs now use motorcycle/lawn mower Lead-Acid batteries. So, same technology, just smaller package. Less capacity.

      The UPSs I'm mainly interested in though have their own rooms. : - )

      And they do typically use deep-cycle marine batteries. Just lots and lots of them.

      1. Anonymous Coward
        Anonymous Coward

        "Battery past useful life is almost always only a light on the front,"

        Not in my experience. When the battery fails the auto test, you can see it from the management tool, and even have message sent.

        I'm fine with a small, cheap, easy to replace battery for a SOHO UPS as long as it can shutdown machines cleanly before discharging. If you need to work fully through a blackout you have needs that should be addressed by a specialist - not a DIY UPS...

      2. Charles 9

        "The problem I've encountered is that for most of the small-ish UPSs, that when the batteries go past their useful life, the UPS starts cycling the power, even if the wall power has been steady on. APC is particularly nasty about this for certain models."

        Most UPS will cycle the batteries periodically to test them. They keep the mains ready on standby in case the test fails, but the idea is that the batteries need an occasional "stress test" to help cycle them and to make sure they still work. It's when this test fails that the UPS reports the problem.

        My issue is that UPS management software sometimes acts funny. I had it completely drain out an UPS once for no apparent reason which is why I don't hook up an UPS's communication channel anymore and simply go by its built-in display.

    5. Alan Brown Silver badge

      "what's wrong with lead-acid?"

      Deep discharge tolerance.

      Lead Acid is cheap/cost-effective and will float forever, but discharge them down to 30% a few times and they're toast.

      Did I mention they're cheap? The cheaper solution than using better chemistry is usually to just use a bigger lead acid bank. There's a reason telcos still use them.

    6. Alan Brown Silver badge

      "rig ubiquitous car batteries"

      Lead-acid isn't lead-acid isn't lead-acid.

      Car batteries are designed to be lightweight and provide _enormous_ discharge currents for a short duration (seconds). The lead plate grids on them are relatively fragile and if you deep-discharge them, or use them as leisure batteries (or UPS batteries), they die in short order. (The plates will sulphate up quickly if left discharged below 80%)

      Likewise if you use leisure (caravan) batteries to start your car, you'll get one shot at doing so, then start wondering why the plates are fragmented all over the bottom of the case. These are a better choice for UPS use though.

      Sealed mat lead acid batteries are a good compromise for most UPS purposes as they don't outgas unless heavily abused.

      If you need large capacity, then deep discharge Traction cells are the way to go, but they're not small or light or cheap and you DON'T want them in the same room as the rest of your power electronics (If they boil, every piece of your wiring insulation goes black. I've seen this after a lightning strike blew up a charger on a remote site) and ideally they go in a completely isolated room with good temperature control, nothing that can spark and decent active+passive ventilation.

      In all cases, the float voltage is critical to a few millivolts (and different for each design) Get it wrong and you'll eventually kill the cells whilst spreading acid fumes all over the place.

      Anyone who tries to sell you a car battery for standby power applications doesn't have the first clue about what they're selling (or is intending to make lots of repeat battery sales).

      It's one of the more annoying things I saw whilst spending time in Outer Bumfuckistan, (you could ONLY get car batteries for standby power purposes) along with selling WD40 as "oil" to people who were wondering why everything kept clogging up with gritty shit and needing a constant dousing with more WD40, but still ended up breaking.

      1. Ole Juul

        It's nice to see that there are indeed a few technical people here. However it is amazing how many armchair critics there are too. The sky is not going to fall with this kind of project. It's not rocket science, it's electrical engineering 101, and those of us with basic skills will not overlook the problems that some so gleefully predict while wagging their nerdy little finger. Regarding my own project, there is no outgassing, no acid spills, ( I mean really!) and the charging circuit handles the deep cycle battery just fine with an initial 13.7v followed by a nice sedate 13.4v float. The unit does get hot under full load but nothing it can't handle, and this will not happen often as power outages here tend to be in the range of a few seconds and only once every year or two go into the hour range. I put some ventilation holes in the box which seems to keep heat under control. The overall concept is to think things through and be sensitive to what it's actually doing. Assuming that other people are unable to do such things is not an impressive trait to display on a technical forum.

  2. Ian Mason

    Lack of clue

    Yeah, the LiPO battery thing is stupid. Lots of little cells that you need charge balancing circuitry, and cell protection circuitry for instead of a single, big sealed lead acid battery? Yeah, that'll keep down your bill of materials - not!

    The problem is that this seems to be being driven by software weanies and there's no evidence of any power electronics folk's input in the specifications. The fact that they find managing a bypass switch challenging is evidence of that:

    "This may have to be a mechanical switch (possibly flipped by a solenoid, but not a traditional spring-loaded relay that resets to a specific state when power is removed). Implementing the option in software seems hard, especially if this UPS is supposed to protect against intermittent voltages and overvoltages as well (i.e. cases where bypass is the wrong response and there’s no software running due to lack of power)."

    i.e. They have never heard of latching contactors or solid state relays (i.e. triacs).

    Further evidence for the supposition that this is just a bunch of programmers are the facts that the current specification has no, zero, zip, mention of the options for power conversion technology to be used, or the type of output waveform to be supported, or any of the basic questions that an Electronic Engineer would ask first. Like, they've said 300W for 15 minutes (based on measurements apparently) but no mention is made of a figure in VA - probably because they have zero clue what a "power factor" is.

    They need to get some power electronics engineering experience on board very fast or this is going to become another "Arduino Maker Project" designed by programmers. Do that on mains attached power electronics and watch your house burn down.

    1. mtnrbq78

      Re: Lack of clue

      I was reading and agreeing with you, but then checked out the project. On the front page...

      We welcome contributors: people with interest in UPSes who have expertise in battery technology, power-switching electronics, writing device-control firmware, relevant standards such as USB and the DMTF battery-management profile.

      So they've got a clue they don't have all the clues.... yet

      1. Martin Summers Silver badge

        Re: Lack of clue

        "So they've got a clue they don't have all the clues.... yet"

        Well maybe Ian Mason might want to contribute there and point them in the right direction. All they're guilty of is saying that UPS's suck in their opinion and they want them to be better. If people contributed to projects like this rather than complain about them doing it wrong or not having a clue then they might get somewhere.

      2. Ian Mason

        Re: Lack of clue

        Granted, but if you were going to build a house you wouldn't pick out the specific joists to use, and say "We'll leave the rest to the architect/civil engineer/builder, we must get one of those."

        BUT they have already chosen a processor board they would like to use (an A20-OLinuXino-LIME2). In practice this sort of thing doesn't use a Linux based commercial board such as they have picked, but a microcontroller picked to have all the right interfaces for controlling the power electronics. Heck, there are microcontrollers dedicated to power electronics control with much of the control circuitry already there for the taking. That's where you start, picking which of the dozens of MCUs in that space best fits your first iteration of power electronics design, not "Let's have something that runs Linux". That Olimex board costs 45 euro, an appropriate MCU would likely be sub $1 in quantity, and the whole bill of materials for the power control board (including MCU, display, some buttons and a USB connector) ought to come to less than 45 euros. Carry on designing like that and you'll have a 500 euro bill of materials before you've added a case and a battery.

        To be fair, the board is listed on their 'strawman' list, but they're so far into Dunning Kruger that they don't realise how far off the mark they are. Very typical programmer hubris (Says this programmer who just happens to know rather a lot more about electronics than the average programmer).

        1. Justin Clift

          Re: Lack of clue

          > BUT they have already chosen a processor board they would like to use (an A20-OLinuXino-LIME2). ...

          Sure. They've definitely jumped onto the bits they feel comfortable with, and obviously have not much clue with the rest.

          However, it does sound like they'd be open to constructive pointers telling them they pieces they need to clue up on. You obviously have depth in areas they lack but need.

          As an idea, maybe point out the electrical bits they need in order to not completely burn their own houses down ;), and see if they manage to get something useful for people happening after all.

          Their skills are stronger in software, so they might turn out to have a decent software side to things anyway. :)

          1. Phil O'Sophical Silver badge

            Re: Lack of clue

            They also seem to be asuming that there's only one set of requirements for a UPS. I have three at home, for my server, desktop & TV/satellite system. Well over 99% of the outages I see are < 5 seconds due mainly to summer thunderstorms, or winter snow-on-cables. All the UPSes happily coast through those, no need for anything fancy.

            The one I use to power the server is over-specified, and does link to a NUT monitor process. I get an email when the power goes out, and if it stays out for a long period the server will do a clean shutdown. So far I've had to replace the (sealed lead-acid) batteries once, when the originals lost too much capacity.

            The fancy one that Raymond is suggesting would be OK for my server, but probably over-priced. It would certainly be overkill for the other two.

            To be honest, the biggest criticism I have of the one I use for my server is the amount of power it consumes itself when in standby (i.e. normal mains operation). That will be my primary concern when I come to replace it.

        2. fidodogbreath

          Re: Lack of clue

          In practice this sort of thing doesn't use a Linux based commercial board such as they have picked

          "I suppose it is tempting, if the only tool you have is a hammer, to treat everything as if it were a nail."

          -- Abraham Maslow, The Psychology of Science, 1966

          1. Stoneshop
            Devil

            Re: Lack of clue

            "I suppose it is tempting, if the only tool you have is a hammer, to treat everything as if it were a nail."

            -- Abraham Maslow, The Psychology of Science, 1966

            I suppose it is tempting, if the only nail you have is ESR, to treat everything as if it were a hammer.

            Me, 2018.

      3. Aitor 1

        Re: Lack of clue

        I might contribute to the project as I have the electronics skills, and I am also fed up with UPSs, and I know I am not alone.

        BUT the problem is that designing the right electronics for modern computers is difficult.. we basically need decent electronics.. and that is expensive.

        Read these papers from ABB:

        https://library.e.abb.com/public/58f96070ebc547799c3cae35025f1e85/White_Paper_Power_factor_UPS.pdf

        https://library.e.abb.com/public/525c273041687f0085257bf3005bbf28/ABB-429-WPO_CyberexDynamicInrushRestraint.pdf

        I have seen data centers have fires for not taking these things into account...

    2. Christian Berger

      Re: Lack of clue

      Actually are LiPO batteries even suitable for standby use? I mean an UPS battery usually sits around fully charged for years at a time, until maybe eventually it's used. That's not particularly healthy for the cells used usually in phones or laptops.

      1. Anonymous Coward
        Anonymous Coward

        Re: Lack of clue

        Actually are LiPO batteries even suitable for standby use? I mean an UPS battery usually sits around fully charged for years at a time, until maybe eventually it's used. That's not particularly healthy for the cells used usually in phones or laptops.

        A modest over-provisioning and planned, automated power drawdown purely for battery conditioning would resolve that. The other important thing about LiPO batteries is that there's a lot of noise at the moment around PV+battery which could provide suitable batteries being made at economic levels of production for this project.

      2. ilmari

        Re: Lack of clue

        It's somewhat alright if the battery is maintained at 80% full, alows down the wear and tear.

      3. Jan 0 Silver badge

        Re: Lack of clue

        RTFA, LiPO wasn't mentioned!

    3. Anonymous Coward
      Anonymous Coward

      Re: Lack of clue

      Well said Ian.

      It appears that they have everything arse about face. They should start with the hardware that does the job and then go on to look at how to control it.

      Without full hardware specifications being well defined everything they are doing is whistling in the dark.

      1. Flakk
        Joke

        Re: Lack of clue

        They should start with the hardware that does the job and then go on to look at how to control it.

        That paradigm doesn't seem very DevOpsy.

    4. cantanko

      Re: Lack of clue

      LiPo wasn't mentioned - rather LiFePO4 - Lithium Iron Phosphate. Used for traction batteries in many EV conversions as you can beat on them all day and rather than have a LiPo temper tantrum, they just sit in a corner and sulk (and bulge a bit). Very different technologies...

    5. Bronek Kozicki

      Re: Lack of reading skills

      Not LiPO batteries. LiFEPO4 - which is a different beast. You make many good points which are probably applicable, but c'mon, confusing these two?

  3. katrinab Silver badge

    Can’t we have whatever is used in laptops as a starting point? They would be a lot lighter, take up a lot less space, and running time would be measured in hours rather than minutes.

    1. Richard 12 Silver badge

      Laptops generally produce 12VDC or 5VDC, and consume around 20VDC.

      As they don't produce or consume the 110VAC or 230VAC of a UPS, they are far simpler.

      1. Adam 52 Silver badge

        Sounds like a good approach then. Work on a external specification for ELV power and build something simpler and better.

        Rather than try to reinvent the wheel, badly.

      2. katrinab Silver badge

        The battery in the UPS puts out some sort of DC voltage. That goes through an inverter to convert it to 230/110V AC. It then goes through a power supply to convert it to 12V DC + 5V DC.

        Would it not be better to connect the UPS to the other side of the power supply?

        1. SImon Hobson Bronze badge

          Would it not be better to connect the UPS to the other side of the power supply?

          At what voltage ? The 12V some of my equipment uses ? The 5V some of it uses ? The 7.5V one bit of kit uses ? The 24V some uses ? The 48V some uses ?

          The theory is sound, IF all manufacturers made kit that would run off one standard DC voltage, AND all UPSs used that voltage of battery - but they don't and they don't (the two UPSs I have at home use 36V and 24V nominal batteries).

          It's been an old debate about data centre power - AC or DC with arguments over which is better/more efficient. A lot of equipment runs off 48V nominal (actually 50-something V as it's float charged batteries) as used in the telecoms world - I suspect things might be changing, but traditionally a telephone exchange has a big battery room, some 'kin big busbars to distribute the power, and everything runs off 48V. But use (say) 48V rather than 240V and you (roughly) increase resistive losses in the cables by a factor of 25 - losses are proportional to current squared, and you're pulling 5 times the current. Ignoring losses, your cabling needs to be a lot bigger to handle the current.

          But even if you distribute DC, there's still a power conversion down to the voltages used internally.

          As an aside, while I was at university <cough> decades ago, I was in the computer club and got involved in running a stand at the freshers fair. We were told that we could use mains power (they even switched off the socket circuits) - but they did agree to us using batteries. I had a look inside my monitor and found it used 12V internally, so soldered a couple of wires in, and borrowed my car battery so we could have a working system (this was in the days when computer power supplies were also simple and the one we had could be made to run off 12V as well with a little fiddling). Not quite the sort of battery they probably had in mind ;-)

          1. katrinab Silver badge

            That's why I am suggesting that the server manufacturer should provide a built-in UPS function. It would be similar in design to laptop battery / power supplies, except obviously with server grade components, power requirements will be higher, and therefore everything will have to be bigger to cope with it - bigger than a laptop battery / charger, but probably still smaller than the several kg of UPS battery.

            1. Stoneshop

              Specs

              That's why I am suggesting that the server manufacturer should provide a built-in UPS function.

              Which would require sufficient room inside the server case for the batteries. And what runtime (i.e. battery capacity -> battery size) would you then aim for?

              Never mind that rackmount servers tend to be rather stuffed.

              1. katrinab Silver badge

                Re: Specs

                Have the batteries located where the PSU is currently located.

                Have the PSU located where the UPS is currently located, with a low voltage cable leading to the computer - something a bit bigger than a typical laptop wallwart, due to the higher power requirements.

                Or you could have both outside the case, and a low voltage cable to the computer.

                1. Stoneshop

                  Re: Specs

                  Have the batteries located where the PSU is currently located.

                  You can't use all of the PSU's space for batteries, as you still have to convert the battery voltage to the other voltages used in the system for which you need about one-third of the space a standard PSU takes. For low-power systems (Mini-ITX) you can use PicoPSUs of the appropriate rating which run off 12..18VDC, so easy enough to run off a battery with a charger attached. But for larger systems, especially with beefier video cards, those won't cut it

                  I'm not sure if for those larger systems the reduction in conversion losses is worth it regarding the various stuff you need to do it this way.

            2. AJ MacLeod

              These (servers with built in UPS) have certainly existed in the dim and distant past - Apricot made some (the one I'm thinking of was MCA so that dates it.)

              1. Anonymous Coward
                Anonymous Coward

                @AJ MacLeod

                Indeed, back in those days you could even buy the Emerson AccuCard that plugged into an expansion slot on an XT/AT/386. It had a built in battery and with some software was designed to backup to floppy.

            3. Alan Brown Silver badge

              "That's why I am suggesting that the server manufacturer should provide a built-in UPS function. "

              Funnily enough, many manufacturers DO provide that option, however it doesn't really matter where the battery is, it still has to be large enough to ride out the outage and you get to pay a premium for such hardware well over and above the cost of using a UPS.

              Of course if you're serious about your UPSes, then you could use a flywheel-based 300kW modular system backed up by multiple diesel generators. They're vastly more reliable than pissing around with battery-based systems (We have about 750kW of such setup, I know of a couple of sites with around 5MW)

              There are parts of the world where unreliable power is the norm. Looking at how the issues are solved there would be a lot more educational than gasbagging about how an idealised UPS would work.

              IE: Battery based systems are intended to only be used for as long as it takes to get the generators running or to safely shut down - and just like generators, battery chemistry/testing/monitoring/alarming is quite well understood already, so reinventing the wheel isn't necessary.

              Just because cheap-arse UPSes don't have these capabilities doesn't mean they're not available.

              Incidentally, one of the most common killers of UPSes I've seen has been dirty power from the mains. Just because volts have come back doesn't mean they'll stay back, or that there won't be a 4000V spike coming in. Even in "first world" countries the standards for electrical supply cleanliless are eye-openingly poor. A couple minutes delay before reconnecting mains is a healthy safety precaution and if it shows the slightest instability in that period, pushing the boat out to 30 minutes or more is sensible.

          2. Roland6 Silver badge

            >At what voltage ? The 12V some of my equipment uses ? The 5V some of it uses ? The 7.5V one bit of kit uses ? The 24V some uses ? The 48V some uses ?

            Doesn't really matter, just create a standard, just as we have with USB charging and universal laptop PSUs that use 'intelligent' adaptors to modify the output voltage!

            The point is that by having the UPS delivering the standard ATX connector motherboard voltages, you can avoid the double conversion we currently have where the UPS delivers 240v to the computers PSU for it to convert into ATX voltages.

            The ATX interface standard would satisfy the vast majority of desktop/server use cases. Yes, some investigation will be necessary to determine if the UPS should deliver all the ATX voltages (12v, 5v & 3.3v) or wither it should simply supply 12v and let the computer's PSU sort the rest out.

            1. ChrisC Silver badge

              Having a UPS that replicates the ATX interface is an interesting idea, and could work nicely for headless systems (such as servers) where you only need to maintain power to the PC itself. But for SOHO type applications, I'm not sure too many users would be happy if their monitor suddenly went dark each time the UPS kicked in even for just a few seconds...

              1. Roland6 Silver badge

                >Having a UPS that replicates the ATX interface is an interesting idea

                I wasn't suggesting replicating the ATC interface, just providing the voltages the ATX interface supports.

                Whilst the 12v supply is probably the easiest, I'm not so certain about the 5v and 3.3v rails (ie. cable and distance constraints on getting a usable supply out of a 1~1.5m cable).

                > I'm not sure too many users would be happy if their monitor suddenly went dark each time the UPS kicked in even for just a few seconds...

                Never seen that issue, for the 15+ years I've run the home office PC off the UPS, but then as UPS's currently output 240v...

                However, yes currently most monitors do directly take a 240v feed and so just like PC's some redesign would be necessary to provide an external low voltage socket that would enable the transformer to be bypassed.

                I think from the conversations here, there are several broad groups of UPS: Home ie. single PC (plus DSL router), small office/single rack, datacenter. Splitting the problem into domains helps to simplify things, because you can then solve a simpler problem set before worrying about the complexity of the general case. Also if the idea of home batteries, as extolled by Telsa and their Powerwall gets taken up, then being able to plug equipment directly into a 12v supply might be helpful.

                1. Alan Brown Silver badge

                  "Also if the idea of home batteries, as extolled by Telsa and their Powerwall gets taken up,"

                  Tesla's powerwall runs at upwards of 144V. You do NOT want to mess with HVDC as it is extremely unforgiving and you can baste in your own juices for quite some time before you expire.

                  1. Roland6 Silver badge

                    >Tesla's powerwall runs at upwards of 144V ... HVDC

                    Agreed, however, I was hand waving at avoiding the DC/AC/DC conversion- suggesting it might be more efficient to directly convert HVDC to LVDC, particularly if it enables the battery to provide power for longer...

              2. Alan Brown Silver badge

                "Having a UPS that replicates the ATX interface is an interesting idea, "

                You mean like this?

                https://www.sunpower-uk.com/products/ATX-Power-Supply-with-UPS-Function/eNSP3-450P-S20-H1V/default.htm

                They've been around for years.

                As for the monitor: most LCD monitors run on 12V - guess what this PSU provides?

                1. Roland6 Silver badge

                  @Alan Brown re: ATX-Power-Supply-with-UPS-Function

                  Good find - who stocks these?

                  Only catch seems to be no management interface, only a basic repurposed RS-232C port: Batt_Low, AC_Fail, Shutdown...

                  [ http://www.berger-stromversorgungen.de/verteiler/berger/dokumente/eNSP3_450P_berger_web.pdf ]

                  I can find several manufacturers offering this PSU, but I can't find anyone actually selling individually - I assume the OEMs only deal in bulk orders...

            2. Stoneshop

              ATX connector?

              The point is that by having the UPS delivering the standard ATX connector motherboard voltages, you can avoid the double conversion we currently have where the UPS delivers 240v to the computers PSU for it to convert into ATX voltages.

              So, what about people (such as me) who want to power their server plus a network switch or two? And the DSL modem?

              Also, for an UPS with an ATX connector fitted you would have to install it inside the computer case. Would there be sufficient room for the battery, and a free drive bay where you can put the monitoring panel? There have been PSUs with a battery backup fitted, but they never went past a few minutes of runtime. Save and shutdown gracefully, and that's it. The last time I had a power cut my setup rode out the entire four hours.

              1. Adam 52 Silver badge

                Re: ATX connector?

                "So, what about people (such as me) who want to power their server plus a network switch or two? And the DSL modem?"

                Fairly sure your monitor and network switch don't run off 240v. Mine have external power bricks.

                "with an ATX connector fitted"

                The proposal was ATX voltages, not ATX physical connector.

                1. Stoneshop

                  Re: ATX connector?

                  Fairly sure your monitor and network switch don't run off 240v. Mine have external power bricks.

                  My server is headless, and the network switches do run off 230V.

                  The proposal was ATX voltages, not ATX physical connector.

                  Given the average motherboard power consumption you would need a multipole connector using several pins per voltage not unlike an ATX connector (so why not an actual ATX connector with some auxiliary ones for whatever else you're hooking up), or a connector capable of that 300VA total, one pin per voltage, so rather beefy. And then some converter cable to adapt that to an ATX connector plus again some auxiliary universal ones, as that will be the most common use case. Six of one, half a dozen of the other.

                  1. Roland6 Silver badge

                    Re: ATX connector?

                    >so why not an actual ATX connector with some auxiliary ones for whatever else you're hooking up

                    Because someone is bound to try and connect the ups directly to a motherboard's ATC socket...

                    However, not a totally daft idea, you only need to look in the typical car's electrical system to see similar connectors, just different shapes.

                    >Six of one, half a dozen of the other.

                    Agree, having looked at my home and the entire issue of using solar panels, batteries, low voltage lighting etc. However, I'm sure some balance can be achieved, given that for many people a system capable of running off a 12v 90w power brick is good enough. So there might be some mileage in your idea of having an adaptor capable of delivering the typical (modern) motherboard power requirements...

                    1. Stoneshop

                      Re: ATX connector?

                      Because someone is bound to try and connect the ups directly to a motherboard's ATC socket...

                      And if the UPS output can supply the voltages and currents an ATX mobo requires, what would be the problem?

                      "I wasn't suggesting replicating the ATC interface, just providing the voltages the ATX interface supports."

                      With an approach like that you actually want to be able to hook up such an ATX-UPS directly to the motherboard, minimising cable and connector losses.

                      1. Roland6 Silver badge

                        Re: ATX connector?

                        >And if the UPS output can supply the voltages and currents an ATX mobo requires, what would be the problem?

                        Sorry, I wasn't totally clear. I picked up on your point about additional/auxillary feeds and thought that some might simply take the physical ATX connector and re-assign 'unused' pins to provide the additional feeds rather than use a separate connector...

                        However, if the UPS can fully support the ATX plug compatible interface I would agree with you.

                2. hplasm
                  Meh

                  Re: ATX connector?

                  "Fairly sure your monitor and network switch don't run off 240v. Mine have external power bricks."

                  Some do, some don''t.

              2. Alan Brown Silver badge

                Re: ATX connector?

                "plus a network switch or two? "

                Huawei sell battery-backed switches in their campus range.

                http://www.router-switch.com/s5700-28p-li-24s-bat-p-15898.html

                The datacentre range come with mains or 48V supplies for the same purpose

                "And the DSL modem?"

                I run mine off the same 12V float-charged battery and distribution fuseboard that's running the security cameras+DVR. The monitoring electronics for the charger+battery has its own network interface and is quite generic, talking to the mains power quality monitoring hardware.

                First define the size of the problem you wish to solve, then decide what equipment you need to solve it.

                FWIW, 12V-to-whatever buck-converters up to about 20W are a few dollars each at most.

                None of this needs a "brave new world" solution. It's all been solved before.

              3. Christian Berger

                Re: ATX connector?

                "Also, for an UPS with an ATX connector fitted you would have to install it inside the computer case. Would there be sufficient room for the battery, and a free drive bay where you can put the monitoring panel? "

                Well but then you need more converters. You'd first need to convert your 12V to some voltage your battery needs to charge (e.g. 14V) and then you need converters for all the voltages your ATX connector has. Which is +12V +5V +3.3V -5V -12V.

                That's a lot of converters which need to work at fairly low voltages making them less efficient. It's far easier to have 2 converters. One from 230V AC to 14V DC, and one from 10-14V DC to 230V AC than all those converters you'd need to have when going from the ATX connector.

                What would work is a dual stage integrated design, where you'd build it into your power supply. Essentially you have a barely regulated conversion from 230V AC to (up to) 28V DC, then you go from 20-28V DC (depending on the battery change) to the voltages you need. Google does that in their servers.

  4. Sorry that handle is already taken. Silver badge
    Joke

    What?

    No blockchain?

    1. Anonymous Coward
      Anonymous Coward

      Re: What?

      That would cause 'cog'nitive dissonance. ;-/

    2. IHateWearingATie
      Pint

      Re: What?

      Post of the day right there!!

  5. Martin Gregorie

    You gets what you pays for with UPS

    I have to take slight issue with ESR, but I'm in the UK and believe that our National Grid is in somewhat better repair than the USAian one.

    Bottom line: the real cheap UPSes are junk, just as his rant says, but if you work out sensible requirements and look round a bit you'll find a UPS that meets all of ESR's criteria and does what you want it to do, though not at a rock-bottom price.

    Back in September 2014, and in the face of a few power flicks that had dumped my house server (which has a soft-touch power switch - these seem specially designed to turn OFF if they see even a microsecond of mains loss). Fortunately none of these corrupted the disks, but I decided that, given the Governmental unwillingness to deal with ageing generating stations and rising power requirements (think charging electric cars and the giant whole-wall flatscreen tellies that the average consumer apparently must have) that a UPS would be a good idea, so I started to look for one.

    The problem: I wanted a UPS that could support my house server (30-50 watts on average) during a power loss of 5-10 mins (most power glitches are shorter than this) plus enough time for the server to shut down cleanly once it had been told to do so by the UPS. Not what I'd consider to be exactly hard to do.

    ESR was partly right: this was completely beyond the 'UPSs' sold by Amazon and friends at the time.

    These couldn't run my server that long, let alone do anything else. In particular, they couldn't tell the server to shut down before it drained their batteries. This was the killer: any UPS that can't do this is just junk.

    So, I widened my search.

    I ended up paying 300 quid for a lowish spec Riello and got a piece of kit that does everything ESR (and myself) wants. It can run my server for up to 50 minutes. It has a UPS monitoring server process that runs under Linux and is easily configurable to do what I needed: shut down the server after a total power loss of more than 5 minutes.

    The end result is that this Rielllo box comfortably exceeds ESR's requirements. It has a very nice LCD display that shows current status and gives direct manual control of it. The UPS monitoring server does everything ESR and I want and maintains a UPS event log. I have always run the UPS with the batteries in circuit. It uses a pair of 12v 7Ah SLA batteries. These are currently being reported as being in better condition than similar batteries I use in my glider and that get slung after 3-4 years when their annual test says they've lost 30% of the original capacity. The UPS's internal fans are no noiser than the server it powers, so quite suitable for domestic use.

    Revised bottom line: after 3.5 years the Riello UPS is still reporting that its original batteries have plenty of capacity to do what the UPS is configured to do.

    1. Richard 12 Silver badge

      Re: You gets what you pays for with UPS

      Exactly.

      Also, this project's BOM cost is currently of the order of £1000.

      Lead acid is used because it's cheap, reliable, easily recycled, easily maintained and cheap.

    2. John Sager

      Re: You gets what you pays for with UPS

      Too true. I've got a APC one now on its second battery, and it gave me fair warning that the first one was on the way out. However, like a lot of other people, I have reason now to hate systemd as it does shutdown in a completely different way, so the first power failure after I upgraded Ubuntu didn't shut the server down. It took me several goes at hacking on the apcupsd scripts to finally get it working reliably. It's still got a 'hope and pray' delay in the process to deal with an inbuilt systemd delay:(

    3. SImon Hobson Bronze badge

      Re: You gets what you pays for with UPS

      but if you work out sensible requirements and look round a bit you'll find a UPS that meets all of ESR's criteria and does what you want it to do, though not at a rock-bottom price.

      Yup. not rock bottom price, but when I was looking for a larger UPS (in the 8-12kW range) a few years back I found it hard to find anything without basic deficiencies. I drew up a list of requirements, and one vendor said "no problem" - except that when the unit arrived, no it didn't !

      SNMP reporting lacked certain vital measurements, or reported them in such coarse units as to be worthless, or has such offsets/span errors as to need correcting (seriously, load power 2kW out !). One of our requirements was to restrict battery charging power to suit our limited supply capacity - no point surviving a power cut if you then blow the power supply when the battery charger kicks in ! One vendor said "no problem" but had no such ability, APC just said we needed to upgrade our supply (not practical) to DOUBLE what we actually needed.

      The first system we got failed after a few years - the power conversion modules starting packing in until we didn't have enough to power the load. So we picked up a second user APC Symmetra LX and got a whole new level of idiocy. Yes there was more information from the SNMP - but still with errors, missing values, coarse resolution, etc. And of course, the non-repairable battery packs !

      Yes, the packs use standard batteries (12V 7Ah), but the UPS will not tell you the status of the batteries other than "this pack has failed" - won't tell you the charging current or estimated capacity remaining (even though they are available internally to the unit) so you can pre-empt one being flagged as failed. But when one is flagged, the in-pack flash memory is updated to show that it's failed so even when you fit new cells it will never go back to "working". Eventually it's power converters all stopped as well.

      So I'm not sure that you can buy a "good" UPS. OK, my sample size is small, but even spending a fair bit of money doesn't seem to buy the right features or reliability.

    4. JeffyPoooh
      Pint

      Re: You gets what you pays for with UPS

      MG offered, "...our National Grid is in somewhat better repair than..."

      Once upon a time, our power flicked off - again - apparently because there was a fluffy cloud in the clear blue sky.

      So I called our local power company to mention the apparent cause ("Hey, it's the fluffy cloud, as far as I can tell..."), and I asked them if they'd be able to maintain the power in the face of such repeated fluffy cloud assault.

      The very nice lady said that fluffy clouds were not usually an issue, and she claimed that power failures were rare. I could hear her 'tap tap tap...' on the keyboard, and then, "Oh My Gawd! You've had about a dozen power failures per year for the past several years! OMG! We'll have to look into that."

      So a crew went out to look at the substation (presumably from a safe distance), and reportedly something exploded precisely on cue. I'll bet that there was a fluffy cloud in the area. They all ducked and said, "Oh that." and then they replaced the still-smoldering device.

      Since then, the power has stayed on. For years straight.

      Fluffy clouds are no longer the rampaging menace that they once were.

      Sometimes it's the littlest thing. The one with thin wafts of smoke curling out, if you're there to see it.

    5. Anonymous Coward
      Anonymous Coward

      Re: You gets what you pays for with UPS

      @ Martin Gregorie

      Good work finding a UPS that covers your requirements, no sarcasm intended, I realized that could sound sarcastic as I wrote that.

      I see the issue as can you then easily hook up more batteries to that system to extend the life of your network if the outage were to be 4 to 8 hours. I know that could be a large battery pack but getting a whole work day out of a UPS set up would be fantastic.

      So, in my house that would include powering the fibre to ethernet converter, modem, and my laptop; so, pretty easy. For others it would include a tower and multiple monitors. Adding additional, maybe, deep cell marine batteries could be the the answer. The problem is, I guess, the proper distribution of that power.

      1. Roland6 Silver badge

        Re: You gets what you pays for with UPS

        >I see the issue as can you then easily hook up more batteries to that system to extend the life of your network if the outage were to be 4 to 8 hours. I know that could be a large battery pack but getting a whole work day out of a UPS set up would be fantastic.

        Well given the home environment, consideration has to be given to the frequency of such outages and the extent to which they impact you. Personally, if the power is out for more than a couple of hours, I simply pack the 'desk' and go visit some friends or a local library; but then having carried my work in a backpack and hot-desked for most of my working life I'm used to this style of working.

  6. FrankAlphaXII

    Sounds like a good idea to me, I'm not one of Raymond's biggest fans but he has a point. UPS as they stand kind of suck, and most people aren't going to shell out a huge amount of money for something decent, which is kind of a shame.

    Then again, my use-case has to do with disasters, and we've had a fairly reliable system* for a long time, the diesel generator and power conditioners. They're also a little on the expensive side, but they tend to last a hell of a lot longer.

    (*-As long as you can get fuel for the damn thing)

    1. Anonymous Coward
      Anonymous Coward

      Cheapo UPS?

      Two years ago, one of my former colleagues who lives in the boonies about 100 miles from Atlanta, GA got really fed up with his power going out (1-2 times a week) that he went the whole hog and is now totally off grid wrt Electricity. He has 12kW of Solar panels and a whopping 71kW of battery backup. Not cheap by any means. The battery is a load of 24v truck batteries all wired up and located in an outhouse.

      The downside is that his old electricity supplier is now threatening to sue him for going off grid. Don't you just love 'Trumpistan', formerly known as the USofA.

      A rather extreme solution maybe but I can see more people here deciding to cut the cord if their supply is 'iffy'...

      Thinking about it, once you have three or four (or more) small UPS's in the home, having one stonking great battery system starts to make sense.

      1. tony2heads
        WTF?

        Re: Cheapo UPS?

        Seriously WTF.

        They can't or won't (for cost reasons) supply him with power, but when he gets his own setup working they want to sue him..

        In many other places he would be lauded for getting his own solar powered system.

      2. fidodogbreath

        Re: Cheapo UPS?

        Thinking about it, once you have three or four (or more) small UPS's in the home, having one stonking great battery system starts to make sense.

        US electric utilities are starting to move residential customers onto Time Of Use billing, where peak-hour electricity can cost 5x-10x more than off-peak*. If the rate differential is high enough, a "stonking great battery system" might pay for itself by storing cheaper off-peak power, and using it to partially offset demand during peak hours.

        * The real point of "smart meters." Laying off all the meter readers was just a goodie grab-bag extra.

        1. Roland6 Silver badge

          Re: Cheapo UPS?

          >The real point of "smart meters."

          There is another interesting side effect of complex and dynamic tariffs: it becomes practically impossible to compare tariffs and thus pretty much destroy the energy supplier switching market...

        2. Alan Brown Silver badge

          Re: Cheapo UPS?

          "US electric utilities are starting to move residential customers onto Time Of Use billing"

          Time of use billing is almost _always_ coupled with a network charge as well.

          Here in the UK, my domestic network charge is higher than the energy charges.

          Once people start ramping usage around to avoid peak charges you can expect the network charges to go up.

          The same will apply to all those home solar panels or wind scams. People think they can make the meters run backwards but they forget there are a bunch of minimum charges built in that can't be undone.

    2. Mike Pellatt

      fairly reliable system.....As long as you can get fuel for the damn thing

      And as long as you don't end up running the system in an environment filled with dust (that wasn't in the design spec), you lose access to it, and the air filters get blocked.

      See 9/11.

      TBF, you did say "fairly reliable". That was an extreme event.

    3. tony2heads

      Diesel generator

      I assume there is a seriously large flywheel in the system to get you therough the diesel startup.

  7. ThatOne Silver badge
    WTF?

    Who will compile my open source UPS?

    The whole project idea has merit but sounds somewhat strange to me.

    An UPS is approximately 80% hardware, 20% software, so how can you open-source it? Give out the circuitry schemas and let people etch their own circuit boards? It would be way beyond the skills of most "consumer/SOHO users" (YMMV, note the "most").

    And what about the batteries, which are a rather important part of the whole thing? What's available in one country might be difficult (or very expensive) to obtain in another, especially since there are shipping restrictions concerning batteries. Whatever those people chose might be optimal, but chances are it won't be easy to buy for "consumer/SOHO users" in other countries/continents.

    (As for those who aren't just "consumer/SOHO users" and can spend as much as it takes, there are already off the shelf solutions out there doing everything required.)

    1. HereIAmJH

      Re: Who will compile my open source UPS?

      Actually, if you published the gerbers for the circuit board, some people would make their own and others would produce complete boards for the ones who can't. And if your focus was on component suitability, rather than the lowest cost solution, some parts could be sourced complete. For example, there are all kinds of smart battery chargers available in all sizes. As far as batteries are concerned, I'd be surprised if you couldn't find common components sourced out of China on eBay. Even without that, batteries with the same characteristics are going to be available just about everywhere. Supply chains are worldwide.

      I've been appalled by my growing UPS graveyard. So much so that I stopped buying APC. Rather than going this route though, when I remodelled my house I isolated circuits. Lights, computers, etc on their own dedicated circuits. Which means I can pull them out of the primary circuit panel and connect them to one supplied by batteries and an inverter. (primary charging by solar, I still need a circuit to initiate charging from a grid powered charger when capacity drops below a certain level and no renewable charging is available) My power requirements have dropped since I moved from compact fluorescent to LED, but a down side (since I am currently still powering lights from grid power) is that state changes in LED is much faster than CF or incandescent, resulting in flickering from dirty power being much more noticeable. Thus my desire to have all my ceiling lights powered by battery full time.

      1. Richard 12 Silver badge

        Where are you?

        That you can have such dirty power that you have visible flicker?

        That's not "dirty", that's actually "gone" - there is a rather large capacitor in a modern LED retrofit, and older ones only draw power as the top of the sinewave, so are generally immune to the common weirdnesses.

        1. ilmari

          Re: Where are you?

          Except really cheap LED bulbs which just have a bunch of LEDs in series straight on AC with perhaps a resistor or current limiting capacitor.

          The retrofits that have a tiny lag in between flicking the switch and turning on, you can be sure has some sort of actual driver circuit..

      2. onefang

        Re: Who will compile my open source UPS?

        There is this thing called open source hardware, look it up. The Olimex board mentioned above is one such bit of open source hardware kit.

      3. ThatOne Silver badge
        Unhappy

        Re: Who will compile my open source UPS?

        > others would produce complete boards for the ones who can't

        That's an ideal situation I wouldn't count upon, except maybe if the project gets enough traction to trigger manufacturing. In practice I'm pretty sure I'd just end up with a downloaded PDF and a copy of "Soldering for Dummies"...

        .

        > batteries with the same characteristics are going to be available just about everywhere

        To specialists. Someone which no clue about battery technology (architect, lawyer, dentist, astronomer, you name it) will have little to no chance translating the US specs to something locally available, and the US project brains can't make suggestions because they can't know what is or isn't available in every country/continent. Don't forget there are restrictions on overseas battery shipping, which would prevent an European (Asian, African, Australian) from buying the right batteries in the USA. At least at an affordable price.

        Downvote me as you like, but this still sounds like an electronics engineer project aimed at other electronics engineers... :-(

        1. Stoneshop

          Re: Who will compile my open source UPS?

          To specialists. Someone which no clue about battery technology (architect, lawyer, dentist, astronomer, you name it) will have little to no chance translating the US specs to something locally available, and the US project brains can't make suggestions because they can't know what is or isn't available in every country/continent.

          But there will be knowledgeable people from other regions who will add to this project[1], as with most open source stuff (both HW and SW) that requires some form of localisation, and a dentist from Dubrovnik would just need to drill down in the list of country-dependent modifications to get the right circuit board plus bill of material, as would an architect from Abidjan.

          [1] or fork it, as there are people who may well be disinclined to cooperate with ESR and Jay Maynard. For, in my view, very good reasons.

        2. Alan Brown Silver badge

          Re: Who will compile my open source UPS?

          "Don't forget there are restrictions on overseas battery shipping, "

          Erm... no. Just on the methods.

          Unless you're in a country with _very_ restrictive import laws (such as outer bumfuckistan), batteries are the least part of this whole deal.

  8. Nate Amsden

    true sine wave?

    I have had almost all electronics in my home on sine wave ups for over a decade now. I started with used APC smartups, though past 9 or 10 years been using cyberpower. I have no complaints about the UPSs themselves. The good ones aren't cheap two of mine were at least 400 bucks a pop i think maybe over 500. My first big cyberpower I had a battery extension pack on it for more runtime(maybe 800 bucks total). I ended up frying it when I hooked up the 3rd party replacement batteries in the wrong sequence(fortunately I got 5 years of service out of it, no damage to equipment other than ups). The newer unit I bought to replace it works fine though the fan is really loud. Both are/were double conversion systems but the previous model the fan only kicked in when on battery. My 2nd newer UPS I opted for a non double conversion but still sine wave, silent when not on battery.

    As for text alerts my upss have had lcd screens for years, even a baby cyberpower for my bedroom entertainment system. I like how i can see the power usage in watts on even the baby ups. Put a bigger battery in it after 3 or 4 years (didn't fry that one only one battery vs 9 in the one i fried). Works great.

    Non sine wave is fine for most computers. Some of my earlier cyberpower upss and computers back in early 00s had some issues. The power supply in the computers would tend to jump from 110v mode to 220v mode when the ups kicked in. Workaround for that was to order the power supplies hard set to 110v (pc power and cooling). Had one power outage which blew a fuse in my stereo at the time and eventually blew the ups (batteries leaking and all). Fortunately no damage other than the ups itself.

    I've been using small upss on my personal computers since 1996. My first real IT job in 2000 I deployed tons of APC rackmount smartups and battery packs. I didn't read or know about the UPS rant myself but maybe most of it comes down to the guy wanting a dirt cheap solution.

    When I first heard about this topic I thought they were wanting to make high end UPSs (e.g. tens of thousands of watts). Then I see the 300W comment and realized it's for toy workloads.

    For ups software mostly I use network ups tools (NUT) for the past 18 or so years. Also used APCs network ups software too especially back in the early days when I had things like hpux and solaris and aix along with linux.

    Kinda funny UPS story. I equipped my org at the time (18 yrs ago ) with tons of upss. Enough runtime for 1 or 2 hrs.

    Then one sunday morning my phone went off. Power outage and batteries were kicking in. I felt good for maybe 10 seconds. Then I realized the AC was off too. So i rushed to the office 2 miles away to force quicker shutdowns. No issues. Fortunately nothing mission critical.

    Since 2003 anything important has gone to a colo so haven't dealt with UPSs outside of my home since.

  9. Anonymous Coward
    Anonymous Coward

    What's the over/under?

    The write up sounds like they are focusing on a big (UPS) batter to charge smaller batteries. That's fine for those who don't want what a good UPS can provide. As Nate Amsden points out, a good UPS cleans the power it outputs. The other thing a good UPS does is regulates the power it outputs to eliminate high and low (brownouts) voltages. To do this, input power needs to go into the battery while output power is taken from the battery.

    1. Richard 12 Silver badge

      Re: What's the over/under?

      Actually it doesn't. Electronic (either controlled or fully electronic) autotransformers are a thing.

      It might be cheapest to do AC-DC-AC, but it isn't necessary.

  10. JeffyPoooh
    Pint

    Comparison to guard dogs

    You're concerned about your family's safety. So you get a guard dog. The dog costs a fortune. It immediately poops on the floor. Then it chews off the entire left side of your Bang and Olfson. It bites the postman's fingers. It then sleeps through an actual burglary. And finally it eats one of your children.

    This is the UPS experience: If they're not preoccupied with smoldering their lead acid batteries, then they're busy buzzing and arcing. Then they blow an internal fuse on the output, and your Great American Novel is suddenly lost, again, for the third time. Then there's an actually power failure (Yay!), so they turn on their patented 387 volt offset square wave, and your PC is instantly corrupted. Meanwhile battery acid squirts out onto the ceiling, again. Then, while you're out trying to buy a replacement PC, the UPS catches fire and burns your house down.

    I'd happily pay $800 to not have one.

    1. ChrisC Silver badge

      Re: Comparison to guard dogs

      Remind me to never, ever, go UPS shopping with you... In the getting on for 2 decades that I've had a UPS protecting my PC at home (ever since a couple of seconds of brownout caused me to then waste several hours of my life reinstalling Windows due to the brownout occurring whilst it was in the middle of writing to something slightly critical such that it couldn't even reach the desktop in safe mode), I've never had a problem with either of the units I've used.

      Yes. Either of them. Nearly 20 years of faultless service from just two different UPSs (both from the APC stable), and the only reason I retired the first one was because it couldn't cope with the increasing power demands of my newer system. The only maintenance I've had to do on either of them was feeding them with replacement batteries every so often, but other than that they've just sat there doing exactly what they're supposed to do, and in all the times they've intervened to keep my PC up and running through brief brownouts, or enabled me to shut down cleanly during longer duration blackouts, I've not once suffered data loss/corruption.

      So for me, the ability to just keep on working without in some cases even being aware of the brownout, or to be able to resume working as soon as the power is restored without having to spend hours recovering the system into a working state again, is worth every penny I've spent (which isn't all that much in total, let alone averaged out over all these years) on these units.

  11. Anonymous Coward
    Anonymous Coward

    At a company I know...

    They have fire drills about once a year. Pleasant, so I've heard.

    Twice they have had actual fire alarms, for real, get out. Smoke, possible fire. Fire Trucks, blinky lights. Both of them were caused by the server room's UPS.

  12. Anonymous Coward
    Anonymous Coward

    300 watts for 15 minutes?

    Talk about setting a low bar, that's hardly more stored energy than a high end phone powerpack provides (something like 85000 mAh based on 115 volts in the US - I've seen powerpacks in excess of 50000 mAh) Granted you can't draw even a fraction of 300 watts out of one, but that's mostly a matter of how the batteries inside are wired together.

    We'd be better off with a spec for a DC UPS, that outputs at multiple DC voltages (5, 12 and 24 volts) since everything you're plugging into a typical UPS ultimately runs off DC anyway. Granted there are a lot of oddball voltages out there like 9, 15, 19.5 etc. but given the wide range that the typical unregulated power supply that comes with a laptop or router outputs, you're almost always going to be OK with using whichever of 12 or 24 volts is closer to your intended voltage.

    Perhaps in a few years a lot of devices will be powered by some form of USB-PD and simplify things enough that an open source DC UPS can handle most stuff you throw at it.

    1. Christian Berger

      It's not that simple

      First of all, converting those 3,5ish volts up to 12 volts which is used by mainboards isn't simple.

      If you want to combine your PSU and your battery backup, do it the Google way and use a 24-ish volt intermediate stage with 2 12V lead acid batteries. The first stage converts the 230 volts to roughly to 24-28 volts to charge the battery, while the second stage converts those to the voltage the rest of the computer needs.

      1. Anonymous Coward
        Anonymous Coward

        Re: It's not that simple

        There are some powerpacks that output at both 5v and 12v. Series wiring would step up the 12v to 24v, if there was a demand for a third output at 24v.

        Might be a bit much to expect it to be useful for a desktop beyond a simple one - it would need a hell of a lot of power to backup a gamer's PC with the ridiculous graphics cards used these days.

  13. Neil Barnes Silver badge

    A point that occurs to me

    There's an awful lot of conversion up and down and up and down in various flavours of AC, DC, switched DC, stepped DC and the like. Seems a terribly inefficient way to go about things... perhaps it would be nice to have a standardised (e.g) 12v LED lighting circuit, powered directly by a lead acid cell (so 11 to 13.8v really), and maybe a 19v dedicated laptop circuit and a 5v usb circuit? Yes, DC resistance losses, but perhaps less than the conversion losses of going up to mains and back at every stage?

  14. steelpillow Silver badge

    BEEEELIONS of $$$ over DECADES

    Off-the-shelf vehicle batteries do make more sense. The use case is similar: long periods of nothing, short periods of massive drain, for low cost and moderate weight. The automotive industry has chucked BEEEELIONS of $$$ over DECADES, at better batteries for low cost and light weight. A backyard UPS project is never going to do better.

  15. Lazy Jack

    He gets some of these right, but...

    Reading ESRs rant on why UPSs suck, i think he gets right that he protocol and the GUI suck big time.

    But the other, I'm not sure. It's easy to bash EEs about the protocol, but the matter is that EEs still do better job at electronics than CS people at programming. I think his views on battery chemistry, battery management, etc. are overly simplified and based on little in-depth knowledge.

    1. Christian Berger

      Re: He gets some of these right, but...

      Well a good electrical engineer can usually code somewhat better than a bad CS person. It's just that good people in both professions are rare, and the fields are dominated by hacks... though there are lots of people who go into CS because they can re-install Windows. You don't really have that equivalent for EE.

      However I think the point behind the rant is clear. There are things that could be done better. For example UPSes could have a mode where they aquire a minimal charge to their battery before turning on again after the power comes back. (to avoid PCs crashing when the power goes out again before they can shut down again)

  16. Lee D Silver badge

    Buy battery charger.

    Buy power inverter.

    Connect one to other.

    Need to monitor it? Put small USB voltmeter in the middle bit and connect.

    I'm always baffled that people think a UPS is anything more than that. Sure, there's a certain amount of passthrough and switchover and so on, but that's the bit that's going to go wrong when people try to wire up their own ones.

    If you're really that worried at the "code" that a UPS is running, when all it does is present a serial interface saying "Please shut down nicely now because I'm about to die", then making something that has no code at all is surely easier.

    1. Natalie Gritpants

      I think you missed out the battery.

      1. Lee D Silver badge

        Quite literally batteries not included.

  17. Grifter

    Still relevant

    http://geekz.co.uk/lovesraymond/archive/gpl-sucker-punch

  18. Anonymous South African Coward Bronze badge

    Cool, nice.

    So if you can get a good diagram, you can start producing your own UPS'es with the correct software.

    Currently we are using RTC ups'es - the software runs on a Java platform, which makes it a real PITA especially with newer browsers and OS'es.

  19. Jason Bloomberg Silver badge

    SLAs for SLA

    One of the advantages of Sealed Lead Acid batteries I suspect may be their robustness should the electronics be blown out by an adverse condition on the mains side; it takes a lot more to destruct an SLA in a devastating way than a LiPo.

    There are good reasons most UPS are designed with KISS in mind.

    I don't have a problem with people wanting an open source UPS but I do worry they have simply embarked on a 'build a better mousetrap' journey.

  20. ianmcca

    Designed by programmers

    The difference between designed by programmers and designed by (hardware) engineers is that programmers know they need help from engineers but engineers almost invariably believe they don't need programmers to do the coding code. Examples of crappy hw-engineer-designed software are legion but here area few I've suffered from recently: domestic heating systems, photo upload from camera, SATNAV map update bloatware, Smart tv menus, any device that says "uploading the wrong firmware to your device may render it inoperable".

    All consumer devices should be designed by programmer-lead teams, IMHO of course!

    1. Stoneshop
      Windows

      Re: Designed by programmers

      All consumer devices should be designed by programmer-lead teams,

      Let me introduce you to the 'programmers' I have to deal with.

    2. Christian Berger

      Re: Designed by programmers

      Well.... actually those pieces of software often have been designed by "software engineers". It's just that there are many really bad ones around. Just look at old devices which were made back when you actually had the firmware designed by hw engineers. Those tend to work for 20+ years without ever needing a firmware update, because they knew what they were doing.

      It's not a question about the field, but about experience.

      Edit:

      Actually here it's also often a question of outsourcing. HW companies tend to outsource the code to SW companies who have no clue about what they are making. Thanks to outsourcing even if the SW people would know better, they couldn't make it better as the specification has already been agreed upon.

      1. ChrisC Silver badge

        Re: Designed by programmers

        "Just look at old devices which were made back when you actually had the firmware designed by hw engineers."

        This is still how it's done in a lot of engineering companies :-)

    3. ChrisC Silver badge

      Re: Designed by programmers

      "Examples of crappy hw-engineer-designed software are legion but here area few I've suffered from recently:"

      There are equally just as many godawful examples of programmer-led software releases - no self-respecting engineer would have allowed stuff like Lotus Notes or the Windows version of iTunes to escape the development labs... Speaking as an embedded systems engineer, I know I can't write front-end code which is as elegant or as pleasing to use as stuff produced by even a half-decent programmer, but what I do know is that whatever I write will at least bloody well work the way the documentation says it'll work, and it won't chew its way through all of the available system resources in the process.

      And as for "uploading the wrong firmware to your device may render it inoperable" - when you see a warning like that it means the user is at least being given the choice whether or not to perform the update. Where was the user choice when MS decided to push out buggy W10 updates which turned a whole bunch of PCs into paperweights? Yes yes, recovering a PC is generally a bit easier than recovering an embedded device provided you've got recovery media to hand, know how to use it, and have the time and energy to devote to the process, but that's still no excuse for forcing an update onto a user if you're not 100% certain it'll leave their system in a fully functional state after the update has been applied.

      There are some truly talented programmers out there whos abilities I'm genuinely in awe of. Unfortunately, most of the software we use in our day to day lives isn't written by people like these, and it shows...

  21. Mr Humbug

    This is something I recall reading on Usenet

    or perhaps on a BBS. Being old, my recollection is hazy.

    Anyway, as the new Battlestar Galactica says "These things have happened before, and will happen again"

    http://www.danielsen.com/jokes/objecttoaster.txt

  22. Stoneshop
    Boffin

    UPSs lack the kind of sensor information that protected car batteries, Raymond wrote

    He thinks car batteries are protected? Shows how much he knows about hardware.

    EV batteries, which are some flavour of Li-ion, are indeed monitored and protected (they have to be, by their nature), but that's still a minuscule fraction of 'car batteries'. Conventional car batteries (lead-acid, trend has been towards SLA for some time now) aren't.

    1. Adrian 4

      Re: UPSs lack the kind of sensor information that protected car batteries, Raymond wrote

      EV batteries (high capacity, long runtime) are far more suitable than car starter batteries (low cost, high surge current, short runtime). The EV batteries are expensive but huge amounts of effort will go into improving them for the electric cars that are supposed to be introduced everywhere. They're the ones to watch, though the technology is likely to change before the cost falls.

    2. Nate Amsden

      Re: UPSs lack the kind of sensor information that protected car batteries, Raymond wrote

      Maybe 2 years ago I had a car battery issue(first time I've ever had to get a jump start). Battery was probably 3 or 4 years old at the time. In the morning started the car, no signs of any problems (one other time on an earlier car I had a battery issue and it seemed to initially manifest itself in very slow power windows, after the battery was replaced the windows went fast again..maybe coincidence I don't know).

      Anyway started the car drove about 100 miles or something. Had to get gas. Stopped to get gas and car would not start again. Tried several times, eventually called AAA (whom I'd never had to call before). They jump started the car and I drove to a dealership closer to home to get the battery replaced. They tested it, and said the battery was "OK", the reading was literally right on the edge of being in the zone for replacement. They didn't understand why the car had an issue starting, especially not showing any other signs of battery problems.

      My friend who was with me has been dealing with cars far longer than I have and he too has never seen that kind of situation before. No warnings lights on the dash (2011 Nissan - by no means an ancient vehicle and I bought it brand new in early 2012), no indications what-so-ever of a battery problem until it just refused to start. Got the battery replaced and no issues after that.

      Weather was fine (not too hot, not too cold), and I kinda would of thought a ~100 mile non stop journey would give the alternator some time to charge the battery up if it was low, but apparently not enough.

      1. J. Cook Silver badge

        Re: UPSs lack the kind of sensor information that protected car batteries, Raymond wrote

        Part of the problem with car batteries is that the alternators that *used* to be on cars could be counted on to provide enough current when running to actually *charge* the battery on top of running the engine; but now? not a chance. My PT cruiser ate it's battery without fail every 18 months, and with little to no warning- I'd go out and try to start it, and instead of cranking over, the dash gauges would just twitch.

        at [RedactedCo], we use APC SmartUPS in the 2KVA range for some of our gear. they have the network management cards installed and they run us about $700 USD each. We have units still in use after five years on their original packs, which we are starting to replace out. (the part of the world I'm in gets stupid hot during summer, and the grid's not quite reliable)

        the data centers are on Liebert APM/NX beasts (three phase, multi-rack) which while I don't know that exact cost on, have never let us down in the 5+ years we've had them. But then, we also keep maintenance up on them.

        Overall, I think they are either trying to find a 'standard' design spec, or otherwise trying to reinvent the wheel, myself.

    3. ilmari

      Re: UPSs lack the kind of sensor information that protected car batteries, Raymond wrote

      The only cars I've seen have 14.5V regulated alternator, that if you're lucky is temperature compensated. Low voltage protection doesn't exist, but sometimes happens accidentally because the starter solenoid just drops out when voltage collapses. Some fancier fuel burning heaters with timers do have low voltage cutoffs, though.

  23. FIA Silver badge

    Old but still worth a read...

    This is old now, but PCs haven't really spiked in power usage in the last few years so probably worth a read...

    http://www.dansdata.com/diyups.htm

    and...

    http://www.dansdata.com/upsupgrade.htm

  24. Omgwtfbbqtime

    hmmm...

    Best thing about opensource projects like this is it's easy to repurpose.

    Once it's at a good enough state I may have to build one with 4 or 5 car batteries and retrofit an arc welding output on it.

  25. Anonymous Coward
    Anonymous Coward

    Good luck, will be nice to see the third iteration

    Very interesting. Yes, UPSes suck. I see three iterations:

    Iteration 1: get it working

    Iteration 2: make it scalable (able to be produced in volume)

    Iteration 3: make it cheap.

    #3 is where I sit up and take notice. In particular, being able to swap out the battery technology for any-damn-thing-I-choose is *very* interesting.

    Have you ever tried building a software economy in a poor country where the power goes out all the time? It's not pretty. So yeah, have fun with your exotic rare-earth battery packs, but when you make it work with ten car batteries I pulled out of a junkyard, you'll really have my attention.

    1. J. Cook Silver badge

      Re: Good luck, will be nice to see the third iteration

      The problem with the idea of 'plug whatever damn battery pack you have handy' is that chargers are generally designed for a specific battery chemistry.

      what's good for charging SLA batteries may not be suitable (or desirable!) for LiPO. What a LiPoFe4 stack would find usable would be unusable for an SLA, or NiCad, etc.

      The 'multi-chemistry' chargers for tools? The charger and packs have circuitry that tells the charger what kind of battery was jammed into it, or the battery pack has a built in protection circuit to cope with whatever trash the charger throws at it. (At least that's my assumption- I'm quite probably waaay off.)

      1. Charles 9

        Re: Good luck, will be nice to see the third iteration

        Given the age of the tools these multi-purpose batteries are supposed to work on (see Ryobi One+, which started with NiCd, moved to related NiMH, and now uses LiPo, and they're ALL supposed to be interchangeable), the battery must have at least similar discharge characteristics so they can work in things from worklights to circular saws. That said, the charger probably needs to be at least be recent enough to detect which of the types of batteries are going into them.

      2. Stoneshop

        Re: Good luck, will be nice to see the third iteration

        what's good for charging SLA batteries may not be suitable (or desirable!) for LiPO. What a LiPoFe4 stack would find usable would be unusable for an SLA, or NiCad, etc.

        So you would want a modular approach, selecting the appropriate charger for the various battery chemistries and the selected capacity. Different chemistry would also mean a few different parameters to monitor (cell temp, charge/discharge current and voltage for all of them, but with different warning thresholds). Then some people would want an 110V inverter (and charger), some 230/240V, and others an ATX-like or SELV output. So modular again.

        In all, I myself would go for an existing UPS of the appropriate config and capacity, then bolt on a monitoring circuit that covers the appropriate parameters, replacing the manufacturer's crap. So, reinventing only that part of the wheel that apparently needs reinventing, leaving the rest, including the parts dealing with mains voltages, to (a selected few) of the existing manufacturers.

  26. Nate Amsden

    server based UPS

    Some folks have mentioned server based UPS. Some solutions do exist for this but note they are for a specific market. I was at a conference back in 2010 (had to refer to a blog post to remember), and there was a presentation on this topic. The one bit that stuck with me was the claim that 90% of power disruptions last less than two seconds(I'm assuming they were specifically referring to data center disruptions).

    So most server based UPSs are only meant to last a few seconds (maybe no more than 30 seconds depends on the server config). To give enough time for the generators to kick in. Or maybe you have massive scale like google and have data center availability -- combined with the fact big scale power outages are rare so the risk of losing a data center is low enough not to be a big enough concern to build out expensive power infrastructure to protect against it.

    In my experience I have experienced one power outage in a data center in the past eight years(tier 4 facilities only). Though there will be another one at the same facility in the coming weeks so they can replace the faulty equipment(surprised they can't do this online, maybe they are not N+1 assuming N+1 provides this ability, I didn't choose this facility but is (now) operated by a very large provider). Fortunately that power outage only took out one of the two feeds, so services stayed online. We did temporarily lose a couple pieces of equipment that were single power supply but they had redundant partners that took over.

    I for one am in the camp where I want longer run times on backup power systems. Server based UPS or flywheel based UPS just doesn't offer long enough run time for me to feel comfortable. But I also don't run massive hyperscale infrastructure either.

    I remember one data center power outage I think 2009, my equipment was no longer in the facility, having moved 2 years earlier. This was a big outage I want to say 40+ hours, had a fire on the busbar(?) system or something like that. Building ran on generators for months while they repaired it. Anyway my 3PAR rep at the time told me that there was one or two (that he was aware of) NetApp systems that were spending maybe 12+ hours past the power outage doing file system checks, maybe the batteries ran out for the cache I don't know. The 3PAR systems by contrast have a different design where the batteries keep the controllers running long enough to dump the cache to an internal hard disk on the controller then shut down. No need to worry about cache retention after that. And since cache is mirrored you have two copies of the dumped data(one per controller) in the event the internal HD or battery or whatever has a problem. Maybe NetApp adjusted their design since I am not sure. I thought it was a cool approach at the time though.

    1. Stoneshop

      Re: server based UPS

      The one bit that stuck with me was the claim that 90% of power disruptions last less than two seconds(I'm assuming they were specifically referring to data center disruptions).

      The VAX9000 had a power system that consisted of a 280VDC bus inside the system cabinet feeding the 5V/-5.5V/12V/whatever PSUs, supplied from a power front end that took about 2 19" racks worth of space. That held a transformer, rectifiers, control circuitry and a serious number of electrolytic capacitors. This was to ride out such short glitches (IIRC it was good for ten seconds or so, definitely not enough to shut down the system)

      1. Mike Pellatt

        Re: server based UPS

        DEC-10 KI10 processor, even better. Absolutely massive transformer in each 38" rack (each processor was 2 of them plus a 3rd for console, DECTapes, paper tape reader/punch, hundreds of blinkenlights) full wave rectification, even bigger caps for smoothing. 8VDC or so fed to series stabilisers down to 5VDC for the TTL along each row. The series stabiliser power transistors were in their own airflow up the end of the row.

        Aircon and humidifier failures totally swamped any power issues (if any, this was South Kensington :-) )

  27. This post has been deleted by its author

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Other stories you might like