UPSide needs a battery technologist

The design of UPSide is coming together very nicely. We don’t have a full parts list yet, but we do have a functional diagram of the high-power subsystem most of which can be expanded into a schematic in a pretty straightforward way.

If you want to see what we have, clone the repo, cd to design-docs, make transactions.html, and view that in a browser. Note that the bus message inventory is out of date; don’t pay a lot of attention to it, one of the design premises has changed but I haven’t had time to rewrite that section yet.

We’ve got Eric Baskin, a very experienced power and signals EE, to do the high-power electronics. We’ve got me to do software and systems integration. We’ve got a lot of smart kibitzers to critique and improve the system design, spotting problems the two Eric’s might have missed. It’s all going well and smoothly – except in one key area.

UPSide needs a battery technologist – somebody who really understands all the tradeoffs among battery chemistries, how to spec battery types for different applications, and especially the ins and outs of battery management systems.

Eric Baskin and I are presently a bit out of our depth in this. Given time we could educate ourselves up to the required level, but the fact that that portion of the design is lagging the rest tells me that we ought to recruit somebody who already knows the territory.

Any takers? No money in it, but you get to maybe disrupt the whole UPS market and and certainly work with a bunch of interesting people.


  1. I typed out a comparison of rechargeable chemistries, but it appears the comment machine ate it. Here’s just the lead-acid stuff, which I found to make the most sense as I was considering while typing it. I’m not the expert you need, but hopefully you find this helpful.

    Lead acid automotive batteries fail after 5-10 years by means of physical deterioration of the lead plates, due to vibration, thermal cycling, and conversion between metallic lead and lead sulphate. So in a stationary, indoor environment with only occasional use, they could reasonably be expected last toward the upper end of that range, or maybe longer. They don’t have the memory effect for which NiCd batteries were notorious, and they put out tons of current as they’re designed to spin your engine up. Moreover, management can be performed by a battery tender, which is not a very sophisticated device, but is quite effective – you can entrust your battery to it, and it’s generally good to go when you get it next year despite being otherwise neglected. A tender is a readily available piece of equipment and it might make sense just to use one in your design as is, although its monitoring abilities are pretty limited and you’d likely want to supplement them.

    Specifying a lead acid battery is usually a matter of ensuring you have enough current, noted as cold cranking amps, to meet the minimum equipment requirements for your mission, and enough storage, measured in amp-hours, to complete all the objectives before energy exhaustion. Both need to be adjusted to compensate for losses in the inverter and other circuitry. You may also be able to find service life measured in years or duty cycles, but rarely both, and I’m not sure in the common use case which metric you’d care about more. The spill-resistance of gel/glass-mat wouldn’t seem to be necessary, and a simple unsealed flooded battery has the advantage of being user-serviceable. Nominally twelve volts, most actually register a bit over thirteen if you take a multimeter to them. This drops somewhat as they discharge, but what you’ll really want to watch for if you’re trying to estimate battery state is the increase in internal resistance (model the battery as a voltage in series with a resistor). Weight may not be critical, but don’t forget to make sure the battery you choose fits in your desired form factor – this is likely a tradeoff with storage capacity.

  2. Reader d5xtgr’s nice summary is dead on.

    I was introduced to lead-acid battery UPSes in the mainframe test room of an IBM plant. There was a well-ventilated side room where several rows of batteries sat on metal shelves. The shelves were arranged like stadium bleachers so that leaking batteries would not spill acid on lower shelves. They provided power to a hefty DC-to-AC converter able to supply enough power for the mainframe. That was in the 80s. Modern datacenter UPSes are integrated and include their own batteries.

    In my (admittedly limited) experience, the only factor that matters for UPS lead-acid batteries is the Ah capacity. The cranking capacity is not a factor, since you are limited by what the DC-to-AC converter can supply.

  3. But the current USPes DO use Lead Acid batteries already and from my personal experience the consumer UPSes batteries rarely last even just a measly 3 years. From what I understood, aren’t we trying to not use these lead acid batteries with the UPSide project?
    But wait, maybe there is a super good lead acid battery that we can just drop right into our current consumer UPS to replace it’s stock battery… but I think not.

    1. >From what I understood, aren’t we trying to not use these lead acid batteries with the UPSide project?

      LiFePo has a better lifetime and such batteries are generally built with an onboard BMS that would simplify the system design a lot.

      I’ve recently learned that the actual reason the sealed lead-acids in conventional UPSes have such a crazily short lifetime compared to other deployments may be vendor malfeasance – they trickle-charge the battery constantly rather than trying to detect when it’s at full charge and cutting current from the charger.

      This choice is (a) cheap to build, and (b) means the UPS comes back from an outage as fast as possible. But the constant charging boils off the electrolytes far faster than if they were allowed to rest. With smart battery management SLA lifetimes can be very, very long – I’ve heard of fixed installations where they ran into decades.

      But there’s a counter-incentive against doing this in consumer-grade UPSes. When an SLA in one of these units craps out, the consumer often reacts by replacing the entire UPS. Profit!

      1. they trickle-charge the battery constantly

        A lot of people don’t grok the difference between a trickle charger and a battery tender. But a UPS maker ought to.

      2. > LiFePo has a better lifetime

        You keep doing this. We don’t need anyone spreading FUD that we’re using Polonium like some kind of Russian assassins. Just keep the left Shift key down a bit longer to type “LiFePO”.

        But the very existence of this post suggests LiFePO4 isn’t carved in stone, even though the Wiki seems to indicate that it is.

  4. I have a GitLab ID now: madrocketsci. May I join your project? Currently, I’d just like to lurk and see what you are working on. (Don’t know what sort of help I can offer, not a EE (yet)).

  5. Would it be possible to have an UPS able to use not just one type of battery but multiple ones (by type, I mean “chemistry” – and no, I don’t mean “multiple ones at the same time”) ?

    Anyway, I hope you will consider NiFe batteries among the possibilities. On paper at least they seem interesting (but hard to find, alas).

  6. I did active rectifier control for a major UPS manufacturer. Megawatt/datacenter scale, not retail/workstation scale.

    It may interest you to know that at the large scale, batteries are starting to fall out of favor. Over the 10 year design life of a big-iron UPS, the batteries are replaced at least once, and at many sites twice. The result is that the batteries are the most expensive part of the system by far. In a high-reliability datacenter (at least, for those customers who aren’t migrating everything to the amazon/google/whatever), you typically already have diesel generators that can come online in a few seconds. So, you only need a few seconds of stored energy in the UPS. Datacenter owners who are actually TCO-aware have recognized that it has become worthwhile to use flywheels or ultracaps for very short holdup-time UPS. A flywheel set will last the life of the UPS. Even though they take up the same amount of floor space for 1/60th the holdup time, it still works out in favor of the flywheels.

    At any rate, the rule for battery replacement in big-iron UPS is still every 3-5 years. This is with a dedicated battery state-of-charge monitoring system that is designed and tested to get the longest practical life out of the cells while still guaranteeing some minimal nameplate-rated uptime to the customer. When you are slinging around megadollars on your megawatt system, backing up gigadollars of customer data you want to make sure that it actually stays up as long as it claims to. Batteries can last “longer” than this. But only if you are willing to accept a significant degradation in their performance.

    Naval submarine batteries are also replaced every 3-5 years. Those systems are several thousand Amp-hr wet-cell lead acid chemistry, monitored per-cell by both voltage measurement and specific gravity of the electrolyte, with a dedicated program to adjust the electrolyte levels in the field. They are also very carefully managed for long life by a state-of-charge management program that has been established by an army (heh) of PhD’s in national government labs plus field trials. At scale. For decades. And the batteries last 3-5 years. In order to make it to the long end of that lifetime for the battery pack as a whole, the ships have sets of busbars that can be used to bypass around an under-performing cell or few. Its normal to cut out a couple of cells before replacing the pack.

    I think at this point that the industry has collectively shrugged its shoulders, embraced recycling, and just accepted that lead-acid batteries are heavy, bulky, reliable, and last for only 3-5 years.

    My recommendation is to pick a gel-cell form factor that you can buy from your local hardware store, and just expect to replace it every couple of years. Put it some nagware into the software to remind the owner to replace his pack from time to time. And be done with it.

  7. So, there’s a small, local (to me) company making high end electric cars, which you may want to contact. The cars do per-cell charging, with some reasonably sophisticated system to maximize lifetime of the batteries, running on an embedded Linux system. If they’re willing to help, it could avoid a fair amount of duplicated effort.

  8. The battery choices are: Lead Acid, NICad,NiMH, and Lithium Ion.
    Yea, don’t use Lithium Ion! if you are not careful with the charge circuitry and your supply chain, you will be creating explosive devices disguised as UPS systems.
    NiCad? No body uses NiCad any more.
    NiMH, these batteries are good, they have a good power density and behave better than nicad. They are more expensive than lead acid and much lighter.
    The best bang for your buck is still lead acid batteries. There are tree basic categories: automotive, marine, and deep cycle.

    An automotive “car” battery is a poor choice, they may last long in a car, but one big drain in a UPS will kill it. The internal lead plates will become warped.

    A marine battery is kind of half way between a deep cycle and a car battery, it can withstand deep cycles without warping the plates.

    The last type, “deep cycle” lead acid battery is what you want, they are designed to charge and discharge. They are used in golf carts, wheel chairs, medical equipment, etc. (Yes, in UPSs too) In this category there are a bunch of types. Old fashioned maintainable cells with caps. Sealed lead acid (SLA), and Absorbant Glass Mat (AGM) and gel.

  9. Maybe you can convince CenturyLink to use these things on their DSLAMS.

    I just had my power drop for about 3 seconds. The UPS on my DSL “Modem” handled it just fine, but the fookin DSLAM reset, dropping me out of the middle of a meeting.

    As Roman Troy Moronie would say “Iceholes! Fargin Bastages!”.

Leave a comment

Your email address will not be published. Required fields are marked *