Yeah, that's an oversimplification. It's a rack with a bunch of ports to plug in whatever 12-volt batteries aren't currently in use elsewhere. So when I retire one from the car, or a solar experiment, or replace a whole pile of 'em from a UPS and most are kaput but a few are still viable, they all end up on that rack, in parallel. (With fuses and stuff, of course. And I precharge each battery so it closely matches the float voltage before plugging it in.)
Presently I think there's like 30-40 amp-hours worth of lead and lithium sitting there, all charged by a little Meanwell RS-25-12 or something. I don't exactly remember, but the photo looks familiar. (25 watts at 12 volts tweaked up to 13.85 volts, so, roughly 2 amps.)
That powers the cable modem, wifi router, service pi, and a RIPE ATLAS probe. The first two eat 12-ish volts and have SMPS input stages so they're not picky about the -ish, so they just run directly from the battery rail. The latter two have DC-DC converters feeding them (the venerable MC34063 or something similar). Combined, their draw is just shy of an amp at 12 volts, so the rack with its current fleet of batteries can run several days with no AC input.
Realistically if power fails, I usually start the generator after just a few hours, so the batteries never actually drain much.
If I can find a good place to mount a solar panel that doesn't annoy the landlord, I might be able to unplug the Mean Well and run it independently indefinitely.
Nah, that's what "float" status is all about. There's a voltage where it's safe to let the battery sit indefinitely. Float is less than full-charge peak stress blow-things-up-if-you-stay-here voltage, but IMHO that's to be avoided anyway. So I set the power supply for float, not peak.
Furthermore, all the lithiums have a built-in BMS that'll dehance the charge-path FETs if a single cell goes over its cutoff voltage. So even if my power supply were to lose its regulation and attempt to overcharge them, they'd just disconnect themselves.
The lead-acids in such an overcharge state would simply convert more of their electrolyte to hydrogen gas and eventually deplete the liquid, but because the power supply is so weak in the first place, it couldn't force that to happen at a problematic rate. The regular drafts and air changes through the house would keep the hydrogen below its LEL unconditionally.
what kind of rack lets you just jack in 12v batteries with no concern about charge/discharge?
what's the spread of their capacities?
do you group similar capacity batteries together? somehow and charge/discharge them together?
won't nearby large capacity batteries spread/smear their voltage with other batteries on the rail? the smallest batteries will just get obliterated, and the large ones will do the heavy lifting, right?
do you have something that disconnects batteries when their voltage drops too low? and does it charge them too?
if you put every battery on its own charger, and then use a high/low voltage disconnect, maybe this works, to connect random batteries together on a shared rail...
am i overcomplicating this? or is your setup dangerous?
are you assuming everything has the same float voltage, and you never need to actually discharge, and then you never charge either?
as soon as you go off mains/float voltage, and actually use this battery system, some batteries will deplete faster than others since they have less capacity.
are you using it for purely backup purposes only, and not recharging?
> are you assuming everything has the same float voltage
Yes. Both 6S lead-acid and 4S LiFePO4 are happy at 13.8 to 13.9v. Lead-acid has a temperature coefficient but my basement temperature is nice and stable so I just set for 13.85 and it's fine. They're all in float state at this voltage.
> as soon as you go off mains/float voltage, and actually use this battery system, some batteries will deplete faster than others since they have less capacity. ... won't nearby large capacity batteries spread/smear their voltage with other batteries on the rail?
No, they're all in parallel. All their voltages are precisely the same at all times, modulo a few microvolts for drop across the wires and fuses if there's a lot of charging or discharging happening. If it's actually floating in steady-state, they're identical within my ability to measure.
> are you using it for purely backup purposes only, and not recharging?
Purely backup. When I add a new battery to the pile, I charge it to 13.85 with a separate benchtop supply first, so there isn't significant current flowing through the fuse block when I plug it into the parallel bank. I fuse each tap at 2A since there's never a reason to move more, though all the wiring could do 20, I'd rather know about a problem early. Some are probably 3A because the baggie of fuses didn't have enough 2's in it, but whatever.
> do you have something that disconnects batteries when their voltage drops too low? and does it charge them too?
It's never run long enough to reach a low-voltage disconnect point, but if it did, the LiFePO4's BMSs would disconnect them around 10.5v, and the lead-acids would continue down to about 5.6v, which is where the MC34063's can no longer control their internal Darlington switch and they fall out of regulation. It's anybody's guess how long the Pi stays online after that, but probably not long.
When power came back, the RS25-12 would try to put 13.85v onto the bus and promptly exceed its overcurrent protection, and go into hiccup-retry mode, burping out little bits of charge every few seconds. It's possible that this would eventually bring the lead-acids up to 11-ish volts where it would restart charging in earnest, and this MAY happen fairly quickly because there's very little useful capacity below there, so it might not take many burps to get there, but I haven't tested it. It's also possible that the remaining loads would keep it down and prevent this from working, so I'd have to add a low-voltage disconnect to avert that. But that adds another possible point of failure, in anticipation of an event that should be exceedingly rare.
Or, I could add a proper charger or a CC/CV-limited PSU that would recover such a state in a sensible and expedient way, or if I add the solar MPPT it'll know what to do with an overdischarged bank. But realistically if power's out long enough for this to die, I probably don't care about the internet anymore.
It works surprisingly well from mobile browsers, too.