UPS

[PlainName]

Currently using an APC 3000RM and the batteries have just died. Replacement cost is anywhere from £200 to £600 (or more) depending on source. I am thinking that perhaps this would be a good opportunity to replace the UPS with a more modern one, perhaps using Li-ion batteries.

Requirements are not excessive - it is supporting approx 1kW of kit (peak - usually it all hums along at <350W) but the longer it can stay up without power the better. No glitch when switching. And network access for config, control and reporting of status. SNMP definitely useful.

Anyone have a suggestion?

[shapirus]

If the old UPS is good at what it's supposed to do and not excessively smart (being an APC, it requires extra attention to the latter), meaning, if it can use a generic two-terminal non-serialized battery, then a LiFePO4 battery made to fit inside it might be good a way to relatively cheaply give the UPS a new life.

Watch out for max discharge current, though, and an active balancer is a must. Working voltages of LiFePO4 make them a perfect fit to replace lead acids in UPS'es.

Otherwise, if getting a new UPS, while I can't recommend a certain brand or model, at least consider getting a modular solution, that is, a standalone UPS without a built-in battery, which would allow to connect any external battery of your choice that meets the required working voltage range and charge/discharge current.

[Jeroen3]

Why would you want an expensive LifePo4 with all the added risks over a cheap and relatively safe sealed lead acid battery?

Also, if lead is already too much, then lifepo4 would double too much.
eg: it looks like sla 24V 17ah is around €150 each. Lifepo4 is around €300 for only 10Ah.

[shapirus]

Why would you want an expensive LifePo4 with all the added risks over a cheap and relatively safe sealed lead acid battery?

YMMV, but in my case with relatively frequent power outages SLA wouldn't last more than about a year (powercom kin-1000ap) with load no higher than 20% of nominal power. Besides, backup power run time with SLA was ridiculously low. After a few replacements in a few years, I got tired and decided to build a proper battery for it.

The LiFePO4 battery with which I replaced the original 2xSLA ones (admittedly of a larger physical size, for which I had to remove parts of the UPS'es metal frame inside) have now lasted over 4 years and is still going strong.

Also, if lead is already too much, then lifepo4 would double too much.

That lead price sounds insanely high, I would assume because it's APC branded.

eg: it looks like sla 24V 17ah is around €150 each. Lifepo4 is around €300 for only 10Ah.

Maybe if you search for the most expensive cells, I don't know. What I have in my UPS is a 8s3p LiFePO4 battery made of decent quality chinese 24x 32700 cells with a total cost of ~US$120. Add assembly costs, wires, connectors, an active balancer, ok, let's say $150 total. This is nominal 24V x 18Ah or, in practice, about 400-450 Wh measured (when new -- but based on run time it hasn't degraded much).

Now, you cannot compare same-Ah SLA and LiFePO4. The latter will provide at least 2-3 times more energy at full charge than the SLA of the same nominal Ah capacity. Capacity degradation over time will be absolutely incomparable as well.

SLAs suck. UPS is one of applications where they suck most prominently.

[PlainName]

Thanks, both.

These batteries are 6 years old and the failure mode is ok, ok, ok, damn - server's rebooted. Although the UPS is telling me the batteries need immediate replacement, it insists there is still 135mins uptime available, and that despite the shutdown caused by the periodic test yesterday. So it would be nice to have some indication of arriving doom (although actually doing something with that warning is a different matter...).

Against the LiPo is the vision of my office doing a good impression of a Tesla and initial expense (I assume ongoing expense would be lower because the batteries would last longer).

Hadn't considered replacing the existing batteries with LiPo on account of the charging needs being rather different. That would be useful in that it would retain the existing setup (which, apart from short time between OK and NOTOK, works well). OTOH, it's an old APC and no doubt on its last legs.

Hmmm.

[shapirus]

These batteries are 6 years old and the failure mode is ok, ok, ok, damn - server's rebooted. Although the UPS is telling me the batteries need immediate replacement, it insists there is still 135mins uptime available, and that despite the shutdown caused by the periodic test yesterday. So it would be nice to have some indication of arriving doom (although actually doing something with that warning is a different matter...).

It might be useful to test the batteries outside the UPS to make sure that it's actually a battery problem and not something in the UPS electronics or relays.

Hadn't considered replacing the existing batteries with LiPo on account of the charging needs being rather different.

LiPo (meaning LiMnCo) is a different animal. I wouldn't recommend them because of the associated safety risks, let alone trying to match the required working voltage range.

LiFePO4 is almost a drop-in replacement for SLA, with a few caveats, among which are price considerations (however price per life expectancy is usually better for LiFePO4). It might or might not be worth it in your case.

OTOH, it's an old APC and no doubt on its last legs.

My Powercom is over 20 years old, and 20 years in active service. Still working like new, except of course for the batteries. Your APC might be similar in this regard. But then again, question is also whether it will accept any new battery or it needs vendor-locked ones.

[PlainName]

LiPo (meaning LiMnCo) is a different animal.

Yes, sorry, brain has room for SLA, Alkaline, NiMH, everything else.

LiFePO4 is almost a drop-in replacement for SLA, with a few caveats

That's looking better
The UPS manufacturing date is 2007 so nearly 20 years here as well. So far as I know it would take anything - the batteries look like normal SLA and this last set was from some third party. They are definitely not APC brand. I'll have a browse and see what's involved.

It might be useful to test the batteries outside the UPS

Yes, good point. Not done yet since they are not easy to whip out at the moment.

[thm_w]

Why would you want an expensive LifePo4 with all the added risks over a cheap and relatively safe sealed lead acid battery?

Also, if lead is already too much, then lifepo4 would double too much.
eg: it looks like sla 24V 17ah is around €150 each. Lifepo4 is around €300 for only 10Ah.

Here you can buy 50Ah lifepo4 12V for $140, 100Ah is $240.
I don't know what this device uses, it has 8x18Ah cells inside or something? that seems like a lot. Maybe the packs can be used externally instead.

[PlainName]

I don't know what this device uses, it has 8x18Ah cells inside or something?

8 batteries arranged as 2 x 24V 14.4Ah

[PlainName]

Since there was nothing reasonably similar near the replacement battery price, I coughed for a set at £169 (approx $214 in funny money). Back to working again now

Just noticed the firmware revision:  666.6.I

[5U4GB]

You can also get LiFePO4's that are drop-in replacements for SLAs, e.g. these ones.

[PlainName]

Couldn't track any down here (UK) in a reasonable time, and they are more expensive. But certainly something to think about in 2030 when this lot will be due to be replaced