UPS lifetime and recommendation

[Ericloewe]

We're closer to that than ever. It's a reality with luggable powerbanks and USB PD. Most laptops can be charged via USB PD, and soon all will.
On the ATX side of things, it's more complicated because there's no consumer-friendly standard for delivering lots of DC power. Hell, there's barely one for enterprise, and it's a bizarre holdover from the analog telephone days that nobody uses unless they're a telecom.

 

[WI_Hedgehog]

My understanding is UPS don't run lithium-based batteries in indoor applications because of the gasses they give off if they rupture, which is almost always due to fire, which is almost always due to over-heating when charging, which in my opinion (as an R/C guy and car guy) is due to crystallization formation that produces sharp pointy crystals that pierce the internal "plastic" insulator and cause an internal short which causes the runaway overheating problem. My understanding is "government regulations in the U.S. do not allow it," though which regulations vary (in Wisconsin, for instance, solar power must only feed the grid, not the home, and no standby batteries can be located in the home nor run the home). So it is not as simple as "tweak the charge cycle."

Separately, I've had APC rack UPS units get the lead acid batteries so hot the batteries swelled and were almost not removable, so the potential for fire is real. Somehow I think temperature sensors wouldn't be expensive to integrate into the units, but that's me and my $1.75 mentality, so no offense to the engineers at APC (who are probably constrained by penny-pinching bean counters).

From what I understand, industrial UPS battery life expectancy is a 2.5 year replacement cycle, and UPS unit replacement cycle is 5 years. (I go about 7.5 years at work and recycle the units by buying replacement batteries bulk and using the units at home, saving the company the disposal fee.)

Lithium-ion batteries found to produce toxic gases

Lithium-ion batteries can produce dozens of dangerous gases when overheated, according to a new study from the Institute of NBC Defence and Tsinghua University in China.

eandt.theiet.org

https://eepower.com/news/100-toxic-gases-from-li-ions-its-not-just-about-fires/#

 

[Constantin]

It really depends on the application. For example, deep-cycle LA batteries can be made to handle many deep dis- and re-charge cycles, all while being juggled around (fork truck). Charging stations are typically in well-ventilated areas and feature Hydrogen-sensors to deal with a cooking battery.

VRLAs are more common in UPS applications because few people want to deal with hydrogen in a computer environment. At the right charge rates, no excess pressure will build up in a VRLA and should the burst disks go, so does your battery as there is no way to refill and they are starved for electrolyte from the get-go, especially the AGM variety.

Most vendors build their UPS to meet a price point and fill them with VRLAs that you might find in a motorcycle or similar small motor application. They are filled with skinny plates good for starting, not deep cycling, which is why most UPSs just like most cars have to have their LA batteries replaced once they’ve been run down to zero a few times.

Skinny plates allow more grids in parallel, which give you more amps but less battery grid failure resistance. Deep cycle LA cells feature thicker plates for that reason. I used a large set of bussed AGMs in the past for multiple hundred amp cranking power, those cables had to be 0000, ie the conductor was as thick as your thumb.

 

[asap2go]

No real advice for Chris, but I'm finding it very frustrating that everything on the market is lead-acid until you get into very expensive commercial units (e.g., https://tripplite.eaton.com/smartpr...fepo4-batteries-2u-lcd-usb-db9~SMART2200RM2UL). LiFePO4 batteries are widely available, have a far greater lifespan (both in terms of cycles and in terms of years), have a much higher energy density, and would need only very slight tweaks to the charging profile. They're more expensive than lead-acid, to be sure, but not greatly so. They really should be the norm these days, or at least available at mid-range capacities.

I also thought about DYIng a UPS for that reason.
48V LiFePo4 Battery, and an Inverter + Charger from Victron. That's a solution often used in camper vans. But I'll do that once my cheap Legrand has fried my network hardware.

 

[danb35]

I also thought about DYIng a UPS for that reason.

As you say, it's a simple enough proposition, at least in principle, particularly if you do online double-conversion. No switching to deal with then, just line -> charger -> battery -> inverter -> load. And I've done this for my wife's long-arm quilting machine, putting everything in a milk crate--it's bulky, and it's ugly, but it works, and will run the machine for several hours if need be. And she doesn't mind that it's ugly, because it sits out of sight.

The problem comes with monitoring the thing. Unless I'm mistaken, nut doesn't know how to talk to a Victron inverter-charger, so your server doesn't know line power has gone out, nor does it know when the batteries are low. For the application of that unit, it doesn't matter so much, but for running your NAS, it's a bit more important.

I'm really surprised that there isn't a good DIY/open-source solution here. esr had been working on one, but the last commit was over 3 years ago, so it looks kind of like a dead project:

Eric S. Raymond / upside · GitLab

Design and implementation of an open-hardware, open-software Uninterruptible Power Supply unit.

gitlab.com

Last week, I put together four 12V-nominal LiFeMgPO4 batteries, fabricated a pair of adapter cables, and plugged them into my rack-mount APC UPS after disconnecting all the lead-acid batteries. Pretty it isn't:

But they power it quite nicely (over an hour on the runtime calibration), and the charge/discharge cutoff voltages the UPS uses are compatible with these batteries. My unit is convenient in this regard in that it has a connector on the back panel (a pair of blue Anderson SB50 connectors, and the color matters) for an external battery pack, and it's designed for a "dumb" pack--it just expects power in/out, no communications.

 

[dak180]

I'm really surprised that there isn't a good DIY/open-source solution here. esr had been working on one, but the last commit was over 3 years ago, so it looks kind of like a dead project:

From what I can tell it is not dead per se, but it is stalled due to a lack an EE power guy to design the parts that need certs.