UPS-"lite"? (Power, not postage)

[Havin_it]

'ning,

I was thinking about getting a UPS to protect my home server, but the darn things all seem to be huge car-battery-filled things designed to keep systems running for a moderate period of time. Noise is often mentioned as well - not what I want for a bedroom-situated unit.

Doesn't anyone make a more simple, fail-safe type device that just comprises a small consumer-grade battery pack and an interface for ACPI or whatever signaling to the OS, just enough to allow (and initiate) an immediate (but graceful) shutdown?

And if not, why not?

[sikpuppy]

You need to work out how much power your server uses, and how long it takes for a shutdown.

As for them "all" seeming to power a system for a "moderate" period of time, where have you seen, heard or read that?

http://www.apc.com/products/runtime_for_extendedruntime.cfm?upsfamily=29 see here, not for the brand but just for some idea of what you want.

As for noise, I don't think that the UPS will make a noise until it switches from "standby" to battery power.

[mvaterlaus]

[energyman76b]

I have a Powerwalker UPS.

Works with nut.
High/width: three half heigt 51/4" slots
length: less than a standard keyboard
600+ VA

keeps my stuff running for more than 20min, sends battery low signal, system shuts down.

No noise in AC mode.
A bit of noise in DC mode.
No fans
Beeps can be turned off.
_________________

[Havin_it]

Thanks all for the replies.

I may have generalised a bit more than is wise, my experience is based on a Belkin unit at work which is also fanless but does emit a very low but still noticeable hum at all times. It might not bother me of an evening, but it probably would my gf who has the ears of a bat (and frequent insomnia). Also I came across many forum posts where people were decrying units that had promised to be silent but really weren't.

My point really is that all these devices seem like overkill for the basic purpose of a safe, immediate shutdown. Even the smallest units I've seen, judging by their size, contain lead-acid batteries. What's wrong with a laptop-grade (hell, smartphone-grade would probably meet my needs) Li-ion battery, coupled with a charge regulator that squeezes the best lifespan out of the battery?

The Belkin unit had two of the mini-car-batteries on board, which lasted about 3 years (I've had phone batteries last longer) and replacement, which had been touted as an easy end-user operation, was anything but: I nearly hosed the unit in the process because the internals were like a frickin' Gordian Knot.

My imagined device would be designed to make battery replacement easy, like tv-remote easy. Dual slots to allow constant operation while replacing would be a bonus, but more than most would need.

Safe, immediate shutdown. That's all. Everything I see still seems over-engineered / over-specced for this simple goal.

[energyman76b]

because Lead batteries can deliver the high currents needed and LI-io can not?

My devise doesn't make any sounds unless it is in battery mode.

And with a online or line interactive UPS you need a pretty big VA number so your starting PSU with its PFC will not overload it.
_________________

[eccerr0r]

actually Li-ion can produce the high currents (or you could just string a lot in series and use voltage instead of current). So why isn't it used?

- Batteries just sit there unused and fully charged: not a big issue but Li-Ion was designed to be used cyclically.
- Li-Ion Batteries cost more than lead acid per unit storage: not healthy for pocketbook.

A 3.7V 2.4Ah cell (18650) costs about $5 for 8.8Wh. A deep cycle car battery costs $80 for about 500Wh. To buy a Li-ion pack for 500Wh costs at least $250. I think it's near 3:1 cost for same capacity across the board for all capacities.

I think lead acid is probably the best cheap battery for UPSes, especially since weight does not matter. It just is a bit larger volume. Lead acid should be very stable, it's a proven technology that has been around for a long time.

(My UPS also is silent when idle (or at least I can't hear it, it doesn't have a fan). It makes noise when it's inverting during a power failure, or warning that the batteries need to be replaced.)
_________________
Intel Core i7 2700K@ 4.1GHz/HD3000 graphics/8GB DDR3/4x500GB RAID5/180GB SSD
What am I supposed to be advocating?

[Akkara]

I've posted before on this topic. Here's a off-the-top-of-my-head discussion on why things are as they are. It comes down to chemistry. Basically, batteries suck. Every last one of them, no matter what kind, there's tradeoffs to be made, and often ugly ones.

- Lithium-ion: they really don't like to be fully charged for long periods of time. From the moment they are made, they gradually lose capacity due to irreversible chemical changes over time. This process accelerates if they are fully charged (I've read numbers stating around 20% per year capacity loss). It also accelerates with warmer temperatures. Hot and fully charged together can kill a battery within 18 months. (That's why batteries in laptops that are mostly kept plugged in don't last.)

And, of course, there's lithium's propensity for spectacular failure. Oh, and don't discharge a Li-Ion too much, either. It can blow up when trying to recharge it from deep discharge.

And since a UPS battery is fully charged 99+% of the time, Li-ion is definitely not indicated for that application. (You could design one that charges the batteries to only 70% or so of max, but then you need more batteries for the same runtime == more expense).

- Lead Acid: These love staying fully charged for long periods of time. Perfect application for a UPS.

But they hate high discharge currents unless specifically built for it (e.g., car starter batteries, but they have their own other set of issues). Realistically, to get a lead-acid battery to last in a UPS, you have to aim for around an hour of runtime just so you're not overstressing the battery. You might be able to get away with 30 minutes if the outages are infrequent, but don't optimize much beyond that point unless you like changing batteries every couple years.

(Ever see the ratings on UPS? a supposed 1000W unit will say something like 40 minutes at 100watts, 15 minutes at 200 watts, 4 minutes at 500 watts, and 90 seconds at 1000W. Notice how the runtime drops fast - much faster than straight-line proportional - as the discharge rate goes up? That's a clear indication of overstress resulting in chemical and electrical inefficiency.)

- NiMH: like to be cycled periodically. Which the UPS won't do. They generally don't seem to work well, tho I don't recall the reasons right now. The newer ones might be better these days, I don't know. They certainly can put out the current so this technology could work in a short runtime UPS. I don't know how memory effect and lifetimes compare tho.

- Supercaps: This, when it gets cheap enough, will finally provide a good solution to a short-runtime UPS. See here: http://www.youtube.com/watch?v=EoWMF3VkI6U .

Until then, stick with lead acid, and aim for at least 40 minutes of runtime and ideally an hour or more - even if you only need 5 minutes. Otherwise you'll be changing batteries every couple outages.

[Havin_it]

Some really good insight here, thanks all

I never considered relative expense of technologies I must admit: just saw it as:

lead-acid = dirty big thing = must be expensive

vs

Li-ion = survives constant use in laptop for many years [if you're a bit careful] + can't be *that* big a price component of laptops/phones/etc that use it - but then that gets covered by economies of scale I guess.

I've certainly experienced the oh-too-fast degradation of Li-ion when I used to leave it constantly docked in past laptops while on AC, but as you (Akkara) say, there are ways'n'means of eking out really good lifetime if you manage the charge level sensibly - certainly better than lead-acid IMHO.

And here's the thing: a well-designed device, with cleverly-written firmware, *could* achieve this I believe. Now I will admit it might be more of a niche offering than I initially thought (as most folk will be happy with the current type of device if they need one), but I'm still designing my dream device in my head.

From what I've read (BatteryUniversity has a great article on this IIRC) Li-ion doesn't really care at all about full charge-cycling, so as you say you could, with smart firmware, keep it ticking along within a narrow band (65-70% charge or whatever; I had heard 55% was optimal myself, so I like yours better in this context ). The battery itself could be entirely idle most of the time, with the device checking the level only very occasionally because it'd be fairly predictable. And as the device, apart from being a power conduit, doesn't have much to do, it could run pretty cool (and insulating the battery from what heat there was shouldn't be rocket surgery).

Once the maths/physics of all that are worked out to be optimal, you could have a very long-lasting device that simply kept your data safe from the mechanical ill-effects of a blackout, with a replaceable part that'd be a lot cheaper to ship.

OK, you can go ahead and shoot holes in it now