Why are computer battery back-ups so unreliable?

[spikespiegal]

Can somebody explain to me why computer UPSs (battery back-ups) are so absurdly unreliable? Here in Michigan the 12volt 'Walmart Special' battery in my car can tolerate several years of extreme temperature changes, several jump starts due to leaving my lights on in the parking lot, and cranking a car engine starter several times a day.

However, the typical desktop computer UPS I've been deploying the past several years (all brands) seems to last an average of less than two years sitting in a cozy office and rarely being tripped more than a couple times over it's lifetime.

I've heard several explanations for this ranging from the poor floor life of gell type batteries to cheap chinese voltage regulation and recharge circuits. Other than paying for 'server farm' class UPSs that weigh (and cost) a ton I'm really looking for an alternative solution that works better and can be trusted for 3-4 years. Any way to incorporate a conventional 12volt car battery into this type of circuit and get more reliability?

 

[CycloWizard]

Lots of reasons. Battery type, cost, current demands, and more factors are all important. UPS supply at a much higher voltage than a car battery, which likely has a lot to do with it as it's harder to generate a very high voltage than a low voltage simply due to the chemistry involved.

 

[degibson]

A computer requires a much friendlier current stream than an automobile starter. Starters don't mind current variation, for instance -- computers do.

 

[mooseracing]

Originally posted by: spikespiegal
Any way to incorporate a conventional 12volt car battery into this type of circuit and get more reliability?

There are home brew solutions for using 12volt auto/farm batteries on google. I for one like the idea of a UPS that has a SEALED battery that is not leaking acid everywhere as it charges. Also a UPS can be at any angle because of the sealed battery. It is also not spewing toxic combustable fumes into my air.

 

[Modelworks]

Car battery NO.
But you can use what I have before.
A rack style ups that was designed for extended run times, internal cooling to allow for hours of running off the battery.
Connected to it is two 12volt marine batteries. (Most UPS are 24VDC or 48VDC so you have to wire the batteries in serial)

Marine batteries are used because they are deep cycle, meaning you can discharge them, then recharge them many times without damage.
Car batteries are meant for a short high current load then a recharge.
If you repeatedly discharge a car battery the plates will buckle and the battery will fail.
Deep cycle batteries are not cheap but can give you run times for as long as you want.
Several hours with a typical pc is common when using two batteries.

Well maintained batteries do not leak everywhere.
They do omit Hydrogen when they charge but its not like its enough to fill a room.
And it is only produced if the battery is recharged.
If its sitting there idle it produces next to none.

There are also sealed marine batteries available to remove the above concerns.

 

[spikespiegal]

No offense to everyone, but besides the reply from Modelworks there was nothing helpful here.

Rant = on

If I ask a theoretical physics question about quantum entanglement in this forum I get somebody with a graduate degree answering for me. If I ask an applied engineering question with some form of practical application I have to sort through answers that sound like a 5th grade science glass.

(Listening to the steady exodus of engineering jobs away from the U.S., and no longer wondering why)

If you repeatedly discharge a car battery the plates will buckle and the battery will fail.

I understand the difference between standard car batteries and deep cycle car batteries. However, if you ever owned a car you've likely run the battery to the ground with the head lamps on, but the plates inside the battery don't "buckle". You can do this a number of times over the life of that battery and while you may shorten it's life, the battery doesn't stop working.

However, an absurdly high percentage of computer UPS's I've deployed over the years fail within a two year period while never enduring a fraction the discharge cycles of a car battery.

Why?

 

[Rubycon]

Two weak points on budget minded UPS' come to mind:

Poor quality SLA (Sealed Lead Acid) battery packs and Charging circuitry. Combine both and your box has a short life. A brand name UPS is worth the extra money. Don't hook up any large battery whether it's out of a car or a deep cycle battery - to a cheap UPS. Even if you get the voltage right (two required in series for larger UPS using 24V) its charger is not designed to replenish the capacity of such a battery bank. SLA and wet Pb also have completely different volts per cell, maintenance and equalization ratios. The end result will be excessive sulfate on the plates and a dead battery when called upon - much sooner than its life expectancy would normally dictate. Finally if and when it would be called upon you must consider the duty cycle of the budget box inverter in such a unit! The built in SLA packs will power a loaded system for minutes. With a huge pack you can theoretically extend the runtime into hours BUT the inverter heatsinks are not designed for such use and their parts can easily reach junction temperatures outside of safe operating area resulting in catastrophic and sudden failure. This is when you HOPE the inline fuse does its job. :Q

If you need extended runtimes, I'd suggest looking for a QUALITY working second hand unit that may need new batteries. Ebay has many (look for higher APC, Deltec or even Ferrups models). You will save on shipping since the batteries are considerably heavy and the unit can be shipped without them. You can purchase decent deep cycle rated marine batteries, mount them safely in boxes, make sure the wire going to the UPS is of sufficient gauge and properly fused with a disconnect at the battery bank itself! This should be able to run multiple computers and if you need really long run times consider purchasing a generator system that can be used to power your computers through the UPS. Connect it before the batteries get too low. A ferroresonant transformer based UPS provides outstanding protection (full galvanic isolation from utility mains) so if the genny is dirty (and most cheap ones are!) this will not be an issue for your sensitive pc equipment!

 

[dkozloski]

This is our local solution. It supplies 27Megawatts for 15 minutes.


http://www.gvea.com/about/bess/

 

[bobsmith1492]

Originally posted by: spikespiegal
No offense to everyone, but besides the reply from Modelworks there was nothing helpful here.

Rant = on

If I ask a theoretical physics question about quantum entanglement in this forum I get somebody with a graduate degree answering for me. If I ask an applied engineering question with some form of practical application I have to sort through answers that sound like a 5th grade science glass.

(Listening to the steady exodus of engineering jobs away from the U.S., and no longer wondering why)

If you repeatedly discharge a car battery the plates will buckle and the battery will fail.

I understand the difference between standard car batteries and deep cycle car batteries. However, if you ever owned a car you've likely run the battery to the ground with the head lamps on, but the plates inside the battery don't "buckle". You can do this a number of times over the life of that battery and while you may shorten it's life, the battery doesn't stop working.

However, an absurdly high percentage of computer UPS's I've deployed over the years fail within a two year period while never enduring a fraction the discharge cycles of a car battery.

Why?

Hey now, I resent that...

Regarding automotive lead-acid batteries, full discharges are the worst things you can do for their lifespan. Do it a few times and it will soon fail. That said, a UPS is completely different than the automotive application and so hardly a valid comparison.

This is hardly a worthy thread for your rant; most other threads here I would agree but you're getting good advice here. It seems like you're just venting steam that all your UPSs are failing? I don't think there's a magic solution other than buying a higher-quality UPS or building your own - that's what I'd do...

 

[firewolfsm]

Computer power has to be EXTREMELY stable and precise. Once small changes show up, it stops working.

 

[dkozloski]

Originally posted by: firewolfsm
Computer power has to be EXTREMELY stable and precise. Once small changes show up, it stops working.

That's BS. Most computers wil tolerate + or - 10% variations on AC supply power. That's why computer power supplies have regulators and filters.

 

[degibson]

Originally posted by: dkozloski

Originally posted by: firewolfsm
Computer power has to be EXTREMELY stable and precise. Once small changes show up, it stops working.

That's BS. Most computers wil tolerate + or - 10% variations on AC supply power. That's why computer power supplies have regulators and filters.

10% is pretty stable if you ask me.

 

[Smilin]

Based on my own subjective observations I would agree. Those things do seem to shit a lot. I would say it's confined to "desktop" backups though. In my experience very large (single unit for a whole server room) backups are fine and run reliably for years.

For the desktop failures a couple factors come to mind:
Cheap batteries
Cheap electronics to handle the switching and failover
VERY cheap software that never seems to do a graceful shutdown (so the battery depletes)
Cheap users who purchase just enough to get by leaving the unit running near full capacity.

I think you add these factors up and you're just rolling the dice too much.

Cheap parts + heavy demand = high probability of failure.

 

[Mark R]

Originally posted by: degibson
10% is pretty stable if you ask me.

Not really. +/-10% is truly terrible, and is outside legal limit for grid power in most places.

Rapid variations of +/- 1% are easily visible in mains powered lighting. +/-2% would drive you nuts.

Computer PSUs are, in general, very tolerant of mains voltage fluctuations - and I wouldn't be surprised if most would continue running, with DC outputs in specification, with mains voltage of +/- 20% of normal. That's not to say running outside the recommended range is acceptable long term.

Both excessively low and high mains voltages cause excessive electrical and thermal stresses in PSUs, which can lead to shortened lifetime, lower reliability, or outright failure/burnout.

 

[TuxDave]

Originally posted by: spikespiegal
No offense to everyone, but besides the reply from Modelworks there was nothing helpful here.

Rant = on

If I ask a theoretical physics question about quantum entanglement in this forum I get somebody with a graduate degree answering for me. If I ask an applied engineering question with some form of practical application I have to sort through answers that sound like a 5th grade science glass.

(Listening to the steady exodus of engineering jobs away from the U.S., and no longer wondering why)

ORRRR... you can say that people enjoy posting even if they are not the best informed people on the topic. :roll:

 

[Foxery]

I may as well ask, what brand of UPS is failing you, and what exactly is happening to it? My APC unit is well past its intended lifetime, (I think they recommend replacing the battery every few years) and still works like a charm... whereas I've been warned never to buy Belkins or generic/store-brand knockoffs under any circumstances because they are not good at supplying a "clean" current.

I mean, my APC came with a crazy warranty that pays you silly amounts of money if their unit ever lets your computer fry, and mine has definitely saved my ass.

 

[degibson]

Originally posted by: Mark R

Originally posted by: degibson
10% is pretty stable if you ask me.

Not really. +/-10% is truly terrible, and is outside legal limit for grid power in most places.

Rapid variations of +/- 1% are easily visible in mains powered lighting. +/-2% would drive you nuts.

Computer PSUs are, in general, very tolerant of mains voltage fluctuations - and I wouldn't be surprised if most would continue running, with DC outputs in specification, with mains voltage of +/- 20% of normal. That's not to say running outside the recommended range is acceptable long term.

Both excessively low and high mains voltages cause excessive electrical and thermal stresses in PSUs, which can lead to shortened lifetime, lower reliability, or outright failure/burnout.

10% is stable relative to the tolerance of external automobile starters, which was the context of the conversation. I'm not very well versed in the tolerances of desktop PSUs -- but 10% RMS input variation is common on server PSUs, subject to the same kinds of issues you mentioned.

Suffice to say that I'm not surprised so many folks have trouble with UPSs ... there's so much variation in electronics that the likelyhood of getting a good desktop PSU and a good UPS and a good whatever-else-is-involved in second- or third-order effects is probably pretty small (on a budget, anyway).

 

[Smilin]

Originally posted by: TuxDave

Originally posted by: spikespiegal
No offense to everyone, but besides the reply from Modelworks there was nothing helpful here.

Rant = on

If I ask a theoretical physics question about quantum entanglement in this forum I get somebody with a graduate degree answering for me. If I ask an applied engineering question with some form of practical application I have to sort through answers that sound like a 5th grade science glass.

(Listening to the steady exodus of engineering jobs away from the U.S., and no longer wondering why)

ORRRR... you can say that people enjoy posting even if they are not the best informed people on the topic. :roll:

ooorrrrrr.... If you're so smart answer your own !#$!ing question

 

[CycloWizard]

Originally posted by: spikespiegal
No offense to everyone, but besides the reply from Modelworks there was nothing helpful here.

Rant = on

If I ask a theoretical physics question about quantum entanglement in this forum I get somebody with a graduate degree answering for me. If I ask an applied engineering question with some form of practical application I have to sort through answers that sound like a 5th grade science glass.

(Listening to the steady exodus of engineering jobs away from the U.S., and no longer wondering why)

Just like it's no surprise that the quality of responses in HT has gone down with the recent influx of knuckle draggers like yourself. If you want a thorough answer, ask a question that would elicit such an answer. I answered your question very succinctly and correctly in the very first response in this thread. If you wanted a more in-depth answer, you should have asked a more in-depth question.

 

[TuxDave]

Originally posted by: Smilin

Originally posted by: TuxDave

Originally posted by: spikespiegal
No offense to everyone, but besides the reply from Modelworks there was nothing helpful here.

Rant = on

If I ask a theoretical physics question about quantum entanglement in this forum I get somebody with a graduate degree answering for me. If I ask an applied engineering question with some form of practical application I have to sort through answers that sound like a 5th grade science glass.

(Listening to the steady exodus of engineering jobs away from the U.S., and no longer wondering why)

ORRRR... you can say that people enjoy posting even if they are not the best informed people on the topic. :roll:

ooorrrrrr.... If you're so smart answer your own !#$!ing question

:thumbsup:

 

[Throckmorton]

Does a UPS continually let the computer draw power from the battery while simultaneously charging it?

 

[Mark R]

Originally posted by: Throckmorton
Does a UPS continually let the computer draw power from the battery while simultaneously charging it?

It depends on the type of UPS.

Cheap ones run the computer off mains power, while charging the battery. When power fails, or a power quality problem occurs (overvoltage/undervoltage), a transfer switch disconnects the computer from the mains input, and transfers it over to the inverter which is powered by the battery.

This is a cheap and simple design - but there is a short power glitch on the output of the UPS. This shouldn't be a problem: A good quality PSU running at 50% of maximum load should withstand about 0.05 seconds of power loss - which is plenty of time for the transfer switch to activate (most transfer switches will transfer in less than 0.01 seconds).

Intermediate level ones, will run the computer off mains power, and will electronically correct power quality problems. However, in a power failure, they will use a transfer switch to switch over to inverter power.

High level ones, convert the mains input to DC, and use that to charge the batteries and run the inverter. The computer runs off the inverter at all times. In the event, of power failure, the inverter will instantly revert to battery power, leaving no glitch, drop out or any detectable fluctuation in power on the output.

These are considerably more expensive - there design is less energy effcient. Cheap UPSs power the comp directly from the mains when power is available. 'Double conversion' UPSs, have to expend energy in 2 conversion processes. Additionally, in a cheap UPS the batteries will run out after 5 minutes, so the inverter will only ever run for 5 minutes at a time, a couple of times a year. A dual conversion UPS will run the inverter continuously 24/7, so it must be built much stronger and have better cooling.

And even then, you can't get rid of the monitoring and transfer system. The risk with a dual conversion UPS is that the inverter burns out - so there has to be monitoring systems, and a transfer switch that will transfer the computers onto direct mains power, if the UPS inverter fails.

 

[davesaudio]

My personal experience is 4 of 4 tripplites have failed less than a year- all were electronics failures
agree with Mark R 's assesment on underdesigned electronics assuming extremely limited dutycycle.
All 4 of my APCs running for years, guess which I buy next time
Who has had bad batteries?

 

[apoppin]

Get a good one

i have a cheap, guaranteed-for-3-years, *dual-battery pack* APC [pro line; yech otherwise] UPS that is 1500va and 850w
- as it is *dual*; if one battery goes bad, i can replace it without shutting anything down and i still get 15 minutes of backup with the remaining one; plenty of time for a RMA or to get a replacement. You need to think ahead. I had an awful time my rig shut down in the middle of a critical update because my underpowered UPS failed. Look at my sig. Each 2900xt can draw near 300w each at max load!!

i can have my rig screaming at max load [~700w+] and pull the plug and still game for about 1/2 to 1 hour before shutting down
- not bad for $110 shipped last month from Frys .. look in hot deals there is a lesser 1300va for $100 at OM i think; you missed another from Frys last night!

*Practical application* of HT^^
i hope you don't mind .. i am just the 'video guy' at ATF

the Moral of the Story - stop buying UPSes from WallyWorld
[wal-mart for the "hot deal" challenged]

EDIT : oops

i forgot this IS "HT"


let me clear my throat

Yes, IF your UPS can be adapted, you *can* use motorcycle batteries or even some marine batteries

now .. am i helpful, spikespiegal ? .. or should i just go back to flaming in P&N ?
- what did you think of my thread on 'Warp Drive' here? It finally turned out to be excellent research for the metaphysical; i appreciated all the answers.

 

[mooseracing]

I think it's been overstated but if you want it to last spend the money, everyone hates to do it yet everyone will complain if it isn't up to our standards. There's a reason we have a 3000VA Rackmount battery for our servers. There's also a reason we have a big Liebert that powers multiple important rooms, has multiple hot swap batteries, qualified electrician has to start it up for warranty purposes, cost over 30k.

I've haven't seen a single desktopp brand I like including APC, Ultra, Tripplite.