Battery backup: problem with my APC?
in [Storage]
|
Prev: OK to leave approx 5% free space on 500GB NTFS partition?
Next: Clone QNX/POSIX HDD via Acronis or Norton Ghost?
From: Mike Ruskai on 19 Oct 2009 01:24 On or about Sun, 18 Oct 2009 03:35:47 -0400 did Yousuf Khan <bbbl67(a)spammenot.yahoo.com> dribble thusly: >Mike Ruskai wrote: >> If you insist that your claim about UPS's being junk is not ridiculous, then >> provide some model numbers or brands/capacities that you've tried. Even the >> very inexpensive 1100VA Belkin unit I started with several years back worked >> better than you describe. > >You sound like you work for a UPS company, otherwise why are you so >invested in it? Just accept that people have different experiences than >you, and move on. No, what I sound like is a person who's incredulous that someone could say something so blatantly ignorant about uninterruptible power supplies. I guess you've never had a file system crash, or lost an important document that hadn't been saved. Your reluctance to provide any detail demonstrates that you probably have little to no actual experience with UPS's of reasonable quality. It's like someone who's driven nothing but 20-year-old used junkers saying that all cars are unreliable junk.
From: David Brown on 19 Oct 2009 03:47 Rod Speed wrote: > David Brown wrote >> Rod Speed wrote >>> Ian D wrote >>>> Rod Speed <rod.speed.aaa(a)gmail.com> wrote >>>>> Ian D wrote >>>>>> kenk <kenk(a)nowhere.com> wrote > >>>>>>> I was sitting at my desk yesterday during a storm when >>>>>>> there was a split-second power outage. Despite the fact >>>>>>> that my battery shows all 5 lights lit, the computer >>>>>>> died. > >>>>>>> Is there a better brand to use than APC? Can the battery >>>>>>> be sub-par even though the test lights say it is OK? >>>>>>> What have others of you chosen? > >>>>>> I would be looking at your PC's power supply. There is a >>>>>> very brief delay between the loss of power, and the UPS >>>>>> switchover to battery power. > >>>>> Not with continuous UPSs, there is no delay at all with >>>>> those. > >>>>> Essentially the PC is running off the UPS output all the time >>>>> and the only thing that changes with the mains failure is >>>>> that the UPS isnt being charged anymore and runs off the >>>>> battery instead. > >>>> That's correct, and those units are the ones with true sine >>>> wave output, > >>> Nope, the type of output is an entirely separate issue. > >> Both types of UPS generate roughly sine wave outputs when the >> inverter is active. > > Pity he was clearly talking about TRUE sine wave output, which only a > small subset of UPSs produce. And it aint even the continuous/online > UPSs that mostly do produce TRUE sine wave output. > I don't know exactly what you mean by "TRUE sine wave" - all UPS's generate a sine wave of sorts when they are active (i.e., all the time for continuous/online UPS's, or during power fail for standby devices). The quality of the sine wave varies from fairly poor (lots of harmonics) to pretty good (few harmonics), but is /never/ pure. If you are restricting discussion only to those UPS's that produce good quality sine waves, then I'll have to take your word for how standby and online UPS prices compare. Baring the worst of the cheapo devices, the sine quality of any UPS is going to be good enough for computer hardware, and is typically better than you get straight from the mains. Thus it's not a restriction that I've considered. >> The difference is that with a standby UPS, the inverter is not >> active unless the power fails, so the output is just a filtered >> version of the input. > > Thats an entirely separate matter to TRUE sine wave output. > True - though again, there is no such thing as "TRUE sine wave output", merely more or less harmonics. >>>> and are relatively expensive. > >>> Not anymore with continuous or online UPSs. > >> Online or continuous UPSs are more expensive than standby UPS for >> the same ratings. > > In theory that is correct. In practice there isnt a lot in it with > the brand name domestic UPSs now. > OK. It looks to me that the difference has been getting smaller in recent years - perhaps we are merely arguing about what sort of percentage is a "big" difference. I also have been looking mainly at APC, rather than a spread of suppliers - I've used them, I have found them reliable, and they work well with Linux, so I haven't bothered looking at many alternative manufacturers. >> Judging by a quick check on APC's website, the difference is >> something like 25% more expensive for "Smart UPS online" compared >> to "Smart UPS" (standby). > > Thats not very much. > In a recent purchase, the online device I bought was actually twice the cost of a standby device with similar ratings. But that was because I needed it to work with a particularly hostile mains supply, thus it was a special case. To be honest, however, the prices I saw when I checked APC's website surprised me slightly - I had expected a bigger difference. But one should never let hard facts get in the way of a good argument. >> Additionally, smaller and cheaper UPSs are mostly standby types, >> while online ones are for more professional markets. > > Thats overstating it, particularly with the stuff out of china. > >> If you are wanting something that can give you 10 minutes at 300W, >> standby UPSs will be half the price. > > Wrong. > Well, that's based on looking at APC's website. There is no doubt that the cheapest APC standby device, which is perfectly good for a single system, is less than half the price of their cheapest online device. For their small devices, there is a huge difference in prices. But as I say I haven't looked at other manufacturers - in particular, I haven't looked at the low-end and unbranded devices. I simply took APC as a familiar professional-level UPS supplier, and I expect other professional-level suppliers to be roughly in line with them. If what you are saying is true of the low-end devices, then that's interesting news. I'm not sure I'd normally pick a no-name UPS - when looking for reliability during a power failure, the supplier's reputation is a factor. But there are certainly situations when a cheap UPS is much better than no UPS. >> Online UPSs are also less efficient for smaller systems - the >> double conversion wastes at least 10% of your electricity. > > Utterly mangled all over again. > No, it's quite simple - AC to DC conversion has some loses, DC to AC conversion has some loses. Simple passive filtering and surge protection, as used by standby devices when they are offline, has virtually no loss. I'm not too concerned about this myself, but other people certainly are, especially in larger setups. > And if you do care about that, you can also get replacement power > supplys that avoid the double conversion and still have the UPS > functionality. > If you are talking about DC supply buses to servers, rather than having an AC supply to each, then I think it's a very good idea. It is quite simply idiotic to take a high voltage AC supply, convert it to low voltage DC for a battery, turn it back to high voltage AC to deliver to a server's power supply, which then turns it back to a low voltage DC. Converting AC to 24V to 48V DC for battery storage, and passing that straight to a server's power supply would be significantly more efficient in energy use, and much smaller and cheaper in hardware. >> That doesn't mean that standby UPSs are a better choice for a small >> user - just that there are different balances to consider and the >> price difference (though less than it used to be) is significant. > > Not anymore. > >>>> I was assuming that the OP was using a regular, run of the mill >>>> APC UPS. > >>> There's plenty of continuous/online consumer grade UPSs now. > >>>>>> The PC power supply should hold during this period. Before >>>>>> I had a UPS, a split-second power outage would cause clocks >>>>>> to lose their time, etc., sometimes, even my monitor would >>>>>> blink, but the computer would hold without a glitch. If >>>>>> you want to test your UPS, pull the power cord with a load >>>>>> on the UPS. > >
From: David Brown on 19 Oct 2009 05:40 Rod Speed wrote: > David Brown wrote >> Rod Speed wrote >>> David Brown wrote >>>> Rod Speed wrote >>>>> Ian D wrote >>>>>> Rod Speed <rod.speed.aaa(a)gmail.com> wrote >>>>>>> Ian D wrote >>>>>>>> kenk <kenk(a)nowhere.com> wrote > >>>>>>>>> I was sitting at my desk yesterday during a storm >>>>>>>>> when there was a split-second power outage. Despite >>>>>>>>> the fact that my battery shows all 5 lights lit, the >>>>>>>>> computer died. > >>>>>>>>> Is there a better brand to use than APC? Can the >>>>>>>>> battery be sub-par even though the test lights say it >>>>>>>>> is OK? What have others of you chosen? > >>>>>>>> I would be looking at your PC's power supply. There is >>>>>>>> a very brief delay between the loss of power, and the >>>>>>>> UPS switchover to battery power. > >>>>>>> Not with continuous UPSs, there is no delay at all with >>>>>>> those. > >>>>>>> Essentially the PC is running off the UPS output all the >>>>>>> time and the only thing that changes with the mains >>>>>>> failure is that the UPS isnt being charged anymore and >>>>>>> runs off the battery instead. > >>>>>> That's correct, and those units are the ones with true sine >>>>>> wave output, > >>>>> Nope, the type of output is an entirely separate issue. > >>>> Both types of UPS generate roughly sine wave outputs when the >>>> inverter is active. > >>> Pity he was clearly talking about TRUE sine wave output, which >>> only a small subset of UPSs produce. And it aint even the >>> continuous/online UPSs that mostly do produce TRUE sine wave >>> output. > >> I don't know exactly what you mean by "TRUE sine wave" > > Should be obvious to even someone as stupid as you. > >> - all UPS's generate a sine wave of sorts when they are active > > Wrong, as always. Plenty generate a square wave instead. > You might know about /buying/ and /using/ UPS's, but you don't know much about sine waves, AC/DC conversion, harmonics, and related technicalities. A square wave (as produced by very low-end UPS's) is just a sine wave with large harmonics. And at the high end, good UPS's use PWM switching to produce something closer to a sine wave - even after filtering, it is still not a "TRUE" sine wave. >> (i.e., all the time for continuous/online UPS's, or during power >> fail for standby devices). The quality of the sine wave varies from >> fairly poor (lots of harmonics) to pretty good (few harmonics), >> but is /never/ pure. > > That last is just plain wrong. > I'll accept that the first part might be wrong - a square wave is a very poor sine wave, rather than just "fairly poor". As I have said, I haven't been looking at such low end devices. However, if you think that UPS's generate *true* sine waves with no harmonics, I'd love to see the circuit diagrams. I'd also like to know which brands go to great efforts to get closest to a sine wave - so that I can avoid them, as it would be a total waste of money. >> If you are restricting discussion only to those UPS's that produce >> good quality sine waves, > > I was JUST commenting on HIS claim about TRUE sine wave output, > fuckwit. So I can't talk about UPS's that don't produce mythical "TRUE sine waves" because a previous poster talked about "TRUE sine wave" UPS's, but /you/ can talk about any UPS's you like? I'm sure that makes sense to you, somehow. > >> then I'll have to take your word for how standby and online UPS >> prices compare. Baring the worst of the cheapo devices, the sine >> quality of any UPS is going to be good enough for computer >> hardware, and is typically better than you get straight from the >> mains. Thus it's not a restriction that I've considered. > > Completely and utterly irrelevant to the comment I made about his > stupid pig ignorant claim about TRUE sine wave output. > >>>> The difference is that with a standby UPS, the inverter is not >>>> active unless the power fails, so the output is just a >>>> filtered version of the input. > >>> Thats an entirely separate matter to TRUE sine wave output. > >> True - though again, there is no such thing as "TRUE sine wave >> output", merely more or less harmonics. > > Wrong, as always. True sinewave is NO harmonics. > Correct - a "true sine wave" has no harmonics. No UPS could possibly produce one. There is no point in an UPS even trying - it makes sense to limit the first few harmonics as much as practically possible, but you get very little gain from going further than that. Surely you are aware that a "true" sinewave exists only as a mathematical concept - anything in real life is only going to be an approximation? <snip the incoherent rodbot responses> >> If you are talking about DC supply buses to servers, > > Nope, its just as true of desktop PCs. > I've only seen discussions of DC supply buses in the context of server racks, where you have lots of machines together. Since desktop PC's are typically spread out, you can't conveniently have a single AC to DC supply for multiple machines. >> rather than having an AC supply to each, then I think it's a very >> good idea. It is quite simply idiotic to take a high voltage AC >> supply, convert it to low voltage DC for a battery, turn it back to >> high voltage AC to deliver to a server's power supply, which then >> turns it back to a low voltage DC. > > It isnt idiotic, its just not as efficient. Many dont give a damn > about the efficiency with a device that isnt taking that much power. > True enough for machines with low power requirements. But the discussion here has moved on to multiple machines - for server racks, the electricity price is often a very big part of the cost. >> Converting AC to 24V to 48V DC for battery storage, and passing >> that straight to a server's power supply > > Just as true of a desktop's power supply. > It is indeed true of a desktop's power supply - but unless a significant percentage of users start using UPSs with their desktops, the economics won't allow anything but AC supplies on desktops. >> would be significantly more efficient in energy use, > > Not very significantly, actually. > For people with big installations, even small percentages are significant. But a rough rule of thumb would be 3-5% energy loss for each high voltage AC to low voltage DC conversion, assuming top of the range converters. By avoiding the extra conversion to AC after the UPS and before the server supplies, you'd save up to 10%. >> and much smaller and cheaper in hardware. > > Nope, its actually more expensive, essentially because that approach > doesnt sell in the same volume that traditional UPSs do. > Economics of scale are certainly relevant. But server rack UPS and supply systems are not huge volumes anyway, and the energy savings would be appealing. And the hardware itself would be smaller and a lot cheaper once volumes are of similar magnitudes.
From: David Brown on 19 Oct 2009 08:15 Ed Light wrote: > David Brown wrote: >> Rod Speed wrote: > > David, alot of us have Rod filtered out. > I know, and I don't often enter discussions with him. However, he has some useful things to say if you can avoid triggering the rodbot switch, as I did for part of my post. Anyway, it's possible that other people find something of interest in what I wrote, even if Rod disagrees with it.
From: David Lesher on 21 Oct 2009 00:54
David Brown <david(a)westcontrol.removethisbit.com> writes: >So you now agree that there is no such thing as a "TRUE sine wave" >output from an UPS? It's a pity that you then contradict yourself >/again/ later on. Well, I know of one design that I'd call same; but not sure you will. The best ferro-resonant UPS's are damn close. Of course, such are neither common or inexpensive. [But then the line itself is never purely sinusoidal, either; it come with flaws such as spikes and dropouts...] The real question is: does the load care if it's not? If all loads are switchers, usually not at all. Linear wall-warts may. Incandescent lamps won't have a clue. -- A host is a host from coast to coast.................wb8foz(a)nrk.com & no one will talk to a host that's close........[v].(301) 56-LINUX Unless the host (that isn't close).........................pob 1433 is busy, hung or dead....................................20915-1433 |