Well ... there is a third type, exemplified by the Best Power Systems FERRUPS which I have on several computers.
These bring the power in through a ferroresonant constant voltage transformer (like the Sola ones) with some extra windings. The line power is also used (separately) to maintain the charge in the battery or batteries (one of mine -- a 2KVA one -- uses four *heavy*
12V gel batteries in series). The ferroresonant transformer does an excellent job of protecting the load from surges and such, as well as keeping the systems happy during brownouts.
When the power goes away, those extra windings are driven by the inverter to maintain the voltage. There are no relays involved, and switchover is fast enough so the load doesn't have a chance to notice it.
These every so many hours will switch on the inverter to verify the state of the batteries, and then switch it back off. Ever so many days it will do a "conditioning" charge cycle on the batteries.
Best Power Systems is long gone -- but their products still show up on eBay -- sometimes for very good prices -- so I have three of them, one powering a two racks full of computers, and the other two powering two Sun Blade 2000 computers, with the larger one also powering a seven-drive tray of Fibre Channel drives running as a RAID assembly. (The racked computers also have two such trays running.)
What is the major hardware failure in today's PC's? Bad electrolytic caps on the motherboard or in the power supply. They are crappy caps with a defective water based electrolyte - to be sure - but it is stated that high ripple currents are responsible for appreciably accellerated failure.
In one office I service we had 8 computers bought at the same time, with identical motherboards and power supplies. The 3 that were on UPS protection from day one are still running I think about 5 years later. I believe one power supply has been replaced. Three that never had a UPS are long gone with swelled motherboard caps.(after at least one power supply each) The other two went onto UPSs after about 2 years, and both have failed over the last several weeks with swelled caps and are on their second or thiurd power supplies. Several hard drives have also failed, but not on the 3 that have spent their whole lives protected.
As for power lines not amplifying harmonics, what exactly is resonance, and how does it affect amplitude??? Resonance downstream from a harmonics generator CAN amplify the resonance, and even feed it back to the source.
The other problem dual conversions are good at solving is common mode noise. And how about floating neutrals? Not out of the ordinary in many commercial apps to have the neutral several (or even tens of) volts above ground.. Ither an isolation transformer or a dual mode UPS can solve this problem. ONEAC made a fortune selling isolation transformers untill the cost of a Exide (now Powerware/ Best?whoever)Prestige UPS got low enough to undersell an Oneac and a cheap miniteman standby UPS.
Played havoc with thesale of the old Sola line stabilizers too - and at least an order of magnitude more efficient.
Every mission critical computer I spec or sell goes with a MINIMUM of a line interactive UPS - andif at all possible a dual conversion (true online) unit.
And somewhere in between a few good midsize dual conversions give you the capacity for some redundancy. You ARE running redundant servers, are you not???? What good are redundant servers if they are taken down by the failure of une UPS??
We have 2 servers, one with redundant power supplies, one without - with one UPS powering each power supply on the dual supply unit, while the single power supply unit takes one UPS while the"maintenance" workstation takes the other.
Who's to know a "better" power supply today. Sure can't go by price - Bling raises the price a lot more than quality.. Can't go by weight either - as some actually have BALLAST installed. to make them appear more "heavy duty" A second fan doesn't mean squat either. I've seen some real "high quality looking" power supplies selling in the mid 3 figure range that were no better than some $49 units. Almost a 10:1 price ratio.
And the ULTIMATE UPS us the Best UBS system. Available in diesel, gasoline, propane and natural gas they are a DC generator unit to match the battery voltage of the UPS that is automatically started X minutes after a power failure.
UBS for "Unlimited Battery System" Fantastic units and were common in emergency services departments in storm-prone areas - Hurricane in the south-east, Blizzard in the NorthWest, and Ice Storms in the North East.
I was under the impression thet the ATX power supply spec came into being sometime in the last 15 or so years - when the PCI buss became standard and the Pentium or Pentium Pro processor... Got me interested, so I decided to check - Anyway - the ATX (AT Extended) was introduced by Intel in 1995.
Must be about 10 or 12 variations since, but all are identifiable by a "software switch" and "constant on" (5 volts always on for keyboard/remote startup function) technology as well as the dual row
20 pin (or on later versions, 24 pin) motherboard connector.
ATX spec was part of a 'system' standard that Intel first defined with the Pentium. Intel also defined standards for video controllers, memory, busses that separated CPU timing from I/O buss timing, the PCI bus, BIOS functions, etc. simultaneously. Why? In the late 1980s, IBM begin stifling innovation; began reverting to obsolete industry practice also found in other uninnovative industries such as domestic autos.
Stifled standards started when IBM 'product people' were replaced by 'finance people; when Cannavino began stifling innovation in Boca Raton. Resulted were numerous product disasters including Microchannel, PS/2, EGA and VGA monitors (that were rescued by innovative companies), OS2, and a long list of other standards that were also closed and so pathetic as to be almost unknown.
What was, at first, could have been a disaster for the PC industry, instead, resulted in a massive new paradigm for all industries. What happened to the industry in 1987 through 1990 should be understood by any computer industry professional who appreciates the hows, whats, whys, and history of innovation. To understand this history, need for, and new paradigms created by ATX specs, trace innovation to its sources.
One early attempt at a new paradigm was Compaq=92s EISA bus. It was not successful as a standard, but helped pioneer that paradigm for innovation across an entire industry. A paradigm that made success also possible in the disk drive industry, video controller industry, memory industry, etc. Innovations so successful that even Apple finally conceded to all of them.
Returning to the ATX power supply standard. It contained 1) the principle of how original PCs worked, 2) in combination with a few innovations, 3) all based in power supply standards that existed long before PCs every existed. Fold back current limiting was always so standard as to be found in all switchers, in linears even long before switching supplies existed, and even in every single chip power supply- bar none. Foldback current limiting was common in every type of supply long before the first PC every existed.
Yes, even the original PC incorporated universal standards such as current limiting and power supplies that worked perfectly OK even when voltages dropped quite low. Any PC of any decade works just fine even when AC voltage drops so low that incandescent bulbs glow at less than
50% intensity. Just another industry standard that any computer always had to meet. Just another old and long required standard that was continued in the original ATX specs.
And yes, my experience precedes and observed all those advancements. I watched it happen from an engineering perspective which meant I also knew 'why' it was implemented. Fold back current limiting is necessary for a long list of reasons - some listed earlier. Intel even went so far as to define an ATX standard number - less than 240 VA for any shorted output. Yes, Intel even defined how large the wire must be to short all power supply voltages and have zero damage. Two more ATX standards that define foldback current limiting for all power supplies.
Contrary to some popular myths, any power supply can have all outputs shorted together and never be damaged. Another industry standard that predates even the original IBM PC.
If harmonics are causing failures, then harmonics are also destoying a UPS power supply - for same reasons. Correctly noted is that harmonics are not a problem to any properly constructed power supply. However, harmonics cause problems elsewhere such as in building wiring and utility power transformer. Harmonics are why a shared neutral wire must be larger than required by code.
Best harmonics solution is a computer with PFC. That eliminates a problem at the source, makes the computer even more resistant to external harmonics, and reduces energy consumption.
A professional? One who learned this stuff by designing things or one who only knows by observing some machines? This poster agrees with jk due to a few generations of actually doing this stuff at the electrical level. Who learned this stuff by actually discovering why electrolytics failed and even which manufacturers were more prone to particular problems? jk is correct. If harmonics were a problem to any computer, then the purchaser must identify himself as a reason for those defective computers.
Harmonics made even less problematic by PFC circuits in that computer's power supply - which are commonly found in Europe where international standards apply, but not required in America.
A computer must be so robust as to make high harmonics from any typical computer grade UPS completely irrelevant.
Cap failure on the mother boards would be a clear indication of crappy power supplies.
WHich would be regulation and filtering in the power supply, and nothing to do with harmonics
It is exactly the same thing it is in ANY R-L-C circuit. Under "ideal conditions" it presents either a 0 impedance or infinite impedance [depending on which sort of resonance].
If you have a source that acts as a constant current harmonic generator, then an infinite impedance at that frequency forces the voltage to increase until the current source saturates.
This is no more "amplification" than a transformer is an "amplifier". The total harmonic energy does not increase in resonance, it just gets converted from a higher voltage- lower current manifestation to a lower voltage-higher current one (or vice versa). THe power line does not "magically " convert it's power into harmonics vis some mysterious "hoodoo" at resonance.
YOu can have a resonant circuit just sitting there all the time without a problem, until you start exciting it with some energy at its resonant frequency. That energy however does not come from the power line directly.
Instead, it is a non linear load (such as your PC power supply) That takes energy at the power line frequency, converts most of it so some purpose, and converts some of it to a different frequency, and pumps it back into the system.
WHich is also a from of a common mode problem.
"tens" would be highly unusual. 3-5 v is pretty common.
They only solve it if they are electrically close to the load.
Yes, but the SOla's are not near so maintenance intensive. [Assuming one does maintenance. So many places don't]
We have UPSes with battery rooms feeding three phase 480V static transfer switches and PDUs, backed by big diesel generators with lots of fuel. Nearly every piece of equipment has N+1 power supplies fed from separate PDUs on separate UPSes. Redundant servers are in different states.
At home I'm a little more lax with single UPSes in my office and on my server rack in the garage, backed by two generators, one as primary which I typically have online in 5 minutes and one as backup which can be online a few minutes later if need be. Server rack UPS gets around
Bullshit. The caps that fail are part of the on bioard swithing supplyt for the CPU and they fail because they are cheap crap.
The harmonics cause problems in older builings using three phase power, and undersized neutrals. This problem was known 25years ago, yet contractors continued to do substandard work. Some offices had electrical fires after a lot of PCs replaced old dumb terminals that had linear supplies, with full wave rectification. The 120,000 sq ft Microdyne facility that used to be near here had constant problems with equipment failures, until they finally pulled a second neutral to parallel the original neutrals. Computer & test equipment failures dropped by over 90%.
Never stated: ATX standard was more than 20 years old. But the requirements in ATX standards come from industry standards that are well older than 20 years. "Industry standards" was not a reference to ATX standards. ATX standards are simply the latest update of "industry standards".
Foldback limiting explains why a shorted memory chip does not cause motherboard damage and also why power supply outputs can be shorted without damage.
My power has a fluctuation at 2:15 AM every day. I see the lights do a dim or if in bed hear some strange sounds out of the box fan I use to make white noise to masq my tinitus. The ups also does the beep beep. Never got an answer out of the electric company when I asked about it.
Wes
-- "Additionally as a security officer, I carry a gun to protect government officials but my life isn't worth protecting at home in their eyes." Dick Anthony Heller
On Thu, 04 Dec 2008 20:30:22 -0500, the infamous Wes scrawled the following:
my tinitus.
Our local transitions are at 5:30ish, 7:50ish, and again at around
8pm. Most don't make the UPS beep, but I see the dimming and hear the UPSes cycle in and out.
I'm guessing that they coincide with local (anywhere on this particular grid) wake-up times where room and water heaters come on, ovens/toasters are warming up, and foundries fire up their induction heaters. ;)
But it has seemingly become worse and worse every year.
-- Smell is a potent wizard that transports you across thousands of miles and all the years you have lived. -- Helen Keller
Well ... I have one serious server (a Sun Fire 280R) with dual power supplies in the server, and in each of the two trays of Fibre Channel drives. One side of each is plugged into the main UPS, and the other side is plugged into the incoming AC line through a surge suppressor only. This way, I am protected if the UPS dies, or if the incoming power does, but not if both. The other two (smaller) BEST UPS are only 1400 VA and 700VA and each support a single machine.
The main UPS also runs two Sun Ultra-10 machines acting as web servers, two Sun SPARCstation 5s acting as mail servers, an Ultra 5 configured as a firewall, the Cisco router/CSU/DSU for the T1 line, and on occasion a Mac Mini.
As for the earlier mention of the Sola style transformers in the FERRUPS' being rather inefficient unless the system is fairly loaded -- each UPS is running about 75% of maximum, so yes, it is loaded.
Oh yes -- on the server, the tape jukebox only has one power line, so that goes to the AC line only at present. If the power is gone, making a new backup is not going to complete before the batteries in the UPS get to the point where it tells the systems to shut down. :-)
But, of course, this is a private system, with no income so I can afford to take some time to rebuild things if necessary.
PolyTech Forum website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.