Uninterruptible Power Supply (UPS) FAQ

VERSION 2.1a, May 16, 2001


TOPIC: What is this document all about?


Q: What is this document?

A: This is a FAQ document on Uninterruptible Power Supplies. It is intended to provide a starting point for those people that want to find out what they are, what they do, and what's available.

Note that most of this document is very US-centric. The power numbers, companies and services all emphasize US consumer needs. Sorry, but that's what I have to work with. All the principles discussed here should be applicable just about everywhere.


Q: How is this document made available?

A: Currently, this document is available on the World Wide Web. It is referenced with most major search engines. It used to be posted to the following Usenet newsgroups: comp.misc, comp.unix.admin, comp.sys.sun.hardware, comp.sys.hp.hardware, comp.sys.sgi.hardware, comp.sys.next.hardware, comp.sys.ibm.hardware, comp.sys.dec, comp.answers, and news.answers. This hasn't been done in years. Should it be? If I post it to every group where UPS questions get asked, that would be a lot of groups. I'm open to suggestions.

This document itself is available via the World Wide Web at: http://www.jetcafe.org/~npc/doc/ups-faq.html.


Q: Who maintains this?

A: Right now, this document is maintained by Nick Christenson. My preferred email address is npc@jetcafe.org, and I would like it very much if questions regarding this document could have the word "UPS" or "UPS FAQ" or some such in the Subject line. Note: I am maintaining this on my own time, so please don't be upset if it takes a while for me to respond to your queries. Also none of the information in here represents the views or has the blessing of any organization whatsoever. The maintainer of the FAQ is to be held solely responsible for its contents.

I've taken a bit of a hiatus from maintaining this as I switched jobs and been very busy. I hope to keep it more up to date in the future. Similarly, the document is going through considerable flux right now. I'm receptive to any suggestions.


Q: Where did this information come from?

A: Thankfully, many people have rallied to my cry to fill in the many gaps in my original draft. This is now a group work, although I claim full responsibility for misstatements and inaccuracies.


Q: How can I contribute?

A: You should mail new information, corrections, suggestions, etc. to the current maintainer of this FAQ. If you provide a suggestion, make sure you reference where the information is located in the document. I guarantee that suggestions of the form "Change the word 'always' to 'almost always' in the part about surge suppression." will be ignored.


Q: Are there any restrictions on distribution of this document?

A: This document is copyright by the author. You are encouraged to distribute this document for any non-commercial purpose as long as the contents remain unchanged and a pointer to an up-to-date version is included.


Q: Got anything else you'd like to add?

A: Yes, now that you mention it. The people who contribute to this document can speak only about equipment they have experience with. This may reflect a bias toward or against certain brands, features, functions, etc.. Please keep in mind that the suggestions, brand names and functions here are by no means exhaustive, or even necessarily applicable to your situation. Also, if you have information that is not in this document, please submit it to the maintainer listed above. If you submit information, please say whether you'd like it to be attributed to you or not. I am more than glad to give credit to the fine people who helped with this document, but I want to respect the anonymity of those people who would prefer it.

One more caveat: While the principles of UPS design and maintenance are likely to be fairly universal, the power figures in this FAQ are very US-centric. Sorry, but they're the only numbers I have.


Q: Glossary

A: This was contributed almost entirely by some kind soul. I just cleaned it up a bit.

Complete loss of power. Some literature considers a voltage drop below about 80V to be a blackout as well since most equipment will not operate below these levels.
Sag or Brownout
Decrease in voltage levels which can last for periods ranging from fractions of a second to hours. Can be caused by heavy equipment coming on line such as shop tools, elevators, compressors etc. Also occurs when utility companies deliberately do this to cope with peak load times.
An instantaneous and tremendous increase in voltage often caused by a direct lightning strike on a power line or when power returns after a blackout.
An substantial increase in voltage lasting a small fraction of a second, often caused when high powered appliances such as air conditioners are switched off.
ElectroMagnetic Interference and Radio Frequency Interference. Caused by, inter alia, lightning, generators, radio transmitters, industrial equipment.
Metal Oxide Varistors used to control spikes. These are common in Power Strips. If you see more than two, you likely have a fairly decent Power Strip. They look like largish disk capacitors.
Circuitry that converts DC battery power to AC power required by most computer equipment.
Surge Protector
Circuitry consisting of MOVs, capacitors, rod-core inductors etc. for suppressing surges and spikes usually embedded in a power strip.
Line Conditioner
A transformer that attempts to smooth out fluctuations in input voltage to provide near uniform output voltage or voltage waveform.

TOPIC: What is a UPS and how does is work?


Q: What is a UPS?

A: An Uninterruptible Power Supply is a device that sits between a power supply (e.g. a wall outlet) and a device (e.g. a computer) to prevent undesired features of the power source (outages, sags, surges, bad harmonics, etc.) from the supply from adversely affecting the performance of the device.


Q: How do you pronounce "UPS"?

A: I pronounce it "ups", but most of the literature seems to favor "you pee ess", since they use "a UPS" instead of "an UPS". This document will try to follow the literature.


Q: Vendor X says that (fill in description) is a UPS, but it's different that what you describe above. Who's right?

A: There really is no standard definition of what a UPS is. Anything ranging from a 9 volt battery backup in a clock radio to a building/compound wide backup generator has been called a UPS by someone. The majority of this document refers to objects larger than a beer can and smaller than a desk that help devices remain temporarily operational when changes to the power they receive would otherwise interrupt their function.

Maintaining power to a minicomputer (like a VAX 11) is beyond the scope of this document. This FAQ deals with UPS equipment that can be installed by a computer owner/administrator. If you have requirements that large, you need to talk to a qualified electrician.


Q: Can you give me some more information on this?

A: (Kindly provided by Don Deal, Don.Deal@oit.gatech.edu, my additions are in [square brackets].)

The UPS industry is made up of many manufacturers, and there is a lack of standard terms within the industry. I think this sometimes borders on deliberate misdirection. (It's a jungle out there!) [ Note, in recent years the whole industry seems to have gotten better, at least mostly agreeing on what the terms listed here mean. This is not true everywhere, but things are getting better. ]

There are basically three different types of devices, all of which are occasionally passed off as UPSs.

  1. Standby power supply (SPS). In this type of supply, power is usually derived directly from the power line, until power fails. After power failure, a battery powered inverter turns on to continue supplying power. Batteries are charged, as necessary, when line power is available. This type of supply is sometimes called an "offline" UPS.

    The quality and effectiveness of this class of devices varies considerably; however, they are generally quite a bit cheaper than "true" UPSs. The time required for the inverter to come on line, typically called the switchover time, varies by unit. While some computers may be able to tolerate long switchover times, your mileage may vary. [ Some articles in the trade press have claimed that their testing shows that modern PCs can withstand transfer times of 100ms or more. Most UPS units claim a transfer time to battery of about 4ms. Note that even if a computer can stay up for 100ms, it doesn't mean that 100ms switchover is okay. Damage can still be done to a computer or data on it even if it stays up. ]

    Other features to look for in this class of supplies is line filtering and/or other line conditioners. Since appliances connected to the supply are basically connected directly from the power line, SPSs provide relatively poor protection from line noise, frequency variations, line spikes, and brownouts.

    [Some SPS's claim to have surge/spike suppression circuitry as well as transformers to "boost" voltage without switching to the battery if a modest voltage drop occurs. An example is the "APC Smart UPS" which claims it will switch to this boosting mode if voltage drops below 103V (from the normal expected 120V) and switches to battery only at 90V and below. This, it is claimed, allows operation of the equipment indefinitely under brownout conditions as long as voltage does not drop below 90V. I have not tested this, and would be interested in independent data. There are other vendors products that make similar claims.]

  2. Hybrid [ or ferroresonant] UPS systems. I only know one vendor who sells them - Best Power, Inc. The theory behind these devices is fairly simple. When normal operating line power is present, the supply conditions power using a ferroresonant transformer. This transformer maintains a constant output voltage even with a varying input voltage and provides good protection against line noise. The transformer also maintains output on its secondary briefly when a total outage occurs. Best claims that their inverter then goes on line so quickly that it is operating without any interruption in power. Other UPS vendors maintain that the transition is less than seamless, but then again it's not in their best interest to promote Best's products. Best has a sizable part of the UPS market.

    [ Note: According to some sources, ferroresonant transformers in an UPS system can interact with ferroresonant transformers in your equipment and produce unexpected results. On the other hand, ferroresonant UPS systems don't kick off a lot of heat, which is important in some environments. The Moral: Again, test before you buy. -npc ]

  3. What I call "true" UPS systems, those supplies that continuously operate from an inverter. Obviously, there is no switchover time, and these supplies generally provide the best isolation from power line problems. The disadvantages to these devices are increased cost, increased power consumption, and increased heat generation. Despite the fact that the inverter in a "true" UPS is always on, the reliability of such units does not seem to be affected. In fact, we have seen more failures in cheaper SPS units. [ Note, though, that given the same quality inverter, you'd expect the one that runs least to last longest. ]


Q: How can it help me?

A: A UPS has internal batteries to guarantee that continuous power is provided to the equipment even if the power supply stops providing power. Of course the UPS can provide power for a while, typically a few minutes, but that is often enough to ride out power company glitches or short outages. Advantages:

  1. Computer jobs don't stop because the power fails.
  2. Users not inconvenienced by computer shutting down.
  3. Equipment does not incur the stress of another (hard) power cycle.
  4. Data isn't lost because a machine shut down without doing a "sync" or equivalent to flush cached or real time data.


Q: What sort of stuff does a UPS do?

A: A UPS traditionally can perform the following functions:

  1. Absorb relatively small power surges.
  2. Smooth out noisy power sources.
  3. Continue to provide power to equipment during line sags.
  4. Provide power for some time after a blackout has occurred.
In addition, some UPS or UPS/software combinations provide the following functions:
  1. Automatic shutdown of equipment during long power outages.
  2. Monitoring and logging of the status of the power supply.
  3. Display the Voltage/Current draw of the equipment.
  4. Restart equipment after a long power outage.
  5. Display the voltage currently on the line.
  6. Provide alarms on certain error conditions.
  7. Provide short circuit protection.


Q: How long can equipment on a UPS keep running after the power goes?

A: How big a UPS do you have and what kind of equipment does it protect? For most typical computer workstations, one might have a UPS that was rated to keep the machine alive through a 15 minute power loss. If you need a machine to survive hours without power should probably look at a more robust power backup solution. Even if a UPS has a very small load, it must still operate it's DC (battery) to AC converter, which costs power. A rough extrapolation from APC's documentation, leads me to guess that a 2000 VA UPS can operate it's own converter (with no extra load) for just over 8 hours. A 1250 VA UPS could run its converter for about 5. These are *very* rough guesses based on information provided by one vendor for one vendor.


Q: Given the same vendor claims, how can I tell a "good" quality UPS from a "poor" quality UPS?

A: Testing, testing, testing. I can't emphasize this enough. There are many good and bad units out there that call themselves UPS's. There are many good units that are wrong for your situation. Caveat Emptor.

Some properties you might look for are:

  1. Sinusoidal power output. In general, the closer the AC output of the UPS is to a sine wave, the better it is for your equipment. Many UPS units, especially the cheaper ones, deviate a great deal from a sinusoidal output. Some of them generate square waves. Waveform effects are dealt with in section 2.12.
  2. Does the UPS have a manual bypass switch? If the UPS is broken or is being serviced, can you pass power through it to your equipment? The last thing you want is for a broken UPS to be the cause of extra downtime.
  3. The more information about a UPS's operation you can get from watching the unit itself, the better. How much power (or percentage load) the equipment is drawing, how much battery life is left and indications of the input power quality are all very useful.
  4. Some newer UPS's can communicate with their monitoring software via network connection and SNMP! This is wonderful *if* your network is on a UPS! Also, beware, I have heard of dealers advertising "Network UPS" monitoring where the network is the normal serial connection (no SLIP or PPP).
  5. Does the UPS vendor offer support/maintenance contracts. If don't even offer them, I would suspect the quality of the equipment.

If you do have a UPS that does not output a sinusoidal waveform, some manufacturers *strongly* urge you to not put a surge protector between the UPS and the computer. The surge protector might mistake the non-sine waveform as a power surge and try to send it to ground. This could be bad for your UPS. I don't know if this has happened or not, but I wouldn't chance it.


Q: Should I make sure I have a support/maintenance contract for my UPS systems?

A: Some people strongly recommend this, but to be honest, I don't know how important it is. I haven't had any UPS's long enough to have enough of them fail to know what the failure modes are likely to be. Some people, with more experience than I in these matters, insist that a UPS support/maintenance contract is as important as your computer support/maintenance contract. I can't argue with them. In any case, it's almost certainly worth pricing at any rate.


Q: What sort of maintenance can I perform myself?

A: One good thing you might want to do is periodically test the UPS's and their failure modes. A good time to do this might be right after after a periodic level 0 backup. Nobody is logged in and you've got full backups of the machines. Throw the circuit breaker with the UPS on it to simulate and outage and see how the transition goes. Note that some UPS vendors suggest that testing an UPS by pulling the plug from the wall is *not* a good idea (Tripp Lite is one of them). These UPS units like to have a good idea of what ground looks like. It is likely that unplugging just about any UPS for a short amount of time would not be too dangerous (don't take my word for it, though!), but in all cases, throwing a circuit breaker would be a better thing to do.

It might be useful to install a GFI (Ground Fault Interrupter) socket to facilitate this testing without having to pull the plug, especially if you don't have your UPS protected machines on an isolated circuit (which you probably should). These are the sockets found in most modern kitchens and bathrooms with a red and a black button. You push the latter to cut power and the former to restore power.

Those UPS units that use lead-acid batteries (that's most of them, I'm told) do not have a battery memory and should be run dry as few times as possible. It's probably not a bad investment to do this once on one UPS out of a largish batch to learn how much UPS time you can expect in a real power outage. Note: depending on the manufacturer, UPS batters can be expected to last between about 1 and 5 years before they ought to be replaced.

As a UPS gets older, its battery life will become shorter. Of course there's no way to reliably test how long it is without running the battery down and you don't want to do that because they have lead acid batteries. <Sigh.> All of these are very good reasons to get a support contract for them that includes periodic battery replacement. At the very least, you can figure that the batteries will still be good at the end of the UPS warranty figure, so that's a good place to start guesswork.


Q: Isn't a UPS just a glorified power strip/surge protector with some batteries and a little power conditioning thrown in?

A: Basically. It's also got a power inverter and some other circuitry. It may also have a timer, thermometer or other gadgets.


Q: How important is the UPS output waveform?

A: That's a good question, and one is worthy of some debate. One school of thought holds that one should always run equipment on the best approximation of sinusoidal input that one can, and that deviations produce harmonics which may either be interpreted as signal if they get through a power supply, or may actually damage the equipment. Another school holds that since almost all computers use switching-type power supplies, which only draw power at or near the peaks of the waveforms, the shape of the input power waveform is not important. Who's right? I don't know. My opinion is that sinusoidal output is worth the extra money, especially for on-line UPS systems that continually provide their waveform to the computer. Also, if you don't *know* that your equipment has a switching-type power supply, you might want to think twice before buying a low quality UPS. [ Some of this information from a great article in the October 1994 issue of LAN Magazine, check it out. -npc ]

TOPIC: UPS monitoring/shutdown software.


Q: If the power is out for a long time, I would like to have my computer automatically shut itself down gracefully before the UPS batteries die. Can I do this?

A: Yes. Most UPS manufacturers support software that will do this for some UPS's on at least some platforms. Ask your UPS vendor for details.

Q: Okay, how about restarting the system for me once power returns?

A: Fewer software products do this, but many do. Again, ask your vendor. I do not know of any freely distributable products that will do this. It doesn't mean that they can't be built, but vendor software is cheap enough (usually) that it's probably not worth building.


Q: How does it work? I'm a starving (fill in the blank) and I really don't want to pay for software unless I absolutely have to.

A: Usually, there is a serial connection running from a UPS into your computer. The UPS sends information along the serial line as it goes. If you can decode which pins contain which information, how the information is formatted and figure out what it wants to hear from the computer side, you're all set. Make sure you have the right serial cable and know how the pins map between DB9 and DB25 as both your computer and your UPS may take either.

Since UPS units with network based monitoring capabilities are appearing on the market, we can hopefully get something that will communicate with those units.

Here is a skeleton script provided by Joe Moss, joe@morton.rain.com. Definitely check this out as a starting point, but don't expect it to do anything meaningful without some work.

        #! /bin/sh

        # Shut down system in case of extended power failure

        # This should be the serial port to which the UPS is connected
        # This port must be set to block on open until the DCD line
        # is asserted - many UNIX systems have this determined by
        # the minor device number, if not, see if there is some way
        # to enable this behavior on your system

        # Ok, this should block until there is a power failure

        : > $PORT

        # If we reach this point, we've lost power
        wall << EOF
        The sky is falling!! The sky is falling!!

        # call shutdown (or init or whatever)
        exec shutdown


Q: Hmmm... that sounds kinda complicated. Has someone already done this?

A: Any solution would almost certainly be vendor specific. However, some brave souls have provided partial functionality for certain vendors' UPS's. I don't know the original source, but I have a copy available for anonymous FTP at navigator.jpl.nasa.gov in the pub/src/UPS directory as upsd.tar.Z. I haven't tried it and I don't honestly know if it even works.

Note: Different UPS's produce different sorts of signals. Just installing this already built package may require a great deal of work. The cabling can be complicated, etc.. I would be interested in hearing where this software does/doesn't work.

Another good example, that probably works straight away for SunOS 4.1.X machines using APC Back-UPS devices, is also available on navigator for anonymous FTP in the pub/src/UPS directory is pf.c. It was written by Ronald Florence (ron@mlfarm.com). It looks like a nice framework for expansion to other OS platforms and UPS implementations. Give it a try.


Q: I can't find monitoring software that will work on my configuration. What should I do?

A: Well, it seems you have a few choices:

  1. Build your own. See item 03.02.
  2. Use something freely distributable. See item 03.03.
  3. Lean on your UPS vendor to port to your platform.
  4. Try a different vendor that supports your platform. See item 05.01.


Q: What other software is out there?

A: Software packages for UPS machines are getting more sophisticated. Most provide some level of power and status monitoring, but lately there are more GUI's, more interactive packages, SNMP support, and even call-out paging. See the software section 05.03 for more info.

TOPIC: How big a UPS do I need?


Q: How are the "sizes" of UPS's determined?

A: Typically, a UPS has a VA rating. The VA rating is the maximum number of Volts * Amps it can deliver. The VA rating is not the same as the power drain (in Watts) of the equipment. Computers are notoriously non-resistive. A typical PF (power factor: Watts/VA) for workstations may be as low as 0.6, which means that if you record a drain of 100 Watts, you need a UPS with a VA rating of 167. Some literature suggests that 0.7 may be a good conversion factor, but this will depend heavily on the machine. WARNING: Don't take my word for it! Note: Some UPS's can continue to deliver power if the VA rating is exceeded, they merely can't provide above their VA rating if the power goes. Some can't provide power above their VA rating at all. Some may do something really nasty if you try. In any case, I strongly recommend not doing this under any circumstances.


Q: How can I tell what VA rating I need for my equipment?

A: First, when possible, get VA rather than wattage ratings. See Q04.01 above. There are a couple of ways:

  1. Direct measurement. You can get equipment to measure the current draw of your equipment directly. You may or may not have access to this. If you are part of an organization that has it's own facilities/electrical type people, they're likely to be able to do this. They might help you out if you ask nice.
  2. Compare notes. If you know someone with the same setup you're using, ask them what they use and how close they are to the maximum VA rating.
  3. Use a chart. Most vendors can help you out for common equipment. If you have an unusual setup, or mix vendors a lot, you're probably out of luck here.
  4. Use the equipment rating. Most pieces of computer equipment have a power rating on some back panel. This number is usually high, as it is necessary for the manufacturer to play it safe or they'll get sued.

Note: Method 1 is by far the best, method 2 and 3 are secondary, method 4 is usually overkill, but pretty safe. There are some examples in section 4.6, but the information is probably worth what you paid for it :-)


Q: Hmmm... seems like a tough thing to determine.

A: Yeah, it can be. It's also very important. If you get a UPS that's too big, then you've overpaid, but your equipment can survive a longer outage. If you get a UPS that's too small, then you could be in deep trouble. Therefore, I recommend that you be conservative in buying these things, unfortunately, this costs money.


Q: What else should I consider?

A: It would be nice to know how long your site's typical power outages are. In some places, with nice weather and a flaky power grid, the power is almost never out for more than 5 minutes, but this could happen quite frequently. In this case, you may as well use a UPS with a VA rating close to your equipment rating with no extra batteries. If your area has longer outages, in the half hour or hour range, as is often the case in thunderstorm country, you can either buy UPS's with multiples of the VA rating of the equipment, since oversizing a VA rating for a UPS has the effect of lengthening the amount of time your equipment can stay up in case of a power outage, or you can buy additional battery units for a smaller UPS. You can probably get away with doing simple math to determine how much longer a larger UPS will keep your equipment running, but I recommend running a few tests before committing to a large purchase order. Also, your UPS vendor will almost certainly be glad to help you size the equipment you need. If all else fails and you guess wrong, or move equipment to a location with different power status, you may be really, really glad if you bought a UPS that can have additional battery packs added.


Q: How about I use one of these UPS thingies for a laser printer?

A: Don't ever do this. If you ever measured the current draw of a laser printer during startup (and during printing) you'd be stunned at what it pulls. All UPS manufacturers I know of tell you not to do this.

Okay, I have to back down from this. I know APC, just as an example, now does rate some of their UPS units for use with certain laser printers. Not that I think this is a good idea, mind you. In general, they are difficult to size and rarely do they require the same level of uptime as servers.

In any case, don't do this without specific approval of your UPS vendor.


Q: So, what sorts of UPS sizes do you use on your equipment?

A: BIG DISCLAIMER. I disclaim everything about these figures. At best, they are very, very rough. Heck, I may be lying. Don't trust them. Here they are anyway.

Note also, this is what the equipment apparently PULLS, not the UPS sizes that are on them. Generally, I've been using UPS's that are about 2X the VA ratings shown. At the very least, I would using UPS sized 1.5X the VA ratings here.

400 VA:
Sparc 2 with 3 600 MB disks, 1 200 MB disk, 1 exabyte 8200 tape drive, 19" color monitor.
600 VA:
HP 750 with 4 1.3 GB disks, internal 4mm tape drive and internal CD-ROM drive, external disk cabinet and 19" color monitor.
500 VA:
SPARC 2GX clone. 1 1.2 GB disk, 4 2.0 GB disks, 2 tape drives, 1 CD-ROM drive, "big" monitor.
300 VA:
Sparc 2 clone with 100W power supply, internal 424 disk, 16" color monitor, external 1 GB disk drive.

These are U.K. numbers, based on 240 V wall current. Most of these VA numbers are very close to "American" VA numbers, but if my caveats weren't strong enough earlier... .

100 VA:
SGI Indigo R4400, 48 MB RAM, 1 GB int. disk, no graphics, no monitor.
580 VA:
SGI Indigo R4400 configured as above with 17" Sony monitor plus 3 19" monochrome X terminals.

Another word of warning, don't assume that power requirements scale with compute power and number of peripherals, ESPECIALLY if they are different architectures. Older equipment, CPU's, disks, monitors, whatever almost universally requires more power than new equipment. For example, it seems that an HP 9000/425e with 1 internal 420 MB disk and 19" color monitor pulls a lot more power than a much more modern and much faster HP 9000/715 with an internal 1.3 GB disk, CD-ROM drive and more modern 19" color monitor. Again, the moral is don't assume.

TOPIC: Specific manufacturer's information.


Q: What vendors are there and what do they produce?

A: Here is a very incomplete list, based only on what I know. Please give me information to expand it. I make no claims as to the accuracy of this information. It is mostly based on personal recommendations and vendor propaganda.

Note: The October 1994 issue of LAN Magazine has a great vendor list. I have used it to update many of the entries here. However, there is a lot of information available there that I don't have space to include here. This article is an excellent starting point for comparative pricing on UPS equipment.

05.02 UPS Hardware (and software) manufacturers:

In earlier versions of this document, I gave contact information and a brief (usually a bit out of date) product listing of all the major UPS vendors I could find. Now, with nearly all of this information on line, it makes more sense just to provide a link to the web pages of the companies and list what sort of general market they're in. I think this is actually more useful and is certainly less likely to drift out of date. Note, I'm including information only on manufacturers, not retailers.
American Power Conversion APC is the largest UPS manufacturer and has a whole line of UPS systems (mostly line interactive), software, and power system accessories which can be purchased directly from them or via many retail outlets around the United States and overseas.
Belkin Belkin makes a lot of computer connectivity products. Its UPS products rate up to about 650 VA.
Best Power Technology Best Power produces inexpensive standby power systems, more advanced line interactive systems, and ferroresonant line interactive systems as well as software, PDUs, and power system accessories. They sell primarily through resellers.
Clary Corporation Clary sells UPS products and specializes in emergency, military, and life support systems. They also sell management software and accessories. Clary does custom work, which they sell direct, but they sell most of their catalog from a network of distributors.
Controlled Power Company Controlled Power produces UPS systems, power conditioners, voltage regulators and transformers. Equipment can be ordered direct.
Deltec Electronics Deltec makes standby, line interactive, and on-line UPS systems. They sell primarily through resellers.
Elgar Corporation Elgar produces 1250 VA and 2400 VA "GUPS" systems designed for harsh environments. They sell direct.
Emerson Electronics Emerson is a big electronics conglomerate. It's claim to fame in the UPS world is that it's the parent company to Liebert.
Exide Electronics One of the bigger players in the data center sized UPS system industry, Exide also makes more modest sized on-line and line interactive systems. Exide products can be purchased direct or from their distributors.
IMV/Victron Based in the Netherlands, this company sells line-interactive and on-line UPS systems.
IntelliPower Inc. Intellipower sells on-line UPS systems and management software. I can't say for sure, but they probably sell direct.
Liebert Probably the largest manufacturer of large (10 KVA +) UPS systems. Also well known for their other data center products including power distribution units and HVAC products. They also make smaller UPS systems (300 VA on up), but these are not nearly as popular. Liebert is a subsidiary of Emerson Electronics.
MGE UPS Systems MGE UPS Systems sells UPS systems from 220 to 3000 VAC and additional power equipment.
Oneac Oneac sells line interactive UPS systems with software in the US and UK.
OPTI-UPS OPTI-UPS makes line-interactive UPS systems ranging from 500 VAC to 1440 VAC, including rack-mount models.
Philtek Philtek makes UPS systems.
SL Waber SL Waber sells mostly lower end UPS systems and a wide assortment of surge suppression and other power accessories.
Tripp Lite Tripp Lite is one of the bigger UPS companies, selling standby, line interactive and on-line UPS systems as well as software, surge protection, line conditioners, and accessories. They predominantly sell through distributors.

Note, there are more companies than are listed here. Please let me know which ones I missed.

TOPIC: Bibliography

There are many good references and review articles on UPS information. Some of the best sources can be found in vendor information. There is great reference material woven into their propaganda. Some other good sources are:

TOPIC: Acknowledgments

I would like to thank Charles Rhoades (Charles.W.Rhoades@jpl.nasa.gov) for his sage remarks on an early draft of this document. I would like to thank Kevin R. Ray (kevin@kray.com) for sending me the freely distributable upsd software and Ronald Florence (ron@mlfarm.com) for contributing the pf program. Thanks also to Don Deal (Don.Deal@oit.gatech.edu) for a great many valuable suggestions and that great section on the types of UPS units.

The following people have all made valuable contributions to this document:

	Scott Pinkerton, spinkert@t4rta-gw.den.mmc.com
	Morris Galloway Jr., mmgall@presby.edu
	David E A Wilson, david@cs.uow.edu.au
	Edward Hartnett, ejh@larry.gsfc.nasa.gov
	Joe Moss, joe@morton.rain.com
	Kurt Hillig, khillig@chem.lsa.umich.edu
	Robert D. Freeman, rdf@thermo.chem.okstate.edu
	Jochen Bern, bern@kleopatra.Uni-Trier.DE
	Dave Gruhn, dgruhn@fuzzy.eskimo.com
	Steve Welch, smw@columbine.cgd.ucar.edu
	Ron Tansky, ron.t@bix.com
	Andrew J. Templin, nosilla@ohionet.org
	Chuck Bennett, chuck@benatong.com
	M.V.S. Ramanath, ram@sclara.qms.com
	Max Hailperin, max@kolmogorov.gac.edu
	Larry Moss, moss@cvs.rochester.edu

Please note that I take full blame for any errors or omissions.