OT a bit - matching UPS with one's pc

Do you have a question? Post it now! No Registration Necessary.  Now with pictures!

Threaded View
I'll probably be buying a UPS for my pc soon and need to learn how to
size and match a UPS with my pc, but there's a lot of unclear info out

Complicated but probably more exact method - add up the amps each
device in my pc uses to calculate the minimum size UPS.  I've found
some calculators out there that follow this method.

Variation to that method - add the number of watts each of the various
devices need, add 10% as a safety margin and divide that number by 12
to derive amps needed from a UPS.

Simplistic method - My brother said use the specs posted for my pc's
PSU regarding amps and pick a UPS based on that number - his
explanation being that all the devices in the pc are ultimately powered
through the PSU and it therefore incorporates everything needed by the
individual components of the pc so that the amps it provides can be
thought of as the minimum amps that a UPS should provide.

My power supply is a 550 Watt PSU.  Its output is listed as +3.3V@28A,
+5V@26A, +12V1@22A, +12V2@22A, +12V3@25A, -1 which doesn't really mean
much to me.  If I add them together, the sum totals 123 Amps, but using
the second procedure yields roughly 50 Amps.  For simplicity, call it a
100% difference.

My pc has an Asus M4A89GTD Pro mobo, an AMD Athlon II x3 440 CPU, a
Radeon 6750 video card, 4x4Gb DIMMs, a DVD and 2 500 Gb hds.

As far as my needs go, I think I want to use a UPS just for my pc and
leave my monitor and cable modem on a surge protector.  I might choose
a UPS that can instruct my pc to shutdown when power goes out.  My
biggest concerns in my neighborhood are rolling brownouts.  It promises
to be a very hot summer.  I don't need the UPS to power my pc for
extended periods of time, just enough time (minutes) to safely shut

My brother's suggestion seems the most practical way (KISS) to decide
how much UPS I need, but I feel like I'm doing something wrong to add
up to 123 Amps using the PSU specs whereas dividing 550 Watts by 12 =
47 Amps.

Are there other considerations I should keep in mind?



Re: OT a bit - matching UPS with one's pc

Yes wrote:
Quoted text here. Click to load it

In simplistic terms, when you buy a 550W supply, it draws
anywhere from 0 watts to 550W from the wall.

Imagine, for a moment, you have the PSU sitting on the bench,
with nothing connected to the wires. We would hope the supply
draws very close to zero, with no load on it. There is little
heat coming out of the back of the PSU.

The PSU converts AC power to DC power, but only converts
enough to supply the load. That's why it draws zero when
nothing is connected.

And at the other extreme, by carefully tweaking a load box,
we can force it to supply up to 550W. Any more than that, it'll
shut down.

Now, your PC parts don't draw 550W. They draw some intermediate
value. And that intermediate value, is the one you'll be buying
the UPS for.

Depending on your budget and level of curiosity, you could buy
a Kill-O-Watt meter, and remove all uncertainty. This one is about
$20, and measures the power for you. You can use it later for
your other appliances. If you run the computer off a power
strip (plug in computer, plug in monitor), then plugging the
power strip into the Kill-O-Watt, totals all the power used.

(Amazon.com product link shortened)

This is a spec sheet for a Kill-O-Watt meter. It
measures both watts and VA, the two units of measure
we need.


    "Also displays Volts, Amps, Watts, Hz, VA"
                                -----      --


The UPS has two specs.

It was watts (real component, no imaginary component)
and VA (ignores phase angle).


    "When a UPS powers equipment which presents a
     reactive load with a low power factor, neither
     limit may safely be exceeded."

The old ATX power supplies, had a power factor of 0.7. The
newer supplies, with power factor correction, are 1.0.
When the power factor is 1.0, watts is equal to VA.
A light bulb is power factor 1.0, and its watts are the
same as the VA.


Just taking the first UPS I'm offered on Newegg as an example...


    750 VA
    450 Watts

You have to work out both of those, to determine if you're exceeding
the limits of the unit.

Here, someone works out the KVA (kilo-volt-amps) by multiplying
the power in watts, by the inverse of the known power factor
of the load (office building full of PC power supplies).
The person doing the estimate, used a PF of 0.8.


     110kW * 1/pf = 110*1/0.8 = 137.5kVA

So it is possible to estimate this for yourself, where PF=0.7
for a very old PSU, and PF is close to 1 (and it may state
in the specs it is power factor corrected).

The UPS companies make a web page available, for picking a UPS,
and it nicely ignores those specs, and picks a unit for you.
Naturally, such a web page is filled with bad assumptions.
But it will likely get you a UPS that works.

If you use a Kill-O-Watt meter, and run a 100% computing load
(i.e. playing Crysis, running Prime95 or the like), then the
meter can simply print out on its screen, the number of watts
and the number of VA. And then you can compare those to
the UPS specs. Another nice aspect of the Kill-O-Watt
meter, is it does the vector math for you, if you
have multiple loads, all with different power
factors. If you connect a desk lamp, the LCD monitor,
the computer, all to the power strip, the Kill-O-Watt
will still read out the total Watts and VA for you.
Very accurately.


Now, what is the danger of doing it this way ? You end up
buying a $50 UPS, because the math says that's all you need.
I didn't buy mine that way. I bought a mid-tier UPS,
to avoid the "bottom feeding junk" in the $50 range.

So on the one hand, doing the math is reassuring from
an "I only bought as much as I needed" point of view.
But, the product quality simply may not be there,
by being so cheap. By the time you move upscale a
bit, you have enough power capacity that the math
is less of an issue.

This one has a runtime rating of 5 minutes at 780W. $160


This one runs for 12 minutes at half load (half of 1500W). $509
Weights 69 pounds.


Now, if you get your Kill-O-Watt, make the measurements, it
is always possible a unit like this will fit the bill. But it
could be a false economy, depending on what the customer reviews

Runs 2.1 minutes at 257W. $52. It might be barely enough to
do the job. I could spend my time estimating your load
(X3 440 95W, HD 6750 86W [bad estimate], 50W mobo, 12W disk drive),
but that'll get me 243 watts / efficiency = perhaps 304 watts
at the plug, and we know the video card really doesn't draw
that amount of power. So estimating is going to be error prone,
due to the people who normally measure the power, not having
measured it for HD 6750. But this might be an example of
what you would buy, if using maths on it. 2.1 minutes is
probably enough time to shut down.


If you're going to skimp that much, I'd use the Kill-O-Watt
to be sure. A couple reviews had problems with that UPS,
which is actually pretty good. I expected to see more
trouble reports. When we bought 100 units of some junk
at work, we had a 10% failure rate. I guess it's all right,
if yours was one of the 90 working ones :-)

So it's about $20 + $50, if you do it the Kill-O-Watt way,
or maybe $160 if you just buy a mid-tier unit. Years ago,
when I bought mine (2002?), they were more expensive than
they are now. I guess that's the Chinese batteries in them,
"paying dividends". Or maybe it's the competition from
CyberPower :-)

Your choice,

Re: OT a bit - matching UPS with one's pc

On 6/10/2013 1:19 AM, Yes wrote:
Quoted text here. Click to load it
Safe way: measure the actual power consumption of the equipment you need  
for minimal functioning and work from that. This equipment would include  
the computer + monitor + whatever equipment is needed to connect to the  
Internet (modem, router, switch or whatever other bits your situation calls  
for). Add at least 20% to that although doubling it won't hurt. Then buy  
yourself a UPS which will support that load. You should not include power  
hogs like printers, especially lasers.

Keep in mind that the power rating of the UPS is not terribly informative  
about how well it will serve your purpose. A massive power output rating  
simply means that it will put out that much power but not for how long. A  
1000W UPS which only runs for one minute is not terribly useful; at the  
very least the UPS needs to keep everything going for long enough to tell  
the computer to shut down safely. A UPS which supports your equipment and  
runs for 20 minutes would be a better value.

I have standardized on 1200W CyberPower units not because any of my  
equipment consumes anything like that (maybe 350W max) but because the are  
have proven to be reliable, have a good number of outlets (both battery and  
surge protected), they are line-interactive and handle brownouts well and  
they go on sale pretty regularly so they are affordable. I program the  
shutdowns so that when the batteries have maybe 5 minutes of support left  
the computer does a safe powerdown. I have 5 of these units as well as a  
couple of 12VDC units supporting network infrastructure but I live in an  
area with obsolete power distribution, lots of storms, and huge trees so  
what seems like overkill is actually just barely enough.

Re: OT a bit - matching UPS with one's pc

On Monday, June 10, 2013 1:19:04 AM UTC-4, Yes wrote:
Quoted text here. Click to load it

  Generally sum the amps or watts.  Then remember a UPS battery typically d
egrades in three years.  So, for example, if your system consumes 350 watts
, then the UPS should be sized about 500 watts.  So the the three year old  
battery can still provide enough power.  And for other technical reasons.

  Now, the power rating on the label for most engineer designed appliances  
is good enough.  However many put 800 watt power supplies into computers th
at consume only 200 watts and never more that 350 watts.  To get a better i
dea, read what others measured with their Kil-A-Watt in:

  BTW, a UPS connects your equipment directly to AC mains when not in batte
ry backup mode.  Many rumors are based in hearsay that is discussing a UPS  
costing on the order of $1000 or higher.  Cleanest power come when the UPS  
connects your computer directly to AC mains.

  And finally, power from a UPS in battery backup mode is so 'dirty' as to  
sometimes be harmful to motorized appliances.  But electronics are so robus
t as to not care.

Re: OT a bit - matching UPS with one's pc

westom1@gmail.com wrote:
Quoted text here. Click to load it
degrades in three years.  So, for example, if your system consumes 350 watts,
then the UPS should be sized about 500 watts.  So the the three year old battery
can still provide enough power.  And for other technical reasons.
Quoted text here. Click to load it
good enough.  However many put 800 watt power supplies into computers that
consume only 200 watts and never more that 350 watts.  To get a better idea,
read what others measured with their Kil-A-Watt in:
Quoted text here. Click to load it
backup mode.  Many rumors are based in hearsay that is discussing a UPS costing
on the order of $1000 or higher.  Cleanest power come when the UPS connects your
computer directly to AC mains.
Quoted text here. Click to load it
sometimes be harmful to motorized appliances.  But electronics are so robust as
to not care.

I glossed over that detail.

Power rating is not the same thing as runtime.

Power rating is the maximum power it will provide, before
shutting off on an overload. A 1200W UPS could power up
to 1200W worth of electronics. You want the power rating,
to be greater than the sum of the loads.

Generally, to have a decent runtime at full load, a
larger battery is used in a 1200W UPS, than would be
used in a 450W UPS.

Then, if less than full load is used (a 200W PC on a
1200W UPS), the runtime could be 20 minutes. Runtime
is a function of the actual PC loading, as well as
being related to the size of the battery. When we
buy a 1200W UPS, not only is the power rating higher,
but most of the time, it comes with a physically bigger

But there isn't a direct proportionality or anything.
The manufacturer can make the runtime anything they

A dumb manufacturer, could use the battery from a 450W
unit, in a 1200W product. The product would be lightweight,
the shipping cost would be minimal, but then it would
only have a few minutes of runtime with a 200W load.
And if the 1200W load was applied (with that flimsy battery
in place), it would run out of juice in less than a minute.

So while there is a proportionality (bigger battery in
higher power UPS), you really need to consult the
runtime graph, to figure out the actual runtime.
Some manufacturers provide a nice graph, while
others say things like "run time 9 minutes at half
load). Which amounts to a single data point.


Re: OT a bit - matching UPS with one's pc

On Monday, June 10, 2013 5:58:25 PM UTC-4, Paul wrote:
Quoted text here. Click to load it

  And that unit would immediately shutdown three year later as the battery  
further degraded.  It would not have power necessary to provide 1200 watts  
long enough to do anything useful.

  All electronics are so robust that a modified sine wave UPS is sufficient
 and ideal power.  Many get hyped by hearsay (as another demonstrates by no
t saying why power must be 'cleaner').  Spend massively for a double conver
sion UPS or a double conversion that still outputs 'dirty' sine waves.  (La
tter costs less, is double conversion, and still has a crappy output.)

  Spending all that additional money accomplishes nothing useful.  Since an
y advantage from a double conversion UPS is already done better inside elec

  UPS is temporary and 'dirty' power during a blackout.  It must be sized s
o that it will provide sufficient power even when seriously degraded in thr
ee years.  Other engineering factors also say why a 350 watt load means a U
PS maybe 500 watts or larger.

Re: OT a bit - matching UPS with one's pc

The double conversion scheme is a power conditioner and UPS at the same  
time. If you understand electronics, it is far easier to absorb a surge  
at DC versus AC.

Seriously, those modified sine wave UPSs are so shitty it is far better  
to suffer a power outage.

Look at the modified sine waveform. It is more like a modified square  
wave. The steep edges will stress the power supply since i=C*DV/DT, and  
the DV/DT of a square wave is infinite. Of course the modified sine  
doesn't have an infinitely fast rise time, but it is faster than the  
design criteria of 2*pi_60Hz.

Re: OT a bit - matching UPS with one's pc

On Wednesday, June 12, 2013 10:44:12 PM UTC-4, miso wrote:
Quoted text here. Click to load it
Quoted text here. Click to load it
Quoted text here. Click to load it

  How many power supplies have you designed?  Anything done by the power co
nditioner is routinely done inside all electronics.  For example, how does  
that 'shitty' power become the rock solid 3, 5, and 12 volt DC.  All shitty
 power is first converted to well over 300 volts.  And then converted to hi
gher voltage radio wave spikes.  Does not matter how 'clean' that UPS power
 is.  Because inside all electronics, the power is then made much 'dirtier'

  What happens next?  Those many hundred volt spikes are converted to rock  
solid  3, 5, or 12 volts DC by superior power conditioners already inside a
ll electronics.

  Why spend so much money on a $1000 UPS when all that 'cleaning' is automa
tically undone inside all electronics?  Many have fears rather than learn h
ow electronics work.  Those fears get the naive to recommend a $1000 UPS ra
ther than the $100 one.

  If 'dirty' UPS power is harmful, then you can cite each specific part ins
ide an appliance damaged by 'dirty' power.  If you cannot, then knowledge c
omes from advertising and hearsay; not from designing electronics.

  'Dirty' UPS power is ideal for all electronics.  However is can be harmfu
l to less robust devices such as small electric motors and power strip prot

  Cure the problem.  Spending excessively on a double conversion UPS solves
 what problem?  Averts damage to what part?  Making subjective claims about
 'shitty' power is why advertising is so profitable.  What a double convers
ion UPS might do is already done better inside electronic appliances.  Now  
reread those numbers.  Any cleaning by a UPS is first completely undone ins
ide electronics.  And then ?cleaned? by superior circuits already insid
e each appliance.

Re: OT a bit - matching UPS with one's pc

On 6/12/2013 10:44 PM, miso wrote:
Quoted text here. Click to load it

If one has a conservatively designed computer then the modified sine wave  
you so fear is not a problem at all. This is especially true since it is  
powering the computer for very short periods of time. As a Unix sysadmin  
I've been down the road with the high-priced always-on sine wave UPS units  
such as the big Toshibas and know that for the home user who suffers a few  
minutes of outage every month such a thing would be throwing money away.  
Doubly so since the always-on units are fairly greedy consumers of power on  
their own.

I have five of the sort of UPS units you preach against and have had them  
for years and from several manufacturers. I have fairly frequent short-term  
power failures and occasional brownouts. I have had zero power supply  
failures during this period of use. If they were as dangerous as you say,  
shouldn't I have had at least one computer going up in smoke?

Re: OT a bit - matching UPS with one's pc

You are making the assumption that the UPS is one of those crappy  
modified sine wave types. They should simply be avoided. I would rather  
lose power than use a modified sine UPS.

I use a double conversion UPS, not all that different from the $500+  
unit the other poster listed from newegg. Double conversion true sine is  
the only way to go. Otherwise, don't use a UPS. Your gear SHOULD be able  
to handle the crappy signal from the cheap inverter, but some devices  
can be damaged. Basically the modified sine, which is practically a  
square wave, causes increased surge current in the capacitors of the  
computer power supply.

The thing to note with a true sine wave double conversion UPS is that it  
is running all the time. It has fans and thus noise. Think of a computer  
server room at a business, then imaging that in your house.

I'm using an optiups 1500b. About 4 or 5 years old at this point and  
thus far no problem. It can run a desktop and a few very low power  
computers. That is, 1500va is a lot of power.

Most of your computer power supplies, unless they are bargain basement  
PCs, will have power factor correction. So they will look resistive to  
the UPS.
Quoted text here. Click to load it

Everything I build uses 80plus rated power supplies. But a lot of the  
low power accessories will not have active power factor correction. That  
shouldn't be an issue since the key phrase here is low power. But it  
wouldn't be surprising for all the wall warts and such to add up to a  
hundred watts.

BTW, I also agree having a Kill-A-Watt is a good idea. Get the cheapest  
model since the fancy ones have features that really aren't all that useful.

  > http://www.newegg.com/Product/Product.aspx?Item=N82E16842107015

I see they still make it. OK unit, but I don't like how the front panel  
attaches. It doesn't attach firmly. That is my only complaint. It is one  
of the quieter double conversion units around since the fans are  
variable speed. Still it is damn loud.

Tomshardware has review of it and a few others. Note the opti can work  
off of serial or usb. The deal is if you have a number of devices on the  
supply, serial is better since that one signal can be bussed around.  
That is, for a UPS to be completely effective, all your attached devices  
need to be able to sense the power failure signal. This can also be done  
via lan instead of serial port, but the serial port scheme requires the  
least brain power.

Quoted text here. Click to load it

Note it has the best voltage regulation because it was the only double  
conversion unit in the test. It was also the most expensive. However, it  
will last hopefully for use over at least two different computer systems  
for the average home user that gets a new PC every 5 years. Maybe three  
if you change computers more often.

Note that the double conversion scheme gives you true surge protection  
since it comes for free given the design. That is, the raw AC is  
rectified then filtered. At the filter stage, you can really do decent  
surge protection because you are dealing with DC. Filter the bejesus out  
of it. Big ass caps are not a problem. Then you convert the DC to AC  
with the true sine inverter.

Re: OT a bit - matching UPS with one's pc

On Sunday, June 9, 2013 10:19:04 PM UTC-7, Yes wrote:
Quoted text here. Click to load it

You can't just add up amps but instead have to add up powers  
(amps X volts), but it's a lot more realistic to plug your computer  
and monitor (but not the printer, scanner, or audio system) into a  
Kill-A-Watt ($15-25) and run a worst-case power test (video game).
Then note the maximum Volt*Amp and watt numbers the Kill-A-Watt  
reports.  Add 50% to those numbers to get the VA and watt ratings  
you'll need from a backup power supply.  Generally backups have
a watt rating that's about 60% of its VA rating.

Re: OT a bit - matching UPS with one's pc

On 06/10/2013 12:19 AM, Yes wrote:
Quoted text here. Click to load it

Yes...UPSes are typically rated in VA   (volt amp) which is not the same  
as watts.  If you assume a power factor of .65    a 1000VA  UPS is good  
for 650 watts.


Site Timeline