UPS

[chrisisasavage]

This is the first time I am looking for a UPS for a small server room.
We need a UPS for our phone system. We do not have a generator, so
we'd like to use a UPS as batter backup for a couple hours. I know
this isn't going to be cheap, but it's our only solution in the suite
we're at.

I opened the side of the voicemail server to see the power usage on the
power supply. It rated 115V/10a. This seemed excessive, but I double
checked it was the actual input rating, and considering it's function
I'm assuming it's correct. When selecting a UPS for a computer should
I use those numbers or the number of Watts the supply is rated at? I
don't have a way to test it to check it's actual draw. A 1150va and
300W Power supply don't seem to correlate very well, any idea why the
input is rated so high?

The next question is with our phone switch. It's rated 117V 6A.
Generally, should I use 702va to figure the power usage of this device?
I've had problems getting help from the manufacturer on how the power
usage would work.

The next questions is on the monitor for the voicemail server, which
happens to be a Compaq v570, which I've seen listed as 75W. The back
of the monitor shows the device voltage as 100-240V 1.2A. If it's 100V
that would be 120VA, right?

Sorry I sound ignorant (and am ignorant) on this one. I've read and
read and read trying to figure out what I need. Is Watts or VA better
to figure power by. It's not going to be cheap (we're looking in the
range of 2 hours), so I'm ready for that. Thanks for any help,
answers, or ANY information you can pass on. Thanks.

 

[Bennett Price]

I think you should spend $20 and get a clampon ammeter from Radio Shack,
model # 22-602 along with a short 3 wire extension cord, preferably the
sort with 3 parallel wires, (glorified zip cord). An air conditioner
extension cord is the sort you want.

Split the three wires apart, without knicking the insulation.
You'll have to power off the equipment to do the following.
Put each of the loads on the extension cord and clamp the ammeter around
each of the 3 wires and let the equipment boot up. You should get
identical readings on two of the wires and zero on the third. Either of
the identical readings is how many amperes your equipment is consuming,
(don't double the reading because you measured it twice.) 120 volts
times the sum of your amperes is the VA.

As you suspect, the ratings on the equipment may bear little relation to
what you measure. They may simply reflect how many amperes the line
cords and power switches are rated to carry, not how many amperes they
are carrying.

Also, if power is out, is the server room likely to overheat due to no
ventilation?

 

[Mike Walsh]

A clampon meter is not necessary. If you have a regular digital multimeter you can cut the hot wire and use the meter to measure the current. The neutral should have the same reading. The one that should not have any current is ground.
Voltage times current equals watts. VA is not watts, but comes from some formula that give an inflated number. This plus the fact that a UPS is rated for peak, and not continuous power means that the VA rating should be twice the actual wattage required, and should be four times if you want to have a decent battery run time. E.g. with a PC that requires 200 watts you should have a UPS with a 800 VA rating.

 

[Bennett Price]

The original poster seemed inexperienced at matters electrical; hence I
suggested a clampon ammeter as much safer than cutting all 3 wires and
then trying to determine H-N-G. Also, some inexpensive DVM's don't
measure AC current.

I don't agree that UPS' are rated at Peak; in general they are rated at
operating conditions - whether in VA or Watts. VA is Volts times
Amperes - not some 'inflated number'. Watts are V x A x Power Factor
(which can be anywhere from zero to one); thus watts are equal to or
less than VA.

Since it is difficult to know the PF or true wattage of a load, purchase
a UPS based on its VA rating and throw in some more fudging factor.
Keep in mind that long run time can be achieved either by buying an
oversize UPS or buying one of the appropriate VA/wattage rating with
extra large or supplemental batteries. The VA capacity of a UPS is
determined in part by its batteries but also by its internal components.
You can put the biggest battery on a 350 VA UPS; it will still only
put out 350VA but it may do so for days. Similarly, if you have enough
flashlight batteries, you can run a 5KVA UPS, but only for seconds.

 

[kony]

The original poster seemed inexperienced at matters electrical; hence I
suggested a clampon ammeter as much safer than cutting all 3 wires and
then trying to determine H-N-G. Also, some inexpensive DVM's don't
measure AC current.

I don't agree that UPS' are rated at Peak; in general they are rated at
operating conditions - whether in VA or Watts.

Definitely the consumer grade units, including APC's, are
not rated at operating conditions. AT least two APC units I
have cannot output 500VA with fresh battery, yet that's
their rating. With a load (system and monitor requiring
roughly 300VA) they will shut down immediately.

 

[chrisisasavage]

Thanks for all the help. I think I understand what we need now. It
sounds like over compensating is the best bet, which I'm assuming
buying based on VA + Extra capacity will achieve. I have a multimeter
and plan on measuring it. I agree with the statement that I "seemed
inexperienced at matters electrical" to some extent.

I found the following on APC's website using their selector:

Sizing and run time calculators are based on Watt ratings. VA
(Volt-Amp) ratings are also a common method for measuring power but
provide less accurate results. On APC.COM, VA values are often shown
for convienience and may be entered as input data. However, VA values
are always translated into Watts. This translation is done using an
approximation formula which assumes that low VA loads have a power
factor of around 0.6 and high power loads have a power factor of nearly
1.0, which is representative of typical, (but not all) computing
equipment. For highest precision, use Watt values where available.

The translator below shows how VA can be converted to Watts using the
APC.COM formula:

My concern is in regards to the power factor. The voicemail server is
rated at 300 watts, but is 10 amp. That's a big difference between
watts and VA. If what APC is saying I should use 300 watts as the
rating. I guess testing it is the only way. I plan on getting a
clamp-on ampmeter today.

 

[kony]

My concern is in regards to the power factor. The voicemail server is
rated at 300 watts, but is 10 amp.

It is not likely to be 300W or 10A.

Where are you seeing those numbers?
They are not usually a rating of what the device uses, only
a maximum rating for some particular portion of the supply,
for example the maximal current a fuse in it could sustain
before tripping, or the maximal power the power supply could
output IF it were loaded such that all rails had a fair
balance (and of course if power supply was rated
accurately).

That's a big difference between
watts and VA. If what APC is saying I should use 300 watts as the
rating. I guess testing it is the only way. I plan on getting a
clamp-on ampmeter today.

For a headless (no monitor) basic server, not some beefy
multi-CPU raided-array large server, you should be able to
factor about 500VA per but a larger margin would be prudent
like 750VA per. You're the only one who has access to the
servers though, it's not impossible to have a server using
300W, just not very common or necessary for a task like
serving email.

To provide more margin or runtime (or both) even more
capacity could be allocated. This is within context of
putting mulitple systems on same UPS. Runtime after power
shuts off can be pretty much ignored though, in most
scenarios, since the primary goal with a server would be to
shut down gracefully and it matters little if it ran for 5
minutes before shutting down or 15 minutes, presuming it's
not common for your 'site to have power outtages that just
happened to conincide with a 5-15 minute window of recovery
and power returning to a normal state.

If you need more uptime during an outtage, thought should
shift from a battery powered UPS to a fuel or gas powered
generator.

 

[John McGaw]

This is the first time I am looking for a UPS for a small server room.
We need a UPS for our phone system. We do not have a generator, so
we'd like to use a UPS as batter backup for a couple hours. I know
this isn't going to be cheap, but it's our only solution in the suite
we're at. snip...

Sorry I sound ignorant (and am ignorant) on this one. I've read and
read and read trying to figure out what I need. Is Watts or VA better
to figure power by. It's not going to be cheap (we're looking in the
range of 2 hours), so I'm ready for that. Thanks for any help,
answers, or ANY information you can pass on. Thanks.

This is probably the easiest way to do the job:

http://www.brandelectronics.com/

Even their cheaper meters do the job well. I've been using one for a few
years and can find no real weaknesses in it.

 

[Bennett Price]

Do you actually need several hours of backup time? What happens to the
phones and PCs on peoples' desks when the power goes out? Little need
to keep the servers running if noone can access them.

 

[Bob]

Do you actually need several hours of backup time? What happens to the
phones and PCs on peoples' desks when the power goes out? Little need
to keep the servers running if noone can access them.

In terms of Internet connectivity, the most important question is when
does your ISP lose power. I do not even know if mine has generator
backup.

 

[Bob]

This is probably the easiest way to do the job:

Even their cheaper meters do the job well. I've been using one for a few
years and can find no real weaknesses in it.

I still say that the best thing is to call the UPS customer support
and ask them to spec the correct size UPS based on the loads you plan
to backup, how long, etc.

I did the math and it did not work out. I kept getting requirements
that were too large. Then I called Tripp Lite and the technician did
the calculation using his estimator software and got a much more
realistic value. The difference was such that I could use the 750 VA
unit I bought and did not have to get a 1500 VA unit as my
calculations indicated.

Let the professionals make the determination. It is apparently a black
art.