UPS Advice

[IdlePhaedrus]

Hi All,

My current 1KVA UPS is dying and I need to get a new one. I don't want to just replace the batteries in the current one.

Reasons:
1KVA is not enough, it takes longer for my AD server to shut down with RAID disk caching disabled than there is power from the 1KVA unit, especially if my dev box is running as well.

The current unit has it's own monitoring software which does not interact with Intel RAID, so caching is not turned off when there is a problem, so when the machine is turned back on, the RAID arrays need to be verified, which can take hours.

So, I need a more powerful UPS / Inverter, which works with the MS UPS software on Win2K3. Any advice appreciated.

Phaedrus

 

[ponder]

APC Smart-UPS (XL)

What is your server wattage? Look into 3-5kVA units.

 

[Gnome]

APC is the only brand when it comes to UPSs IMHO. All the rest are just cheap in comparison (quality wise). Although going up in kVA *usually* means a larger battery that may not be the case. If looks aren't a problem the two internal batteries can be removed and it can be hooked up to 2 higher Ah batteries that sit external.

For example: One of the APC unit I recently replaced the batteries on has 2x7Ah (amp hours) 12v batteries in series. There isn't enough space to fit a larger battery, however 2x100Ah 12v batteries can easily be used by simply running the 2 wires outside the UPS. Those batteries aren't cheap (you need deep cycle batteries). And it'll probably cost near R2000 for a single battery but you'll have some serious up time.

Still 1kVA isn't much, taking inefficiency into account you can easily exceed that on a single server. 1kVA does not mean 1000w. Who decided on 1kVA for the server?

 

[ponder]

Who decided on 1kVA for the server?

I suspect somebody sucked their thumb

Eaton, Liebert etc are also good brands but I don't know if they are available locally.

 

[Pada]

You can take the sum of all the power supplies's ratings that it has to feed and divide it by their efficiency and divide it again by their power factor.
eg. if you have 2 power supplies of 350W each, with 75% efficiency and power factor of 98%, then you should get an USP of at least: (350 + 350) / 0.75 / 0.98 = 952VA
If the server's power supply is rated for 350W, it doesn't imply that it WILL consume that much though.

I'm not sure how you're going to calculate the power factor if the PSU's doesn't have PFC (power factor correction) AND it has to be specified on their data/specifications sheet.

 

[IdlePhaedrus]

Thanks everyone for the advice, I will look at APC.

The 1KVA unit was sufficient given my hardware and software configuration at the time of purchase, and given we were into two weeks of rolling blackouts it was also the only one I could lay may hands on at short notice.

Phaedrus

 

[Gnome]

You can take the sum of all the power supplies's ratings that it has to feed and divide it by their efficiency and divide it again by their power factor.
eg. if you have 2 power supplies of 350W each, with 75% efficiency and power factor of 98%, then you should get an USP of at least: (350 + 350) / 0.75 / 0.98 = 952VA
If the server's power supply is rated for 350W, it doesn't imply that it WILL consume that much though.

I'm not sure how you're going to calculate the power factor if the PSU's doesn't have PFC (power factor correction) AND it has to be specified on their data/specifications sheet.

True Pada but also, many UPS have much higher kVA ratings than their rated sustainable power output (in Watt). I don't really feel like searching now but last I checked there were UPSs with 1kVA rating and 700W sustainable power output. So the kVA rating is simply for the sake of peak output (and maybe power factor) rather than constant output. Might have been a cheapo UPS tho but it is worth making sure when buying..

Disclaimer: My terminology may be a bit off, I haven't worked with this stuff in a long while but I still remember the theory I keep wanting to say RMS power but somehow in my mind that applies only to sound.

EDIT: Wasn't so hard to find, example:
http://www.apc.com/resource/include/techspec_index.cfm?base_sku=SC1000I

1kVA unit with 600W maximum power output. Not sure if that is real power (eg. in Pada's example, the real power would be 700w) or if reactive power should be considered instead (reactive being 952VAR in Pada's example) when it comes to maximum power. Probably only real power needs to be considered (so long as the VA Reactive is lower than the VA rating of the UPS).

hehe, hope my theory is correct, maybe some power guru will comment.

 

[Archer]

True Pada but also, many UPS have much higher kVA ratings than their rated sustainable power output (in Watt). I don't really feel like searching now but last I checked there were UPSs with 1kVA rating and 700W sustainable power output. So the kVA rating is simply for the sake of peak output (and maybe power factor) rather than constant output. Might have been a cheapo UPS tho but it is worth making sure when buying..

Disclaimer: My terminology may be a bit off, I haven't worked with this stuff in a long while but I still remember the theory I keep wanting to say RMS power but somehow in my mind that applies only to sound.

EDIT: Wasn't so hard to find, example:
http://www.apc.com/resource/include/techspec_index.cfm?base_sku=SC1000I

1kVA unit with 600W maximum power output. Not sure if that is real power (eg. in Pada's example, the real power would be 700w) or if reactive power should be considered instead (reactive being 952VAR in Pada's example) when it comes to maximum power. Probably only real power needs to be considered.

hehe, hope my theory is correct, maybe some power guru will comment.

VA (volt amps) refers to total apparent power, W (watts) refers to real power. Has nothing to do with peak/sustained power. Essentially you will get more real power out of 1000VA if efficiency and power factor are good.

 

[Gnome]

VA (volt amps) refers to total apparent power, W (watts) refers to real power. Has nothing to do with peak/sustained power. Essentially you will get more real power out of 1000VA if efficiency and power factor are good.

I disagree, why would they post the VA and Watt rating of a UPS?

As I understand it power factor is not a measure of real power. If you have a duty factor of 0.2 on a 200w load you still only using 200w of real power. Naturally some power is lost in transmission because of the low duty factor but considering the wire length and current between UPS the loss due to low duty factor would be fairly low.

I highly doubt the VAR would influence the maximum power output of the UPS as that power isn't actually real power. So long as you do not exceed the VA rating (Calculated using the power factor and real power) and your total load (in Watt) does not exceed the real power rating it should be fine.

 

[Archer]

I disagree, why would they post the VA and Watt rating of a UPS?

As I understand it power factor is not a measure of real power. If you have a duty factor of 0.2 on a 200w load you still only using 200w of real power. Naturally some power is lost in transmission because of the low duty factor but considering the wire length and current between UPS the loss due to low duty factor would be fairly low.

I highly doubt the VAR would influence the maximum power output of the UPS as that power isn't actually real power. So long as you do not exceed the VA rating (Calculated using the power factor and real power) and your total load (in Watt) does not exceed the real power rating it should be fine.

Disagree all you want, thats how it works... P² (Real power (W)) + S² (reactive power (VAR)) = Q² (Total apparent power (VA)). Has nothing to do with peak/sustained power.

And power factor directly relates to how much real power you can get out. Apparent power x PF = Real power

VA + W ratings are given since its easier than listing one with a power factor which would then require a calculation. When sizing a UPS you can not exceed either of the numbers.

 

[Gnome]

Disagree all you want, thats how it works... P² (Real power (W)) + S² (reactive power (VAR)) = Q² (Total apparent power (VA)). Has nothing to do with peak/sustained power.

*sigh* I didn't SAY it has anything to do with peak/sustained power. WHAT I SAID WAS: If your PEAK power is 1100W on a 1kVA unit then you will most likely blow the break/damage the UPS because the peak power likely cannot exceed 1kVA limit (regardless of power factor). Hence the VA rating can be considered the absolute maximum power (1000watt) that can be loaded by the UPS (peak or otherwise). As you should know most power supplies (UPS or otherwise) can for a short period supply more power than their sustained rating (peak power) but the point I was trying to drive across was that your peak load cannot (or rather should not) exceed the VA rating.

Then I said that sustained usage by the load on the UPS should not exceed the rated power (In Watts) of the UPS. And I also said at the same while not exceeding real power, your reactive power cannot exceed the VA rating of the UPS.

Once again I disagree with your point that it just makes it easier, you cannot use just the VA rating. If you have a 900W load with 1.0 Power factor you still couldn't power it with the UPS above which is rated at 600W. The maximum REAL power provided by the UPS is 600W, your duty factor therefore can be low, so long as you stay within 1kVAR and 600W ratings.

EDIT: Do you realize we are both saying the same thing regarding the ratings. LOL. nvm, we have totally derailed the thread.

 

[Archer]

*sigh* I didn't SAY it has anything to do with peak/sustained power. WHAT I SAID WAS: If your PEAK power is 1100W on a 1kVA unit then you will most likely blow the break/damage the UPS because the peak power likely cannot exceed 1kVA limit (regardless of power factor). Hence the VA rating can be considered the absolute maximum power (1000watt) that can be loaded by the UPS (peak or otherwise). As you should know most power supplies (UPS or otherwise) can for a short period supply more power than their sustained rating (peak power) but the point I was trying to drive across was that your peak load cannot (or rather should not) exceed the VA rating.

So the kVA rating is simply for the sake of peak output

Phrases like that confused me...

Then I said that sustained usage by the load on the UPS should not exceed the rated power (In Watts) of the UPS. And I also said at the same while not exceeding real power, your reactive power cannot exceed the VA rating of the UPS.

You most likely mean real power (that cannot exceed the W rating) PLUS the reactive power should not exceed the VA rating.

Once again I disagree with your point that it just makes it easier, you cannot use just the VA rating. If you have a 900W load with 1.0 Power factor you still couldn't power it with the UPS above which is rated at 600W. The maximum REAL power provided by the UPS is 600W, your duty factor therefore can be low, so long as you stay within 1kVAR and 600W ratings.

When sizing a UPS you can not exceed either of the numbers.

Yes, I clearly said you can just use the VA rating

EDIT: Do you realize we are both saying the same thing regarding the ratings. LOL. nvm, we have totally derailed the thread.

No we were not.