How to choose UPS size?

[KennethW]

I am a first time PC builder that has pretty exclusively used laptops my whole life. My neighborhood has had an uptick in brownouts/15 minute blackouts in the past couple of months and I have lost work that was not autosaved/generally had a bad experience. Picking a UPS seems to require an electrical engineering degree to understand what to even buy, I write software and leave the voodoo magic of hardware to others. 

I have a 1000W seasonic power supply, with a 3900xt, 3070 and 2 IPS monitors I would like to not immediately shut off. No overclocking because I use this computer to make money coding. Any suggestions. I do not even know what to budget. 

 

Thanks!

[Mighty_Dork]

Sorry for not having useful advice myself but Linus talked about this like a week ago FYI, here (timestamped)

 

[IPD]

This question keeps getting asked every week because the moderators decided not to sticky previous threads on the topic (mine included).  And no matter how many times I point this out--none of them will ever sticky a thread to help out the forum users.

 

/rant

 

In any event, add 10% onto whatever your PSU is rated for, because that's the output not the input rating.  Next, add up all the wattage of every device you intend to run on the UPS that the PSU isn't powering.  Add that to the figure you just calculated for your PSU.  Then take that sum, and add 10% overhead for the UPS itself.  That's the grand total of wattage you need the UPS rated for.

 

VA rating will be higher than the rated wattage, sometimes by up to 100% higher.  Worst case, if you take your grand total wattage requirement and find a UPS that has double that number in VA--you'll be fine.

[wseaton]

Unless you are running a data center and just have a desktop get the biggest you can afford and stop wasting brain cells on it. 

 

This is typically a 1000-1500 VA unit. For a basic desktop unit I prefer Cyberpower. This will give you 20-45minutes of typical runtime during a power outtage. All depends on what you are running. Big difference in power consumption between a laptop in a docking station vs an over clocked 12core workstation with a 27" monitor.

 

Battery UPS units are over rated for power continuity and should never be trusted for such. Their key role is to allow for a smooth shutdown during a power outtage, which any of the 'smart' units do reliably enough with a USB and windows built in power management. Power goes out, UPS kicks in, batteries start running down, and UPS sends signal to operating system via USB that it needs to shut down, or else. Even my Windows SQL servers do this plug and play without fault. Also gives you a chance to shut it down gracefully and close all apps. 

 

Hard power faults are the #1 cause of SSD errors in my experience. Wish I could shout this from a hill top. 

 

Biggest pet peeve I have with virtualization in small business is right here. I go into a company and see a rack of servers with a couple of beefy UPS units. Problem is they only have power reserves for a couple of hours. Thunderstorm goes through and knocks out power for half a day and they are screwed. 99% of the time consultants don't configure virtual machines (guests) to integrate with their UPS, so when the batteries die there's nothing to gracefully shut down the VMs. Everything just hard faults as if there were no UPS in the first place, which raises the question, what's the point? Network based UPS units do the trick, but they typically cost 10x vs a simple USB based cyberpower.

 

 

[KennethW]

3 hours ago, Mighty_Dork said:

Sorry for not having useful advice myself but Linus talked about this like a week ago FYI, here (timestamped)

 

Nice to know even linus finds the rating system bad

[IPD]

2 hours ago, Caroline said:

Why 10%? out of all numbers 10%? that's not how efficiency works. If you have a 500W DC output power supply and it's only 80% efficient then it'll draw 625W AC if it was working at full load, and in this equation of yours the rest of the devices are 100% efficient for some reason... weird to say the least. A monitor "brick" is also a power supply and it's not 100% efficient, the wattage written on the label is always the output, and the input is expressed in A.

 

I won't even ask about this.

True.  Assuming you are using 99.9999% of the rated output of your PSU...then you would be correct; you'd need to add a 20% headroom as your PSU is only rated at 80% efficiency.  I don't know of anyone who runs a PSU at 99% capacity (or thinks it's a good thing).

 

I also wouldn't pull data off a monitor brick (they actually make bricks for these things?)  The rated power draw should be whatever the MFG specified in the data specs for that device.  That number is what I add 10% headroom to--not some nebulous idea of whatever a power-brick says.  No devices are 100% efficient, which is why if a monitor is listed at 80 watts power consumption by the MFG, I will calculate based off 88-90 watts. 

 

As far as VA vs. Watts:

https://www.electronicdesign.com/markets/energy/article/21801657/whats-the-difference-between-watts-and-voltamperes

http://www.differencebetween.net/technology/difference-between-va-and-watts/

 

This unit for example:

https://www.eaton.com/ca/en-gb/skuPage.5SC1500G.html

 

Has a VA of 1500 but a wattage of 1050.  VA is almost always higher.  The ratio between VA and Wattage indicates Power Factor of the UPS and therefore is an indication of its efficiency.  While I'm no electrical engineer, I would venture that pure-sine-wave UPS will be inherently less efficient.

 

https://www.fluke.com/en-us/learn/blog/power-quality/power-factor-formula

https://www.riello-ups.com/questions/53-what-is-ups-efficiency-and-how-is-it-calculated

https://www.riello-ups.com/questions/49-what-does-power-factor-mean-with-a-ups

 

[Guest]

13 hours ago, Mighty_Dork said:

Sorry for not having useful advice myself but Linus talked about this like a week ago FYI, here (timestamped)

 

They're close to knowing the answer.  So close.

 

Because VA can be the maximum capability of apparent power.  Since the UPS can only guarantee the duration of the battery under apparent load, they sell under apparent power.

 

The wattage is suggested based on the real power, which can vary based on the power factor.  But because not everything has 98% power factor, they cannot guarantee the real power.  So therefore, the UPS has to be sold based on apparent power, and you'll find that the same UPS with the same VA routing may have different wattage ratings, because they're using different power factor ratios.

 

The fact that they have problems without UPS and with UPS may come down to poor filtering of the AC.  But has nothing to do with the fact that you're using a UPS.  A UPS is completely passive unless you get an online UPS.  But the UPS can add additional EMI filtering that other devices on the circuit cannot filter out. But a good line conditioner does the same thing, which is what I used as a reviewer. I had a rack mount Tripp-Lite line filtering system.