UPS for unraid


Recommended Posts

38 minutes ago, Sabot said:

Any thoughts on a stepped approximation to a pure sine wave UPS? From what I read so far, a pure sine inverter is the better option. How important is pure sine for use with current computer power supply for the shutdown process?

 

A few years back that was a real problem with some of the new green PS's  not being compatible with the the stepped approximation wave forms.  It seems that this issue has been resolved.  And the stepped approximation waveform will cause the transformers in the PS to run a bit hotter which is a concern.   In my opinion, if you are in the situation where the number of times that you on the UPS is only a few times a year and you actually initiate shutdown with a two minutes of power outage, the use of a stepped approximation UPS will not cause any problems.  You are looking at the time to when the batteries are exhausted.  That type of calculation is a waste of time in the real world.  That run time is usually wildly optimistic  and will decrease with time as the batteries age. 

 

You should be concerned more with the instantaneous peak power that you server requires when it spins up all of your drives simultaneously.  Measuring it is very difficult as it last less than two seconds.   (You can try to find what the peak current is for your drives but many manufacturers don't provide it.  A rule of thumb is about 2amperes per drive.  Why is that number important?  Because if the power rating of the UPS is exceed for a few milliseconds, it will instantly shutdown to protect its power switching transistors.  My suggestion in determining when to shutdown your server after a power outage is this.  How long is the power out when it comes back on within a reasonably short time and stays on.  In my area, if it is out thirty seconds, it will be out for, at least, a couple of hours.  So I have my time on battery set at thirty seconds. 

 

One more factor to be considered.  Let's say you have the time set for 70% of remaining battery.  You have a power outage and the server shuts down as a result.  The power comes back on, you decide the things are back to normal and restart the server.  And the power down goes out again a few minutes later.  The UPS battery will probably not have enough reserve to shut the server down again.   (It takes more than eight hours to completely recharge those batteries.)

Edited by Frank1940
Link to comment

Ok, I don't know what the numbers mean here they are...  First I had to remove the insulation from the three wires on the power cord between the wall and the power supply so I could clamp around either the black or white wires. I have plenty of power cords, so I used my oldest one that I could find.

 

The meters highest read was .82 amps at power up & .33 amps at shut down. I was watching during the cycle. I tested it three times and these are the highest numbers I saw. Plus the meter has High/Low settings anyway.  

 

Question: is this really .82 amps or 8.2 amps. I left the meter on default which has the decimal two places to the left. There is an option to move the decimal one place to the right (Call Range). The attached photo is of the tool used and it's set to default settings as shown. 

 

Any idea on what a computer power supply's power factor is? I used 1 in various calculators but not sure if this is correct... 

 

I am leaning that it's 8.2A since the power supply label gives us 100-127V (I went with 120V) and 8A. Which if the online calculator is correct, 960 Watts for a few seconds during startup and 396 Watts during the shutdown. 456 Watts while the server is running. This is the max amount that I saw in my time watching the meter, for a few minutes. 

 

Now which this knowledge (cough...lol) How do I go about calculating how much time I would have with a 500 watt UPS?

 

20170501_130729.jpg

TS140 Power Supply.jpg

Edited by Sabot
Link to comment

Pretty sure you are doing it wrong. I'd be more willing to bet you are currently seeing .82A for startup, .33 running. However... I don't trust your meter. Since you've already killed and skinned a power cord, separate out enough insulated wire in it to get several turns around the clamp, and try again. Each conductor that passes through the meter loop multiplies your value, so if you manage to get 5 turns you would divide the face value of the meter by 5. That should get the meter into the meat of its measuring range, where it is more accurate. 10 turns would be nice, but may not be doable. You need to also measure the voltage accurately, not just guess. Volts * Amps is the number we want, not watts.

 

Now, the second part of your question. UPS's aren't commonly quoted in watts, as the VA number is bigger, and it's actually more accurate for what's being measured. The stated rating of the UPS is the maximum peak power it can supply, but has little if any correlation to runtime. Runtime is a measure of usable battery capacity and efficiency, and is rarely stated clearly because the variables are too hairy.

 

Battery chemistry in most UPS's is pretty basic (actually very acidic9_9) and hasn't changed in over a hundred years. Because lead is the heaviest ingredient in the battery and is responsible for overall capacity, you can bet that if you have 2 UPS's rated for the same rough peak capacity and one is twice as heavy, it will run your system for significantly longer. Bigger, heavier batteries = longer runtime. Conversely, if 2 UPS's are the same rough size and weight, the higher peak capacity will probably have shorter runtime under equivalent load because of inefficiencies in power conversion.

Edited by jonathanm
Doh!
Link to comment

The way to count 'loops' is the number of wires in the center when the clamp is closed.  (I.e., when you put the wire through when you made your first measurement that was a 'one loop' coil.  If you would take that wire, wrap around and put it through the center portion again, that is two loops.)  @jonathanm said much the same thing but this concept is rather confusing to most laymen. 

Link to comment
1 hour ago, Sabot said:

If i understand you correctly, skin the power cable from plug to plug. Then loop it around the meter loop 5 to 10 times and report the number plus the voltage read at that time?

Yep, like @Frank1940 explained, you are still only capturing 1 of the 3 wires, either one of the white and black or brown and blue depending on country of origin, just not the green. Open the clamp, and wrap as many turns of that one wire as you can comfortably fit around the clamp, up to 10. Close the clamp fully or unwrap wire until you can, and start logging reported amps. Take the meters face value AMP number, and divide it by the number of wires that actually pass through the clamp. Voltage should be extremely stable, just take a reading at the plug you are using before and after you are done playing to be sure.

 

Oh, if you want to take a current reading on all three wires one at a time, it could be educational if your house wiring is wrong. The two main conductors should yield identical results, and the green conductor should measure exactly zero at all times. If not, something is wired VERY wrong, and I'd have an electrician inspect it ASAP. :)

Link to comment

I'd forget the clamp on meter and just see what you're drawing with all drives spun up.

 

Simply use your Kill-a-Watt (set for watts);  click the Spin Down button (so all drives are spun down) ... wait a few seconds for the reading to stabilize (this will be your draw with the system idle (no drives spinning);  and then click the Spin Up button and watch the Kill-a-Watt.   It will show the spin-up draw current for a few seconds; then will stabilize at the max "normal" draw for your system -- with all drives spinning.   You could then start a parity check to see how much this goes up with all drives not only spinning but actively being read.     These results will give you all you need to make a reasonable decision about just how much power you need from your UPS.

 

Link to comment

Skinned the power cord, she looks so naked... :) I can do as many wraps as you wish now!  I did ten wraps around the clamp, trying to stay between the little marks on the clamps. I assume that is where they want you to place the wire that I am testing.

 

Loops around the Clamp: 10 

Starting: 9.45A (Highest)

Starting UnRaid: 3.99A (Highest)

Idle with UnRaid: 3.85A (Highest) 

Shutting down: 3.84A (Highest)

Voltage: 117.5V  Voltage was in motion during monitoring but almost always came back to 117.5. 

 

Are these numbers workable to inform you what my system needs in terms of a UPS? I'm looking for an absolute, just confirmation. :)

 

20170501_200806.jpg

Start.jpg

Unraid started6.jpg

20170501_201930.jpg

Link to comment
59 minutes ago, Sabot said:

Using the online calculator, it gave me 468.825 VA. Is this the number that I should work with?

Close, but no. You forgot to divide the amps by the number of passes through the clamp, 10 if you counted correctly, and you want the absolute worst possible scenario, so the highest amp draw you actually measured was 0.945 Amps, giving you a max draw of around 112VA. It looks like your meter was off by about 10%, since you originally measured only 0.82 Amps before multiplying the signal with the wrap trick.

 

That's only the server itself, you need to account for all the other equipment that you would possibly connect to the UPS for peak load capacity. That said, if you truly were ONLY connecting the server to the UPS, you could get away with one of the little beefy power strip looking ones, but you would probably get less than 5 minutes runtime before the batteries were completely dead.

 

Now that you have a VA figure and can calculate loads for what you need to power, you can get a better idea of runtime by using the % full load figures. I'd recommend at least starting with 500VA as a minimum UPS capacity, and find the runtime charts for the models you are interested in to see what kind of minutes you can get draining the batteries. Then figure half that runtime to avoid abusing the batteries.

Link to comment

Do you recall reading this instruction on how to count the 'turns' or 'loops'

 

4 hours ago, Frank1940 said:

The way to count 'loops' is the number of wires in the center when the clamp is closed.  (I.e., when you put the wire through when you made your first measurement that was a 'one loop' coil.  If you would take that wire, wrap around and put it through the center portion again, that is two loops.) 

 

1 hour ago, Sabot said:

Skinned the power cord, she looks so naked... :) I can do as many wraps as you wish now!  I did ten wraps around the clamp, trying to stay between the little marks on the clamps. I assume that is where they want you to place the wire that I am testing.

 

 

54 minutes ago, Sabot said:

Using the online calculator, it gave me 468.825 VA. Is this the number that I should work with?

 

 

You have got me thinking.  You said you did ten wraps around the clamp.  That would put 11 wires inside of the closed clamp.  So you should be dividing the reading by 11 NOT 10!  (You do the loops/coils because most of these units have very poor accuracy at the low of their measurement range!)  So your Max VA =   9.45/11    X 117.5   = 101VA    That is the figure, I would be working if I truly believed that your meter was capable of reading instantaneous peak current.  What I would use would be this figure 114VA.  Where did I get that.  A very conservative estimate of peak starting current for hard drives is 2amperes per drive.  You have three drives  so   

 

2A/drive  X  12V  X  3drive   = 72VA      and    3.85A/11  X 117.5  = 41.125VA       Total instantaneous VA requirement on drive spinup is 72VA   +  42VA   = 114VA  

 

So the numbers are actually in fairly close agreement.  (Your newer drives probably have a slightly lower peak current draw than drives with older designs did.)  But you have to ask yourself about what plans you have for expansion of this server and how are you going to do it.  Are you going to add more drives  or   Are you going to replace the current drives with larger capacity drives?   That 230W PS is going to limit what you might want to with regard to the first option. You also have to keep in mind that 230W limit when you are looking at the ampere ratings on the individual voltage buses as their total wattage often exceeds the power rating of the PS and that rating is a hard limit.  Exceed that rating and most PS simply shutdown! 

Link to comment

Using your Kill A Watt EZ, hit the MENU button until Volt is showing, like in your pic.  Then use the UP or DOWN button to cycle through the readings until you get power factor (PF), Watt, and VA.  Test and record peak power-on VA and peak Watt.  Record for idle with drives down and idle with drives up.  UPS's are rated in VA (apparent power), with usually around 0.6 to 0.7 power factor being used in the calculation.  They will also list real power in Watts in the specifications.  Looking at amps is not going to tell you much without knowing power factor.  Computer power supplies are inductive and resistive loads.  Once you have your VA and Watt numbers you can plug that into a web site like APC and they will calculate approximate run times for a given UPS.  Your 80+ Bronze should be fine with a stepped approximation UPS, but if you go 80+ Gold or better the active power factor correction circuitry in the power supply may or may not work well with the high current draw and THD when using a stepped approximation UPS and a sine wave UPS would be a better choice. Some supplies are better at handling it than others. APC recommends 25% above power supply rating when using a stepped approximation UPS and a power supply with active power factor correction. Ex: 280W x 1.25 = 350W. Something in the 650-700VA range.

 

 

 

 

Link to comment

You guys are awesome! Thank you for joining me on my little journey. :)

 

As for the wires on the inside of the coil, there was 10. Attached is a slightly better picture showing the wire in question. 

 

Wow, never knew that the Kill a Watt could provide so much information. It's currently in use between the server and the wall. UnRaid is running at idle during these readings:

122.1 V

.87 PF

.36 A

38 W

44.2 VA

59.9 Hz

 

I thought about powering the network gear but I decided against it as I learned more about what the UPS in my case is for 1) Allow the server to shut down properly 2) Surge Protection 3) Power Correction. I use to think that I didn't need to protect the server for it was not worth much compared to my other systems. Then I realized, it's the data that sits within her that is very important. 

 

What's next for this server, well not much. It's a simple little box without much room for additional toys. It is a tough little guy, I keep coming back to it for these projects. The last project was a bit hard on her, handling BlueIris and 16 HD cameras. I had to move that project to a more powerful box. The little server then sat idle until this project. I think this will be her final project until her death. My next project is to see if I can do a 10Gb peer to peer with it. I am sure I will be seeing the same helpful faces again on that project! ;)

 

Idel1.jpg

Link to comment

I am now thinking of purchasing the CP1000PFCLCD. It will give the little server plenty of time for shut down and still have some battery life left in case of follow-up outages as mentioned above. 

 

Next question, will the server be able to boot on its own upon restoration of street power? How could this be achieved? 

Link to comment
16 minutes ago, Sabot said:

Next question, will the server be able to boot on its own upon restoration of street power? How could this be achieved? 

Often this can be accomplished with a BIOS setting. Whether it is a good idea or not is debatable.

 

I usually prefer to be there when my server boots so I can take a look and make sure everything is working well. Some even go so far as to not have the array autostart on boot until they have checked on it, to make sure all the disks were found, for example.

 

Which reminds me, I need to take care of this myself so my housesitter can watch Plex without complication while I'm on vacation. Seems we often get a power out in this situation.

Link to comment

Also keep in mind that if you get power cuts back to back where your server auto starts when the ups battery is already low, and you get another power cut, your server may not have enough time to shut down safely. I like to start the server manually (or through ipmi) after making sure the ups battery has enough juice for another power cut

Link to comment

Hi guys 

 

I am going to build my first NAS based on my old rig and now I am trying to choose compatible UPS for is. It will be used only as NAS, 1WM (just for testing and dowloading, not gaming) and for some dockers.

Config will be: Intel 2500K, 16GB RAM, ASUS P8Z68-V, couple of 3TB WD RED drives (3-4), SSD, Seasonic X-660. I suppose that power consuption will be below 250w in load.

 

Can you please take a look on follwing UPS https://www.alzashop.com/apc-back-ups-700-bx-eurozasuvky-d2356824.htm?o=2 and try to tell me if it will be compatrible and sufficient? (10min running would be more than enought).

 

Thank you

Link to comment
16 minutes ago, Ghost_sk said:

Hi guys 

 

I am going to build my first NAS based on my old rig and now I am trying to choose compatible UPS for is. It will be used only as NAS, 1WM (just for testing and dowloading, not gaming) and for some dockers.

Config will be: Intel 2500K, 16GB RAM, ASUS P8Z68-V, couple of 3TB WD RED drives (3-4), SSD, Seasonic X-660. I suppose that power consuption will be below 250w in load.

 

Can you please take a look on follwing UPS https://www.alzashop.com/apc-back-ups-700-bx-eurozasuvky-d2356824.htm?o=2 and try to tell me if it will be compatrible and sufficient? (10min running would be more than enought).

 

Thank you

 

That UPS is fine.   Plenty of power and has AVR (which I consider a mandatory feature).

 

Link to comment
  • 5 months later...

I feel really dumb asking this, but I have this UPS: https://www.newegg.ca/Product/Product.aspx?Item=N82E16842101123CVF&nm_mc=KNC-GoogleAdwordsCA-PC&cm_mmc=KNC-GoogleAdwordsCA-PC-_-pla-_-UPS+(Back+Up+Power+Supply)-_-N82E16842101123CVF&gclid=CjwKCAjw64bPBRApEiwAJhG-fh02SfTGHMzxd7s3j9Smal8XxVcaBv3wjAcmfKf9SsoLu-pRGA63eBoC6AMQAvD_BwE and I don't know how to plug it into my server. There's that data port on the back, but how would I plug that into my server if the ethernet port is being used already, or will this UPS just not work because it doesn't have a USB port?

 

Thanks!

Link to comment
  • 2 months later...

Looking to purchase an APC 1500VA Back-UPS Pro Sinewave (BR1500MS) for my HP Microserver Gen 8.  I know that this is overkill for a server of this size, but hoping to avoid buying another UPS too soon in the future.  Any thoughts on this model for this server and compatibility with UnRaid 6?  Thanks for any suggestions.

Link to comment

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.