[ale] OT need 600-1000W power protection for 3 minutes - cheap

Ron Frazier (ALE) atllinuxenthinfo at techstarship.com
Thu Jul 11 02:09:45 EDT 2013


Hi Alex,

You obviously have an engineering background and probably recent experience with this type of equipment.  I also have a EET engineering degree, from the 80's, and less recent experience.  I have more questions and some minor disagreement with some of what you said.  If I'm fundamentally wrong, feel free to correct me (nicely).  I may remember just enough about this to get in trouble, and it's been a while since I studied it.  I'm also trying to put it into somewhat laymen's terms.  I know you know about what I'm trying to describe.  I'm doing that for others.  Why bother, because I learn too.

In your other thread to me, you made a statement that VA and watts (W) are the same thing.  I think I have to disagree on that.

Borrowing from here:

https://en.wikipedia.org/wiki/Electric_power

<quote>
The portion of power flow that, averaged over a complete cycle of the AC waveform, results in net transfer of energy in one direction is known as real power (also referred to as active power).
</quote>

So, watts measures real power.  Some descriptions say this is energy flow that can do actual work, run your motor, etc.

When you introduce energy storage elements, such as capacitors which store and release electrons, or coils (inductors) which store and release magnetic fields (which induce electron flow), you also introduce additional oscillations in energy flow, which cannot do actual work for you but which may be useful or required for the operation of the system.  Try operating a motor without a coil.

Actually, I've never heard a really good explanation of this, but, the effect of this extra oscillating energy that doesn't do work but shuffles between the source and the energy storage components is reactive power, and is measured in volt-amperes-reactive, or VAR.

So, the effect of the combination of the real power doing the work (W), and of the reactive power not doing work (VAR), is called apparent power, and is measured in volt-amperes (VA).  This apparent power is a real pattern of energy flow in the system.  It heats up the components, and affects the system components.  It just doesn't turn your motor or paint graphics on your computer screen.  Since W and VAR are separate and distinct energy flows, then VA is also a separate and distinct quantity from W.  The W rating on a UPS and the VA rating are two separate and distinct types of energy flow patterns that the device can support without exceeding its capabilities.  Now, whether they rate the ups by using a simple mathematical ratio of VA vs W, or whether they test it for each type of energy flow, I don't know.

The power triangle, and the trigonometric equations that go with it, allow us to determine the relationships of these quantities.

Power Factor is the ratio of W to VA, or real power to apparent power.  So, PF  = W / VA.

If you have a APC BR1500G ups, which is rated at 1500 VA and 865 W (I think), then it an simultaneously support those maximums of those types of energy flows but cannot exceed either.  Those numbers imply a certain amount of reactive power (VAR), which could be calculated.  Suffice it to say that adding more reactive power, either capacitive or inductive (VAR) also increases the apparent power (VA), but the increase is not linear, and must be calculated with trigonometry.  Actually, I did the math, and if I did it right, the reactive power could be 1225 VAR.

Let's say you connect a light bulb.  Assume it draws only real power and there is no inductance and no reactive power.  Your light bulb can draw a maximum of 865 W, since that's the ups limit.  The apparent power (VA) in this case, is also 865 VA, is well below it's limit, and is not an issue.  Now, let's replace the light bulb with a motor.  The motor requires real power (W) and reactive power (VAR) which combine to create apparent power (VA) to run it.  It is conceivable that this motor is drawing large amounts of reactive power (VAR) and lesser amounts of real power (W).  The combination of the two could create an apparent power (VA) which will exceed the 1500 VA rating of the ups.

In the first case, the real power rating of 865 W is clearly limiting what you can do with the device.  In the other case, the apparent power rating of 1500 W is limiting what you can do with it, which is the combination of whatever real load is there and whatever reactive load is there.  Note that it is entirely possible that, in the second example, the real power being drawn is nowhere close to the 865 W limit.

In a residential setting, you are typically billed for the real power only, the up to 865 W.  In an industrial setting, you're typically billed or penalized for using reactive power.

(For extra credit, those interested can look up power factor correction capacitors.  Hint, running a motor uses inductive reactive power.  Adding a capacitor uses capacitive reactive power.  If you do it right, they cancel out.  This lowers your bill in an industrial setting.)

So, we see that we can exceed the real power rating of the ups, or we can exceed the apparent power rating, and either one is a no no.  Trust me, the ups will scream about it, or shut down.  Don't ask me how I know.

You also said that you didn't think I was running with a high power factor.  I've done some testing and I still think I am.

I'm taking readings from the Kill-A-Watt EZ and from the APC Smart UPS display panel itself.  There is more stuff on the ups than just the pc, and the Kill-A-Watt measures just the pc, but the data is still useful.

Test 1: Idle PC
Kill-A-Watt: 193 W, 195 VA, .98 PF
UPS Display: 265 W, 277 VA,  .96 PF (calculated)

Test 2: Loaded PC
Kill-A-Watt: 445 W, 448 VA, .99 PF
UPS Display: 485 W, 502 VA,  .96 PF (calculated)

Presumably, the ups measures VA more accurately than the Kill-A-Watt, so it will know when to shut down if overloaded.  However, all my tests show that I'm running in excess of .95 power factor in both idle and loaded conditions.

I don't think the advice to do calculations all in VA is a good idea for a couple of reasons.

First, is accessibility to the data.  I looked at the name plates on a power brick for my laptop, my monitor, and my pc power supply.  Granted, this is a limited statistical sample, and granted, you might not want to attach your laptop to a ups, but you get the idea.

Laptop: no VA rating, no W rating, V = 100 - 240, A = 1.5
This makes no sense to me in that the A rating is not different for 240 V.  It should be substantially less.  In any case, 120 V * 1.5 A = 180 W.  That sounds way off to me, but it's what the label says.

Monitor:  V = 100 - 240, A = 1, calculate 100 V * 1 A = 100 W 

PC Power supply, in the PC, no ratings visible, box says 750 W.

Basically, there is not a VA anywhere in site.  Nada.

Also, there are many wall warts running phones and hubs around that it would be almost impossible to reach and read the labels on.

The second reason I don't think you should gauge by VA especially using the .6 conversion factor, is that you get the wrong numbers.

Let's take the figures above.  They add up to 1030 W.  Using your conversion factor of .6, this equates to 1716 VA.  So, you might buy an 1800 VA / 1100 W ups, if one existed.  This would be pretty pricy, but would probably work.  It would probably be overkill.

Let's take another example.  Let's say I know I'm pulling 800 VA based on the Kill-A-Watt, including pc and all the stuff I want to keep running.  Let's assume the reading is fairly accurate.  So, we take 800 VA, and multiply by .6, and conclude that I can buy an 800 VA / 480 W ups.  This assumes I'm running at .6 power factor.  This is totally wrong.  I have demonstrable evidence that I'm running at .95 power factor.  So, I don't need 480 W, I need 760 W.  So, I've just dramatically undersized my ups.

This is exactly the situation I'm in, although the load grew to exceed the ups.  I need 540 W minimum, and the ups only has 500W.

So, if you start with W, convert to VA, and buy that VA of ups, assuming they use the same conversion factor, you'll be OK.

But, if you start with VA, and don't account for the W, your ups equipment will be way too small, especially if running at a power factor of greater than .6.  The greater the power factor, and hence the more real power, the more undersized your ups will be.

Here's how I would size a ups.

Figure out everything you want to run during a power failure.  PC, monitor, hubs, keyboards, external hdd's, phones, whatever.  
A) Add up all the name plate watts ratings.  Add a fudge factor for growth if you like.  Buy a ups with that many watts.

or

B) Connect all those things to a Kill-A-Watt AND run them all at full blast.  Max out the CPU, max out the GPU, let all the fans spool up, have the drives accessing.  As you saw, the difference between my idle numbers and my loaded numbers is 250 W, and I'm not even taxing the cpu or one gpu.  The only reason you wouldn't do this is if you KNOW you'll never need all that power during a power failure.  Add a fudge factor for growth.  Buy a ups with that many watts.  This may be similar to the name plate method but may be a number that's easier to get.  Buy this size ups.  You could buy smaller if you know you don't need it.

Now, this ASSUMES you can live with a 3 - 5 MINUTE runtime.

Yup, that label on the box that says "120 minute maximum runtime" etc. is totally USELESS.  That's only true if you're running it at about 1/20 of it's maximum load.  But we've calculated the size based on running at the maximum load.

You have to search for the run time data.  If you want more run time, decrease the load or increase the ups so you are running it at say 30% of its maximum capacity.  Although it's not totally linear, you'd get about 3 times the runtime that way.  As Alex said, you can also add batteries.  Sometimes the ups has a built in connector for this and sometimes they sell extended batteries.

Perhaps you can dump some loads when the power fails, either manually, or automatically.  For example, if I immediately shut down my pc during an outage, my wife's monitor will keep running MUCH longer if it's on the ups.

In terms of runtime philosophy, I think Jim Kinney said in a prior thread, he has the system shut down after 1 minute.  Be sure to allow a few minutes for the actual shut down to take place.  You may have to time that.

Also note, the general settings for power management software, like in your laptop, will be set to shut down at 10% or 5% of battery power left.  The problem here, is that you're going from 100% to 0% in 3 minutes.  That's a 33% drain per minute.  If your system shuts down at 10%, it will have only 20 seconds to shut down.  Either set it to shut down based on time, or set it to start the shutdown at 75% of battery power left.

My personal attitude is that, if I can afford it, I prefer 10 - 15 minutes of runtime.  If I'm in front of the pc, I'll monitor things and shut down myself.  Many power failures correct themselves in less than 15 minutes, and I may be able to continue working.  If not, the system still gets shut down even if I'm not there.

Hope this helps.

Sincerely,

Ron



Alex Carver <agcarver+ale at acarver.net> wrote:

>The information on Volt-Amps was mostly esoteric and aimed at
>describing 
>why things are rated in VA instead of Watts for AC devices (in other 
>words there's a real, engineering reason to do it.)
>
>What is fundamentally important in sizing your UPS are the following
>things:
>
>1.  Total power consumption of your devices.  You can do a multitude of
>
>things to determine this.  The Kill-A-Watt is one way but you can also 
>use the nameplate method (read the sticker on the power supply, assume 
>that's the load and go with it.)  The nameplate method usually results 
>in overestimation of your power but that's ok, it just extends your run
>
>time unless you're trying to save every last penny on the UPS.
>
>The nameplate will either have Watts (and should properly also specify 
>the power factor if it lists Watts) or it should list Volt-Amps.  I
>took 
>a look at some things around my desk (power bricks, an ATX power
>supply, 
>etc.) and they all are rated in Volt-Amps (one only lists Volts and
>Amps 
>separately so just multiply.)
>
>If the power supply is a switching type and is rated in Watts rather 
>than Volt-Amps (improper but possible) then divide by 0.6 to get an 
>approximate Volt-Amp value.  That's an average power factor of a 
>switching supply, most aren't quite that bad, but you'll get some
>margin 
>out of the underestimation.
>
>Do all your calculations in Volt-Amps, don't worry about Watts.
>
>2.  Decide what kind of runtime you want for your system.  Do you want 
>to ride out a blackout for 10 minutes (long enough to shut down
>cleanly) 
>or do you want to stay up and running for over an hour?  Your choice 
>here is reflected one of two ways, either you end up with a low power 
>UPS that has a massive battery bank or you end up with a UPS that can 
>handle a very large load and you'll only use it to a fraction of its 
>capacity (which is the equivalent of a really big battery bank.)  So a 
>1kW UPS with 1000 Amp-hour battery pack would be similar to a 10kW UPS 
>with the matching 1000 Amp-hour battery pack.  The inverter rating
>gives 
>you how much load you can hang off the UPS, the battery pack Amp-hours 
>gives you the run time.  But battery packs scale with the inverter 
>rating so really big inverters must have really big batteries.  Small 
>ones don't need big battery packs but can use them (cooling is
>important 
>here.)
>
>3.  Select a UPS with the appropriate rating.  Your minimum is going to
>
>be the lowest tier UPS that is larger than your combined load.  Your 
>final selection will depend on your choice of run time (small inverter 
>with cooling and a giant battery pack or a big inverter with a built-in
>
>big battery pack.)  Usually the easiest solution that doesn't require 
>modifications is the big UPS.  If you don't mind the tinkering, just
>get 
>a small one (with fans) and put a huge battery on it.
>
>On 7/10/2013 09:23, Scott Plante wrote:
>> Wow, the third paragraph on mostly went right over my head. I realize
>> you were responding to a post, but is that esoteric background or do
>> I need to wrap my head around it to decide what size UPS I need.
>> Coincidentally, we also need a new UPS and we're trying to figure out
>> exactly what we need. I don't have a Kill-A-Watt, but I see I can
>> pick one up for ~$20 if it'll help. The blurb on it claims:
>>
>> * Large LCD display shows kW, volts, amps, watts, Hz and volt amps
>>
>>
>>
>> So how would I use that device, or another method, to size the
>> correct UPS? And do I need to perform a calculation involving PI?
>>
>>
>> Scott

<snip> <snip>



--

Sent from my Android Acer A500 tablet with bluetooth keyboard and K-9 Mail.
Please excuse my potential brevity if I'm typing on the touch screen.

(PS - If you email me and don't get a quick response, you might want to
call on the phone.  I get about 300 emails per day from alternate energy
mailing lists and such.  I don't always see new email messages very quickly.)

Ron Frazier
770-205-9422 (O)   Leave a message.
linuxdude AT techstarship.com
Litecoin: LZzAJu9rZEWzALxDhAHnWLRvybVAVgwTh3
Bitcoin: 15s3aLVsxm8EuQvT8gUDw3RWqvuY9hPGUU




More information about the Ale mailing list