[ale] help providing stable power to pc's to ride through storms

Ron Frazier (ALE) atllinuxenthinfo at techstarship.com
Wed May 23 02:04:23 EDT 2012


Hi guys,

I want to thank everyone for the information you shared with me regarding this topic.  I ended up buying an APC Smart UPS 750 VA / 500 W unit.  I've had some interesting experiences installing it and testing it and the batteries in my laptops, which are attached only to a surge suppressor.  The presumption is that they can run on their internal batteries if there is a power failure.  I've added an APC premium A/V surge suppressor, since, as I mentioned in a previous thread, one of the other APC UPS's I looked at had only 354 joules of surge protection.  This additional surge suppressor, which I've placed between the wall outlet and the UPS, can absorb over 4000 joules of energy.

This UPS has several features which I really like.  It has a pure sine wave output, and Automatic Voltage Control (AVR) which can either boost low voltage or limit high voltage.  Their cheaper units with AVR only boost, not limit.  I think Jim Kinney mentioned this feature in another post.  It also has a sophisticated menu which gives you lots of status data and control over various settings.  It comes with a couple of pieces of software, which I haven't tried yet.  It comes with communications cables for serial and USB.  I plugged it into a USB port, and Windows recognized it immediately.  I don't know how to manipulate the power settings in Linux, so I'll describe my Windows experience.  Those of you who know how can describe how to do similar things in Linux.

I actually had a less sophisticated APC UPS of the same size capacity already, but it didn't have the advanced features I mentioned.  Those, along with the new surge suppressor, will give me confidence to run through storms.  I haven't solved my runtime problem in the event of a power failure.  So, I get about 7 minutes with everything on the table running.  Not ideal, but it's enough time to shut down.  Getting hours of runtime at a 300 W load costs thousands of dollars, which I'm not prepared to spend.

Now, I remember why I hate dealing with batteries.  They're all different, all finicky, and generally don't do what you expect.  After rerouting all my electrical cables and configuring the system.  I set about testing it.

The battery was almost fully charged already, I let it finish charging and pulled the plug.  I've gotten burned this way before, but since this unit was new, I thought it should work.  I let it run until it thought it had 1 minute left, for all of 6 minutes, and plugged it back in.  Worked like a charm.  Then I decided to test the laptop batteries.  This is where the fun began.

My objective, which may differ from other people's objectives, is for each machine to run as long as it can before shutting down, then hibernate in the last 5 minutes of battery life.  The problem is that every battery is different.  The other problem is that the PC's battery meter is sometimes wrong.  The other problem is that, in a high load short runtime situation, like mine with the desktop and two monitors, the normal auto shutdown settings, like 10% charge remaining, won't work.  In my case, 10% of 7 minutes is 42 seconds, even assuming that the battery meter is right and the battery is performing up to par.  In any case, 42 seconds is not enough time to hibernate the PC, so the settings must be changed.

By the way, for anyone reading this who might not know how UPS's are rated, here are the basics.  First, the runtime on the box, without other data, is useless.  A valid rating must give power drain and runtime together.  Runtime ratings without power drain usually are listed at about 1/10 the capacity of the unit, so for realistic runtime, you may have to divide by 5 or 10.  There will be a rating in watts (W) which is  basic power usage.  Then, there is a rating in VA (volt amperes), which allows for the back and forth fluctuation in the power line caused by "non linear" loads (like PC power supplies).  So, in my case, the unit can support loads of 750 VA, but only 500 W.  Neither of these tells you what the runtime is.  This depends entirely on the energy capacity of the battery, and a 750 VA / 500 W unit could have various different sized batteries.  In my case, at a 300 W drain, the battery lasts for about 7 minutes. 

Here's how I worked everything out for each computer.

Note that depriving a computer of power to test the battery can be dangerous if the battery fails prematurely.  I crashed a machine once a few years ago this way and had to rebuild it.  But, the only way to know if a battery works is to run on it.  One option is to do battery testing with a live cd.  That way, if you have a crash, it won't toast your hard drive.

The TAZ computer, my tower, plus 2 monitors, plus a few accessories.  These are the only things on the UPS battery.  Power drain at full load is 300 W or so.  Since the total run time is 7 minutes, and I want to leave 5 minutes for shutdown, I have to set the critical battery limit at 70%.  When the charge reaches that level, the hibernate procedure starts.

Windows has settings in the power options for critical battery action, critical battery level, low battery notification, low battery level, low battery action, and (in Windows 7) reserve battery level.  That last one is where it warns you to plug in a power supply.  I have low battery notification on, low battery action to nothing, and critical battery action to hibernate if on battery power but not if plugged into ac power.

So, for the TAZ, I have low battery set to 85% charge remaining, which notifies me at about 1 minute after the power failure, and the critical battery level is set to 70% charge, which is about 2 minutes after a power failure.

DELL computer, a 3 - 5 year old laptop.  This one was loads of fun.  I pulled the plug to test its battery.  It said it was fully charged.  It didn't even make it through the boot.  I plugged it back in and booted and let it charge up, since it now knew the battery was empty.  After that, I unplugged it and it discharged normally, but a bit rapidly, since the battery is old.  I got about 68 minutes of maximum run time from this one.  So, I set the low battery level at 25% and the critical  battery level at 10%.  This still leaves about 5 minutes of time to hibernate.

TOSH computer, a 10 year old laptop.  This old battery actually lasts longer than the one in the DELL, with an interesting quirk.  It has a maximum runtime of about 1 hour and 40 minutes.  I was testing this one with Linux running.  I watched the charge graph abruptly drop from 58% to 3% in a few seconds.  Something is wrong with the charge meter.  So, for this one, if this trend continues, I'm going to have to set the critical battery level to 60% or 65%.  I don't know how to do that in Linux, but I do in Windows.

ASUS computer, a 1 year old laptop.  My testing shows this battery to be running well and predictable.  It has a maximum runtime of about 1 hour and 50 minutes, similar to TOSH.  However, in this case, the battery meter works.  So, I have the low battery level set at 10% and the critical battery level set at 5%, which gives me the 5 minutes I want at the end.

Warning, old batteries can fail well before the charge meter shows 5% or 10%.  I think the settings in Gnome are way too low for my TOSH battery's condition.

So, if you want to maximize your runtime on battery, this shows you how to test your batteries and adjust each machine accordingly.  Of course, if you just want to shut down or hibernate as soon as there is a power failure, you can do much more minimal testing, just to make sure the battery works, and set the shut down or hibernate level at 90% or something.  But, you don't want it set so high that a momentary flicker of the lights causes a shutdown.

Hope this info is helpful.  I'll be glad to answer questions if anyone has any.

Sincerely,

Ron






--

Sent from my Android Acer A500 tablet with bluetooth keyboard and K-9 Mail.
Please excuse my potential brevity.

(To whom it may concern.  My email address has changed.  Replying to former
messages prior to 03/31/12 with my personal address will go to the wrong
address.  Please send all personal correspondence to the new address.)

(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




More information about the Ale mailing list