William H. Magill on 20 Aug 2004 18:34:02 -0000


[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]

Re: [PLUG] UPS, 2s failures


On 19 Aug, 2004, at 13:49, Paul wrote:
plug-mailing-list-member@gnuisance.net wrote:
apcupsd is logging 2 second power failures often. Is there something odd about apcupsd, my config, or my UPS?

Wed Aug 18 18:14:54 EDT 2004  Power is back. UPS running on mains.
Wed Aug 18 18:14:52 EDT 2004  Power failure.

Can you confirm these weren't actual brown-outs? I have similar entries
in my logs from all the brown outs here yesterday.

Well, the logs cover a period of 13 days.  This just happened:

Thu Aug 19 13:37:35 EDT 2004  Power is back. UPS running on mains.
Thu Aug 19 13:37:34 EDT 2004  Power failure.
Thu Aug 19 13:37:25 EDT 2004  Power is back. UPS running on mains.
Thu Aug 19 13:37:23 EDT 2004  Power failure.

I actually saw it happen. This WinXP PC rebooted and the lights blinked off for a second. My VoIP phone and GNU/Linux PC stayed up. Wow, I hope this isn't what has been happening. It might be time to get an extension cord and add the WinXP PC to the UPS. (I just haven't wanted to deal with apcupsd master/slave mode.)

In general, AC power is "dirty." That is to say it's parameters fluctuate all over the place.


If you have access to a recording voltmeter (a big ticket item) you will see "spikes" and "sags" happening constantly. You will also see frequent zero voltage times (power outages). The same is true for the "actual voltage" supplied. It used to be that everyone talked about 110v AC now it's 120vAC, But it doesn't much matter as they are only averages anyway. Moreover, "electrical equipment" (aka power supplies) are well aware of these issues and are designed to "smooth" out the events and deliver a more constant and consistent voltage to the "delicate electronics" on the other side.

With the "dirty power" events I mentioned above, their duration is normally in micro seconds. They don't become a visible event (like your lights flickering) until they hit a duration of about 1 second. The microsecond events are normally handled by the capacitors in a devices power supply. The longer ones are those which a battery-backup unit (what most people call a UPS) smoothes out.

All of these situations are directly effected by the environment.
Even if you live IN the city, you are probably served by arial lines at some place in the grid. Here in Philadelphia, most folks think their power is "underground" when in fact it is daisy chained along the front or rear of their row houses. Other areas have visible power poles and the issue is more obvious.


While the power distribution system is often referred to as a grid, it is really no different in topology than the Internet or your local Cable provider. The "backbone" of the grid may be interconnected, but the "spokes" are not "spokes in a wheel" (implying alternate, duplicate feeds from opposite ends), but "points on a spoke." You are "down-stream" from your feed on one end, and not connected on the other. In the event of a failure of your primary feed, the secondary feed will (usually) automatically energize -- about a second later ... your lights blink!

In other cases, you "take a hit." In the massive storm that hit Center City about two weeks ago, we took a lightning strike on the transformer that feeds our block. The "lightning arrestor" worked, and popped the transformer off line... oops we had no power for 4 hours until the breaker was manually reset. It was actually a fascinating event in topology -- there are 5 poles supporting our block. Houses on the block are fed from all 5 poles. The transformer is on pole 2. Anybody fed from pole 1 had power -- anybody fed from poles 2-5 were dark.

As for the power in the house being effected by things like the AC, Fridge, Freezer, Dryer, etc. There are two issues here.

All are rotational devices with large startup current requirements. Rotational devices do nasty things to the power phases when they draw large amounts of current -- it looks like a power outage, typically on one phase. The other issue is classic "supply and demand." Depending upon your dwelling's vintage and wiring, you have 40, 60 100 or 200 (and now 400) AMP service. The older the service, the lower the total current available to the building. Just as with a single circuit where you cannot power-up both the computer and the monitor at the same time without popping the circuit breaker (or fuse), you can power up the computer then the monitor in series, then they will both run on that same circuit... startup voltage requirements are higher than operational voltages.

The issue is not that the UPS is that sensitive -- it is supposed to be -- Logging a 1 second outage does not appear to be particularly sensitive. As you noted, it does cause your other device to reboot. Before you had the UPS log, you simply had no idea that it happened since it happened when you weren't around.

The UPS log is diagnostic tool. It DOES tell you what is happening when you are not around.

In your case, you are realizing that you probably want to connect the PC to the UPS as well.

However, you probably don't want to connect the monitor! The monitor is a large power draw and will dramatically impact the power (i.e. time) available from the UPS batteries for no good reason.

Power issues are fun. They are something that used to be designed into "big iron." But once "RISC" appeared on the scene, power supplies were "downsized" to cut costs. As outfits like Dell started slashing prices -- one of the quickest ways to cut costs was to dramatically cut down the "robustness" of the power supply. The design assumption today is that your "electronic appliance" will receive "clean power" and therefore all of the components associated with "robustness" (which are the really expensive ones) are moved out into somebody else's device -- like your UPS, and Surge Protectors. Note that this applies to EVERYTHING electronic in your home today -- TV, DVD, IP alive refrigerators, etc.! PECO offers a "whole house surge protector" to address some of this issue. Urban areas have different issues and requirements than Suburban which are different than Rural.


T.T.F.N. William H. Magill # Beige G3 - Rev A motherboard - 768 Meg # Flat-panel iMac (2.1) 800MHz - Super Drive - 768 Meg # PWS433a [Alpha 21164 Rev 7.2 (EV56)- 64 Meg]- Tru64 5.1a # XP1000 [Alpha EV6] magill@mcgillsociety.org magill@acm.org magill@mac.com

___________________________________________________________________________
Philadelphia Linux Users Group         --        http://www.phillylinux.org
Announcements - http://lists.phillylinux.org/mailman/listinfo/plug-announce
General Discussion  --   http://lists.phillylinux.org/mailman/listinfo/plug