[time-nuts] Line Voltage

[Hal Murray]

A week or two ago, there was some discussion of line voltage.  I've been 
collecting data for several years.  That was a good reminder to see if I 
could find anything interesting.

Executive summary: Boring at my house.

I get a low voltage glitch every 10-20 days.  Most last less than a second 
which is the limit of my setup.  No brownouts.  Mostly my PCs ride through 
it.  (But who knows what they really do.)

--------------

Background:

Here is PG&E's DESCRIPTION OF SERVICE:
  http://www.pge.com/tariffs/tm2/pdf/ELEC_RULES_2.pdf

Nominal 120 volt service turns into 114 to 126 at your service entrance.
(Page 4: max is 120 for Class A, 126 for Class B.)

American National Standard, C84.1 says equipment should be prepared for 110 
to 125.  The difference between 114 and 110 is allocated to drop in your 
house wiring.  I don't know how they got from 126 to 125.  (Page 6)

-----------------

I started collecting data in late May 2008.  I didn't get the software tuned 
up until late 2008 so the data for 2008 isn't as good as the rest of it.

I'm using a UPS by APC.  It's got a serial port with a simple 
command/response protocol.  The voltages are returned with 1 digit to the 
right of the decimal point, but the step size is 0.7 or 0.8 volts.  I'm not a 
volt-nut.  It's been pretty close the few times I have checked with a Fluke 
meter.  I just checked again.  The UPS is reading a volt lower than my meter.

It has commands to read the min and max input voltage since the last time you 
asked.  I poll them.  If either has changed by more than 2 volts or the clock 
has crossed a 5 min mark, I write a line to the log file.

For the first part of 2008, the line voltage was a bit high.  It looks like 
PG&E tweaked it a bit.

I started by making a histogram, but that was boring.  Mostly, my line 
voltage is fine.  I don't have any brown outs.  The only thing interesting is 
the glitches.  So I counted the times where it dropped below 110 or 114 volts 
and the times it went above 126 or 130.  Later, I added below 105 and 100.

There were 2 glitches below 100 volts in 2011.   One was a 1-sample spike 
down to 97.6 volts.  The other dropped to 0 and bounced around for 5 seconds.


Here is a summery.  (It works better in a fixed pitch font.)

year   lines    <100   <105   <110   <114   >126   >130
2008:  69198      13     19     28     32    234      0
2009: 106365      14     15     17     36      2      3
2010: 105566       6      6     10     28      3      0
2011: 106089       2      2     10     19      4      0

Graphs are here:
  http://www.megapathdsl.net/~hmurray/time-nuts/line/

There is a pair of graphs for each year.  The second of each pair starts at 
100 volts.  The min voltage is actually shifted up 1/4 volt to get it out 
from under the max.

events.txt is a listing of the individual events.  There are a lot of 
duplicate info but I didn't bother trying to clean that up.  (I'll say more 
if anybody is curious.)

I remember the big event in late 2009.  When I got back from the grocery 
store, a bunch of my neighbors and a fire truck were out in front of my 
house.  It was a windy day.  A few houses down a tree branch was occasionally 
blowing into the power lines and making minor fireworks.  A few minutes 
later, a bucket truck from PG&E arrived.  He pulled the fuses on the pole in 
front of my house, moved over and pruned the tree, then came back to 
reconnect the fuses.  Power was out for 10 minutes.

Capturing the line as audio is on my list so I can investigate quirks in the 
60 Hz clocking.  That same data should also shed some light in this area.  
(But it means the PC doing the collection has to be on the UPS.)


-- 
These are my opinions, not necessarily my employer's.  I hate spam.