PR1.1 and unusual battery drain (not wifi related) - Page 11 - maemo.org

jer006 | # 101 | 2010-01-27, 03:40 | Report

Removing the Conversations widget and firefox has fixed the problem for me, Since I removed it on saturday I've started to get upwards of 36 hours on a single charge now - with heavy browsing for about an hour each day using 3G, SMS messaging and several calls!

pyllyukko | # 102 | 2010-01-27, 07:31 | Report

Originally Posted by econan View Post

Then I used some rrdtools to chart.

And here's a script that uses gnuplot (available in extras-devel):

Code:

#!/bin/sh
# plot.sh -- for n900
# pyllyukko at maimed dot org -- 18.1.2010
[ ! -x /usr/bin/gnuplot ] && {
  echo "error: no gnuplot!" 1>&2
  exit 1
}
[ -n "${1}" ] && INFILE="${1}" || INFILE="/root/battery.log"
[ ! -f "${INFILE}" ] && {
  echo "error: input file \`${INFILE}' does not exist!" 1>&2
  exit 1
}
PIC_DIR="/home/user/MyDocs/.images"
NOW=$(date +%s)
TODATE=$(date +%Y%m%d)
# BEFORE = 24 hours ago
BEFORE=$((${NOW}-60*60*24))
gnuplot -persist <<-EOP
        set terminal png
        set output "${PIC_DIR}/battery-${TODATE}.png"
        set title "n900 battery status"
        set xdata time
        set nokey
        set timefmt "%s"
        # enable this to plot only the last 24 hours
        #set xrange ["${BEFORE}":]
        set yrange [0:]
        set xlabel "time"
        set ylabel "mAh"
        set y2label "load avg"
        set y2tics
        plot "${INFILE}" using 1:2 with lines,\
             "${INFILE}" using 1:6 axes x1y2 with lines
EOP
exit 0

To create graphs like this:

The Following User Says Thank You to pyllyukko For This Useful Post:
econan

radiowc | # 103 | 2010-01-27, 20:58 | Report

Originally Posted by pyllyukko View Post

And here's a script that uses gnuplot (available in extras-devel):

Code:

#!/bin/sh
# plot.sh -- for n900
# pyllyukko at maimed dot org -- 18.1.2010
[ ! -x /usr/bin/gnuplot ] && {
  echo "error: no gnuplot!" 1>&2
  exit 1
}
[ -n "${1}" ] && INFILE="${1}" || INFILE="/root/battery.log"
[ ! -f "${INFILE}" ] && {
  echo "error: input file \`${INFILE}' does not exist!" 1>&2
  exit 1
}
PIC_DIR="/home/user/MyDocs/.images"
NOW=$(date +%s)
TODATE=$(date +%Y%m%d)
# BEFORE = 24 hours ago
BEFORE=$((${NOW}-60*60*24))
gnuplot -persist <<-EOP
        set terminal png
        set output "${PIC_DIR}/battery-${TODATE}.png"
        set title "n900 battery status"
        set xdata time
        set nokey
        set timefmt "%s"
        # enable this to plot only the last 24 hours
        #set xrange ["${BEFORE}":]
        set yrange [0:]
        set xlabel "time"
        set ylabel "mAh"
        set y2label "load avg"
        set y2tics
        plot "${INFILE}" using 1:2 with lines,\
             "${INFILE}" using 1:6 axes x1y2 with lines
EOP
exit 0

To create graphs like this:

Does not make sense at all, even no load, looking at the graph, the battery still drain...so what does it tell you what cause the battery drain????

pyllyukko | # 104 | 2010-01-27, 21:19 | Report

Originally Posted by radiowc View Post

Does not make sense at all, even no load, looking at the graph, the battery still drain...so what does it tell you what cause the battery drain????

Don't know, I used the example log by econan (http://talk.maemo.org/showpost.php?p...0&postcount=88).

But hopefully people can track their battery draining a little bit better with the help of the script.

The Following User Says Thank You to pyllyukko For This Useful Post:
radiowc

econan | # 105 | 2010-01-28, 18:25 | Report

Great script pyllyukko!

Thanks for that.

Whoever has the initial drain issue, I suspect that the charging process does not understand that the battery is fully charged. Try this when your battery is fully charged:

unplug the phone.
shutdown the phone
remove the battery
(optional) push the power button 10 secs (without the battery, this might not even do anything, i might be imagining things )
put the battery back in
plug the phone immediately after boot finishes

Now whenever I did this, the phone continued to charge for another 20-30 mins.

I also recorded the percentage and the charge (mA) from the hal-device before and after the shutdown.

The fully charged battery before the shutdown was around 97% and ~1200 mA
After the reboot it goes to 87-90% battery state. (i don't remember the amps)

Either I'm missing something (maybe boot drains a lot of battery, or some enormous discharge happening somewhere in between) or the measurement of the battery charge is somewhat skewed in the OS.

egoshin | # 106 | 2010-01-28, 19:45 | Report

econan (charging process) - it is a usual behavior even with my battery powered saw. LiON does some "restructuring" process after it reaches a full charge.

marmota | # 107 | 2010-01-30, 12:04 | Report

I am having the battery drain problem after the update. I tried to unistall firefox but without results. I do not have desktop widgets and top, conky or htop do not show strange processes using CPU.
The phone is often very warm.
Thank you for the nice and well documented discussion.

b666m | # 108 | 2010-01-31, 12:06 | Report

Originally Posted by lasala View Post

Does anyone know that with pre PR.1.1 the Xorg logfile /tmp/Xord.0.log was set to -logverbose? It now is and i have been seeing constant cpu usage at 6%.

i tracked this problem for about 2 hours now.
it constantly uses 1 - 6 % in idle, mainly around 4 %.

if i do something the value raises to about 10 %.

any solutions?

b666m | # 109 | 2010-01-31, 14:45 | Report

killing the process reboots the n900.
(i knew this would be happening - but it was worth trying ^^... not)

Catacylsm | # 110 | 2010-01-31, 15:28 | Report

I also found that Xorg was constantly doing something, although i don't quite understand the process itself.

As for battery, i've only ever gotten 1 day total out of the device, but then i've been on the internet 12 hours a day running flash based content while i browse so i expect that.

CPU is strange with the device, but i wouldn't expect xorg to drain it too fast!