I got my Kobo Touch two months ago, and, admitedly, it was a second-hand one. The battery was manufactured in september 2011, so it must have had about a year of use before I got my hands on it.
But the battery behaved strange... or so it seemed to me. The charge would at times show 106% or 104% after I disconnected the charge cable. The first drop would come quite quickly: the battery would drop to 90% quite soon. Then it would hold charge in a way that I perceived to be "normal". But upon reaching ~40% it would just turn off and demand to be charged!
So I did a discharge test on a fully charged.
First of all, I added telnet capability to my Kobo Touch.
Then I launched a script with a gzip process that would keep the CPU occupied about 96%. The insides of the script:
Code:
sleep 600;
#PIRMAS.BIN is just a large file
gzip -c /mnt/onboard/PIRMAS.BIN > /dev/null
Then I launched a script that would get a file form my router in regular intervals. This script also would record wifi and telnet/ftp processes state (so I were sure these processes did not die) The insides of the script:
Code:
while [ 1 = 1 ]
do
echo "`date`:`ifconfig|grep inet`" >> /mnt/onboard/wlan_statusas.txt ;
cd /mnt/sd/ ;
wget http://192.168.1.1/ -O index.`date '+%m%d%y'`.html
wget http://192.168.1.1/ -O index.`date '+%m%d%y'`.html
wget http://192.168.1.1/ -O index.`date '+%m%d%y'`.html
wget http://192.168.1.1/ -O index.`date '+%m%d%y'`.html
echo "`date`:" >> /mnt/onboard/telnet_statusas.txt ;
ps -A|grep telnetd|head -1 >> /mnt/onboard/telnet_statusas.txt ;
ps -A|grep ftpd|head -1 >> /mnt/onboard/telnet_statusas.txt ;
sleep 300 ;
done
Then I killed the nickel (it is nickel that shuts the wifi off in some time, and wifi is the biggest drainer of battery, so I needed it).
Finally, I launched a script that would record the battery state every five minutes.
Code:
while [ 1 = 1 ]
do
echo -n "`date` , charge:" >> /mnt/onboard/baterijos_statusas.txt ;
cat /sys/devices/platform/pmic_battery.1/power_supply/mc13892_bat/capacity >> /mnt/onboard/baterijos_statusas.txt
echo -n "current: " >> /mnt/onboard/baterijos_statusas.txt ;
cat /sys/devices/platform/pmic_battery.1/power_supply/mc13892_bat/uevent |grep CURRENT| awk -F"=" '{print $2}' >> /mnt/onboard/baterijos_statusas.txt
echo -n "voltage: " >> /mnt/onboard/baterijos_statusas.txt ;
cat /sys/devices/platform/pmic_battery.1/power_supply/mc13892_bat/uevent |grep VOLTAGE| awk -F"=" '{print $2}' >> /mnt/onboard/baterijos_statusas.txt
sleep 300 ;
done
These scripts are messy and not optimal, but they did the job for me.
I repeated the test with another battery in the same kobo touch (manufactured the same year, same month).
Well, first of all, I did drain the battery in both cases in a little bit more than four hours
Here is what the battery discharge looked like (both batteries on one graph - so two colours):
The interesting thing is, that both batteries drop charge dramatically at the end of the period. It is not like I managed to see before (it does not just turn off when the charge is 40%; or maybe it would, if nickel were running and controlling the process - I do not know which is right), but still, it is about 30% of charge drop in the last 10 minutes of operation. Do your Kobo's behave this way as well?
Here is what the discharge current was like; since there were scripts running that would take more energy at sertain moments, I do not attach a lot of significance to this image:
Initial lower level of current is due to the fact that my gzip script slept for the first 10 minutes (otherwise it jams the CPU and it is hard to do everything else via telnet).
However, I calculated the average discharge current and this allows me to guess that both batteries have about 84 to 86% of original capacity (the factory capacity is 1000 mAh and voltage - 3,7 V)
And finally, here is what the voltage curve was like:
Nothing that I can comment on here.
So, what do you think, does what I describe resonate with what you are experiencing? Or maybe my batteries are too depleated already?