The first look at battery life considered cpufreqd and ways to control the CPU frequency. Two and a half years later, that battery is now almost completely dead (the battery indicator in the task tray says "no battery"), so I looked around to buy a new one.
On the Dell website the mail order price is over 250 CHF, but I found another supplier called akku-handel.ch which offered a compatible replacement part for only 140 CHF. This arrived by mail after just a few days (payment in advance by transfer), and it turned out to work well. After giving it a couple of cycles of warming-up time, I was interested to see what the life of this battery is compared to the original battery.
These tests were done with Mandriva 2007.1, with no applications running apart from gkrellm. The screen brightness was on minimum and the network unplugged.
|Graphs of reported battery percentage against time for the two batteries|
So the behaviour of the new battery is still linear, and doesn't last quite as long as the old battery did, but not much less. The maximum life we can expect from the new one is about 3 hours 10 minutes, compared to 3 hours 30 minutes for the old one.
Actually this is not too far off what one would expect because the old battery has "4700 mAh" written on it, whereas the new one says "4400 mAh". So all other factors being equal the new one should last 6% less, or around 3 hours 20 minutes.
The next question, although the percentage drops quite linearly, can we trust the "remaining time" indicator? Using the same data, we can compare the predicted remaining time with the actual.
|Graph of reported remaining time and actual remaning time for the new battery|
Overall the estimates are not bad but tend to overestimate how much time is left (probably because of that final dive from 10% to nearly empty). The estimates seem to jump around a bit at the start, apparently more than the old battery did, but maybe that's just a quirk of the KLaptop meter.
As a rough rule, subtracting a quarter-hour from the estimate gives a pretty good guide to when the battery will run out. Obviously that will be less accurate when the load is variable.
Lastly, we'd already seen the effect of charging with the laptop on or off in the first investigation, so I didn't repeat that with the new battery - I just wanted to know how long it would take to charge it and whether it was linear or not. Here is a comparison of the reported percentage charge as a function of time, immediately after the graphs above were produced and again without any load on the system.
|Graphs of reported battery charge against time, during charging for the two batteries|
The new battery shows a much more extreme 'kink' in the charging, charging rapidly up to about two-thirds capacity and then flattening markedly towards 90%. Then finally it jumps instantly from 90% to "fully charged".
The most annoying thing about this is the time it takes to charge - instead of an hour and half, this new battery takes two and a half hours to charge up - not so good if you're in a hurry.
The duty cycle then obviously depends strongly on the amount charged - it seems like this battery can be charged from flat to half-full in around 40 minutes, which should give around 1 hour 20 minutes life (1 minute charging for 2 minutes offline) - this is around what the old battery offered. But charging it fully takes 2 and a half hours for only 3 hours 10 minutes life (1 minute charging for 1 minute 15 seconds offline) - much less convenient.
Given that the laptop is still perfectly usable from the mains with the old (and dead) battery in, the new one will be kept on standby and not over-exerted. It remains to be seen how long the new one lives if it's only brought out for special occasions. I don't think I fancy keeping it in the fridge though...