Right now the power statistics show that:

Energy when full: 25.5 Wh  
Energy (design):  93.2 Wh

And indeed the battery doesn't seem to be lasting as long as it used too.

My question:
Is this data reliable? Does it really indicate that I should replace the battery, or could it be the charger, laptop, or OS that is stopping the battery from fully charging?

Is any way of validating the battery is indeed to blame? I'd like to be sure before shelling out 90$ for a new battery.

(If it helps, the battery is a 3 year old dell 9 cell rated at 90 Wh).

  • 1
    I always find that what Ubuntu reports is not so reliable at least in my case. Laptop is brand new. Atleast for me, Windows gives much accurate (but not so accurate) figures – pratnala Nov 12 '12 at 11:35
  • @PratyushNalam - So how would you recommend deciding whether or not to get a new battery? – nbubis Nov 12 '12 at 11:38
  • 1
    Anyway since in this case, 25.5 is an extremely low number when full, I think it is better to change it. 3 years is a long time for batteries anyway IMO – pratnala Nov 12 '12 at 12:11

For x86 laptops, the battery information you see in /sys/class/power_supply/*/uevent is as accurate as the data provided from the battery. Typically this smart battery data comes over something like an i2c bus via an embedded controller and firmware allows one to gather the dynamically changing information via the ACPI battery status _BST control and the hard-set static battery information via _BIX or _BIF controls. Since these are the standard ACPI interfaces and are accessible to userspace via the /sys/class/power_supply/*/uevent data I suspect the data you see here is exactly what Windows will be seeing too.

The only thing which will be different is the way the data is presented to you in higher level applications and widgets. Who knows what kinds of data averaging tricks will be done on the data in these applications. In my opinion, if you want to see what the battery is actually reporting, look at the /sys interface.

  • So you're saying that if the /sys/class files show the battery is empty they should be believed? Isn't there a chance that the OS / charger is undercharging it? – nbubis Nov 12 '12 at 12:09
  • The data that is presented from the /sys interface is what the operating system is provided from the firmware. It has no way of knowing if this data is valid or not. The kernel will not lie, it just reports what it is being told. If you get data that is different from reality it could be the ACPI implementation of the _BST, _BIX or _BIF controls that are broken, or the reading of the smart battery data which is faulty. So there are various levels (embedded controllers, firmware, etc) that we have no control over. – Colin Ian King Nov 12 '12 at 12:38

As a person with an experience with battery failures, here is a strategy:

  1. For two consecutive days work with the charger plugged in. On the third day, work with the battery. When the battery strength goes upto 5% set it to charging.
  2. One a month, completely discharge your battery overnight by starting it in the expert command prompt mode.

A good solution to get the maximum out of your battery. Furthermore, a battery has a certain number of cycles (charging and discharging). Near the end of the number of cycles, the battery would go down. The lesser the charging and discharging cycles go, the battery would last the longer. But make sure that you DO discharge the battery completely once in a month.

Regarding the answer to the question: the battery power status is displayed differently in M$WINDOWS and Ubuntu Linux. I have remained ever confused with which one is the correct one.

  • 1
    Good advice maybe, but not an answer to the question. – Nanne Nov 12 '12 at 7:57
  • I left the battery to completely discharge in a Macbook once and it never charged again - battery completely dead :( – NoBugs Jan 8 '16 at 7:05

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.