926215 - Geeksphone Keon runs out of power within 24 hours

Status: RESOLVED INCOMPLETE

(Firefox OS Graveyard :: General, defect, P3)

Description

[Martijn Wargers (dead)]

I'm not sure if this is an issue with the Geeksphone Keon (or only this particular phone), the OS, but with a freshly installed FirefoxOS 1.2 OS on this Keon device, no simcard in it and only wifi enabled, the phone has no power left after 24 hours. Something is draining battery power massively.

This is with a build from http://downloads.geeksphone.com/keon/1.2/latest_v1.2.html

Some numbers here:
100% charged at 11:33pm
97% charged at 12:37pm
81% charged at 3:00am
75% charged at 5:50am
22% charged at 4:07pm
17% charged at 7:12 pm
8:57pm shut down as soon as I tried to read the battery status

So within 22 hours, and apart from reading the battery status not using the device, the battery is empty.

Comment 1

[Jon Hylands [:jhylands]]

I can have a look at this most likely first thing next week. I'll be getting the new PCBs for the Keon harness on Friday (reference bug 937833), and assuming everything works I'll be good to run a power test on my Keon on Monday.

Comment 2

[Jon Hylands [:jhylands]]

I can definitely correlate that my Keon, with the screen off and idling, with wifi enabled (and connected to my router) and no sim card, seems to very between consuming 17 mA for 25 seconds, and then about 85 mA for abour 65 seconds. My Keon is running 1.0.1, so clearly this behavior is not something new. I will update my Keon to 1.2 (on Friday) and run the power test again.

Comment 3

[Jon Hylands [:jhylands]]

Attachment: Screenshot of ammeter software

This screenshot shows the transition from ~85mA down to ~17mA for about 25 seconds, and then cycling back up to ~85mA again.

Comment 4

[Martijn Wargers (dead)]

I think this is wfm nowadays, this presumably got fixed by bug 917070.

Comment 5

[Mike Lee [:mlee]]

Jon please confirm if this is really fixed using the FxOS Powertool: http://mzl.la/1erl33E

Comment 6

[Jon Hylands [:jhylands]]

Attachment: Screenshot from 1.5

Ammeter showing screen-off, sleeping power usage is still way too high

Comment 7

[Jon Hylands [:jhylands]]

This is definitely not fixed in current 1.5/master. I'm still seeing the same cycling between ~15 and ~85 mA (see attachment), which is problematic in two ways:

- the 85mA cycles are long, and consume a huge amount of power
- the sleep power usage should be < 1mA (like the Hamachi)

Comment 8

[Dave Huseby [:huseby]]

I'm not sure this is blocked on, but certainly related to Bug 989715, Bug 995301, and Bug 995311.

Comment 9

[Jon Hylands [:jhylands]]

This test was run with wifi turned off, and no sim card. I'm going to run the profiler with the ammeter and see if I can figure out what's happening.

Comment 10

[Jon Hylands [:jhylands]]

So, it appears that this is not CPU-related, which means it isn't a process that is running in the background and thus consuming power. Given that, this must be some piece of hardware on the phone that is consuming this power. I enabled airplane mode, and power usage dropped to ~0.2 mA, so that means this must be RIL related.

Comment 11

[Jon Hylands [:jhylands]]

So interestingly, with an no-longer-active SIM card inserted, and airplane mode turned off, the behavior is similar to no SIM card, except instead of cycling between 85 and 15 mA, it cycles between 85 and 0.2 mA.

Comment 12

[Jon Hylands [:jhylands]]

As a further piece of information, when the phone has been on for ~10 minutes, the cycling pattern is happening with ~30 seconds of low power draw, followed by ~45 seconds of high power draw.

Comment 13

[Nicholas Nethercote [inactive]]

This bug is ancient and unlikely to produce any useful results at this point.