starting the idle battery benchmark again

July 4, 2008

I am starting the idle battery benchmark again: that is I am charging my 8200 up to full charge and will start it IDLEing with GMail this afternoon, and will leave it going until the battery runs out. I think some of the recent changes will have reduced the battery use when idling.

I have also written a little application that reports battery charge so that I can more accurately record the battery use each hour. There is a script running on a server that sends an email to my gmail account every two minutes. This is intended to simulate a heavy load. Vibrate to alert for new messages is on. I am not going to use the device in the mean time, so the screen backlight will not be a significant factor.

Also, I am running this test using my T-Mobile pay as you go SIM, which has the advantage of being able to see how much data is used, so not only will I have a view of battery life, I will have a better understanding of real data use when idling.

Results follow

After three hours, the battery is reporting 89%

After about 2 and a half days the handheld battery is below 10% and so tiggit mail has quietly closed the connection.

In truth, there were a number of interruptions to the test, with sessions needing to be re-established. There is still plenty to do on the session re-establishment algorithm, but at least all the messages were delivered and for the most of the time the idle session was maintained.

The real problem is that I did not have a “control” example of a completely idle BlackBerry and one receiving the same number of messages via BIS/BES push.


IDLE Battery Life

June 15, 2008

Some tiggit mail users have been experimenting with tiggit’s IMAP IDLE functionality. One of their concerns is that the BlackBerry battery life is severely compromised as a result of enabling IDLE.

I have started running a test to benchmark the impact of maintaining an IDLE session on battery life.

The test is set up as follows. I have a BlackBerry Curve 8210 that uses GPRS over Vodafone in the UK, on a regular data package. It is running tiggit mail 0.1.48 and is connecting to gmail over IMAP. Obviously tiggit has useIDLE set and has no value in idleRefresh – so it is using the default.

I also need to simulate email activity and so have set up a cron job on a webserver to send an email to my gmail account five times an hour on an irregular schedule. Actually at these minutes past the hour 10,20,40,45 and 47.

I started the test running at 11:00 this morning with a fully charged battery, and it is now nearly 3pm, and the battery shows about 5% used. (I guess a tool to tell me precisely how much battery has been used would be handy, and should only take a few minutes to write.)

As an aside, there have been a few questions about data use, so this test will probably be repeated with some additional data use tracking enabled in the code.

UPDATE At 8pm: I have written a little tool that tells me the battery use reported by the handheld. After 9 hours of running in IDLE mode, the BlackBerry Curve is reporting 83% battery.

UPDATE At 11pm: The Curve has been running for 12 hours in IDLE mode, and the battery is reporting 77%. I have shutdown the test for now, and turned off the BlackBerry for now.

No UPDATE on Monday, as I have been fixing packaging issues all day. Will restart the tests tomorrow.