Under Settings->Display, you can increase the screen timeout. I have the same issue when I'm running tests using the terminal prompt.

adam

Another option is to disable sleep by selecting the the following option.
Settings > Applications > Development > Stay awake

Well I cannot disable the auto-sleep completely. Remember I am only waking the CPU, not the LCD when I send data through its UART pins. I just want to delay the auto-sleep part a bit longer. Either that or allow me to control it in software manually. i.e. in android software, I say I need to stay awake mode and when I finish doing what I need to do, I say auto-sleep mode again.

OK I solved the problem.

I used this:

https://developer.android.com/training/scheduling/wakelock.html

Sorry to bring this topic back up, but it seems like the wakelock in the android code doesn't work if the device is battery powered. I have tested and verified that when the device is USB powered, then it artificially keeps the device on even if I press the power button to put it to sleep. But when I power the device from battery, this does not work.

The quick solution was to disable the sleep in the init.rc, but that will result in high power consumption during sleep when the screen is off.

Is there another solution? Do I write some other values to the /sys/power/wake_lock?

Again I am using Android 2.3.4.

Having the wake locks keep the system on when connected by USB is normal because it detects the USB cable and that prevents it from sleeping. The device is supposed to sleep when running off the battery. Are you saying you would like to disable the sleeping when running for the battery even when the USB cable is disconnected?

adam

One of my colleagues suggested this link:

http://elinux.org/Android...lock_from_user_space

You can write to a file and have it prevent sleeping.

Thanks Adam and your colleagues who recommended the link. I will look into that. Worst case, I might have to write to the powerlock file manually.

OK so I tested this and it doesn't not quite work. I think it has to do with how long the ARM is awake after the UART port wakes it up. Sending 1 character is not sufficient to keep it awake long enough for my service in my app to handle and process the other UART characters. KeyEvents and mmc_ wakelock is set, but they are cleared too quickly before my service can run. If I send quickly in succession, then my service has a chance to run and analyze the UART message.

Note my service writes to the wakelock so this does indeed keep the ARM awake until I finish my process, which is nice to know the wakelock method works... provided you get to that point to write to it before it goes to sleep again.

Obviously we lose the first character that goes to the UART. I will try to send a few UART characters spaced out to determine what the spacing is and the number of characters required to wait it up properly to keep it awake.