MobileRead Forums - View Single Post - Deep sleep in glowlight 4/4e and modern nooks

Tialex · 09-13-2023, 04:46 AM

Hi,

jumping from NST to "glowlight 4" (4e actually) I was disappointed at how little a single charge were allowing to read. Direct tests revealed something like 8 hours of reading from 90% to 10%. This was mostly with third-party readers. Then, I installed BetterBatteryStats (BBS) and made some observations, hoping that as I expected, the system would fall into Deep Sleep from time to time outside of the locked state.

It was to no avail except when I noticed that the BN reader app did fall into DS between page turns. Since then I never noticed any other app to activate DS. Actually the deep sleep behavior of BN app was also non-constant. The primary observation is when I made different tests when the delay between turns were different and I calculated the time spent in the full wake state, there was a constant wake time about 6-10 seconds that could never be decreased. But for 30 seconds page turns it was already ok for saving the battery life.

I have plans to investigate whether an app can force the system to fall into DS without going into the lock mode or changing the screen contents, but I would like to be sure the things I noticed are the same for all devices in the wild and it's not specific to my particular instance. If somebody from the community may find time and take some tests, that would be very useful. I will explain how I did mine, but I'm sure there are other apps and other means to achieve the same.

So, the questions are:

Can someone detect deep sleeping outside the locked state
Is the BN app the only one that effectively utilizes it?
Is it the same for all nooks after NST. I'm excluding it because NST probably had another system for DS trickery, because I could use any third-party apps for weeks. So the third question probably is what models are affected

So my testing steps :

My device is not rooted (yet) so the following tests are possible with any device with adb access.
To avoid background processes involving in the tests, I changed three settings is the Developer Options. "Don't keep activities" was switched on. This effectively stops any activity that the user leaves. "Background process limit" was set to one. I'm not sure whether this is important or not. "Background check" lists several services from the installed applications. I disabled all except the Android keyboard.
Installed BetterBatteryStats (BBS). The community version is available at XDA The app interprets the power statistics that virtually any android device are collecting. In order to work one should have either rood access or to give it three permissions listed in the forum thread (using adb). There are proabably other apps effectively doing the same (Wakelock detector comes to mind), but I'm not familiar with them
The app is not so scary after all. Basically it shows different states of the devices and their corresponding time and percentage. When we know the usage style like how much was spent with the screen on, with heavy apps performing something or in the locked state, the numbers should correspond or point out to the reason why they are different to what is expected.
The program is flexible in what the starting point for observation is. We're mostly interesting in using Custom feature. When you go to the menu and tap "Set Custom ref." then the app for the Custom kind (in the drop in listbox containing also Unplugged, Boot) would show the numbers for the period from the moment you clicked to the moment you returned to the app. For instance, for the scenario "investigate a reader app for half an hour" we should set the Custom reference, return to the reader app, after half and hour return to the BBS and look at the "Custom" set of value in order to see whether something important took place (and time) or not.
What we are interested mostly in BBS whether the Summary contains the item Deep Sleep and what percentage it shows. If you lock the device and unlock it after some time, you will see this item because the DS in the locked state is a norm. This is when almost all on-board devices are turning into a low-energy mode (and CPU also) until some approved interruption (like buttons click or screen tap) wake it up. But we would like to find it when the nook is not in the locked state. This is important point about eInk devices. Since they don't need energy to keep the screen image, any eInk device might go into DS while still "pretending" like it continues to work, this is what takes place between the page turns when you use BN reader app (the general smartphone with LCD screen should switch the screen completely in order to preserve energy with such efficiency). The funny things is that early eInk readers sometimes listed not the time for a single charge, but the number of page turns due to this specifics.
The other interesting item is Waklocks-PowerManagerService.Display. Its percentage should be around (100% minus DS percentage) and the counter should probably correspond to the number of times page were turned.

Any results are appreciated. Thanks.

09-13-2023, 04:46 AM	#1
Tialex Member Posts: 13 Karma: 11102 Join Date: Aug 2023 Device: Nook Glowlight 4e	Deep sleep in glowlight 4/4e and modern nooks Hi, jumping from NST to "glowlight 4" (4e actually) I was disappointed at how little a single charge were allowing to read. Direct tests revealed something like 8 hours of reading from 90% to 10%. This was mostly with third-party readers. Then, I installed BetterBatteryStats (BBS) and made some observations, hoping that as I expected, the system would fall into Deep Sleep from time to time outside of the locked state. It was to no avail except when I noticed that the BN reader app did fall into DS between page turns. Since then I never noticed any other app to activate DS. Actually the deep sleep behavior of BN app was also non-constant. The primary observation is when I made different tests when the delay between turns were different and I calculated the time spent in the full wake state, there was a constant wake time about 6-10 seconds that could never be decreased. But for 30 seconds page turns it was already ok for saving the battery life. I have plans to investigate whether an app can force the system to fall into DS without going into the lock mode or changing the screen contents, but I would like to be sure the things I noticed are the same for all devices in the wild and it's not specific to my particular instance. If somebody from the community may find time and take some tests, that would be very useful. I will explain how I did mine, but I'm sure there are other apps and other means to achieve the same. So, the questions are: Can someone detect deep sleeping outside the locked state Is the BN app the only one that effectively utilizes it? Is it the same for all nooks after NST. I'm excluding it because NST probably had another system for DS trickery, because I could use any third-party apps for weeks. So the third question probably is what models are affected So my testing steps : My device is not rooted (yet) so the following tests are possible with any device with adb access. To avoid background processes involving in the tests, I changed three settings is the Developer Options. "Don't keep activities" was switched on. This effectively stops any activity that the user leaves. "Background process limit" was set to one. I'm not sure whether this is important or not. "Background check" lists several services from the installed applications. I disabled all except the Android keyboard. Installed BetterBatteryStats (BBS). The community version is available at XDA The app interprets the power statistics that virtually any android device are collecting. In order to work one should have either rood access or to give it three permissions listed in the forum thread (using adb). There are proabably other apps effectively doing the same (Wakelock detector comes to mind), but I'm not familiar with them The app is not so scary after all. Basically it shows different states of the devices and their corresponding time and percentage. When we know the usage style like how much was spent with the screen on, with heavy apps performing something or in the locked state, the numbers should correspond or point out to the reason why they are different to what is expected. The program is flexible in what the starting point for observation is. We're mostly interesting in using Custom feature. When you go to the menu and tap "Set Custom ref." then the app for the Custom kind (in the drop in listbox containing also Unplugged, Boot) would show the numbers for the period from the moment you clicked to the moment you returned to the app. For instance, for the scenario "investigate a reader app for half an hour" we should set the Custom reference, return to the reader app, after half and hour return to the BBS and look at the "Custom" set of value in order to see whether something important took place (and time) or not. What we are interested mostly in BBS whether the Summary contains the item Deep Sleep and what percentage it shows. If you lock the device and unlock it after some time, you will see this item because the DS in the locked state is a norm. This is when almost all on-board devices are turning into a low-energy mode (and CPU also) until some approved interruption (like buttons click or screen tap) wake it up. But we would like to find it when the nook is not in the locked state. This is important point about eInk devices. Since they don't need energy to keep the screen image, any eInk device might go into DS while still "pretending" like it continues to work, this is what takes place between the page turns when you use BN reader app (the general smartphone with LCD screen should switch the screen completely in order to preserve energy with such efficiency). The funny things is that early eInk readers sometimes listed not the time for a single charge, but the number of page turns due to this specifics. The other interesting item is Waklocks-PowerManagerService.Display. Its percentage should be around (100% minus DS percentage) and the counter should probably correspond to the number of times page were turned. Any results are appreciated. Thanks.