Koreader running on Kobos (all of them, hopefully!)

[Ken Maltby]

This problem has some twists and turns that make it hard to pin down what is really happening. One odd result came from some of Tshering's testing; it appears that the suspend can work for a time with normal very low self discharge, and all functions suspended - then at some point, for no established reason, it wakes up or at least partially wakes and excessively drains the battery. His testing was with overnight suspends. I now look at it as suspend is only for relatively short periods, like cleaning the screen or getting something to drink. I poweroff when not reading for any more extended period of time.

Luck;
Ken

[Frenzie]

@Alan_S
Well, you'd have to define "drained". If I leave my device suspended for a week it'll discharge at least a few percent, but that's to be expected. But if you mean e.g. 10% in a night then that means the device didn't actually suspend at all (which is unrelated to whatever message may be drawn on the screen prior to attempting to suspend).

Originally Posted by Ken Maltby

This problem has some twists and turns that make it hard to pin down what is really happening. One odd result came from some of Tshering's testing; it appears that the suspend can work for a time with normal very low self discharge, and all functions suspended - then at some point, for no established reason, it wakes up or at least partially wakes and excessively drains the battery. His testing was with overnight suspends. I now look at it as suspend is only for relatively short periods, like cleaning the screen or getting something to drink. I poweroff when not reading for any more extended period of time.

That doesn't reflect my extensive log-file based testing (including overnight etc.).* Once properly suspended, the only thing that'll wake up the device is a button press or some such. I do however agree that you might just as well turn off the device when you know you're not going to be using it for several days or more, although I generally don't.

* And not log-file based, reading a few thousand pages with overnight suspends etc. over the past month depleted my H2O's battery from a claimed 90% or so to a claimed 47%. I consider that acceptable and not subjectively significantly better or worse than what I got out of Nickel.

[WolfBack]

Originally Posted by Ken Maltby

Have you tried the Advanced settings? I find it easier to edit "defaults.lua" in Notepad++ where there are helpful comments (in green) about the settings.

You can change the values used for "Lighter" "Default" and "Darker" to match your preference.

Luck;
Ken

Originally Posted by NiLuJe

@WolfBack: Depending on the quality of the font you're using, switching CRe to use native hinting instead of freetype's autohinter might help (i.e., set font.hinting.mode to 1 in koreader/data/cr3.ini).

Thank you Ken & NiLuJe for your replies!
I try both solutions with no success...
First I was change the values of the default.lua with something beyond of 25 in DCREREADER_CONFIG_DARKER_FONT_GAMMA = 25 :

-- crereader font gamma
DCREREADER_CONFIG_LIGHTER_FONT_GAMMA = 10
DCREREADER_CONFIG_DEFAULT_FONT_GAMMA = 15
DCREREADER_CONFIG_DARKER_FONT_GAMMA = 25

I try with values of 30, 50, 100, 150, even 200, but the brightness are still the same as before...

Second, I leave the default.lua as it is and then I change the the hinting mode to 1 in koreader/data/cr3.ini and leave the CRe do its things. I use some sideloaded fonts and experiment with these with no effect at all...

Anyway, I can live with that...

Think I will try the manual way and I'll change the hinting on my fonts instead, but I don't know how I can do something like this. I'm very confusing with FontForge and it's autohinting options, or I'll seek the "semibold" approach (if I can do something like this, either...)

[Ken Maltby]

Originally Posted by WolfBack

Thank you Ken & NiLuJe for your replies!
I try both solutions with no success...

Think I will try the manual way and I'll change the hinting on my fonts instead, but I don't know how I can do something like this. I'm very confusing with FontForge and it's autohinting options, or I'll seek the "semibold" approach (if I can do something like this, either...)

I normally use Stone Serif font and I just installed Stone Serif Semi Bold, and you can certainly tell the difference. This might be an easy answer to match your preference.

Luck;
Ken

[Alan_S]

Originally Posted by Ken Maltby

I now look at it as suspend is only for relatively short periods, like cleaning the screen or getting something to drink. I poweroff when not reading for any more extended period of time.

Basically, this is the same as my way of using it now. If I know that I would read in very short time (within an hour), I suspend it. If I know that it would probably wait till next day, I shut it off.

Originally Posted by Frenzie

@Alan_S
Well, you'd have to define "drained". If I leave my device suspended for a week it'll discharge at least a few percent, but that's to be expected. But if you mean e.g. 10% in a night then that means the device didn't actually suspend at all (which is unrelated to whatever message may be drawn on the screen prior to attempting to suspend).

First to explain my reading pattern.

I usually read in one block of time per day. With, if ever, short pauses (till one hour, usually) within that block. After that, reader is suspended (now shut down) till next day, or in some cases it can wait few days. Then I again read in one block.

Reading time of course depends, but let's say it's 2-3 hours.

Battery readings for older version of koreader (before this change in suspend script and before I started shutting it off) didn't changed while I was reading, or very, very rarely.

Between days it changed 3-5%, but it also depends, there were times with 7-8%.

Now, with shutting it down, it usually changes 1-3%, but basically on lower side of change. And sometimes it can be the same. Also, I sometimes notice even a bit strange thing of having more battery, but it may be just my imagination, or lack of memory what battery status really was. I just ignore it, it's not that important.

Also, at least for my H2O, lower part of battery readings (under 50%) drains faster, which isn't anything alarming, battery reading doesn't have to be linear, so I don't worry about this.

One time when I was putting device in standby it completely drained itself from about 20% or somewhere in that part. I thought I would have at least hour of reading time and didn't put device to charge itself, but when I tried to read, it didn't started. So, this was one time when battery drained in less than day. After that incident I started charging device when battery is under 50%, just to be on safe side and can read.

Also, after that I started shutting device off and stopped trusting standby for longer periods of time.

As I said, this all is for older versions of koreader, now I would need to start use suspend if my version (2015-11-3 stable) has suspend tweaked.

[Frenzie]

Originally Posted by Alan_S

One time when I was putting device in standby it completely drained itself from about 20% or somewhere in that part. I thought I would have at least hour of reading time and didn't put device to charge itself, but when I tried to read, it didn't started. So, this was one time when battery drained in less than day. After that incident I started charging device when battery is under 50%, just to be on safe side and can read.

The same happened to me a few times under 30%. I really don't think that indicated a sudden and miraculous failure of the device to suspend properly, but merely that it's bad at indicating the battery value in a way that's useful to the user.

As I said, this all is for older versions of koreader, now I would need to start use suspend if my version (2015-11-3 stable) has suspend tweaked.

It looks like NiJuLe added another 2 seconds. Who knows, it might help.

[Alan_S]

Maybe it is bad battery reading from devices firmware, I surely noticed that lower battery percentage drains faster than higher, so it really maybe isn't koreader's problem. That's why I charge it sooner.

And I see that stable version has this two second change in suspend script. Will see if it helps.

Also, it would be good if someone can confirm that CoolReader suspend script works better for them, just in case.

[WolfBack]

Originally Posted by Ken Maltby

I normally use Stone Serif font and I just installed Stone Serif Semi Bold, and you can certainly tell the difference. This might be an easy answer to match your preference.

Luck;
Ken

The Stone Serif (Semi Bold) font is awesome for my taste!
The only problem is that font family cannot display my native language letters.
BTW, is a excellent choice for reading English novels that I have sideloaded on my reader.
Thanks, Ken!

[Ken Maltby]

Here is a listing of Greek Fonts that might be worth looking into. There might be a semibold there.

You also might want to check on what "fall back fonts" are in effect.

Luck;
Ken

[WolfBack]

Originally Posted by Ken Maltby

Here is a listing of Greek Fonts that might be worth looking into. There might be a semibold there.

You also might want to check on what "fall back fonts" are in effect.

Luck;
Ken

Thank, again, Ken!
The link is bookmarked already! I saw a few nice fonts to experiment with.

One more thought. Is there any change to add someone any time schedule suspend (or, even better, shutdown) in future versions of Koreader? I was very tired last night and I fall asleep reading only one page of my book and the battery level was 39%... Wake up after 6 hours, the same page is waiting for me to read -at least- a few more words and the battery level was 24%... (the frontlight was off).
The same incident occurs when I'm reading paper books and I'm very tired! Thankfully, I wont seek the reader on the floor, under the bed the next morning... -yet....!

[Ken Maltby]

Cronx has suggested that, since the Kobos power management is not offered up as an OS function (it is actually handled by separate calls to an I2C bus, as I understand it.) A power management function would need to be added to the KOReader startup before KOReader actually starts. Not sure if anyone is working on that now.

Luck;
Ken

[Alan_S]

As I found out that latest koreader has a bit different suspend script, I started testing suspending device.

And first day of testing had a bit of "luck" with it. After suspending reader with 100%, next day it was on 82%. This means that latest suspend script (with added two seconds) doesn't do the trick.

Conditions were:

-I read pdf file
-I used light for reading
-I have magnetic cover that was closed after I suspended device with power button (and I would love to have option to turn off magnetic cover having any effect on koreader)
-I don't use book cover as suspended screen.

As I had light turned on, I would saw if device exited suspend normally even under cover. It seems that something irregular happens that makes battery drain, as device wasn't waked up in a way that would trigger light.

Now I'm testing suspending with CoolReader suspend script and see what happens.

EDIT: I didn't closed magnetic cover at all after suspending device next time, it didn't affect battery percentage. It doesn't mean that there's correlation for sure, but it might be some problems with using magnetic cover...

[Ken Maltby]

Originally Posted by Alan_S

As I found out that latest koreader has a bit different suspend script, I started testing suspending device.

And first day of testing had a bit of "luck" with it. After suspending reader with 100%, next day it was on 82%. This means that latest suspend script (with added two seconds) doesn't do the trick.

Conditions were:

-I read pdf file
-I used light for reading
-I have magnetic cover that was closed after I suspended device with power button (and I would love to have option to turn off magnetic cover having any effect on koreader)
-I don't use book cover as suspended screen.

As I had light turned on, I would saw if device exited suspend normally even under cover. It seems that something irregular happens that makes battery drain, as device wasn't waked up in a way that would trigger light.

Now I'm testing suspending with CoolReader suspend script and see what happens.

EDIT: I didn't closed magnetic cover at all after suspending device next time, it didn't affect battery percentage. It doesn't mean that there's correlation for sure, but it might be some problems with using magnetic cover...

I don't use a magnetic cover, so I haven't keep up with its effects, but I remember when there was an issue where if you used the power button also, the magnet switched it again.

Luck;
Ken

[church mouse]

In my experience Ken is correct that however you suspend in Koreader, because it has no override of the magnetic cover function that function is always active if one is using a magnetic cover.
My testing to date has shown me 3 things on my 2 readers:
1. Koreader has a significantly higher battery drain in suspend mode than nickel or Coolreader.
2. Using the Coolreader suspend script made no improvement on my readers.
3. Having a magnetic cover seems to lead to even higher drain + the reader seems prone to waking up with movement when the cover is closed.

I have wondered whether the fact that the magnet recognition is always 'On' in Koreader might be a factor in its higher battery drain (whether using a sleepcover or not), but that is pure conjecture.

[Ken Maltby]

Someone might try the fix in these posts https://www.mobileread.com/forums/sho...postcount=2479
https://www.mobileread.com/forums/sho...postcount=2480
https://www.mobileread.com/forums/sho...postcount=2481

(If you click on the "Thread:" shown in the top right on the single post display, you will go to the thread and be at that post's position.)

If it works, someone might be motivated to make the PR at Github.

I don't expect that this would fix the whole power draining issue, but it might make things more predictable.

Luck;
Ken