Glo Bad battery?

[TechniSol]

I think you have to perform the occasional deep discharge for the battery metering to properly determine the remaining power capacity of the battery. The normal method is to account for power usage levels and uptime and build up a total current or watts available profile and then subtract during discharge. You can get fancier, but it's probably not worth the work.

The occasional deep discharge to shutdown should only help to recalibrate the profile. The shutdown should occur prior to any internal low voltage shutdown circuitry in the battery kicking in as long as Kobo left the proper wiggle room.

As to how the chapter size might effect current draw during sleep mode, what if the RAM being used is segmented and can be independently powered down if not being used? Buffer a small amount of text in memory and you have minimal current draw during sleep mode. STUPIDLY buffer multiple chapters and keep a lot of dynamic RAM refreshing and kiss your battery life goodbye. It's just a theory from a hardware dinosaur, but it might be relevant...

OTOH, if you're trying to get maximum battery life out of a device like these, and the processor is fast enough, why even bother to buffer text in between page turns? Just preserve a few pointers and grab the text and render on the fly. It would seem like the lowest power consumption based method?

[jusmee]

Originally Posted by TechniSol

I think you have to perform the occasional deep discharge for the battery metering to properly determine the remaining power capacity of the battery. The normal method is to account for power usage levels and uptime and build up a total current or watts available profile and then subtract during discharge. You can get fancier, but it's probably not worth the work.

The occasional deep discharge to shutdown should only help to recalibrate the profile. The shutdown should occur prior to any internal low voltage shutdown circuitry in the battery kicking in as long as Kobo left the proper wiggle room.

As to how the chapter size might effect current draw during sleep mode, what if the RAM being used is segmented and can be independently powered down if not being used? Buffer a small amount of text in memory and you have minimal current draw during sleep mode. STUPIDLY buffer multiple chapters and keep a lot of dynamic RAM refreshing and kiss your battery life goodbye. It's just a theory from a hardware dinosaur, but it might be relevant...

OTOH, if you're trying to get maximum battery life out of a device like these, and the processor is fast enough, why even bother to buffer text in between page turns? Just preserve a few pointers and grab the text and render on the fly. It would seem like the lowest power consumption based method?

It seems to me that, by paging forward as fast as you can, page turns slow down after some number of pages - perhaps indicating that they buffer a few pages ahead, but that might be just rendering buffer, the whole or any part of the file itself may or may not be held in main memory.

One small nitpick, you mention watts available when speaking about battery life. Watts are the rate of energy usage, and what you actually want to know is watt-hours, the energy left in the battery, I think.

[Mrs_Often]

Originally Posted by jusmee

Maybe it's just the battery meter. It is said that the calibration of the battery meter can take a few cycles to "learn" how the battery behaves and become accurate. If you've only had a few charge discharge cycles, maybe it is still learning?

Thanks for that. I really hope you're right. It has had a few cycles already, but then again, those cycles were messed up with the bad kepub.

Originally Posted by GeoffR

I wonder if it is possible to tell from the time it takes to charge whether there is a battery problem? I.e. an abnormally long or short charge time might indicate a dud battery, normal charge time but short battery life might mean battery OK but a problem with the device using too much power. I don't know what the normal charge time is though. Mine takes about 2.5 hours from fully drained to fully charged.

Originally Posted by voidoid

Same here for my Touch with a good battery.

Thanks GeoffR, I was wondering about charge time. I thought mine was really quick last few times which would indeed point to a bad battery, BUT, thanks to your and Voidoid's time indication maybe it wasn't that quick. I'll have to time it next time, now that I know that 2.5 hours is normal.

[davidfor]

Originally Posted by Mrs_Often

Thanks for that. I really hope you're right. It has had a few cycles already, but then again, those cycles were messed up with the bad kepub.

That won't matter. Calibration for Li-Ion batteries is simply two or three complete discharge/recharge cycles. All it is doing is calibrating the the fully charged and discharged voltages. Unless Kobo has out much more intelligent battery management software in the device than most laptops have, the charge level is simply a linear scale over the difference.

It doesn't matter how the discharge happens. The bad epub will just have drained the battery quicker which is handy when doing the calibration.

[TechniSol]

Jusmee,

You're correct, it was a nitpick... Bwah, ha, ha...

Basically, since regulated voltage would remain the same, you're really accounting for how many milliamps or Amps were consumed while the processor had uptime which could be expressed as Amp-hours, but more likely as millisecond or microsecond counts of varying power rates being consumed. If you multiplied that by voltage, you'd get watt/hrs or an equivalence for power, if you'd like.

Since regulated voltage should remain constant, it's probably sufficient to just account for uptime for each power state in a variable and multiply each of those by the respective amperage draw for that operation mode on a per uptime basis and subtract that from the total estimated capacity when first starting the device. Later as the battery was cycled you'd re-evaluate that total estimated capacity and replace it with measured capacity based on the method espoused above.

If all they're doing is putting an A/D converter on the raw battery voltage and estimating remaining charge from raw voltage under varying load, they'd better have a battery curve available to increase linearity if they want any kind of reasonable accuracy at all.

The methods where you estimate load based on what's active or actually measure current draw are considerably more accurate. OTOH, it's an ereader, not a life safety device...

[jusmee]

No, not watts/hr, it's watt-hours, same as how you buy electricity - only then it's kilowatt-hours. Constant voltage allows you to quote amp-hours (or ma-hrs for most batteries).

[voidoid]

Originally Posted by TechniSol

Jusmee,

Since regulated voltage should remain constant, it's probably sufficient to just account for uptime for each power state in a variable and multiply each of those by the respective amperage draw for that operation mode on a per uptime basis and subtract that from the total estimated capacity when first starting the device. Later as the battery was cycled you'd re-evaluate that total estimated capacity and replace it with measured capacity based on the method espoused above.

No wonder these devices are running short of power -- if they have to do the above then 90% of their power usage is calculating battery capacity left. [Just joking -- but still...].

[Mrs_Often]

Originally Posted by TechniSol

OTOH, it's an ereader, not a life safety device...

Yes it is!!

Originally Posted by davidfor

That won't matter. Calibration for Li-Ion batteries is simply two or three complete discharge/recharge cycles. All it is doing is calibrating the the fully charged and discharged voltages. Unless Kobo has out much more intelligent battery management software in the device than most laptops have, the charge level is simply a linear scale over the difference.

It doesn't matter how the discharge happens. The bad epub will just have drained the battery quicker which is handy when doing the calibration.

Oh. Gah. If it is calibrated, it's being strange deciding to hang around 47% for a few days. Doesn't seem like a linear drain at all (with hardly any irregular use, I've left it in sleep mode as much as possible). I suppose I really will not know until it loses its charge again.

[stethorn]

When you are just reading, does the battery % update ? It never seems to move when I am reading. I am hoping its because the battery is far superior to the Kindle, as that's what I want the most. But I am also concerned that I will be happily reading away with 38% battery and then it goes off on me at a bad time..

[VelvetElvis]

Originally Posted by GeoffR

Might not help your GF, but my problem turned out to be the external SD card, I removed it and battery life improved dramatically. Still investigating whether it is the card itself or the Glo at fault.

Edit: I wonder if it is possible to tell from the time it takes to charge whether there is a battery problem? I.e. an abnormally long or short charge time might indicate a dud battery, normal charge time but short battery life might mean battery OK but a problem with the device using too much power. I don't know what the normal charge time is though. Mine takes about 2.5 hours from fully drained to fully charged.

Just recharged it from depleted, and 2.5 hours was about it for me too. Going to see how it goes this time. Thanks for your input.

Update. Just got home from work, 6 hours after device showed full charge and was put immediately to sleep. Pic shows where now. Just sent an email to support.

[ib4l]

I recently found out I can drain my battery in a few hours if I have WiFi on, so I only switch it on when I'm synchronising/downloading books when using the computer is not an option. Saying that, the battery usually jumps level usually jumps up a few tens of a percent when I switch the WiFi off, so the battery indicator probably isn't accurate under high usage.

Aside from that, the battery seems near infinite, and I'm really picky when it comes to advertised specs not being fulfilled. At the moment I'm reading The Idiot, a long piece of Russian literature that, if it were a hardcopy, could be used to bludgeon someone to death. I'm around half way through the book and approximately half of my reading time has been with the light on (lowest setting). The reader was charged before I started and now the battery is at 94%, so it's holding up well. With my font settings, that equates to 600 page turns, or about 10,000 turns if the battery indicator is accurate. That's stunning, considering the light has been on for a fair amount of those turns.

It definitely took a couple of charge cycles for it to start performing like that - the first week I had to charge it way too often, but I was only putting the reader to sleep, so perhaps that was the issue. Now I turn it off completely, turning it on only once a day for a couple of hours of reading.

[Mrs_Often]

Thanks for the detailed report ib4l!

It's beginning to sound like the Glo uses a lot more power in sleep mode than the Touch does, also considering VelvetElvis' post above. Hmm. Or maybe it's the firmware version?

[Mrs_Often]

Okay, so, the cycle I was testing is now over. During reading, the battery didn't seem to drop at all. The battery was up around 95% for a while, then dropped to 54% and then to around 47% where it stayed for quite a while... and then dropped to 0% over night.

Averaged the battery dropped about 0.87% each hour which included mostly sleep mode and three reading sessions of about 45 minutes, and one batch of loading & reloading a book with processing and restarts.

Overall that means the battery lasts 116 hours, or 5 days. And that's being generous because I don't actually know how far into the night the battery lost its charge.

That's just crap.

Details in the spoiler for who's interested.

Spoiler:

Date & time      	Batt%	Action after	Sleep time	%d	%d ph	Avg%dph	Total%dph	up time	
5 Jan 2013 10:22	100%	on, sleep							
5 Jan 2013 12:51	98%	on, sleep	2h 29m      	2%	0.81	1.53	1.65	60	hours
6 Jan 2013 01:50	85%	on, sleep	12h 59m      	13%	1.00			3	days
6 Jan 2013 11:10	59%	on, sleep	9h 20m      	26%	2.79				
	Got rid of bad kepub							
									
6 Jan 2013 13:42	100%	on, sleep							
6 Jan 2013 14:42	100%	on, sleep	1h 0m        	0%	0.0	0.89	0.86	116	hours
6 Jan 2013 16:46	100%	chnge k*	2h 4m        	0%	0.0			5	days
6 Jan 2013 18:11	95%	on, sleep	1h 25m      	5%	3.53				
6 Jan 2013 23:21	95%	on,read**	5h 10m      	0%	0.0				
7 Jan 2013 00:00	95%	sleep         	0h 39m      	0%	0.0				
7 Jan 2013 09:30	95%	on, sleep	9h 30m      	0%	0.0				
8 Jan 2013 09:30	55%	on, sleep	24h 0m      	40%	1.67				
8 Jan 2013 14:30	54%	on, sleep	5h 0m        	1%	0.2				
8 Jan 2013 22:45	49%	on,read**	8h 15m      	5%	0.61				
8 Jan 2013 23:19	47%	l.off,sleep†	0h 34m      	2%	3.53				
9 Jan 2013 14:47	47%	on, sleep	15h 28m      	0%	0.0				
10 Jan 2013 09:18	46%	on, sleep	18h 31m      	1%	0.05				
10 Jan 2013 16:44	46%	on, sleep	7h 26m      	0%	0.0				
10 Jan 2013 23:03	45%	on,read**	6h 19m      	1%	0.16				
10 Jan 2013 23:55	45%	l.off,sleep†	0h 52m      	0%	0.0				
11 Jan 2013 10:04	0%	nothing       	10h 9m      	45%	4.43				

*change kepub, restarting and processing etc
**on, reading+light
†light off, sleep

[Mrs_Often]

So, testing the charing...

It went from 0% to 50% in an hour (or less, I didn't check until an hour after starting the charge).

Then it stayed at 50% for two (!!) hours... I suspected the battery meter wasn't updating. So I unplugged the cable, reconnected it, and the battery percent changed to 96%... and instead of showing the "plug" icon in the battery, it's now showing the "check mark"... so it stopped charging at 96% (I now vaguely recall it won't charge unless it's under 95%).

So what's up with that? Three hours in the charger, didn't update the meter and it didn't get further than 96%?? (Using the 2.5 hours charge as 'normal'.)

I unplugged the Glo and synced it to try and get the battery below 95% so it might charge... once the sync had completed, the battery read 98%!

I would have guessed a bad battery would charge extremely quickly... is my guess wrong? Could a bad battery also take forever and/or not get to its maximum charge? (Hmm that also sounds reasonable...)

[jusmee]

Originally Posted by Mrs_Often

So, testing the charing...

It went from 0% to 50% in an hour (or less, I didn't check until an hour after starting the charge).

Then it stayed at 50% for two (!!) hours... I suspected the battery meter wasn't updating. So I unplugged the cable, reconnected it, and the battery percent changed to 96%... and instead of showing the "plug" icon in the battery, it's now showing the "check mark"... so it stopped charging at 96% (I now vaguely recall it won't charge unless it's under 95%).

So what's up with that? Three hours in the charger, didn't update the meter and it didn't get further than 96%?? (Using the 2.5 hours charge as 'normal'.)

I unplugged the Glo and synced it to try and get the battery below 95% so it might charge... once the sync had completed, the battery read 98%!

I would have guessed a bad battery would charge extremely quickly... is my guess wrong? Could a bad battery also take forever and/or not get to its maximum charge? (Hmm that also sounds reasonable...)

No reason a bad battery might charge quickly. In fact, if it is prone to self-discharge, the 'leakage' would make it take even longer to charge, if anything.