Power problems with pandora we should avoid


ZetaNeta

Member
Joined
Feb 27, 2014
Messages
173
I like low-power mode, but...

Its unstable. Not only wifi goes down, but most processes which been left running sometimes may go crazy.
Its just totally useless when you use you pandora as music player. I never tried, but 14Mhz seem too low for anything to play. On other ARMs i got, music stops being beautiful at ~150Mhz, and finally dies at ~90Mhz.
You cant just go eject the battery, as that may corrupt data. And as i understand, it does not unmount the SD cards, meaning they cant be removed either.

I dont say we should remove it, but i am saying that nothing should force the user to stay in this mode.
I ask to remove the "i cant imagine why anyone needs it" HOLD button, and make a secondary power button, so the user can choose if he wants "LP - Halt" "Halt - Reboot" or "LP- Reboot".
As well, as making the switch more protected. Its very common for my pocket to wake up the OP, and start draining the battery faster.

And i dislike that charging requires main CPU to be up. There should be some separate charging management for this.
Some people are very used to shutdown the device if they want it to charge faster.

Also, a full discharge SHOULD NOT be a trouble for the user. I didnt check, but does OP hold the time when it got no power? (For ex. battery removed)
Full discharge, is a thing we will get here and there. Sometimes we forget to charge things to 100%, sometimes we use it too actively...

And when the time of "Warning: Battery level is less than 5%" comes, alot of people goes "No, you are not taking away my music... not now... please". (I used to swear a lot when my Bada phone didnt let me use the media player below 10%)

More than that, when the system gets down because of the dead battery, some dudes, go shake the device, and try turning it on, to enjoy 10 more seconds of "i almost opened the media player!" before the screen goes black, getting the battery totally drained. Its more common than you think.

Another trouble, which could probably be fixed with the new case design, is that the battery is too small for the slot its been made for.
It may disconnect, because the cover doesnt hold it, and that maybe fixed with some spare paper.
But not only this, the slot is wider and taller than the battery is.
Any shock. Including the classical "i just thrown it on the bed".
Its not funny when it transfers few gigs between the cards, and suddenly stops because "i am holding it wrong", like it did yesterday.
We should make sure that the new case, fits the battery better than OP.


Anyone got any opinion?
 
That's a lot of grievances, almost all of which have been addressed in the past but I'll summarize as best I can.

i am saying that nothing should force the user to stay in this mode.
What is currently forcing you to stay in this mode? How are you using the low power mode? Make sure that you've updated to the most recent version of the firmware because the low power mode has been replaced with a much more efficient sleep mode in most cases, only falling back on the old mode when it can't sleep.
Its very common for my pocket to wake up the OP, and start draining the battery faster.
Flip it to HOLD if that's causing problems ;) More than likely though, I suspect you haven't updated the firmware and it isn't actually going to the proper sleep mode, just basic low power which was always kind of flakey about when it would wake up.

I didnt check, but does OP hold the time when it got no power?
No, the hardware clock is powered by the Pandora's battery. If it is removed or goes dead then the clock stops. This may seem like an oversight but may not be worth doing otherwise: having a dedicated button battery takes up a surprising amount of space on the board, and while you have the ability to remove the battery it's not expected that you frequently would.
And i dislike that charging requires main CPU to be up.
It was a requirement at the time, in fact for anything that used lithium-polymer batteries. Charging a lipo battery isn't as easy as any previous battery, it needs to be properly regulated or you risk explosion, and at the time charge chips just didn't have the brains to handle it. Even cell phones would boot to a kind of mini-mode where the CPU kept tabs on the charge progress even while off. There are better charge chips now that don't need any CPU intervention, they're smart enough to manage themselves without fear, I'm hoping that gta04 has selected one of those.
Some people are very used to shutdown the device if they want it to charge faster.
That's mostly in your head. The Pandora is so efficient that the difference would only be 8% while idle. If it takes 4 hours to fully charge with the CPU it'd take about 4 hours and 20 minutes to fully charge while off (assuming it could be charged while off, which it can't in any way, shape, or form).
Its more common than you think.
It shouldn't be. When the battery is dead, accept that it is dead. Trying to turn it on to get 10 more seconds out of it will only result in damage to the battery. At 5% you can put it to sleep for almost 20 hours and the battery will still have a charge, or shut it off completely and it'll still be good for days. This is a user problem, not a hardware problem: when the battery is almost dead, save your progress and shut down, there is nothing more that can be done. Get used to it.
Another trouble, which could probably be fixed with the new case design, is that the battery is too small for the slot its been made for.
ED has addressed this, saying the battery will definitely fit the Pyra much better. In the meantime you can take a thin strip of paper and just fold it a few times to make a sponge to hold the battery in place if it causes you trouble. Some people have problems and others don't.
 
That's a lot of grievances, almost all of which have been addressed in the past but I'll summarize as best I can.

i am saying that nothing should force the user to stay in this mode.
What is currently forcing you to stay in this mode? How are you using the low power mode? Make sure that you've updated to the most recent version of the firmware because the low power mode has been replaced with a much more efficient sleep mode in most cases, only falling back on the old mode when it can't sleep.
Its very common for my pocket to wake up the OP, and start draining the battery faster.
Flip it to HOLD if that's causing problems ;)
More than likely though, I suspect you haven't updated the firmware and it isn't actually going to the proper sleep mode, just basic low power which was always kind of flakey about when it would wake up.

I didnt check, but does OP hold the time when it got no power?
No, the hardware clock is powered by the Pandora's battery. If it is removed or goes dead then the clock stops. This may seem like an oversight but may not be worth doing otherwise: having a dedicated button battery takes up a surprising amount of space on the board, and while you have the ability to remove the battery it's not expected that you frequently would.
And i dislike that charging requires main CPU to be up.
It was a requirement at the time, in fact for anything that used lithium-polymer batteries. Charging a lipo battery isn't as easy as any previous battery, it needs to be properly regulated or you risk explosion, and at the time charge chips just didn't have the brains to handle it. Even cell phones would boot to a kind of mini-mode where the CPU kept tabs on the charge progress even while off. There are better charge chips now that don't need any CPU intervention, they're smart enough to manage themselves without fear, I'm hoping that gta04 has selected one of those.
Some people are very used to shutdown the device if they want it to charge faster.
That's mostly in your head. The Pandora is so efficient that the difference would only be 8% while idle. If it takes 4 hours to fully charge with the CPU it'd take about 4 hours and 20 minutes to fully charge while off (assuming it could be charged while off, which it can't in any way, shape, or form).
Its more common than you think.
It shouldn't be. When the battery is dead, accept that it is dead. Trying to turn it on to get 10 more seconds out of it will only result in damage to the battery. At 5% you can put it to sleep for almost 20 hours and the battery will still have a charge, or shut it off completely and it'll still be good for days. This is a user problem, not a hardware problem: when the battery is almost dead, save your progress and shut down, there is nothing more that can be done. Get used to it.
Another trouble, which could probably be fixed with the new case design, is that the battery is too small for the slot its been made for.
ED has addressed this, saying the battery will definitely fit the Pyra much better. In the meantime you can take a thin strip of paper and just fold it a few times to make a sponge to hold the battery in place if it causes you trouble. Some people have problems and others don't.
Updated to latest few days ago.

Under LP i mean whatever it does when i slide the switch. It turns off instantly and turns on from instantly to 3sec.


Well. HOLD cant really help when the Pandora either takes the pocket to its limits, either require a "bigger pocket mod".


I get it. I just provided a bad example of a problem.


Another problem is that you should play magic for CPU to startup, and do the charging.


I also do.


Not just mine. Also, did you mean that it would take 4:20 to charge with CPU on, and 4:00 without? Or you meant "as you wrote"?

I am just stating that its a mental problem of myself, and some other people i know. As the practice says, its unpatchable.

Thats just a example of why the user will get his battery very empty. There are some more.

(Been talking about a different device, for which 5% means a completely different thing from what 5% means for OP :) )

I already used cotton disks, but it doesnt fix the problem. The slot is too WIDE, not too deep.
 
Under LP i mean whatever it does when i slide the switch. It turns off instantly and turns on from instantly to 3sec.
And what is wrong with it? I don't understand what the complaint is. What is it doing (or not doing) that you think it should be (or shouldn't be)?
Thats just a example of why the user will get his battery very empty. There are some more.
I'm sure there are. It doesn't matter how many scenarios you come up with, you can't do anything to just make there be more power. When the battery is empty it is empty. The only thing that can be fixed is the user, you'll just have to accept it.
Not just mine. Also, did you mean that it would take 4:20 to charge with CPU on, and 4:00 without? Or you meant "as you wrote"?
Whatever I wrote, I meant that difference between completely off and just sitting idle is only 20 minutes over 4 hours. That's 5 minutes in an hour. Yeah, there's some, but not any sort of huge gain that you're likely to notice. Also that it doesn't matter because it is impossible to charge without the CPU being there to monitor the charging. The Pyra may be able to fix this.
 
I never powered on the Pandora when it was in my pocket. The switch is in a cavity !

If your battery is a bit small, you can put some adhesive padding to keep it in place. It's actually quick and easy to do that.
 
Only time I've had my Pandora switch on or come out of low power mode in my pocket is from before I padded the battery correctly, and the system would reboot. Yes the battery slot size is a defect which couldn't be fixed without costing a lot unfortunately, and will be better in the Pyra.


Low Power mode is arguably badly named. It's not for when you want to do something that doesn't need much power, to eke out the battery life. The CPU already has very good scaling capabilities, and so the best you can do is close the lid (or turn off the screen some other way) and turn off all connectivity. Low Power mode was designed to emulate stasis as closely as possible by clocking the CPU as low as it will go (and doing all the above), and if that's changed since it's going to be even better at simulating suspend. I got six days out of low power mode with occasional use - it's not going to do much using less than 27mA (assuming 4Ah battery). I'm slighty surprised it doesn't cut power to the headphone amp too, but I guess that's hardwired into the battery.


My own gripe is that it lets you go to low power mode when charging. If that's not a problem, why does it wake you out of low power mode when you plug it in to charge?
 
My own gripe is that it lets you go to low power mode when charging. If that's not a problem, why does it wake you out of low power mode when you plug it in to charge?


It's because when it's charging it needs to have some CPU cycles to charge properly.. However when you flick the switch without it being plugged in it goes into a suspend to ram mode.. when you plug in the charger it needs the CPU to charge properly so it wakes up. So maybe it should go from suspend to RAM to low power mode vs all together waking up instead...  


However the Pyra may never have this problem so it seems silly discussing these issues here.
 
Last edited by a moderator:
Under LP i mean whatever it does when i slide the switch. It turns off instantly and turns on from instantly to 3sec.
And what is wrong with it? I don't understand what the complaint is. What is it doing (or not doing) that you think it should be (or shouldn't be)?
Thats just a example of why the user will get his battery very empty. There are some more.
I'm sure there are. It doesn't matter how many scenarios you come up with, you can't do anything to just make there be more power. When the battery is empty it is empty. The only thing that can be fixed is the user, you'll just have to accept it.
Not just mine. Also, did you mean that it would take 4:20 to charge with CPU on, and 4:00 without? Or you meant "as you wrote"?
Whatever I wrote, I meant that difference between completely off and just sitting idle is only 20 minutes over 4 hours. That's 5 minutes in an hour. Yeah, there's some, but not any sort of huge gain that you're likely to notice. Also that it doesn't matter because it is impossible to charge without the CPU being there to monitor the charging. The Pyra may be able to fix this.
Under this line, i tried to describe what happens, in case i actually take sleep mode as LP.


I am telling WHY the user would have the battery totally drained. I am not saying that there should be some magic to make more power. I am saying that there should be no serious troubles if the battery ever reaches zero, gets removed, or in any other way, there is a power fault. Time could be provided by NTP over Wi-Fi, but having to "wake up the CPU" every time is already a serious matter.


And no, i said that they will also will try turning it on a bunch of times, to explain that we should not rely on "last amp" to keep stuff working, as the battery may just get removed.


CPU takes alot, unless it been projected to serve as a charging controller too. A charging controller takes not much.

20 minutes, sometimes maybe quite useful.

I never powered on the Pandora when it was in my pocket. The switch is in a cavity !


If your battery is a bit small, you can put some adhesive padding to keep it in place. It's actually quick and easy to do that.
The switch seems not all that hidden actually.


Its the dirty fix for Pandora. But we should have no troubles in Pyra.

Only time I've had my Pandora switch on or come out of low power mode in my pocket is from before I padded the battery correctly, and the system would reboot. Yes the battery slot size is a defect which couldn't be fixed without costing a lot unfortunately, and will be better in the Pyra.


Low Power mode is arguably badly named. It's not for when you want to do something that doesn't need much power, to eke out the battery life. The CPU already has very good scaling capabilities, and so the best you can do is close the lid (or turn off the screen some other way) and turn off all connectivity. Low Power mode was designed to emulate stasis as closely as possible by clocking the CPU as low as it will go (and doing all the above), and if that's changed since it's going to be even better at simulating suspend. I got six days out of low power mode with occasional use - it's not going to do much using less than 27mA (assuming 4Ah battery). I'm slighty surprised it doesn't cut power to the headphone amp too, but I guess that's hardwired into the battery.


My own gripe is that it lets you go to low power mode when charging. If that's not a problem, why does it wake you out of low power mode when you plug it in to charge?
Isnt it was ED who said that the 14Mhz was for OP to wait for "some alarms" he didnt list.


The main reasons behind some people wanting to have a full off, all the time:

1) The system is down. There are alot of the reasons you dont want to bootup you OS. Like if booting it up, or later turning it off involves special troubles.


2) Charging is kinda faster. Those 20 minutes can be very important if you got no time.

3) Some people cant trust sleep mode, because it may do some bad to their system. LP is out of the question in here. When the system gets downclocked to 14Mhz, and back to 1000Mhz, it can get "weird" very often.

My own gripe is that it lets you go to low power mode when charging. If that's not a problem, why does it wake you out of low power mode when you plug it in to charge?


It's because when it's charging it needs to have some CPU cycles to charge properly.. However when you flick the switch without it being plugged in it goes into a suspend to ram mode.. when you plug in the charger it needs the CPU to charge properly so it wakes up. So maybe it should go from suspend to RAM to low power mode vs all together waking up instead...  


However the Pyra may never have this problem so it seems silly discussing these issues here.
We are discussing it for the glory of satan, because people still got questions for some reason.
 
there should be no serious troubles if the battery ever reaches zero
The clock stops. That is it. That's the extent of any trouble that could be caused by the battery going dead. If that's your definition of "serious" then your life is a rose garden.
CPU takes alot, unless it been projected to serve as a charging controller too. A charging controller takes not much.
No, the CPU does not take a lot, it takes almost nothing. The wifi and the LCD are the biggest drains, an idle CPU is barely a blip.Yes, the CPU is the charging controller as is every single other device until recently that used lipo batteries. If your phone is more than a couple years old it definitely turns on when you plug it in to charge, even if you think it is off. Until recently there did not exist a dedicated charge chip that could charge a lipo battery under dynamic load.

Isnt it was ED who said that the 14Mhz was for OP to wait for "some alarms" he didnt list.
Low Power mode hasn't been 14Mhz for years. Where did you get that?Besides which, a flip of the power switch doesn't use low power mode anymore, it uses sleep mode. The only time it does low power is if it was charging, and that's only so it can shut off peripherals like the wifi and USB host.

Those 20 minutes can be very important if you got no time.
If the 20 minutes is so important, then you can stop charging at 92% and go off with only 9 hours and 15 minutes of use instead of the full 10 hours.
Some people cant trust sleep mode, because it may do some bad to their system.
We can't protect people with trust issues for everything. If they don't trust it, don't use it.
LP is out of the question in here. When the system gets downclocked to 14Mhz, and back to 1000Mhz, it can get "weird" very often.
You keep saying this but you haven't explained what "weird" means. Also it doesn't get downclocked to 14Mhz anymore, no idea where you got that information.The current low power solution only uses less power than being on by turning off the wifi, bluetooth, USB, and screen. You get the same performance out of manually turning these things off so if you want to charge without whatever "weird" you think is happening then just turn off the wifi and close the lid: it is exactly the same.

edit: accidentally a few words.
 
Last edited by a moderator:
there should be no serious troubles if the battery ever reaches zero
The clock stops. That is it. That's the extent of any trouble that could be caused by the battery going dead. If that's your definition of "serious" then your life is a rose garden.
CPU takes alot, unless it been projected to serve as a charging controller too. A charging controller takes not much.
No, the CPU does not take a lot, it takes almost nothing. The wifi and the LCD are the biggest drains, an idle CPU is barely a blip.
Yes, the CPU is the charging controller as is every single other device until recently that used lipo batteries. If your phone is more than a couple years old it definitely turns on when you plug it in to charge, even if you think it is off. Until recently there did not exist a dedicated charge chip that could charge a lipo battery under dynamic load.

Isnt it was ED who said that the 14Mhz was for OP to wait for "some alarms" he didnt list.
Low Power mode hasn't been 14Mhz for years. Where did you get that?
Besides which, a flip of the power switch doesn't use low power mode anymore, it uses sleep mode. The only time it does low power is if it was charging, and that's only so it can shut off peripherals like the wifi and USB host.

Those 20 minutes can be very important if you got no time.
If the 20 minutes is so important, then you can stop charging at 92% and go off with only 9 hours and 15 minutes of use instead of the full 10 hours.
Some people cant trust sleep mode, because it may do some bad to their system.
We can't protect people with trust issues for everything. If they don't trust it, don't use it.
LP is out of the question in here. When the system gets downclocked to 14Mhz, and back to 1000Mhz, it can get "weird" very often.
You keep saying this but you haven't explained what "weird" means. Also it doesn't get downclocked to 14Mhz anymore, no idea where you got that information.
The current low power solution only uses less power than being on by turning off the wifi, bluetooth, USB, and screen. You get the same performance out of manually turning these things off so if you want to charge without whatever "weird" you think is happening then just turn off the wifi and close the lid: it is exactly the same.


edit: accidentally a few words.
Most of you questions were asked, and most of them been answered.


Not going to answer today, as i got some things to do right now.
 
Last edited by a moderator:
Most of you questions were asked, and most of them been answered.
There's only one important question that I asked, and I only asked it because you seem to be having trouble and I want to help: what do you mean when you say low power does something "weird"?
 
I could imagine adding a capacitor to hold the RTC for a while (like <1minute) while changing the battery. Wouldn't take up as much space as a full battery.
Syncing time over wifi should be already possible, right? (BeagleBone Black has a linux-script fix, as it doesn't hold RTC when turned off)
 
Syncing time over wifi should be already possible, right?
Yes, ntpdate is included in the firmware.
sudo ntpdate pool.ntp.org;sudo hwclock -w

Of course, there is a GUI version of that in Menu, Settings, Date and Time, Sync time over Internet. (does fix your timezone too)
 
Last edited by a moderator:
Latest OS version can also automatically sync time whenever a WiFi connection is established (enabled as default )
 
Back
Top