A real life example of battery life

[codifies]

I was interested to get a real world view of battery endurance...

I've basically hung off this poor board everything that I thought might be useful in a desktop type system, way more than you might need in some kind of remote unattended control / logging project but it gives you an idea of what's possible in the extreme

First here is the system

powering the system is a 6600mAh LiPo battery with the following devices connected
a 7" lcd display
a four port USB unpowered hub
zippy mini usb keyboard
optical mouse
usb wifi R5370 with antenna
intel sata SSD (320 series 120GB) :-o - rootfs device
GPIO connector 2 +5v pin powering a max232 chip...

now with both an LCD *and* SATA connected this is obviously a rather (over) loaded system!

while under test the system was running top on the debug serial port and was being used to browse the internet and do other simple tasks, for half an hour it was running a screen "saver" which was loading one core to 50%

The total time running was just over 2 hours, which given what's hanging off it is not too shabby!

However after the 2 hours while the battery capacity was at 50% the power controller shut down the system, presumably because the LiPo was unable to provide enough current.

if you graph the included data you can see that the capacity drops in an almost linear manner...

Code Select Expand


       16:41:04 100   16:42:04 100   16:43:04 100   16:44:04 100   16:45:04 100   16:46:04 100   16:47:04 100   16:48:04 100   16:49:04 99   16:50:04 99   16:51:04 99   16:52:04 99   16:53:04 98   16:54:04 98   16:55:04 98   16:55:37 97   16:56:37 97   16:57:37 97   16:58:37 96   16:59:37 96   17:00:37 95   17:01:37 95   17:02:37 94   17:03:37 94   17:04:37 94   17:05:37 93   17:06:37 93   17:07:38 92   17:08:38 92   17:09:38 92   17:10:38 91   17:11:38 91   17:12:38 91   17:13:38 90   17:14:38 90   17:15:38 89   17:16:38 89   17:17:39 88   17:18:39 87   17:19:39 87   17:20:39 86   17:21:39 86   17:22:39 85   17:23:39 85   17:24:39 84   17:25:40 84   17:26:40 84   17:27:40 83   17:28:40 83   17:29:40 82   17:30:40 82   17:31:40 82   17:32:40 82   17:33:40 81   17:34:40 81   17:35:40 80   17:36:41 80   17:37:41 80   17:38:41 79   17:39:41 78   17:40:41 78   17:41:41 78   17:42:41 77   17:43:41 77   17:44:41 76   17:45:41 76   17:46:42 75   17:47:42 75   17:48:42 75   17:49:42 74   17:50:42 74   17:51:42 73   17:52:42 73   17:53:42 73   17:54:43 72   17:55:43 72   17:56:43 71   17:57:43 71   17:58:43 70   17:59:43 70   18:00:43 69   18:01:43 69   18:02:43 68   18:03:43 68   18:04:43 67   18:05:44 67   18:06:44 66   18:07:44 66   18:08:44 65   18:09:44 65   18:10:44 65   18:11:44 65   18:12:44 64   18:13:44 64   18:14:44 63   18:15:44 63   18:16:44 63   18:17:44 62   18:18:44 62   18:19:44 62   18:20:45 61   18:21:45 61   18:22:45 61   18:23:46 60   18:24:46 60   18:25:46 59   18:26:46 59   18:27:46 59   18:28:46 58   18:29:46 58   18:30:47 57   18:31:47 57   18:32:47 56   18:33:47 56   18:34:47 56   18:35:47 55   18:36:47 55   18:37:47 54   18:38:47 54   18:39:47 54   18:40:47 54   18:41:47 54   18:42:47 53   18:43:47 53   18:44:48 53   18:45:48 53   18:46:48 53   18:47:48 52   18:48:48 52   18:49:48 52   18:50:48 52   18:51:48 52   18:52:48 51   


I know not a lot about LiPo's but note that the AXP202 is a "Single Cell" charger, so presumably multiple cells in parallel is a no no !

[dave-at-axon]

The 6600mA battery is actually 3 parallel cells but they appear as 1 cell to the charger and they have a built in circuit to handle this.

I got around 5 hours from a 6600mA batter on an A20 but I wasn't using it as heavily as you where. I still need to test with the GPRS modem online and see how long it last.

I really only need an hour max as the system I am monitoring will actually be powered off (40kW motor) so only need some time to alert the operator that power has failed. I will then send them a message after 1 hour and do a controlled power down. When power is reconnected, the device will reboot and send them a message to let them know it has resumed operation.

[codifies]

Dave - out of interest did you notice did battery capacity go down as you'd expect ie down to say 2-3% before the power managment chip shutdown?

Just a thought for the remote monitoring project I've managed to get another Arm SoC to do "wake on alarm" I've not actually had time to fully look at the power managment system on this platform, but something like this could be quite useful, if you're in a scenario where you only need to poll infrequently...

[dave-at-axon]

Hi there,

It went all the way down to 3% before it switched off.

My remote needs to be running 24/7 so I have simply configured the boot so that if it does power off then on re-applying main power, it boots up again and my application starts. Actually, I am looking to install a custom app that gets run at boot which monitors the system and I can configure it to start one or more apps. It will handle the power control in the background, leaving the foreground app to go about it's business. I have 3 uses for this one design so this gives me the flexibility I need.

I am working on the networking just now as I see that it still tries to connect to 3G even when WiFi is active and connected. Something is missing in the configuration somewhere.

[codifies]

I've previously implemented things using androids IOIO board but its so expensive (even the clones) that by the time you add a stripped down phone / tablet its not worth doing, my experience with Android is that you always seem to be fighting to work with it.

I'm interested to learn of your reasons for not just using Debian on this board? would it not be simpler, or is it just from my perspective of knowing GNU/Linux better than Android/Linux ?

[dave-at-axon]

I've gone with Android for a number of reasons.

1. I can programme in Android and I have working I2C, UART serial, GPIO etc so hardware access is easy.

2. Development done on the PC and runs on the target and debugging requires nothing more than plugging in a USB cable.

3. The play store allows me to put the client apps on there and have them updated without any intervention from the user. (It should be fun to see the phone users trying to get it working but I might use the beta option for these apps but it's no real issue as without my custom hardware, the apps are basically useless to anyone else)

4. I can update the OS in a similar way (although that part is still to be developed after I reverse engineer the Allwinner update app)


I'd like to have a go at Linux/Debian as I have an A10S I first got for experimenting with these boards and it has the XFCE desktop. Lack of knowing how to develop on this and debugging on the target means a steep learning curve. I'd love to be able to develop on the desktop PC, hit a button to transfer the app to the A10S and see it running and have breakpoints and single step in the same way I have with Android.

I am also used to this type of development with NETMF and using JTAG emulators with AVR, Cortex M0 etc.

I'd love to learn to do Linux apps, especially GUI stuff as that would expand by capability, but the above limits and not knowing where to start is putting me off the idea, but I am willing to listen if there is information on how to do this.

[tks]

interesting what you are saying.
I'm looking into the job of getting lazarus on the A20. So apps are not a problem then.
also commandline aps arent then.

For the hardware acces there a options.

[codifies]

Let me know how you get on compiling Lazerus, its a kind of boot strap process for the compiler, which isn't exactly fun...

look on their wiki for cross compiling for arm

you could use this to cross compile Lazarus, once you have it running on the board, further updates/extras should all be able to be compiled on the board.

I like Lazarus (reminds me of the days before they wrecked Delphi with ActiveX) so please do let me know if you get it working...