Beaglebone Black - Power supply and shutdown process.

[michael.ja...@gmail.com]

Hello,

I have inherited a project which contains a Beaglebone Black within the design. I have no particular knowledge of the beaglbone.

I am attempting to design a power backup for the beaglebone to ensure corruption does not occur when power is lost. I understand this can happen. Apparently it has happened with this product.

For that, I'm looking to use a relatively new device from LT being LTC3643. Basically, it charges a capacitor/super-cap arrangement during one mode, discharges the supper-cap by regulating to the load in another mode when the power is lost. Seeming quite nice.

I've been looking through the pages and pages of information to determine the following questions about the beaglebone.

Q1. In the real world, how long does the beaglebone black take to shutdown?

Q2. Apparently the power button is held down for 8 seconds. Does this shutdown period commence from the end of that 8 seconds?

Q3. From the notes left to me, I'm assuming the continuous current consumption to be 500mA at the 5V supply. Would anyone have an idea of what that may be during the shutdown period.

Ultimately, I'm attempting to determine some supercap sizing here.

Regards,

Michael Kearsey

[Dennis Lee Bieber]

On Mon, 4 Sep 2017 16:28:47 -0700 (PDT),
michael.ja...@gmail.com declaimed the
following:

>
>Q1. In the real world, how long does the beaglebone black take to shutdown?

In my limited experience... It depends upon the OS in use. No idea for
"Stretch", but "Wheezy" and "Jessie" showed almost opposite behavior in
terms of boot-up and shutdown -- one being relatively fast at shutdown, but
took "forever" to boot up to where a connection could be made to it.

>Q2. Apparently the power button is held down for 8 seconds. Does this
>shutdown period commence from the end of that 8 seconds?

I've avoided the power button for shutdown; I've not had good results
with it, and rely instead upon a shell alias for "sudo shutdown -h now"
(Hmmm, I may need to modify that -- I think more recent builds trigger a
password prompt if sudo hadn't been used fairly recently)

--
Wulfraed Dennis Lee Bieber AF6VN
wlf...@ix.netcom.com HTTP://wlfraed.home.netcom.com/

[Tarmo Kuuse]

On Tuesday, September 5, 2017 at 3:12:44 AM UTC+3, michael.ja...@gmail.com wrote:

I am attempting to design a power backup for the beaglebone to ensure corruption does not occur when power is lost. I understand this can happen. Apparently it has happened with this product.

My experience shows that the ext4 file system with journaling enabled does occasionally become corrupt when pulling the plug, although I've tried to avoid writing to the eMMC (no read-only root, just redirected all disk-hoggers to tmpfs-s in RAM). Still, during first half of the year I've had 3-4 cases of file system corruption among a set of ~10 devices which get their plug pulled daily (they are powered by street lighting circuits which are off during daytime). 

The easiest software "fix" I've come up with so far is mandatory fsck (the file system checker) on boot. This requires adding "fsck.mode=force fsck.repair=yes" to the kernel command line in /boot/uEnv.txt. On my desk it revives the corrupt ext4 file system automatically. Definitely not ideal, but the probability of ending up with an non-booting system is reduced (can't give stats, been only running this for 2 months). 

 

Q1. In the real world, how long does the beaglebone black take to shutdown?

For me, around 8 seconds with a recent Debian 8.6 or 9.1 from https://rcn-ee.net/ (some services, some networking). I've seen bad network configuration delay it by another 5-10 seconds.

 

Q2. Apparently the power button is held down for 8 seconds. Does this shutdown period commence from the end of that 8 seconds?

Pressing and immediately releasing the power button will trigger a software-initiated shutdown. Holding the power button down will probably do a "hard" shutdown, i.e. power will be cut.

 

Q3. From the notes left to me, I'm assuming the continuous current consumption to be 500mA at the 5V supply. Would anyone have an idea of what that may be during the shutdown period.

You'd have to measure that. I doubt it'll be any different from normal operation, unless shutdown triggers some externally connected HW to consume more. 

--

Kind regards,

Tarmo Kuuse