Chip power down and PMIC_POWER_EN

Hi Matt,
thank you for this important informations.

How can set these actions automatically on boot? Can I use uEnv.txt that uboot read? How?

Thanks,
Daniele

Matt Porter (mdp) worked with me this morning in IRC to track down the
power draw culprits causing the 570 uA observed draw with the Bone in
"off" mode.

It turns out that the TPS65217 only should draw about 6 uA in this
mode, which is rather good.
But the NCP349 (U1, schematic page 2) will draw 140 uA typical, 250 uA
max with no-load, in the configuration used on the Bone.
And then R1, R2, and R3 (schematic page 2) together equate to about 17
k-Ohm, drawing another 290 uA or so.

Together, this could lead to >500 uA current draw with the PMIC in
"off" state when powered by 5V external power source.

When powered by a lithium rechargeable battery, the NCP349 will not be
in the circuit (and thus draw 0 current), and the SYS_5V output of the
PMIC should be disabled (so that R1, R2, and R3 also draw 0 current).
This should drop power consumption to be more inline with the TPS65217
"off" state levels of sub 10 uA.

-Andrew

Andrew: Thanks for verifying the steps for me and doing some current tests. I dont have the patch or board file posted anywhere yet. The rest of the work is not very kernel specific, just C code. The issues left is propagating adding a second up to min/hour/day/month/year (all in BCD) and compensating for the race condition if the system time seconds increment after we read but before we increment and write.

Daniele: These arent settings for uboot that can be added, they are an example of what to write to the tps65217 and am335x to power down the device. I am working on writing this same process now that it is working into the kernel board file for shutdown. Kernel patch will have to be added and recompiled for these changes to work.

Hi Everyone,

Thanks for looking into this issue and laying out the procedure to shut down the Beaglebone.

I noticed that people had discussed creating a Kernel Patch, but as of last night I had not seen one. So I decided to do it this afternoon. The attached patch hooks into the Linux “poweroff” command and shuts down the board. The patch baseline is taken from the the OE/Angstrom Linux 3.2 build.

I’m not sure who exactly I should send this patch to in order to get it pushed upstream. Koen? So I would appreciate any advice on that front.

Thanks!

-Coln

0071-beaglebone-poweroff.patch (4.24 KB)

Thanks! The patch was a big help.

Because the BeagleBone doesn’t shut down the PMIC after ‘shutdown’, if there is a a battery backup connected to the TPS65217, the system will NOT be reset if 5V input is restored via DC or USB input. One has to disconnect the battery first before reconnecting power. This is a very bad side effect, but this patch fixed that issue.

I second the need to get this patch either in the Arago branch or Angstrom recipe.

You are very welcome. This has been incorporated into the latest TI PSP release as well. I have some other TPS65217 drivers (ADC Mux/ PMIC power path control/ push button input etc.) that I am going to try to get into mainline when things slow down a little.

Colin

Hi Colin,
I'm very interested to this driver, but I'm very young in Linux word so, if
you can help me , I'm looking for the method to control TPS65217 functions
from user space: in particular I have enabled WLED functionality from
u-boot, but I want control it from my c++ code. How can y do that if i2c is
controlled from kernel? I'm not able for the moment to create a Linux driver
...
I need also to read TPS65217 ADC, to know the level of the battery used for
backup.
Have you any idea of how to make RTC work under battery supply in sleep
mode?
Whit your PowerOff service BeagleBone should go in sleep mode? Because I see
that the VDD_3V3EXP is high after powerOff...

How many question... excuse me, but you patch have opened me a word... so
new for me!

Thank you again, and itf you can help me again with some suggestion...

Francesco.

-----Messaggio originale-----

The RTC in the am335x doesn't work in TPS65217 SLEEP mode due to a
silicon errata. It'll be fixed in the second rev of the silicon which
is sampling real soon now. If you need an RTC and SLEEP or OFF mode,
get an external one.

-Andrew

Hi Cecco.

How about this matter? Have you get it working?

I need directions in order to implement the PMIC SLEEP mode after gcc and, if possible, I’d like to know how you could do that.

Best regards,

José Henrique