AM335x RTC

Hai Biser,

If I am using RTC from power management then why do I need to connect CLK32KOUT from PMIC to Processor? Processor can read the time and date values directly from PMIC itself on the I2C bus right?

Regards,

Erfan

Hello,

I stumble into this problem yesterday. I set up the date and time on my AM335x evm board expecting the system to keep the setting after rebooting, knowing that there is a backup battery.

Well, this morning the date and time were wrong. I'm trying to get some information around the net, but it seems that this matter has not really been tackled.

I don't really want to use the RTC from the PMIC and I rather keep the 32 Khz crystal feeding the AM3359.

However, Erfan brought an interesting point about reading the RTC from the PMIC using I2c.

Then my question is, how do we do that and especially how can we change the setting?

Thank you very much.

Thank you for your prompt answer.

Just to clear my mind on how the PMIC and AM335x are supposed to work. I can see on the TPS65910A documentation and EVM schematics that the backup battery is connected to VBACKUP which should maintain a VRTC voltage (during power OFF). VRTC is connected to the AM3359 VDDS_RTC pin. Is it how the MPU keep his RTC running?

We are currently developing a new product that will be release March-April next year. Do you think the revised AM3359 will be available?

Thank you.

Thanks, got you.

To make sure we can deliver the unit with date and time (which is really critical because our unit have to have a calibration cycle and need to prompt the user when is the next calibration), I'm trying to read those information from the PMIC via i2c. However, in the /dev and /sys, I can see i2c-1 and i2c-2, I can communicate with all the different sensors (light and temperature) but there is no i2c-0, which is connected to the PMIC (@ 0x2D).

I tried to create the node i2c-0 with "mknod /dev/i2c-0 c 89 0 but, I can't open the device.

Do you have any suggestion or guidance about that?

Thank you very much.

Hi Laurent,

I have similar problem. My system has i2c devices 1 to 3, so the EEPROM and PMIC are on /dev/i2c-1

# i2cdetect -l
i2c-1   i2c             OMAP I2C adapter                        I2C adapter
i2c-3   i2c             OMAP I2C adapter                        I2C adapter

Here you can see a dump of PMIC registers.

# i2cdump -y -f 1 0x2d
No size specified (using byte-data access)
     0  1  2  3  4  5  6  7  8  9  a  b  c  d  e  f    0123456789abcdef
00: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00    ................
10: 00 00 00 00 00 00 00 00 00 00 00 00 9f 0d 01 00    ............???.
20: 05 0d 00 00 05 00 00 04 00 00 00 00 00 00 00 00    ??..?..?........
30: 09 0d 01 0d 0d 0d 09 01 0d 00 00 00 00 00 3b 60    ?????????.....;`
40: 35 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00    5...............
50: 13 02 00 00 00 00 00 00 00 00 00 00 00 00 00 00    ??..............
60: 0a 0a 0a 0a 0a 0a 00 00 00 00 00 00 00 00 00 00    ??????..........
70: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00    ................
80: 01 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00    ?...............
90: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00    ................
a0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00    ................
b0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00    ................
c0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00    ................
d0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00    ................
e0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00    ................
f0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00    ................

The first thing I see, that reset values are not equal to the default ones. The next problem, I cannot write. The values won't be saved. After write to the RTC_CTRL_REG (0x10), i2cdump shows the same picture. I've asked in several forums/mailing lists, but still have no answer concerning real RTC driver for tps65910 RTC. The only clue was this post http://support.criticallink.com/redmine/boards/28/topics/1999. I tried to contact the developers of this company, but still have no answer.

Please share your i2cdump.

i2cdump, i2cdetect etc. are part of http://www.lm-sensors.org/wiki/I2CTools

Yegor

I've got RTC working. First step is to activate RTC domain in DEVCTRL_REG (0x3f). I've chosen an external 32KHz clock source:

i2cset -f -y -m 0xf0 1 0x2d 0x3f 0x00 b

after that you can activate RTC in RTC_CTRL_REG (0x10)

i2cset -f -y -m 0x01 1 0x2d 0x10 0x01 b

making i2cdump you can now see seconds increasing.

Hi,

I'm glade you were able to make it work.

What board and OS are you using?

I'm using AM335x evm board and the PMIC is on i2c-0. The problem is I don't have access to i2c-0 once Linux boot, I do have access to i2c-1 & i2c-2.

I'm using http://buildroot.uclibc.org/ and kernel from http://arago-project.org/git/projects/?p=linux-am33x.git;a=shortlog;h=refs/heads/v3.2-staging

What does "i2cdetect -l" say?

What do you mean saying, you don't have access to i2c-0 after Linux boot?

So I'm using AM335x EVM board from TI with Linux. During the boot Linux go get the IDs from the 3 stacks of boards via i2c-0. However once Linux boot, there is no file i2c-0 in /dev and no entry in the /sys/class folder.

Here is what i2cdetect display

root@am335x-evm:~# i2cdetect -l
i2c-1   i2c             OMAP I2C adapter                        I2C adapter
i2c-2   i2c             OMAP I2C adapter                        I2C adapter

I can't bring the i2c-0 to "live" to read and write the PMIC register (which is connected to i2C-0).

Execute "i2cdump -y -f 1 0x2d" and let me know what is says.

root@am335x-evm:~# i2cdump -y -f 1 0x2d
No size specified (using byte-data access)
     0  1  2  3  4  5  6  7  8  9  a  b  c  d  e  f    0123456789abcdef
00: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00    ................
10: 00 00 00 00 00 00 00 00 00 00 00 00 9f 0d 01 00    ............???.
20: 05 0d 38 00 05 2e 00 05 00 00 00 00 00 00 00 00    ??8.?..?........
30: 09 0d 01 0d 0d 0d 09 01 0d 00 00 00 00 00 3f 70    ?????????.....?p
40: 35 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00    5...............
50: 13 02 00 00 00 00 00 00 00 00 00 00 00 00 00 00    ??..............
60: 08 0a 0a 0a 0a 0a 00 00 00 00 00 00 00 00 00 00    ??????..........
70: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00    ................
80: 01 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00    ?...............
90: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00    ................
a0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00    ................
b0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00    ................
c0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00    ................
d0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00    ................
e0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00    ................
f0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00    ................

So, it seems that I have the same i2cdump as you, so I executed the two command lines and it seems to be working.

Thanks for the tip. However I'm still confused about the i2c line. The command line use i2c-1 but according to the documentation and the schematic the PMIC is connected to i2c-0.

http://processors.wiki.ti.com/index.php/AM335x_General_Purpose_EVM_HW_User_Guide#I2C_Address_Assignments

There is something that I'm missing somewhere!!!

Ok, after looking at the uboot pinmux configuration and trying few stuff I finally figured it out.

Even though there are three i2c on the am335x-evm only 2 are used. I2C0 and what TI called in the user guide I2C1/I2C2.

But what is confusing is that in the /dev and /sys directories I2C0 is renamed i2c-1 and I2C1 is renamed i2c-2.

Hope this help.

The numbering will be defined in the board file and must start with 1. I don't know why it will be made this way.

 omap_register_i2c_bus(1, 100, am335x_i2c0_boardinfo,
                                ARRAY_SIZE(am335x_i2c0_boardinfo));

I also encountered similar issue: http://comments.gmane.org/gmane.linux.ports.arm.omap/81231

Now we have a working RTC, but still no driver ... Are you going to write one?

Hi,

I can definitively start writing a driver however it will be on my spare time, we have to release our new product in a few months and I don't really have time. I'm probably going to use i2c to read and write the registers.

However, I think I will have to use an external crystal (right now I don't have one and I'm using the internal RC), I noticed that the clock is 4min behind already.

We keep in touch for the driver.

Thanks for the help

Hai Yegor,

I just wanted to know if you were able to change the contents of the RTC registers on PMIC. The reason I am asking for this is I am using RC oscillator on the PMIC which does not provide the accurate values and I will be using the RTC from Processor which has dedicated 32KHz connected to it. I want to write PMIC RTC registers in continuous intervals to make sure the time variance can be minimized and accurate timing can be maintained.

Regards,

Erfan 

Hi Erfan,

I was able to change the values. So far I've touched following registers:

RTC_CTRL_REG

BCK1_REG..BCK5_REG - they maintained their value after power off.

In my design I have external 32KHz source connected to PMIC. Have you looked at RTC_COMP_LSB_REG and RTC_COMP_MSB_REG?

Hi Laurent,

I was blind ;-) Look here: http://git.kernel.org/?p=linux/kernel/git/torvalds/linux.git;a=blob;f=drivers/rtc/rtc-tps65910.c;h=7a82337e4deee1ae9d1648ec6df5dbbcfe344ba4;hb=HEAD

And this thread: http://comments.gmane.org/gmane.linux.kernel/1340005

One still has to back port, but this should be a minor problem. That's why I hate to use old PSP kernels :-(

Yegor

i2cdump  can show i2c device values of  register.  On  AM335X EVM ,PMIC on I2c-1, not at  i2c-0.  Because i2ctool  and  linux on  userspace start  one index,  not zero.

root@am335x-evm:~# ls /sys/bus/i2c/devices/
1-002d 1-0040 1-0051 2-001b 2-0048 i2c-2
1-0035 1-0050 2-0018 2-0039 i2c-1

I have the same problem with WinCE 7 on AM335X.  I can read the registers via testi2c but I am unable to write to, for example, 0x3f:  testi2c reports success but the value does not change, and I am unable to enable the RTC as a result.  I tried writing many other registers but no success there either.

Hi,

I have seen that silicon version 2.0 and 2.1 do not have this issue? Right?

Are these already available

Johan

Hi Biser,

Can you comment on how big the cap must be to store eg data for a power outage of 1 day?

Johan

Biser-

Can you provide the link in this thread?   When I look in the Technical Documents tab of TI Home > ARM Processors > Sitara ARM Cortext-A Processors > ARM Cortex-A8 Core > ARM AM335x ARM Cortex-A8 I find nothing whatever related to the AM335x chip (or the power sequencer TPS65910A3 that I referred to), nor do I see anything dated yesterday or any errata since  2011.  The website is not at all helpful here.  Can you clarify exactly what problem is being solved and whether the erratum petains to the AM335x or the TPS65910A3?

Thanks,

Bruce