TPS65910Ax Built-in RTC

For switching to the crystal/external clock please check the section below in the TPS65910 datasheet.

http://www.ti.com/lit/ds/symlink/tps65910a3.pdf
32-kHz RTC CLOCK (Page 46)

Therefore, I guess that other EEPROM bits too can be changed.
However, I worry that the registers are reset in OFF state and Backup state.

Best regards,

Daisuke

When TPS65910Ax is in OFF state or Backup state, are not Built-in RTC and the crystal/external clock available?

Best regards,

Daisuke

When TPS65910 is in Backup state:

 - VRTC regulator is in backup mode.
 - Only the 32-K oscillator and RTC module are operating (RTC_PWDN bit is not reset).
 - All other resources are off or under reset.

 - Is the CK32K_CTRL bit too reset?

When TPS65910 is in OFF state:

 - All power supplies are in OFF state except VRTC.
 - DEV_SLP  bit is cleared in OFF state.
 - DEV_OFF  bit is cleared in OFF state.

 - Is any bit of all other registers not reset?
 - Is VRTC_OFFMASK bit too not reset?

http://www.ti.com/lit/ds/symlink/tps65910a3.pdf
STATE-MACHINE (Page 41)
Table 62. DEVCTRL_REG (Page 74)

Best regards,

Daisuke

"Switching from the 32-kHz crystal oscillator or external square-wave clock to the RC oscillator is not supported."

http://www.ti.com/lit/ds/symlink/tps65910a3.pdf
32-kHz RTC CLOCK (Page 46)

CK32K_CTRL bit default setting (EEPROM Configuration) is 1 (Clock source is internal RC oscillator).

http://www.ti.com/lit/ug/swcu093b/swcu093b.pdf
Table 2. EEPROM Configuration for TPS65910Ax

After CK32K_CTRL bit was reset, is 32-kHz RTC CLOCK not switched from the 32-kHz crystal oscillator or external square-wave clock to the RC oscillator?

Best regards,

Daisuke

Hi,

Any updates on this topic since November 2012?  

As I also consider using the TPS65910A3 RTC with a battery on VBACKUP pin.

According to another thread on this forum, the EEPROM Bits are reloaded from a OFF to Active state transition.

(http://e2e.ti.com/support/power_management/pmu/f/43/t/225987.aspx)

As I want to use an external crystal, my code will clear the CK32K_CTRL once powered-up.  So I'm also concerned about the transition from external crystal oscillator to internal RC when the TPS65910A3 pass from OFF to Active state (EEPROM bits being reloaded) if this mode is not supported by the TPS65910A3 (according to page 46 of the datasheet).

I’m also concerned about the value of the bit RTC_PWDN since it’s default EEPROM value is 1 (off).  I already know that this bit is not cleared from a Active to Backup state, but from a OFF to Active state is this bit reloaded with the default EEPROM value?  If so, the RTC counter would skip many clocks before my code can clear this bit again.  This isn’t what we expect from a battery Backed-up RTC.

Regards,

 Frank

Hello Frank,

The following response is from our part expert:

The device cannot switch from external 32k to internal 32k, but the other way around is possible. By default, it will boot up with internal RC if set in the EEPROM. As long as there is a valid backup battery this bit will not lose its value. This will be reset only if you remove all the power supplies from the device or change it with software.

RTC_PWRDN bit gets reset only when there is no supply on the device. If the backup battery is present this bit does not change when the device goes from off to active. What exactly do you mean by off state? Is this the same as mentioned in the datasheet or some other definition? As long as there is a valid backup battery this bit will never lose its value.

Thanks,

Dan

Hi Daniel, 

On our design, we plan on using the TPS65910A3 RTC when there is no power to the board.  A small battery connected to VBACKUP should maintain the RTC active until main power returns.

My question is exactly as described by Daisuke:

http://www.ti.com/lit/ds/symlink/tps65910a3.pdf datasheet , state transition diagram on page 42 doesn’t show direct transition from BACKUP to ACTIVE mode.  It shows a transition from BACKUP to OFF mode (when V Main Battery (i.e. VCC7) goes above VMBHI).  Then the PMIC wait’s for a “POWER ON” condition.  This will trigger a OFF to ACTIVE transition and as per Flore’s (TI employee) answer on this post: http://e2e.ti.com/support/power_management/pmu/f/43/p/225987/797357.aspx#797357  will reload the EEPROM bits (CK32K_CTRL and RTC_PWDN will both be set to one i.e. Clock source is Internal RC oscillator and RTC is OFF).

If this is the case, the RTC will be OFF until some code re-enable it (several seconds later)...

If those bits aren't modified from a OFF to ACTIVE state, then some information is missing.

Regards,

Frank 

Hello Frank,

Content of the EEPROM will be uploaded on the first power up when coming from no supply. Then, if you change the content of the backup domain register and backup battery is present, after you go from off to active state you will still keep the values of those registers, it won't be reset with the content of the EEPROM. Only the registers that are in the turnoff domain will be reset (example o/p SMPS output voltages).

If there is no supply at all present, then the RTC will be reset. As long as there is some form of supply, even in off/backup/active states the RTC will be present.

Best Regards,

Dan

Hi Dan,

Thanks for your support and those precisions.

Regards,

Frank

Hello Dan,

We've designed a board that uses a TPS65910A to power an AM3359. Our plan was to use the RTC in the TPS device, as the one in the 3359 doesn't work.

Unfortunately, no matter what register settings I use, the RTC always dies when the main power (VBAT) is removed, even if VBackup is present. Among other things, VRTC_OFFMASK is always reset to 0, even though the EEPROM bits should only be reloaded after a VBackup failure.

As far as I can see no other driver is modifying the RTC registers.

Below are the relevant registers after power cycling, and before the RTC driver sets them up (but after the regulator driver):

reg[10]: 0x00
reg[11]: 0x00
reg[12]: 0x00
reg[13]: 0x00
reg[14]: 0x00
reg[15]: 0x00
reg[16]: 0x00
reg[17]: 0x00
reg[18]: 0x00
reg[19]: 0x00
reg[1a]: 0x00
reg[1b]: 0x00
reg[1c]: 0x9f
reg[1d]: 0x0d
reg[1e]: 0x01
reg[3f]: 0x70
reg[40]: 0x35
reg[41]: 0x00
reg[42]: 0x00
reg[43]: 0x00
reg[44]: 0x00

And these are the values after the RTC driver init:

reg[10]: 0x81
reg[11]: 0x82
reg[12]: 0x00
reg[13]: 0x00
reg[14]: 0x00
reg[15]: 0x27
reg[16]: 0x00
reg[17]: 0x00
reg[18]: 0x00
reg[19]: 0x00
reg[1a]: 0x00
reg[1b]: 0x00
reg[1c]: 0x9f
reg[1d]: 0x0d
reg[1e]: 0x09
reg[3f]: 0x10
reg[40]: 0x35
reg[41]: 0x00
reg[42]: 0x20
reg[43]: 0x00
reg[44]: 0x00

I tried lots of variations, but the result is always the same. Is there something I've missed?

Best regards,

Fabio

Hello again,

Please disregard my previous post.

It turns out that the problem was caused by the solid state backup battery not being able to cope with the current surge that occurs when the device switches to backup power. It works fine with a lithium cell.

We're going to investigate this further, but anyone planning to use the RTC in the TPS chip would be well advised to use a lithium cell as a backup battery.

Best regards,

Fabio

Hi Fabio,

i m trying to use the rtc for timekeeping on a customer board (am335x+tps65910a3) and i want to have your feedback about the circuit.

- did you try the rev 2.0 of am335x where the bug of RTC is fixed ?

- in case of using RTC of PMIC, and when the system is off, what's the lifetime of the battery ? 

- what's the configuration Hw/Sw needed for using the RTC of tps65910a3 ?

thanks Fabio

Hello Mokh,

We haven't tried rev2 silicon yet.

To use the PMIC RTC in Linux, all you need is the driver and the I2C bus connected to the AM335x. Do not drive the AM335x VRTC from the PMIC, use an external LDO, otherwise you'll drain you backup battery within hours.

Unfortunately the PMIC RTC still takes too much current, even in sleep mode, so we're now using an external RTC. Even if you're going for rev2 silicon, I'd still put down an external RTC as a get out of jail card.

Fabio

thanks again Fabio,

Nice to use an external RTC but this is make additional cost $ on the BOM :)

In the rev 2.0, the am335x takes about 40mW in RTC-only mode. when the PMIC is in sleep mode, how much current did you find ? did you try to use the low power mode on pmic to reduce current ?

Hi Miokh,

If you use the pMIC internal RTC and in backup mode, pMIC only consumes 16 uA max.