Forums

Hi,

A customer using our OMAP3530 EVM with TPS65950 onboard, found that the battery backup mechanism is not working well. Evidently the backup status register has not the right value when booting after going to BACKUP state.

- Ideas? Please advice. Thanks.

/Magnus

First questions:

Taken from SWCA025 chapter 4:

Power cut means a momentary disconnect of the main battery (for example, when the phone is dropped

and the main battery is removed momentarily). In such a case, the capacitors connected to the main

battery can keep the TPS65950 in ACTIVE state for some time, depending on the load on the application

processor.

To support this feature, perform the following minimum steps:

1. Unmask the MBCHG interrupt of the TPS65950 power subchip by setting the PWR_IMR1 MBCHG bit to 0

2. The MBCHG interrupt is generated when VBAT falls below 3.0 V or rises above 3.2 V. Configure the

MBCHG interrupt to be generated for both of these cases by setting the MBCHG_FALLING and

MBCHG_RISING bits of the PWR_EDR2 register to 1.

If power is cut and if 2.7 V < VBAT < 3.0 V, the host processor receives the MBCHG interrupt. The host

processor must take necessary action (save data, shut down applications, etc.). If the main battery is

connected before the VBAT capacitors discharge below 2.7 V, the TPS65950 remains in ACTIVE state. If

the VBAT level increases to more than 3.2 V, the same MBCHG interrupt is recognized by the host

processor.

If the main battery remains disconnected for a longer period and the capacitors on the VBAT are

discharged below 2.7 V, the TPS65950 enters BACKUP or NO SUPPLY state (assuming VBAT < 2.1 V

and VBACKUP < 1.8 V).

We has followed the above description, but has not seen the chip indicate it has been in BACKUP state.

How long time does it take the TSP65950 to enter BACKUP state ?
Is MBCHG interrupt nessary for the TPS65950 to enter BACKUP state?

My response:

I do not see from documentation that enabling or not the MainBattery alert interrupt MBCHG to the CPU (when VBAT falling below 3.0V) , would affect the chip going from ACTIVE or SLEEP state to BACKUP state (when VBAT falls below 2.7V).

- The time it takes should be instantaneous at 2.7V, but the slope 3.0V to 2.7V depend on your capacitors.

If the VBAT rises above 2.85V the state changes to WAIT-ON, and usually some reset would invoke the system

Could you describe how you expect the "chip indicate" that it went to backup mode? (Register bit STS_BOOT BACKUP?)

I read from Section 2 in the appsnote that it is usually at boot-time when you detect this?

Second questions:

"When I cut the power, and wait some time, then power up again, I find:

Register bit STS_BOOT BACKUP is always zero and the RTC is not running.
BACKUP_REG_(A..H) is always zero.
SECURED_REG_(A..U) has sometimes the values I put into them, sometimes zero.

We have testet the RTC on the linux, that the EVM board has on board at delivery, and found the RTC is not running when power is removed.

The WinCE BSP to the EVM board (rev G) does not switch on the charger for the backup battery, I have done that and the voltage is now 2V9.
I have scaned the BSP code and found no reference to STS_BOOT,
to me this indicates that the BSP is not finished on this subject. What else is missing?

I have put my "detector" in the RTC startup code, it is run when Windows is starting, prior to this the XLDR bootloader and the EBOOT bootloader
has run, in the bootloaders I have not found any code accessing RTC or BACKUP registers.
What do I miss?"

/Magnus

[TPS65950] Battery BACKUP state indication missing, and WinCE handler missing