Customer observation on OMAP+TPS65930 system:
The OMAP starts correctly, i.e. the program is loaded correctly and it starts. After a random time (it may happen after a few minutes or an hour) , the Companion Chip stops giving it the clock (26 MHz).
The Companion Chip freezes: it does not reply any more to I2C read commands, and sometimes it happens that it switches to a Scaled Voltage (from 1.3V to 1.05 vcore to OMAP) even if nobody asked with HW pin assertion or I2C SW request and as I said the clock provided to the OMAP processor is not present any more.
We have verified that the voltages provided to the Companion Chip are stable , before during and after the freeze event.
The clock provided to the chip is also stable and compliant to the requirements in the latest version of the Datasheet.
Temperature is well below the critical threshold.
Customer found the solution after checking the schematics/board design:
The instability was caused by a 100 K pull-up resistor connected to the A1 pin of the Companion Chip. Because internally there is a 100 K pull-down resistor, the situation was unstable.
Note: This may happen in any of the devices in this family - TPS65950 and TPS65920. Please check your schematics with the recommendations provided in the schematic checklist document - http://focus.ti.com/general/docs/litabsmultiplefilelist.tsp?literatureNumber=swca081