In my design, the device is battery powered via the DCDC_BATT pin of iMX280. Instead of using the iMX280 charger, we used an external charger chip which has better power-path control. For this reason, the BATTERY pin of iMX280 is NOT connected to the battery and is left NOT connected. This approach has been checked with Freescale support and was told that it is not a problem.
In my design, when the battery is first connected to DCDC_BATT, the iMX280 should be in the OFF state (until the PSWITCH input is activated). However, what I observed is, iMX280 is in an unknown and unstable state with ~1.5V on the DCDC_VDDIO rail and ~0.8V on the DCDC_VDDA rail. There is also a leakage current from the battery (~12mA). The 24MHz crystal is also found oscillating occasionally in the OFF state and when it happens, it will draw extra ~3mA leakage current. When the PSWITCH input is activated, the iMX280 will be turned on and all the power rails will be normal. In other words, this unstable state only occurs in the iMX280 off state.
What I eventually found out is, this behavior is caused by the BATTERY pin being left un-connected. After I wired the BATTERY pin to DCDC_BATT, iMX280 will behave normally in the OFF state, i.e. 0V on all the power rails and no leakage current.
Has anyone encountered the same issue before???