I studied the variation of the sleep current. We took a total of 39 measurements with 3 devices at different times and on different days. The results showed a variation of nearly 1uA for LightSleep, probably due to the RTOS running, but very little variation was seen for DeepSleepd.
(Note that the sleep currents shown here include the current drawn by the on-board 3.3V LDO).
