The RAK5802 is powered through 3V3_S which you can control with WB_IO2, but you might need to de-initialize the UART, as both RX and TX are connected with pull-ups to 3V3_S, which can lead to powering 3V3_S through the RX or TX lines when they are kept on HIGH.
RAK4631 + RAK19007 + RTC RAK12002 (in SLOT-A) gets me into the 30uA at deep sleep
but when I use also the RAK5802 (IO_SLOT) i have to use WB_I02 … which is also used for interrupt on the RTC board.
So trying to use WB_I01 as interrupt did not go as expected… (RTC or something make it reboot right away and not because RTC)
Then using RTC (only without RAK5802) in SLOT_C worked with WB_I04 (WB_I03 not), but the current was not low enough compared to using only the RTC in SLOT_A.
is there really a way to use SLOT_A with RTC and IO_SLOT with 5802 and keep the deep sleep into the 30uA ???
SLOT_C for RAK12002 with minimal code changes worked now and achieved low current (same as SLOT_A). Awesome!
WB_I02 as IO_SLOT switch (ON/OFF) for the RAK5802 worked as expected, but as you recommended, UART stop = serialX.end() helped a lot
disable pins later on the RX and TX pins of the serial, also the WB_I02 pins and all the LoRa pins did the trick.
RAK4631 + RAK19007 + RAK12002 + RAK5802 is doing ~30uA at deep sleep (with trickle charge of the RTC working fine and waking up the board at expected intervals)
PS: the WisBlock Mapper does not work with open source apps like Libre Office and Open Office