Risto

I had a look with in-circuit-debugger into the memory and indeed no _paint_-pattern 0x55555555 can be found at _main_stack_start  address:

And the _main_stack (from 0x1000FC00 to 0x1000FFFF) is nicely "painted" on debugger reset as can be seen here: 

For testing purposes, I changed main stack size in linker script on the following line: https://github.com/opentx/opentx/blob/f8498ddf86f37ecf4a1eb962dec0776d220984dc/radio/src/targets/horus/stm32f4_flash.ld#L28 to 2048 bytes and what I can see, is indeed, that the main task...

@bsongis Were you able to replicate the issue? Thanks!

I increased the stack to 2048 (see above), I see that it is used over the first 1024 byte border, thus the most important message to me here is that...

For reference, here the original IMU PR that turned TX16S AUX1 from serial into I2C: https://github.com/EdgeTX/edgetx/pull/617

How about MAVLink as the "UART protocol"? The channel data could be transmitted with https://mavlink.io/en/messages/common.html#RC_CHANNELS and an aux of https://mavlink.io/en/messages/common.html#MANUAL_CONTROL could be mapped to various switches.

Yes, but the code is there: https://mavlink.io/en/getting_started/use_libraries.html (here especially https://github.com/olliw42/fastmavlink would likely be the one to go with). Especially the integrated MAVLink router in fastmavlink lib could be IMO of...

Here the "latest" state for EdgeTX: https://github.com/EdgeTX/edgetx/pull/150/files For truly latest state, see Olli's own repo (committs on top of OpenTX): https://github.com/olliw42/mTX

See option 3: https://headtracker.gitbook.io/head-tracker-v2.2/getting-started/wiring#option-3-bluetooth-bridge