How to make a LoRaWAN End Node with STM32WLE5J8I ?

[SHATE.1]

Hello

i made a custom board using MIDATRONICS module that integrate the STM32WLE5J8I and i don't find how to make a lorawan end node project !!! when i import the end node project from the nucleo-WL package, it doesn't work (don't make a join and don't send data!!) and because the STM32WLE5J8I HAVE ONLY 64K of flash i'm at 96% of memory !!!

Can you help please ? is anybody make a lorawan end node project directly with cubeMX for the STM32WLE5J8I without any import from the Nucleo project?

Thank you in advance

Bests Regards

[jrgert]

I'm afraid I do not have an answer for you. I cannot get the the Midatronics module to join.

FWIW, I'm moving to the SEEED Lora E5 or RAK3172 next. Charles from TTN built boards for them and are in GitHub https://www.thethingsnetwork.org/forum/t/custom-boards-based-on-lora-e5-and-rak3172/49947.

[YBOUV.1]

Hi @SHATE.1​ ,

when you have time, can you share the solution of this post?

thank you

[SHATE.1]

Hi YBOUV.1

Sorry i was in vacation and i'm returning now

I will share my experience soon when i have time.

But i have always the problem to see the level of RSSI on my object as explaining by email to evaluate the strengh of signal.

Thank you to give me a solution

Sincerely

[Andrew Neil]

Have you seen the videos:

https://community.st.com/s/question/0D53W0000182SEJSA2/full-exampletutorial-for-icubelrwan-v210-and-the-things-stack-ttn-v3-

"because the STM32WLE5J8I HAVE ONLY 64K of flash i'm at 96% of memory !!!"

I would strongly suggest that you, at least, start with a larger-memory variant - so that you can at least get something working. Then you can look at optimising that down to a smaller memory footprint

As the well-known saying goes,

"Premature optimisation is a root of all kinds of evils"

I would also strongly suggest that you, at least, start with the Nucleo board - again, get something working on a known supported platform first, and then look at porting it to something else ...

[jrgert]

Andrew,

In my case, I did start with a larger-memory variant - and got it working. The NUCLEO WL55 board.

I then tweaked the ST code and rewrote a couple lines per information from the manufacturer, Midatronics. Only the RF switch interface pin #defines and removed a couple lines in the RF switch subroutine. None of the mission critical LoRa stack or task switcher code was modified. From what I've seen on github, essentially the same modifications required to use other STM32WLE5 modules.

From my experience, using 96% read-only memory doesn't negate the code's effectiveness. It still links (fits), downloads and executes. For evaluation purposes, it should work. Production code is another matter.