I’m currently working on a project to build a custom LoRaWAN gateway and I’m looking for some advice. I have previously built a gateway using the Waveshare SX1302 Raspberry Pi HAT and a Raspberry Pi 4, which worked well. However, for this project, I’m interested in using an STM32 microcontroller instead of the Raspberry Pi.
I’ve been researching and trying to find proper documentation or suggestions on which STM32 and Semtech module I should use, but I haven’t found anything definitive yet. I’m particularly interested in using the SX1302 or SX1303 for the LoRaWAN gateway.
I found this P-NUCLEO-LRWAN2 - STM32 Nucleo pack LoRa™ HF band sensor and gateway - STMicroelectronics
Which used sx1301.
I found the STM32MP157F MPU and I think it can be used for building the gateway along with a Waveshare/RAK/WM SX1302 concentrator. I would like to know your thoughts on this.
I intend to use The Things Network (TTN) as I have used it in my previous Raspberry Pi gateway. So, the new gateway should be compatible with TTN.
I would greatly appreciate any advice or suggestions on which STM32 and Semtech module would be best for this project. If anyone has experience with this or knows of any resources that could help, please let me know.
The original gateway / device pack you refer to has hardware design issues that stop the gateway from acting properly, is not at all welcome on TTN and should not be used as a reference design.
Nor should you use TTN until you have a verifiable good gateway as this will impact on the stability of the local network used by the community.
As for using an STM32 for the host MCU, Mr Why wants to know WHY? Why use the STM32? What is it about it that you want to use it? Because that information will help us advise you. Without an answer this topic will have a very short shelf life because it will become very academic: Sure, you can use an STM32 if you port the appropriate gateway software and provide it with a TCP/IP connection.
Mr Why would ask how many gateways you’ve found that use an STM32.
And for the concentrator - do you want to build your own or use a pre-made one?
@descartes The reason for selecting the STM32 is based on our client’s suggestion. However, if there are better SoCs available for this purpose, we would appreciate any recommendations.
I have found mainly the following STM32-based solutions:
STM32 Nucleo Pack LoRa™ HF Band Sensor and Gateway
STM32MP157F-DK2 Discovery Kit I have read this can be used for LoRaWAN gateway.
We do not plan to build our own concentrator. We are looking to use a pre-made concentrator.
So this is a commercial project? And you want free consultancy on ‘how to’ without having clear reason why? Hummmm…
Which as noted
Basically it appears they cut corners and didnt follow original SMTC ref design implementation and recommendations so doesn’t comply as a field deployable solution without issues and should only be used as a development platform and on your own private LNS instance. You could take as a start point and correct their errors but
So best forget you ever found it
You want a management and control plane processor sub-system that can run one of the various flavours of packet forwarder that exist in the market (have you selected one?) - you may need to write your own (do you have the skill-set and knowledge to do this?) if you can’t licence (for commercial use!) one of the open versions. It will need to provide a ready solution for the internet backhaul interface (Wi-Fi/Wireline Enet?), will need suitable SPI support (USB can be adapted), and enough grunt and memory space for management interface implementation (likely GUI based if to be commercially competative and easy to use/set-up - so will need to be able to run a simple local webserver & and perhaps have readily available libraries & modules that can be pulled in to support these functions?). Do you know why your client has suggested STM32? Do they know how many implementations are in the market based on this? Do you? why are there that many?
What capabilities and requirements have you listed for your potential GW design/build? How does they map to the features and capabilities of this device?
Depending on use case and deployment you may also want to include option to grab GPS/GNSS location information and/or utilise GPS base PPS signal.
BTW which LW version are you looking to support? When Looking at ESP32 based devices previously you were a bit confused by support for 1.02/1.03 vs 1.04? (Inthings)
I’d start running - you have a client that wants to influence the choice of MCU with little or no consideration to what’s generally in use. Their job is to set business objectives, your job is to find the best way to fulfil them. Choosing a vendor MCU that’s only seen in a borked implementation is not sound engineering. If you search the forum, you’ll find plenty of comments about the issues with the ST hack, so no points to you for research on the forum before posting.
Mr Why wants to know why you want to build your own gateway? What is the benefit?
He’d also be quite interested in why you’d want to take on such a project. And whilst not a why, if you have much LoRaWAN experience?
Its not uncommon and also allows seperation of personal vs professional vs academic vs hobby interests. Some post or read on behalf of company or other entity but may also have interest as a hobby etc. Though as here it can blur…
why are there that many?