STM32WLE5CCU6 LoRa Antenna design review

[manuelambaum]

Hi,

I want to design my own LoRa-Board based on an STM32WLE-Chip. It is my first time handling with RF-related topics which is why I'm creating this post. I created a design which I would want to have reviewed for major flaws, so I can order my first prototype. I know that antenna impedance matching is not an exact science but I just want to correct the worst errors.

Where did I get the LC-Values from:

The LC-Values are according to AN5457 from ST (for 868MHz LoRa LowPower mode). 

Where did I get the PCB-Antenna-Design from:

The PCB-Antenna-Design is from Design Note DN023 from Texas Instruments. However, the length of L6 is 2mm longer in my design in order to opimize the PCB-Antenna by shortening it.

What else did I take into consideration:

• RF-tracks surrounded by Ground-plane with vias included to the bottom layer
• Smooth tracks and track-size transition (this is why my RF-traces are rounded off)
• Almost no traces underneath RF-track (except for two traces underneath the RF-switch)
• 10mm clearance between the literal PCB-Antenna and the closes SMD-parts

My questions:

• My RX-path from the RF-switch to the RX-input on the uC is rather long. Is my design approach okay like this? I have seen many designs where there is no RF-switch and therefore the RX-path can be shorter.
• Some designs include a PI-Filter between the antenna and the RF-switch, however the filter does not look populated in most ST-designs. Should I consider a filter to be able to optimize the circuit with these C-L-C-parts?
• I am going to order the PCB over JLCPCB. There is a calculator for defining the track width for a 50Ohm matching. According to this calculator, for a 2-Layer, 1.6mm thick PCB and 0.254mm spacing between RF-trace and the surrounding ground, my RF-trace needs to be 1.2mm wide. Is this accurate? I haven't seen any layout with such wide RF-traces. I ended up, using a 0.254mm RF-trace width instead of a 1.2mm one...
• I have seen a reference design of the SX1261 / SX1262 for 868MHz with differing L-C-values. Where do these differences come from? The STM32WL is based on the SX LoRa-chip series which is why I wonder why different values are needed? I am too unexperienced to tell if the differing values are just slightly or completely different. Does it even matter?

I know it is a lot to ask but I would really appreciate if some of you reviewed my circuit and PCB-design so I can order my first version of it. All the relevant documents are in two .zip-files attached. Under "Resources.zip", you can find all the documents I referred to in my post and in "EndResult.zip", you can find my actual design.

Thanks in advance

Manuel