433MHz PCB trace for Antenna

[maurosmartins]

Hello all,

I've been playing around with LoRa related stuff (@434MHz) and I've layout some PCBs where I would like to clarify some doubts.


On my first board (rectangular shape) the trace going out to the antenna connector has random width and clearance, nevertheless I had good results in terms of range with it.

on my second board (the round one) I've decided to use a calculator (saturn PCB toolkit) to calculate the required width and clearance for 50Ohms impedance with the following results:


afterwards I've soldered a 49.9Ohms smd resistor between the "ant" pad and GND on both boards and using a nanoVNA measured the s11 parameters with the following results:


the blue trace is from the new board while the golden trace is from the old board, oddly it seems that the old random parameters worked better than the new ones.

a more complete list of measures for the new board can be found on this image:


In either case and since my RF related knowledge is not that great my questions are:

• Is this test valid, or should I make other setup to measure the RF impedance of the PCB trace.
• In case it is ok, are the values any good? I see that they are not a complete match to 50Ohms but they are very close to.
• The RF trace is very small maybe it isn't affected that much, what is your opinion?

looking forward your reply,
Best regards, Mauro.

[schratterulrich]

I think the trace itself has the least impact on the result.
-There are greater differences between the boards due to the different connectors.
-There is a much shorter trace on the old board.
-The new connector makes a via stub.
-Also the GND return is somehow undefined - There are no nearby vias at the GND Pin to connect directly to the bottom GND plane, so there will be a undefined inductance...

[TheUnnamedNewbie]

Your trace is so short it shouldn't really matter all that much.

The free-space wavelength at 500 MHz is about 60 cm. Both of these traces are so short that the tolerance and variability easily accounts for the results you are seeing. The performance is just fine, I wouldn't worry about it too much.

What I would worry about is that on your round PCB the trace goes underneath what i assume will be an SMD can or module. That could actually hurt performance, esp if there are metals on the bottom of that module.

You also need a via near the ground of the lora module pin to connect ground pad to the ground plane

[maurosmartins]

Hello schratterulrich and TheUnnamedNewbie,

thank you very much for your replies! in future revisions I'll add layer stitching to better connect GND on both layers. in the meanwhile I'm waiting for RFM98W modules so I can test the new boards.

Best regards, Mauro.