RF 2.4GHZ through Via

[24602]

Have done a board design with a 2.4ghz module on the bottom of a PCB (placed due to size restrictions of board). The RF receive of the board has been fine but when transmitting the signal strength is basically non existent.

The RF TRX of the board is a castellated edge terminal then sent through a via to the other side of the board goes to a RF switch to either a SMD antenna or SMA connector (Schematic attached)

PCB Top and bottom attached also, I have deleted other components around the area for clarity.


Running RF through a via isn't ideal but because I have to, what is the best way of going about this? What would improve the TX Rf performance of the design?

[szoftveres]

Are you sure the TX problem is due to the via?

In general you'll need to absorb the parasitic reactances of the via and pad into a pi-network; i.e. the via would be the series inductor and the pads would be the shunt capacitors; tune everything to 2.4GHz and 50 ohms. This needs to be simulated or modeled for any good results.

[24602]

What would the best next step, any recommendations for RF simulators?

[uer166]

At 2.4GHz the via is likely not the problem. That antenna layout is crap-tastic. When you say you have low Tx, did you measure it at the SMA connector conducted using a spectrum analyzer? What dBm do you expect and what did you get?

[24602]

Haven't measured it but if TX RX antennas are right next to each other there's basically no reception.

[tszaboo]

I don't think the via is the issue.

I see two immediately obvious issues on the layout, that should be addressed first:
1) Your antenna has no matching network. Place at least a PI in between the antenna and the switch. Someone also need to connect the antenna (and the added pi network) to a VNA, and measure it, and select the right value components. This needs to happen 100% on every PCB antenna designs, and it's not something that you can just simulate or copy.
2) I don't think the antenna layout is OK. You copied the ground plane clearances, but you didn't place the antenna in an optimal orientation.
3) Your SMA might not be 50ohm with this layout.

But even without both of these fixed, you should have some reception, which makes me think there is another, more fundamental issue with your system. Meaning that having both these fixed will improve your signal maybe 20-30dB, but two boards next to each other should 100% be able to communicate with each other, since they have something like -90dB link budget. Something else is blocking the RF completely, like timing, channel selection, software...

[ConKbot]

The via with the grounds right there is good enough that it shouldn't significantly impede the signal. Perfect swr? Probably not. Still signal? Yep. Is there a ground plane immediately on the bottom layer of the module? Running under the module with it loading down the transmission line would reflect way more power than the via mismatch.