I am using the Si4468 for a 915Mhz and a 390Mhz device. With 915Mhz I can close the link with a 110dB of attenuation between transmitter and receiver using 2GFSK at 500kbps. I use WDS and change the frequency to 390Mhz and then change the PA matching network according to table 5 in AN648. What I have noticed is that at 915Mhz my power output is around +12dBm, at 390Mhz my power output is +18dBm however at 390Mhz the link will not work with anything more than 90dB of attenuation. I have checked the matching networks and even changed parts but I can not get better than 90dB. I figured the problem is not the transmit side of the link as I have +18dBm output and must be on the receive side.
Has anyone seen such a problem before or does anyone have a suggestion of what to try?
The RX path matching network should also be updated for the proper frequency for sure.
On the other hand, a 30MHz crystal harmonic falls into the RF band (13*30MHz=390MHz) which may also cause RX sensitivity degradation, so it could be suggested to try to use another crystal frequency - or check the RF link in a slightly different RF band.
Would you expect 20dB of reduce sensitivity for the 13th harmonic?
Sensitivity measurements at a crystal harmonic could have pretty large spreading in results. Also, together with a mis-matched LNA it can easily become significant.
I have tested at 393Mhz with the same results, is 3Mhz shift enough?
At 385Mhz I can get around 100dB but no where near as good as with the 915Mhz.
I'd recommend to double-check the LNA match you're using at 390MHz.
And, if you're still observing the issue then I'd suggest opening a SF case in the portal system where you can share your design files for review.
I have built several pairs of boards with parts from different suppliers to ensure that I did not get a bad part in the match.
I will see if I can open a support case.