I’m trying to do some mesh algorithm testing/tweaking which means I need to play with attenuation to get a multi-hop network.
I’ve used every pad in my radio kit, put the nodes at opposite ends of the house, used a directional antenna pointing away… and I still get -4.75dB SNR (26% signal).
That’s 140dB of hard attenuation! I guess the 154dB link budget for #LongSlow is a conservative number. Lora is quite impressive!
I did a similar experiment with dummy loads as antennas, only to find out that the board leaks RF 
Perhaps you could put the board in a metallic enclosure and see if the SNR changes.
Hello colleagues. It is known from measurement practice that an open printed circuit board cannot be insulated by more than 80dB. Leakage of high-power RF currents to the case, the size of which is commensurate with or a multiple of the wavelength, is emitted very efficiently. For 900MHz, 1/4W is about 8cm, which is the length of the PCB. The area of the PCB emits very high RF power. To achieve attenuation 150 … 160dB (+ 20dB - (-140dB) = 160dB) Full blank RF shielding, including battery + external attenuators, will help you with this.
I didn’t consider the PCB leaking RF because the lora module is shielded but I suppose the trace to the sma connector would act as an antenna for sure.
You could also just dial down the output power with the API:
meshtastic --ch-set tx_power 2
…to set the maximum transmit power to 2dBm, for example.
I ended up just using a dummy load. Knocked it down far enough to simulate a multi-hop mesh.
