Having fun with the RFM96W

LoRa | 2020-11-09, 01:16:00

There is a patented modulation technology called LoRa, originally invented in France and bought by the capacitor manufacturer called Semtech.
It's a spread-spectrum transmission tech, targeted at getting a longer range than a random shift-key sampling of the signal.

I've bought the HopeRF's rebranded chip-on-module (SX1276) called RFM9XW, for which I have had designed my own board.

This design has a very bad antenna path (I assumed it wouldn't have significant losses), which I ended up changing by carving the board out (you can download it from my github).

I have had the board made in blue and I'm quite satisfied; it stacks perfectly with the power supply,

Wei Lin has written an article and re-wrote some Adafruit code for the module to work with the Micropython interpreter running on an ESP8266. You should have a look at his repo and my fork of it to acquaint yourself with how it works.

I've ran the LoRa module at 868MHz with a half-wavelength dipole antenna.

After setting up a working circuit, took it for a walk:

Google Maps has a 'ruler' function which allows to estimate the distance traveled on ground.

1.16 km is the maximum distance where I can get a signal in summer, without line of sight, with a lot of trees and buildings in-between, north to south. The signal to my house isn't even direct, I live in a shadow zone.

In winter, I get about 1.27 km from an entirely different area, south-to-north. Funny enough, during the summer, I don't get the signals from that area.

The quality of the signal isn't very great, -113 dBm at that distance, well below the -100 dBm cutoff for transmission on the gsm or wireless networks.

Be aware that signals take time to propagate and there should be some delay between the messages sent.

When both antennas are near, some of the signals get reflected or are simply unprocessed, for some reason it simply misses some of the sent packages. I've put a 1.5 minute distance for them but, still, after five or so, it starts missing packages.

At 1km, about 2 out of 3 messages are lost. I suspect it has to do with the lack of line-of-sight but I can't prove it.

The modules can self-report the signal strength and the SNR.

For example, -25 dBm at 40 cm distance, just right out of the near-zone (tx 17 dBm).

The quality of the modules themselves isn't very great either, I haven't managed to get them working below 125khz bandwidth, not even by changing the crystals.

Categories antennae, design