Update on the measurements:
I’ve taken everything out of the steel box, except the 230V AC → 12V DC adapter, as it is likely a noise source that affects the shielding in the coax cables. I’ve also removed the bandpass filter, as it did not seem to consistently reduce noise levels. For these tests, the SensorStation was inside a small Pelican case. A 6-element Sirio antenna was used, with a VSWR of ~1.12. There are ferrite clamps on various locations along the coax cable, as well as at the point where the coax cable connects to the antenna.
I was able to get noise levels of ~-80.5dB by closing the door to the steel box. Noise would be ~4dB higher with the door opened. With the door closed, the noise levels were quite consistent. A few averages of signal/noise at distances:
- 100m: signal -35dB, noise -65dB
- 400m: signal -56.5dB, noise -80dB
- 800m: signal -70dB, noise -80.5dB
- 1000m: signal -73.5dB, noise -80.5dB
- 1100m: no tag detections. Live pulses were detected at -74dB with -80.5dB noise.
1000-1200m was the range I expected for a setup with our antenna at ~4m and the tag on top of the dike. It likely would have ended up as at least 1100m, but the signal dropped a bit faster at the 200m to 400m range than in some previous measurements. We ended up with 2-3dB lower signals, but also ~5dB lower noise, yielding a net improvement in detection range.
From all the testing I’ve done so far, the culprit in my case definitely seems to have been the power supply. Moving it in certain ways can easily swing the noise by up to 10dB in some cases. Completely isolating it inside the steel box has given me the best results so far. Luckily, for our actual motus stations we won’t have to deal with an adapter so close to the coax cables, but this noise source can definitely mess with testing setups and waste a lot of time if you’re not aware of it!