An 8 km Bridge with Trees in the Fresnel Zone
A utility company needs a 5.8 GHz point-to-point bridge between two substations 8 km apart. Each end uses a 24 dBi parabolic grid with 2.5 dB of cable and connector loss, fed by a radio transmitting 20 dBm. The link must carry SCADA polling at 24 Mbps and has been engineered for a 15 dB fade margin. The surveyor's path profile shows clear line of sight between the two antenna centres, but a stand of trees at the exact midpoint has a canopy that reaches to within 4 m of the straight line joining the antennas. Heavy rain is not a factor in this region.
1. Compute the received signal level and the fade margin. Does the link meet its design target?
Read the worked answer
Free-space path loss: FSPL(dB) = 32.44 + 20·log10(8 km) + 20·log10(5800 MHz) = 32.44 + 18.06 + 75.27 = 125.8 dB
EIRP per end: 20 dBm - 2.5 dB cable + 24 dBi = 41.5 dBm
Received level: 41.5 (transmit EIRP) - 125.8 (path) + 24 (receive gain) - 2.5 (receive cable) = -62.8 dBm
Fade margin: with a -77 dBm sensitivity for the 24 Mbps rate, margin = -62.8 - (-77) = 14.2 dB.
That is roughly 1 dB short of the 15 dB design target - and this calculation assumes free space only. It does not yet account for the foliage obstruction, so the link is worse than the arithmetic says. It will probably run on a dry day and fail in wind (mast movement), foliage growth, or a wet canopy.
2. Quantify the Fresnel zone problem and state what clearance is actually required.
Read the worked answer
First Fresnel radius at the midpoint (d1 = d2 = 4 km):
r = 17.32 · sqrt(d1·d2 / (f·D)) = 17.32 · sqrt(16 / 46.4) = 17.32 · 0.587 ≈ 10.2 m
Practical rule of thumb is 60% clearance: 0.6 × 10.2 ≈ 6.1 m of clear space around the line of sight at the midpoint.
The canopy intrudes to 4 m, i.e. only ~39% of r1. So the trees are well inside the required zone and will cause diffraction loss on top of the 14.2 dB free-space number - likely pushing the link below its margin intermittently.
Judgement points:
- An obstruction at 60% of r1 contributes almost nothing; at 40% it contributes a few dB and rises steeply as the canopy closes.
- Trees grow and attenuate more when wet; a link that passes on survey day is not a link that passes in three years.
- Raising one mast by ~3–4 m, or moving the antennas to a shorter path, is cheaper than discovering the failure during an outage. Simply pointing harder will not fix diffraction.
3. The integrator proposes replacing only the far-end grid with a 30 dBi dish. Evaluate whether that solves the problem.
Read the worked answer
A 30 dBi dish adds +6 dB of receive gain at the far end, which improves only the downstream direction (near-end transmit → far-end receive). The reverse direction is untouched: the far end still transmits through 24 dBi, so the near end still hears it at -62.8 dBm.
Consequences:
- It makes the link asymmetric. If SCADA is poll-heavy (large downstream, small ACK/uplink), that may be tolerable - the uplink is smaller but not absent, and acknowledgements still need to arrive. If the uplink ever carries alarms or firmware, it governs.
- A 30 dBi dish is far narrower (roughly 3–4° beamwidth versus 5–6°). Alignment on an 8 km path becomes much more sensitive, and mast sway on a hot day can push the main lobe off the far antenna.
- It adds 6 dB to the far-end EIRP, which must be re-checked against the regulatory EIRP limit for that band.
It is a partial fix at best. The tree occlusion affects both directions equally; the budget should be corrected at the obstruction, not by buying gain at one end.