Why line of sight is not enough

Radio does not travel in a line. It travels in a fat ellipsoid, and the ground is closer to it than it looks.

The signal is wider than the line

Point two antennas at each other with nothing visible in between and it is natural to assume the path is fine. It often is not.

Energy arriving at the receiver has taken every possible route, not just the direct one. Routes that arrive close to in phase with the direct signal add to it; routes arriving out of phase subtract. The region containing the constructive routes is an ellipsoid stretched between the two antennas — the first Fresnel zone — and it is much fatter than intuition suggests.

On a 20 km hop at 6 GHz it is about 16 metres in radius at the middle. Thirty-two metres across, in a region most people would describe as empty air.

Put something into it and you remove energy that was contributing constructively. The link does not fail cleanly. It runs several decibels worse than the budget says, which is considerably harder to diagnose than a link that never worked.

Why 60%

The usual planning rule is to keep 60% of the first zone clear. That figure comes from the point at which diffraction loss over a smooth obstruction becomes roughly negligible.

It is a convention, not a law. Meeting it does not guarantee anything, and a path at 55% is not suddenly broken. It is a threshold chosen because it is comfortably on the right side of where losses start mattering, and because a round number that people remember is more useful in practice than a precise one they do not.

The zone pinches at the ends

The ellipsoid is widest at mid-path and narrows to nothing at each antenna. This has a practical consequence that surprises people: an obstacle near a site matters far less than the same obstacle in the middle.

A tree line 500 metres from the tower may barely touch the zone. The same trees ten kilometres out, halfway along a twenty-kilometre hop, can take a serious bite out of it. This is why a site survey that walks the fence line and declares the path clear can be badly wrong.

The Earth gets in the way too

Over any real distance, the ground between two antennas bulges upward relative to the straight line between them. On a 20 km path that is nearly six metres at mid-path under normal conditions — comparable to the Fresnel radius, and not something to leave out.

The atmosphere helps, up to a point. Air density falls with height, so radio bends very slightly towards the Earth, making the Earth appear flatter than it is. That is what the K factor captures: an effective Earth radius equal to K times the real one. K = 4/3 is the standard temperate-climate planning value, and it is where nearly every calculation starts.

The number to check is not 4/3

This is the part that matters most and gets left out most often.

K is not a constant. It varies with the atmosphere, and under sub-refractive conditions — K below 1 — the Earth appears more curved than it really is, pushing the effective ground up into the path. The bulge that was six metres becomes thirteen.

That is what takes down a link that has worked perfectly for three years. Not a fault, not degradation — a particular kind of night, a few times a year, when the atmosphere layers the wrong way. Everything is fine at 4/3 and the path is blocked at 0.6.

So a marginal path is worth checking twice: once at 4/3 for the normal case, and once around 0.6 for the bad night. A path that clears at both is genuinely clear. A path that clears only at 4/3 is a path with an intermittent fault waiting to be diagnosed.

Raising an antenna helps less than expected

If an obstacle sits a quarter of the way along a path, raising the far antenna by ten metres lifts the line of sight at that obstacle by only two and a half.

The line of sight is a straight line between the two antenna tops, so moving one end rotates it about the other. The lift at any point is proportional to how far along that point is from the fixed end. Obstacles near one site are cheap to clear from that site and expensive to clear from the other, which is worth knowing before pricing a taller mast.

What this does not calculate

How many decibels an intrusion actually costs. That needs a diffraction model, and the answer depends on the shape of the obstruction: a knife-edge ridge behaves differently from a rounded hill, which behaves differently from a forest canopy. Clearance geometry tells you whether you have a problem, not exactly how expensive it is.

Frequently asked questions

Can a path with clear line of sight still be bad?

Yes, and this is the commonest surprise in path planning. The signal occupies a fat ellipsoid, not a line — about 16 metres in radius at mid-path on a 20 km hop at 6 GHz. Something inside it that is nowhere near the visual line still removes energy, and the link runs several decibels worse than the budget predicts.

Why keep 60% of the first zone clear?

It is roughly where diffraction loss over a smooth obstruction stops mattering. It is a planning convention rather than a physical boundary — a path at 55% is not broken, and one at 60% is not guaranteed.

What is the K factor?

Air density falls with height, so radio bends slightly towards the Earth, making it appear flatter. K expresses that as a multiplier on the Earth's radius. K = 4/3 is the standard temperate-climate value.

Why check K = 0.6 as well?

Because K varies. Under sub-refractive conditions it drops below 1 and the Earth appears more curved, pushing the effective ground up into the path. That is what takes down a link that has worked for years — a particular kind of night, a few times a year. A path that clears at 4/3 but not at 0.6 has an intermittent fault waiting to happen.

Why does an obstacle near the site matter less?

Because the zone pinches to nothing at each antenna and is widest in the middle. A tree line 500 metres out may barely touch it; the same trees halfway along can take a serious bite. It is why walking the fence line and declaring the path clear can be badly wrong.

Why does raising the antenna help so little?

The line of sight is a straight line between the antenna tops, so raising one end rotates it about the other. The lift at any point is proportional to how far along it sits. An obstacle a quarter of the way along gets only a quarter of the height you add at the far end.

Does this tell me how many decibels I lose?

No. That needs a diffraction model, and the answer depends on the obstruction's shape — a knife-edge ridge, a rounded hill and a forest canopy all behave differently. Geometry tells you whether there is a problem, not its exact price.