Where should the repeater antenna go in a multi-storey building — rooftop or mid-level?

In a genuine high-rise, neither. A single repeater antenna — rooftop or mid-level — has roughly 137 dB of path loss to spend, and a reinforced-concrete floor slab costs 15 to 25 dB. That budget is exhausted after five to nine slabs, before you spend a single decibel crossing the floor plate. Rooftop is the right answer when the coverage target is horizontal: a yard, a campus, detached buildings on one site. Mid-level buys you a few more floors in a six-to-ten-storey building. Above that, the question stops being where one antenna goes and becomes how many antennas go on which floors.

What does the link budget actually allow?

Work it through with our standard figures. A 40 W repeater is +46 dBm at the connector; subtract about 3 dB for jumpers and feeder and add 2 dBi for a unity-gain omni, giving roughly +45 dBm EIRP. Against a −95 dBm design threshold at the portable, with 3 dB of body loss for a radio worn at the hip, the downlink can absorb about 137 dB of path loss. The uplink from a 4 W portable comes out at about 139 dB nominal, so the two paths sit within a couple of decibels of each other — and in practice the uplink fails first, because a rooftop head-end sits in a noisier RF environment than a handset does and 3 to 10 dB of noise-floor degradation wipes out that margin immediately.

Now spend it on concrete:

• Reinforced-concrete floor slab: 15–25 dB each

• RC lift core or stairwell shaft: 25–35 dB

• Total budget available: approximately 137 dB downlink, 139 dB uplink

• Design threshold: −95 dBm, with a 10 dB standard fade margin already included

• Slabs before the budget is gone: 5 to 9

Five to nine floors of vertical reach. From a rooftop antenna on a forty-storey tower, that puts the lowest usable floor somewhere around level 31 — and nothing below it. This is the same arithmetic behind why a walkie-talkie cannot cover a 40–50 storey building on its own.

Doesn't a mid-level antenna halve the problem?

This is the advice circulating on consumer radio sites: in a sixteen-storey building, put the antenna on the seventh or eighth floor, near the vertical midpoint, so the signal travels half as far. The geometry is right and the conclusion is wrong, because the binding constraint is not distance — it is slab count.

Halving the building gives you eight slabs to reach in each direction. You had five to nine slabs of budget. On good construction, with thin slabs and a leaky core, you just make it. On the 200 mm slabs and dense rebar mesh typical of Singapore commercial and residential towers, you do not, and the top three or four floors and the bottom three or four floors sit below threshold. Worse, you have taken a system that failed predictably at one end and made it fail quietly at both ends, which is much harder to diagnose from a complaint log.

There is a second-order reason the midpoint rule sometimes appears to work, and it is worth understanding before you rely on it. Signal does not actually punch through eight consecutive slabs. It leaks — out through the facade, up the outside of the building, back in through the windows above, and up the lift shafts, stairwells and service risers. When a mid-level repeater covers more floors than the slab arithmetic predicts, those leakage paths are doing the work. They are a property of that building's facade and core, not a design method, and they disappear the moment a tenant fits solar window film or a fire-rated riser door is kept closed.

Why does a higher-gain rooftop antenna make vertical coverage worse?

Because gain on an omnidirectional antenna is bought by squeezing the vertical beamwidth. A 6 dBi collinear at UHF radiates a flattened disc — around 20 degrees of vertical beamwidth is typical, though you should verify against the specific datasheet — with peak gain aimed at the horizon and a null directly underneath it. Published figures on high-gain collinears put that null as deep as 15 dB.

Mounted on a rooftop, the entire building sits inside the null. You have paid about 4 dB of extra gain toward the horizon, where there is nothing you need to cover, and given up as much as 15 dB straight down, where every one of your users is standing. Nor can you correct it: electrical downtilt is not available on an omnidirectional pattern, and mechanically tilting an omni simply swings the null around, adding gain on one bearing and deepening the null on the opposite one. We have written separately on why more repeater power or a higher-gain antenna usually solves the wrong end of the link.

So where does the equipment actually go?

Once the building is taller than about ten storeys, the repeater's physical location becomes an access and servicing question rather than an RF one. Put the rack where it can be reached, powered, cooled and maintained — typically a plant room or ELV room — and distribute the RF properly. Antennas go on every floor, or every second floor where construction permits, fed from directional couplers tapped off a riser trunk rather than a cascade of splitters, with tap values staged so every antenna port lands within ±3 dB of the others. A donor or rooftop antenna is then only needed if you are working off-air from an external system rather than an on-site repeater.

Related questions

Does a rooftop antenna still make sense for anything?

Yes — whenever the coverage target is horizontal rather than vertical. Port yards, construction sites before the structure rises, multi-building campuses and detached warehouses are all rooftop cases, and there the high-gain omni that fails in a tower is exactly the right choice.

Can a mid-level repeater cover a basement carpark?

Almost never. The basement sits behind the ground slab plus a transfer structure, and the loss through an RC core or shaft is 25–35 dB on its own. Basements get their own antennas or a radiating cable, regardless of where the repeater lives.

Does putting the antenna in the lift shaft solve the vertical problem?

It helps, because the shaft is one of the few continuous vertical paths in the building, but it is not a substitute for floor antennas — coverage leaks out at the lobbies and dies a few metres into the floor plate. It also brings a lift isolation permit and shaft access into the programme.

How Suneast approaches this

We do not answer the rooftop-or-mid-level question from a plan. We measure: a signal-level survey on a grid, taken in both directions, so the uplink and downlink are separated before an architecture is chosen. If the slab arithmetic clears on a low-rise building, a single well-placed antenna is the cheaper answer and we will say so. Where it does not, we design a distributed antenna system off the ELV risers with a full coupler and tap schedule, so the contractor can build it and the client's consultant can recompute every figure in it. All repeater installations in Singapore require an IMDA licence and type-approved equipment, which we handle as part of the works.