Everything here comes out of one subtraction
A radio link works when what arrives at the far end is stronger than what the receiver needs. Start with the transmit power in dBm, subtract everything the signal loses on the way, and compare what is left against the level you wanted. That difference is the link budget, and every figure on this page is a consequence of it.
Losses come in two kinds. Free space loss happens continuously with distance and rises with frequency: doubling the distance costs about 6 dB, and so does doubling the frequency. Obstruction loss happens in lumps whenever the path crosses something, and a wall costs the same handful of dB whether it sits at ten feet or fifty. The calculator counts walls by dividing the distance by your average wall spacing, which is crude but honest, and it is the part you should tune to your own building.
Why the band changes the answer so much
Higher frequency means shorter range and worse penetration. Both halves matter and people usually only hear the first. The free space term rises by 6 dB every time the frequency doubles, which on its own is a modest cost. What is less forgiving is that shorter wavelengths are absorbed and reflected more by ordinary building materials, so the same partition takes a bigger bite out of a 6 GHz signal than a 2.4 GHz one. That is why the loss per wall in the field above should not be the same number when you change the band.
The practical consequence is the complaint that shows up every time somebody upgrades. The new radio is genuinely faster, measurably so, standing next to it. Two rooms away it is worse than what it replaced, because the extra data rate was bought with spectrum that does not travel as well. Neither device is faulty. They are answering different questions.
| Band | Reach through the same walls | What it is good for |
|---|---|---|
| 2.4 GHz | Longest | Far corners, sheds, sensors, anything slow that has to work everywhere |
| 5 GHz | Middle | The general purpose band for most of a house |
| 6 GHz | Shortest | Rooms with a radio in them, and clients that can use wide channels |
Notice that none of those rows is a recommendation to disable a band. A house usually wants the low band available for reach and the high band available for rate, with the client choosing. Turning off the low band to force devices upward is a common and generally regretted move.
Measuring the wall loss instead of guessing it
The loss per wall field is the input that most changes the output, and it is also the one you can measure in ten minutes without buying anything. Put the radio in a fixed place. Stand a set distance from it in clear line of sight and note the signal level your phone or laptop reports. Move to the same distance but with one wall in the way and note it again. The difference is your wall, in dB, for that band.
Do it for the worst wall in the building rather than the friendliest. Plasterboard on studs is mild. Masonry, tiled bathrooms, plaster on metal lath, foil-faced insulation board and anything with a mirror on it are severe, and a single one of those can matter more than the whole distance term. Mechanical rooms, water tanks and stacks of books are all wall-shaped as far as a radio is concerned.
Where the access points should go
The calculator lays cells out as hexagons because that is the arrangement that covers a plane with the fewest circles and the least waste. What follows from it is that the spacing figure is centre to centre in a staggered pattern, not a row down a corridor. Two radios in a straight line along one wall overlap heavily on one side and leave the other side uncovered.
Height and clutter matter as much as position. A radio on the floor behind a television is radiating into furniture and a metal chassis; the same radio high on a wall or ceiling in the open has a much cleaner path to everything. If a unit has to live in a cupboard or a media cabinet, treat that cabinet as another wall and add its loss to the budget.
Vertical coverage deserves its own thought rather than being treated as a bonus. Coverage area goes with the square of the radius, so a floor assembly that costs a modest number of dB can remove most of the area upstairs. Two modest radios, one per storey, beat one strong radio in the middle in almost every building that is not open plan on both levels.
What a coverage figure cannot tell you
Coverage is not capacity. A client can sit at a perfectly healthy signal level and still be slow because thirty other devices are sharing the same airtime, or because the channel is shared with four neighbours. Those are different problems with different fixes, worked through in the Wi-Fi channel width planner and, on the wired side, the household bandwidth calculator. If you are not yet sure whether the wireless or the internet connection is the slow part, the Wi-Fi improvement guide starts with the test that separates them.
The model here is also deliberately simple. Real propagation includes reflection, waveguiding down corridors, constructive and destructive interference, and antenna patterns that are nothing like a sphere. A measured survey with the actual hardware in the actual building beats any calculation, including this one. What the calculation is for is deciding roughly how many radios to buy and roughly where to run cable to before you own any of it, which is a question a survey cannot answer because there is nothing yet to survey.
Questions people ask
How many access points does a typical house need?
There is no typical house, which is exactly why this page asks for the loss per wall rather than assuming one. The same square footage can want one radio or four depending on whether the partitions are plasterboard or brick, whether there is a tiled bathroom in the middle of the plan, and how many storeys the signal has to cross. Run the numbers with a wall loss you measured rather than one you guessed, then treat the answer as a starting point for a walk around with a signal meter. The reliable pattern in multi-storey buildings is at least one radio per storey, because the floor assembly costs more than people expect and coverage area falls with the square of the radius.
Why did my new router make things worse in the back bedroom?
Almost certainly because it moved your devices onto a higher band. Higher frequency carries more data and travels less far, and it loses more in every wall on the way. Standing next to the new unit the improvement is real and measurable. At the far end of the house the client is now trying to use spectrum that never reached there. The fix is not more transmit power, which is capped and which does nothing for the return path from the phone anyway. It is either letting the far devices fall back to the lower band, or putting a second radio nearer to them.
Is a mesh node as good as a wired access point?
A radio you feed with a cable is unambiguously better than one that has to relay over the air, because a wireless relay spends airtime receiving and then spends it again retransmitting. Whether that penalty matters depends on how much spare airtime there was to begin with. A dedicated backhaul radio reduces the cost considerably; a single-radio repeater does not, and roughly halves throughput on that hop. If a cable can reach the location, run one. This page is agnostic about which it is, because the coverage arithmetic is identical either way; only the capacity behind each radio differs.
What signal level should I be designing for?
The default here is -67 dBm, which is a widely used target for a link that has to carry video reliably, and it is a design target rather than a threshold. Devices associate and pass traffic at far weaker levels, they just do it slowly and with more retries. If you set the target to -75 the calculated radius grows noticeably, and everything inside that larger circle will connect; a good part of it will connect badly. The honest way to use the field is to run it twice, once at a strict target and once at a loose one, and treat the ring between the two as the area that works until somebody starts a video call in it.
Does adding an antenna or a stronger transmitter extend the range?
Less than the marketing suggests, for a reason that is easy to miss: the link has two ends. Making the access point shout further does not make the phone in the far bedroom shout further back, and the phone has a small antenna and a battery to protect. A link that is only strong in one direction shows the familiar symptom of full bars and no throughput. Directional antennas genuinely help when the coverage you want is a shape, such as a long narrow building or a run out to a workshop, because they trade coverage you did not need for coverage you did. Raw power increases are also limited by regulation, and what is permitted differs by country and by band.