Heat map was green, the room is dead
common
“The design said full coverage. Stand in that corner office and you have got nothing.”
Likely causes
-
Predictive model used the wrong wall losses
30%Model assumed drywall and the wall is block, foil-backed insulation, or low-emissivity glass. Take a reading 1 m either side of it: 15-25 dB across one wall instead of 3-5 dB. The signal gradient through the space is a cliff, not a slope.
-
AP is not where the design put it
20%Compare design drawing to as-built. The AP ended up above a hard lid, inside a metal ceiling grid, above ductwork, or 10 m off plan. Coverage is skewed away from the design in one consistent direction.
-
Wrong AP or antenna for the mount
15%A ceiling-pattern AP screwed to a wall, or an AP hidden above a solid tile, so the main lobe is aimed into the slab. Strong at distance, weak underneath.
-
The survey was RSSI-only
15%Report has a signal-strength layer and nothing else: no SNR, no data rate, no capacity, no second-AP overlap. The dead room measures -70 dBm but the noise floor is -80 dBm, so SNR is 10 dB and nothing works.
-
Survey adapter far better than the real clients
10%Survey done on a laptop with external antennas; the users are on handhelds, scanners or badge phones at one spatial stream and lower gain. Everything measures 8-12 dB worse with the real client in hand.
-
6 GHz expectation gap
10%5 GHz works in the same spot, 6 GHz does not. Indoor low-power 6 GHz is power-limited by rule and loses more through building materials, and the client transmits lower still. Check the SSID is on a preferred scanning channel (a client that only scans those will never find a 6 GHz BSSID parked elsewhere) and that the client can meet the mandatory 6 GHz security, because one that cannot will never appear on the band (see Some devices work, some do not: WPA3 and frame protection). Standard power needs AFC. Get the numbers from the AP's regulatory table for that country, not from a forum post.
What to bring
- Survey tool with active and passive modes
- the actual client device in use on site
- hand-held spectrum analyzer
- design drawings and the as-built
- spare AP, PoE injector and a pole or tripod
- ladder or lift
Safety
Ladder and overhead work for every tile you lift: ladder and fall rules apply (29 CFR 1926.1053, 1926.501). Before disturbing anything above the grid in a pre-1990 building, confirm the ceiling tiles, sprayed fireproofing and pipe lagging are not asbestos-containing. If nobody can produce a survey, stop and ask; do not break the tile to find out.
Steps
-
Step 1: Measure the hole properly, with the client type in use
RSSI, noise floor, SNR and real throughput at the complaint spot, both directions. Common working targets: about -67 dBm with 25 dB SNR for data and voice, better for handhelds and scanners. Write the numbers down: this is the before shot.
If that doesn’t do it
It measures fine where you stand: get the user to point at the exact spot and posture, including inside a lift lobby, a stairwell or a metal-clad cool room.
-
Step 2: Compare as-built to design, with the tile off
Physically find the nearest AP. Above a hard lid, inside a metal enclosure, behind ductwork or wrapped in foil-faced insulation will each cost you 10-20 dB on its own. While you are in the controller, check that AP's country code and channel list match its neighbours: a mis-set regulatory domain reads exactly like a coverage fault (AP will not join, or joins in the wrong country).
If that doesn’t do it
AP is exactly on plan and in clear air: the model's assumptions are the problem, go to step 3.
-
Step 3: Walk the boundary and find the offending wall
Readings 1 m either side of every wall between the AP and the dead spot. The wall with the 15 dB or greater step is the one the predictive model got wrong. That is also the number you put in the report.
If that doesn’t do it
No single big step, just a long slow decay: the AP is simply too far away. Go to step 5.
-
Step 4: Spend the free moves first
Reposition or re-aim the AP, drop it below the tile, get it out of the metal, correct the mounting plane. Re-measure the same points after each change before you propose hardware or cable.
If that doesn’t do it
Best achievable reading is still under -70 dBm or SNR under 20 dB at the complaint spot: you need another AP. Validate first, step 5.
-
Step 5: Put an AP on a stick
Temporary AP at the candidate location, same model, same antenna, same power, same channel as the proposal, and survey the room with the real client. This is what turns an argument into a defensible answer before anyone pulls cable.
If that doesn’t do it
AP on a stick still cannot cover the space: the room needs a directional antenna or a dedicated AP inside it, not a better neighbour.
-
Step 6: Then pull the drop and re-validate
Cable to the validated location, terminate and certify to the current published edition of the TIA cabling standards your spec references, and re-survey both directions after cutover with the same tool and the same spots as your before shot.
If that doesn’t do it
New AP in and still failing: recheck for a raised noise floor at the new location, and for co-channel with what you just added.
References
- ANSI/TIA-568.1 series (current edition)
- TIA TSB-162-A
- 47 CFR 15.407