Mesh Systems, Extenders and Dead Zones

Why this matters

Connected equipment lives where people do not: the attic air handler, the basement boiler, the crawlspace leak sensor, the garage opener, the pool pad. Those are the weakest-signal spots in nearly every house, and when a device there drops, the customer's next question is "what do I buy to fix it?" The tech who answers "an extender" without understanding how it works often makes the device worse, not better. This card explains the three ways to push Wi-Fi into a dead zone, what each one actually does to the signal, where to put it, and which options to leave off the list.

Why equipment rooms are dead zones

A dead zone is any spot where the signal is too weak or too unstable for a device to stay connected, in practice weaker than about -75 dBm on the band the device uses (the measuring sibling owns the thresholds). Equipment rooms get there for predictable reasons:

• They are far from the router, which sits where the internet line enters, usually a living area.
• They are behind the heaviest materials in the house: a slab floor, a fire-rated garage wall and steel door, a foundation wall, foil-faced attic insulation, ductwork, the equipment's own sheet-metal cabinet.
• They are on a different floor. Signal travels poorly straight up or down through floor assemblies, and attics and basements are by definition a floor away.
• They hold water and metal: tanks, pipes, appliances, all of which absorb or reflect 2.4 GHz energy.

The equipment's radio is usually 2.4 GHz only and has a small antenna inside a metal box. So the worst spot in the house hosts the weakest radio in the house. Fixing that means getting a strong access point close to the device, and there are three ways to do it.

Option one: the extender (repeater)

An extender, also called a repeater, listens to the router's signal and rebroadcasts it. It is the cheapest fix and the one most likely to disappoint, for three reasons:

• It needs a strong signal to start with. An extender can only repeat what it hears. Placed in the dead zone itself, it hears a weak, noisy signal and rebroadcasts a weak, noisy signal with more power. The useful placement is roughly halfway, at a spot where it still hears the router well (around -60 to -65 dBm) and is close enough to the device to give it a strong signal.
• A single-radio extender halves its own throughput, because it must receive and retransmit on the same radio, taking turns. For a thermostat that sends a few hundred bytes a minute, that does not matter. For a camera streaming video, it often does.
• It often creates a second network name, such as the home name with "EXT" appended. A device that joins it stays on it even when the router would be better, and a device that joins the main name may never move to the extender. Either way the customer now has two networks and does not know which one each device uses.

An extender is a reasonable answer for one low-bandwidth device in one predictable spot, placed correctly. It is a poor answer for a whole-house coverage problem.

Option two: a mesh system

A mesh system is a set of access points, called nodes, sold to work together. One node connects to the modem and replaces or sits behind the router; the others are placed around the house and talk to each other. Compared to an extender:

• One network name across the house, managed centrally, with devices handed from node to node as they move (a process called roaming, covered by the IEEE 802.11 Wi-Fi standards). Fixed equipment does not roam, which matters below.
• Many mesh systems use a dedicated radio for node-to-node traffic, called the backhaul, so the halving penalty mostly disappears.
• Nodes still need to hear each other well. The halfway rule applies exactly as it does to extenders. A node pushed into the garage to be next to the opener, hearing the next node at -78 dBm, is a node with a bad link feeding a device that inherits it.

The sticky-client trap. A fixed device joins whichever node it hears best at the moment it connects. If it was set up while a far node was briefly the strongest, or it joined before the near node was installed, it can stay locked to the far node at -76 dBm while a node twenty feet away reads -50. This is called a sticky client. Many mesh apps show which node each device is connected to, and let the owner nudge it. Power-cycling the device after the mesh is in place usually makes it rejoin the closest node.

Option three: wired backhaul

The strongest option is to run a cable. A wired backhaul connects a mesh node, or a separate access point, to the router by an Ethernet cable rather than by radio. The node then broadcasts a full-strength signal from wherever the cable lands, regardless of the walls between it and the router.

• Ethernet is the standard choice. A run of one cable to the garage, the basement or the hallway outside the utility room solves the problem permanently. Cable types and termination are owned by their own siblings.
• Existing coaxial cable, the round TV cable already in many walls, can carry network traffic using adapters built to a standard for that purpose. Where an unused coax jack sits near the dead zone, this is often the least invasive wired option. It does share the cable with any TV or internet service on it, so the provider's equipment has to be considered.
• Powerline adapters, which send data over the house electrical wiring, work in some houses and fail in others, depending on whether both outlets are on the same electrical phase, how old the wiring is, and what noisy loads share the circuit. Treat them as a "try it, measure it" option rather than a recommendation.

Wired backhaul costs the most labor and wins every time it is practical. In a new build or a remodel with open walls, it is the default answer.

A worked example: the attic air handler

A two-story house, router on the ground floor in a front office. The air handler for the upstairs zone is in the attic above the back bedrooms, with a Wi-Fi thermostat on the upstairs landing and a Wi-Fi module on the air handler itself for the manufacturer's diagnostic service. The module keeps dropping.

Readings on 2.4 GHz with a phone signal app (illustrative of a typical case):

• Front office, next to the router: -42 dBm.
• Upstairs landing, by the thermostat: -63 dBm. Solid. The thermostat has no complaint.
• Back bedroom, under the attic hatch: -70 dBm.
• In the attic at the air handler, with the phone held against the module's access panel: -81 dBm. Unreliable.

The attic reading is taken standing on the platform boards or joists, never on bare drywall between them, and in summer before mid-morning, because attic air over a hot afternoon can bring on heat illness within the hour. From the bedroom to the attic, the signal fell 11 dB (from -70 to -81) through one ceiling assembly with foil-faced insulation over it. That one layer cost about half of what the whole trip from the office to the landing did (21 dB, from -42 to -63), a trip that crossed a floor and two rooms.

The tempting fix: put an extender in the attic. In the attic it would hear the router at about -81 dBm and repeat that weak signal, so the module would see a strong signal from a starved repeater and drop just as often.

The right placement for a radio fix: the back bedroom, high on a wall, reads -70 dBm from the router, which is weaker than an extender or node wants to hear (about -60 to -65). The upstairs landing reads -63 dBm. A node on the landing hears the router acceptably and is one room and one ceiling closer to the attic. Measured with a node temporarily set on the landing shelf, the attic reading at the module improved from -81 to -72 dBm, from unreliable to marginal.

The better fix: the homeowner's network had an Ethernet run to the upstairs office for a desktop computer. A wired node in that office, which is directly below the air handler, put the module at -61 dBm, solid, with no radio hop between node and router at all.

What would have changed the answer. If the attic had no foil-faced insulation, the radio-only node on the landing might have reached solid on its own. If there were no wired path upstairs and the owner would not pay for one, the landing node at -72 dBm is a defensible marginal result, recorded as marginal, with the owner told the module may drop in summer when attic heat stresses the electronics. If the manufacturer offered an external antenna kit for the module, mounting the antenna below the attic insulation line could recover several decibels without any network equipment at all.

What deliberately stays off the list

Some fixes sound reasonable and should not be offered by a tradesperson:

• Installing a node or extender you supplied without the owner's explicit consent. You have changed their network, and when it breaks something else at night, it is yours. That boundary is owned by its own article.
• Raising the router's transmit power or swapping its antennas. The device still has to transmit back, with its own small radio, so making the router louder fixes one half of a two-way link.
• Moving the router into a closet or cabinet near the equipment. It moves the dead zone to the rooms people live in.
• Recommending an internet plan upgrade. Internet speed has nothing to do with signal strength inside the house.

How to verify the fix held

After any coverage change, measure at the device again, on the device's band, with any cabinet doors closed as they will be in service. Confirm the device is connected to the node you intended, if the mesh app shows it. Then power-cycle the device and confirm it rejoins the same node. A reading taken with the attic hatch open and the panel off is a reading of a house that will not exist once you leave.

References

• IEEE 802.11 family of standards (Wi-Fi radio behavior, roaming)
• Mesh system and access point manufacturer documentation (backhaul options, node placement guidance)
• See related: Home Wi-Fi for the Trades; Measuring Wi-Fi Signal Where the Equipment Lives; Ethernet Cable Categories and Terminations
• See related: The Customer Network Is Not Your Job Until It Is; The Customer Network Change Authorization SOP