A Zigbee sensor does not know that your house is inconvenient. It only knows whether the next radio in its path can hear it. When a motion sensor works perfectly on the kitchen counter but misses events in the garage, the instinct is to blame ZHA, Zigbee2MQTT, or the sensor. I start with placement instead.
This is the room-by-room method I use to build a boring Zigbee network in Home Assistant. It applies to both ZHA and Zigbee2MQTT, and it complements my Zigbee device roundup and channel-planning guide. The goal is not the prettiest map. The goal is an event arriving every time.
Start With the Coordinator
The coordinator is the one placement decision you cannot fix with a later route. Put it near the middle of the occupied part of the house, not necessarily beside the Home Assistant server. A central first-floor location is often better than a basement utility closet at one end of the floor plan.
Give the radio air around it. Avoid metal network racks, electrical panels, refrigerators, boilers, and the back of a desktop computer. If the coordinator has to live in a rack, put it on an extension cable outside the rack rather than treating the rack as an antenna enclosure.
USB 3 interference is the most common self-inflicted problem I see. A USB 2.0 extension cable moves the dongle away from the host, SSDs, and hubs. I would do this before changing channels or re-pairing a single sensor.
For a new installation, a modern coordinator such as the SONOFF Dongle Plus-E (Amazon fallback) is a sensible baseline. The software stack matters, but a current radio in a quiet location matters first.
Design the Mesh, Do Not Scatter Sensors
Battery devices are end devices. They sleep to preserve power and usually do not forward traffic. Mains-powered plugs, bulbs, and wired switches are routers. That distinction explains why buying ten sensors does not necessarily improve coverage, while adding two strategically placed plugs can transform it.
I place routers at transitions: near the stairs, on the other side of a thick wall, at the entrance to an addition, and before an attached garage. Think of them as stepping stones. You want overlapping paths, not one heroic hop from the coordinator to the farthest room.
A Zigbee smart plug is useful because it is powered continuously and easy to move. Do not buy a plug solely because a listing says repeater. Confirm that the exact model is supported by your chosen stack and that its routing behavior is known to be good.
More routers are not automatically better. Cheap bulbs can be poor routers, and some devices behave badly with sleepy Aqara sensors. Add known-good routers one at a time. Let the network settle, then test the room you were trying to improve.
Walls and Floors Are Radio Decisions
A floor plan is more useful than a device list. Mark the coordinator, every powered router, and each battery sensor. Then mark the obstacles: concrete, brick, foil-backed insulation, water tanks, ductwork, and appliances. Zigbee at 2.4 GHz passes through ordinary drywall reasonably well, but dense and metallic materials are very different.
Try to cross obstacles at an angle or through a doorway rather than straight through a utility stack. A router in a hallway can be more useful than one inside the room with the weak sensor because it creates a path around the barrier.
Detached structures deserve their own plan. Do not assume a Zigbee signal will cross an exterior wall, a driveway, and another wall reliably. Put a router near the house-side opening and another in the outbuilding if power is available. If the distance is genuinely too large, use a network-connected coordinator or another technology rather than forcing a marginal mesh.
Pair Devices in Their Final Homes
Pairing on the desk and carrying a device across the house gives you a misleading success. The device chooses parents and routes based on what it can hear during commissioning. I pair fixed devices in place, or at least within a few feet of their final location.
For a battery sensor, wake it after joining and trigger a real report. Then leave it alone long enough to select a route. Moving every sensor immediately after pairing creates a network that looks healthy in the UI but has never had to operate under its final RF conditions.
Powered routers also need time. After adding one, wait several minutes before judging the map. End devices may continue using an old parent until they check in, and different integrations expose route information differently. A single link-quality number is not a verdict.
Test the Event Path
I test coverage with the automation that matters, not a ping-shaped dashboard card. Open the door, walk past the motion sensor, press the remote, or create the leak test event. Confirm the timestamp in Home Assistant and check that the downstream automation ran.
Test from the room where the device lives. A sensor that reports beside the coordinator proves only that the sensor and coordinator can communicate. Walk the route between the device and the nearest router. If the event succeeds near the router but fails in the final position, placement is still the problem.
Repeat at different times. WiFi traffic, televisions, USB devices, and household activity can expose intermittent interference. Leave a questionable sensor overnight before deciding that a move worked. Reliability means repeated reports, not one successful test.
Troubleshoot in the Right Order
When one device fails, start local. Check its battery, orientation, and distance from metal. Move the nearest router before touching the coordinator. If several devices in one area fail, inspect the shared wall, floor, and route. If the whole network fails, inspect the coordinator cable, host USB environment, power, and channel.
My order is deliberately conservative:
- Replace or test the battery and wake the device.
- Check coordinator placement and add a USB 2.0 extension.
- Add or move one known-good powered router.
- Wait for routes to settle and test real events.
- Check the channel against nearby 2.4 GHz WiFi.
- Only then consider re-pairing, firmware work, or changing the Zigbee stack.
This avoids turning a small coverage problem into a whole-house migration. My ZHA versus Zigbee2MQTT comparison explains why switching stacks is rarely the first fix.
A Practical Placement Checklist
Before I call a network finished, I walk this checklist with the floor plan in hand. The coordinator is on a USB extension, its indicator is not trapped behind metal, and it is not sharing a shelf with an access point or an active USB 3 drive. At least one powered router sits between the coordinator and each difficult area. A router is not hidden behind a refrigerator, inside a panel, or at the far edge of the signal it is supposed to extend.
Every fixed device is paired in its final location. Battery sensors have fresh batteries and have been woken after the mesh had time to settle. I have tested the actual event from each critical room, including the garage, basement, and exterior doors. The automation history shows the event arriving, not merely a device card showing as available.
I also record the boring details: coordinator model, integration, channel, device count, and the location of each router. That record saves time six months later when a new appliance changes the radio environment. If I need to change the channel or replace the coordinator, I can change one thing and know what the old system looked like.
The best Zigbee installation is the one that does not require constant map watching. Build the physical path first, then let ZHA or Zigbee2MQTT do its job.
When to Stop Adding Hardware
Once events are reliable, stop optimizing the map. A mesh with a few strong routes is easier to understand than one packed with routers that compete for parents. I would rather have a sensor report correctly for a week than see every room painted green immediately after an installation.
If a problem returns seasonally, look for a physical change: a new refrigerator, a moved access point, a metal storage shelf, or a coordinator cable routed beside a power supply. The house changes, so the RF environment changes. A small placement correction is usually safer than replacing every device.
The Bottom Line
Reliable Zigbee coverage is mostly a placement problem. Put the coordinator in a quiet central location, separate it from USB 3 noise, and build a chain of known-good powered routers across floors, thick walls, and outbuildings. Pair sensors where they will live and verify real Home Assistant events over time.
Do not chase a perfect mesh diagram or add hardware at random. Change one physical variable, wait for the network to settle, and test the automation your family depends on. A modest, deliberately placed network beats a large pile of sensors with no route plan.