ZigBeeForDomoticZ plugin Wiki

Troubleshooting - A device shows a low LQI: a worked case

This page uses the device details window of Management > Device Management, available from plugin version 9.1.009, and the Network > Mesh graph page, available from version 9.1.008. With an older version, update the plugin first (see Plugin update).

This page follows one real investigation from start to finish. Two radiator thermostats showed an LQI of 25-30, and the question was: are they routed through bad routers?

The short answer turned out to be no. Both thermostats had a good link to their parent router. The low number described a different link: the one between that router and the coordinator. The steps below show how to reach that conclusion on your own network, and what to check before you move or replace anything.

The network is a production ZNP network (CC2652, Z-Stack 3.30+) of 62 devices, the same one used for the screenshots of the Mesh graph page. The device names are the ones its owner gave them. The topology figures come from three scans taken in October 2026. The device LQIs and the plugin log are from 1 November 2026.

Contents:


The symptom

In Management > Device Management (see Device Management), two battery thermostats showed a red LQI bar. Their device details gave the last 10 readings:

Device Model Latest LQI Last 10 readings, oldest first
BT Nico #2 (a7b1) Tuya TS0601 thermostatic valve 29 36, 32, 32, 32, 32, 32, 32, 32, 32, 29
BT Sdb #2 (59b6) Danfoss eTRV0100 25 25, ten times

Other thermostats of the same house were at 60-150. The obvious reading is “these two are badly placed, or they hang off a bad router”. The first part of that reading is wrong.

What the LQI of a device measures

The LQI of the device list and of the device details is measured by the coordinator’s radio, on each frame it receives from the device. A frame from a device that is not a direct neighbour of the coordinator arrives through one or more routers, and the coordinator can only measure the last hop: the link between the last router and itself.

The LQI of a device is the quality of the last radio hop into the coordinator. It says nothing about the link between the device and its parent, unless the device talks to the coordinator directly.

What the LQI and RSSI measure explains why, and how the plugin files the value. For this case, the consequence is simple: a battery device behind a router that the coordinator hears badly shows a low LQI, however good its own link is. To find the weak link, you need the neighbour tables of a topology scan, which is what the Mesh graph shows.

Step 1 - Find the parent of the device

Run a topology scan (see How a scan is performed) and click the device in the Mesh graph. In the node panel, Attached to names the router that declares the device as its own. The neighbour table rows show how well that router hears it.

In this case, the same in all three scans:

Device Parent (Attached to) Parent hears the device at
BT Nico #2 Volet Fifi (80da, Profalux shutter) 255
BT Sdb #2 Switch Buanderie (ea10, Tuya TS0001) 97-112

Both thermostats are well heard by their parent. Neither of them has a radio problem of its own.

A battery device is not a router and has no neighbour table, so its own reading of the link is not in the scan; only the parent’s is. See Battery devices are expected to be missing from the link views.

Now click the parent, and find the row in which the coordinator hears it. Open the parent’s device details too: as a router, its own LQI is often its direct link with the coordinator.

Router Coordinator hears it at Its LQI in the device details (last 10)
Volet Fifi 35, 38, 29 (three scans) 29-43
Switch Buanderie 17 (last scan) 14-29

This is the match:

So the weak links are real, but they are router → coordinator links, not thermostat → router links.

Step 3 - Read both directions, and know whose scale you are reading

The Mesh graph keeps the reading of each end of a link (see Direction, and why both are drawn as one link). Click the link between the coordinator and Volet Fifi, and the two readings disagree completely:

Router It hears the coordinator at The coordinator hears it at
Volet Fifi (Profalux) 254 29
Volet Nico (Profalux) 254 21
Volet Cuisine (Profalux) 255 20
Volet Parent (Profalux) 254 23
Prise Aspirateur (Computime SPE600) 254 17
Switch Buanderie (Tuya TS0001) 26 17

The Profalux shutters and the Computime plug report 245-255 for nearly every neighbour they have, the coordinator included: the median of what each of them reports is between 245 and 254. The Philips and IKEA bulbs, the frient plugs and the Tuya switch spread their readings over the whole range, as one would expect.

This is not a link that is excellent one way and poor the other. Each radio maker computes the LQI its own way (see What the LQI and RSSI measure), and some radios, Silicon Labs ones in particular, report close to 255 for any link they can decode at all. Volet Nico’s IEEE address starts with 00:0D:6F, a Silicon Labs prefix. A reading of 254 from such a radio means “I can decode this”, not “this link is excellent”.

What to take from it:

Step 4 - Look at what the coordinator hears overall

Click the coordinator and sort its neighbour table rows on LQI. In this case, the coordinator listed 23 routers as neighbours:

The coordinator hears them at Routers
above 100 4: Lampe Entree 199, Lampe Salon 148, General Counter 132, Volet Salon 109
45 to 100 8
under 45 11, including Volet Bureau at 6 and Prise Bureau Info at 1

Most of the mesh reaches the coordinator through a weak last hop. That is why many devices, not only the two thermostats, show an orange or red LQI in Device Management. The weak point is the coordinator’s reception, not one router.

Step 5 - Check whether it actually hurts

A low LQI is a reason to look, not proof of a failure. Before changing anything, check how the device behaves:

Here, both thermostats worked normally at an LQI of 25-30. On this ZNP coordinator, that is a weak but working link. The colour bands are a rough guide, not a verdict: Device Management shows red under 40, and the Mesh graph bands everything up to 100 as Poor (see A note on LQI).

A ZNP coordinator reports no RSSI. On a Silicon Labs (EZSP) or deCONZ coordinator, the RSSI chart of the device details is worth a look too: being in dBm, it compares better between radios than the LQI.

What it came to, and what to do

The actions, cheapest first:

  1. Improve the coordinator’s placement: a shielded USB extension cable, a USB 2.0 port, away from the host computer, USB 3.0 devices and the Wi-Fi access point. See Build a Zigbee mesh network.
  2. Add a good router near the coordinator, preferably one whose LQI readings spread over the range, so the shutters and plugs get a strong link into the coordinator.
  3. If one area stays weak (here, the Fifi and Nico rooms), add a router between that area and the coordinator.
  4. Run a new scan after each change, and compare what the coordinator hears from each router, rather than the LQIs of the devices.

Do not re-pair the thermostats or replace their parent routers: neither changes the router → coordinator link.

Checklist

When a device shows a low LQI:

  1. Remember that it is the last hop into the coordinator, not the device’s own link.
  2. Find the device’s parent (Mesh graph, node panel, Attached to), and read how well the parent hears it.
  3. Read how well the coordinator hears that parent. If it matches the device’s LQI, the weak link is parent → coordinator.
  4. Read both directions of each link. Set aside the readings of a router that reports almost everything at 250+.
  5. Sort the coordinator’s neighbours by LQI. If most are low, work on the coordinator, not on the device.
  6. Check Health and the log before acting. A low but stable LQI with no delivery errors is a working link.

See also