Overview
According to the Zigbee specification and certification requirement, a direct transmission distance ranging from 10 to 100 meters line-of-sight depending on power output and environmental characteristics, between a Zigbee Coordinator (call it Zigbee gateway, Zigbee hub, Zigbee dongle, Zigbee stick, etc.) and a Zigbee device (Zigbee router device, Zigbee end device) is acceptable.
But we want to know the transmission distance limit of SONOFF Zigbee Dongle series. And we also want to verify that the LQI (Link Quality Index) has no significant correlation with operational performance in daily use.
Note: regarding LQI, it is misleading and has been replaced by LQA (Link Quality Assessment) in Zigbee Specification Core R23. Different Zigbee chip vendors use different algorithms to calculate LQI. This is even true for their product variants. Zigbee2MQTT has clearly stated Unless you are a Zigbee specialist yourself or are guided by one, please ignore those values.
At the same time, to provide users with a more intuitive understanding of the communication capabilities of the SONOFF Dongle series, we organized a distance test and fully documented the test methodology and raw data. This approach ensures that the results are reproducible and verifiable, while presenting the communication performance of different Zigbee Dongle models across varying distances.
After a series of tests, we reached the following conclusions:
- All SONOFF Zigbee Dongle series are capable of achieving a 350-meter control range in open environments.
- The Dongle Max is able to control devices at distances of up to 450 meters. (Due to site limitations, distances beyond 450 meters were not tested.)
- Even when the LQI value falls below 90, 60, or even into single-digit levels, device control can still be achieved with a 100% success rate, while maintaining fast response times with latency under 100 ms.
This test served as an internal "exam" for our own devices, and we aim to present both the process and the results of this evaluation transparently to you.
Test Plan Overview
Before introducing the test setup, we defined several core testing principles:
-
Real products:
All devices used in this test are brand-new retail units, identical to those available on the SONOFF official website, Amazon, and other marketplaces. -
Real data:
All results presented are based on actual recorded measurements during the test. -
Reproducibility:
Under the same conditions, our test methodology should yield comparable results.
Test Equipment
|
Dongle & Zigbee Sub-device |
|||
|
Model |
Product Name |
Firmware Type |
Firmware Version |
|
SONOFF Zigbee 3.0 USB Dongle Plus V2 |
Zigbee NCP |
7.4.4 |
|
|
SONOFF Dongle Lite MG21 |
Zigbee NCP |
7.4.5 |
|
|
SONOFF Dongle Plus MG24 |
Zigbee NCP |
7.4.5 |
|
|
SONOFF Dongle Max MG24 |
Zigbee NCP |
7.4.5 |
|
|
Zigbee Smart Switch (Neutral Wire Required) |
Official Firmware |
1.0.4 |
|

|
Additional Test Equipment |
|
|
Device Type |
Details |
|
Router |
Xiaomi Router 4A Gigabit Edition |
|
Raspberry Pi |
Running Home Assistant with Zigbee2MQTT |
|
Power Banks |
Two units: one powering the router, Raspberry Pi, and laptop; the other powering a ZBMINIR2-modified downlight |
|
Downlight |
Converted into a smart downlight using ZBMINIR2, used for visual confirmation of on/off status |

Test Locations
This test was conducted using two defined points:
-
Starting Point
The router, Raspberry Pi, laptop, and the Zigbee dongle under test were placed at this location.
The dongle was mounted on a stand at a height of 1.2 m, and its position remained unchanged throughout the test. -
End Point
A downlight modified with ZBMINIR2, also fixed at a height of 1.2 m.
The end point was gradually moved to distances of 150 m / 250 m / 350 m / 450 m.

Test Procedure
At each distance point, control tests were performed using a unified process:
-
Channel configuration
Zigbee channel: 26
Wi-Fi channel: 1 -
Control method
Remote device control via MQTT commands -
Control frequency
One command every 2 seconds -
Number of attempts
30 consecutive control attempts at each distance
During testing, we simultaneously recorded:
- Control success and failure
- Control latency (ms)
- Zigbee link quality indicator (LQI)
Note:
Each time a different Dongle was tested, Zigbee2MQTT was fully reset by removing the existing configuration and reinitializing the system before starting the next test.
Real Testing
*For a detailed view of the distance test, please refer to the video published on the SONOFF YouTube channel, which documents the testing process in greater detail.
On February 4, 2026, the SONOFF test team traveled from Shenzhen, China (the headquarters of SONOFF) to Huizhou, China (the location of the SONOFF manufacturing facility). The tests were conducted on an open, straight road, in accordance with the test plan described above.

With the starting point (Dongle) and the endpoint (ZBMINIR2) positioned at the same height of 1.2 meters, the test distance was gradually increased. At each distance interval, we evaluated the performance of ZBDongle-E / Dongle-LMG21 / Dongle-PMG24 / Dongle-M, focusing on device pairing and device control behavior under varying ranges.

After a full day of testing, we obtained a set of reliable and repeatable results, forming the basis for further analysis.
Data Analysis
|
Dongle |
Distance (meter) |
Zigbee Device |
Control Methods |
Control Indicators |
Average Value |
|
ZBDongle-E |
150 |
ZBMINIR2 |
MQTT |
LQI |
96.53 |
|
Control Latency (ms) |
58.33 |
||||
|
250 |
ZBMINIR2 |
MQTT |
LQI |
59.20 |
|
|
Control Latency (ms) |
61.93 |
||||
|
350 |
ZBMINIR2 |
MQTT |
LQI |
59.73 |
|
|
Control Latency (ms) |
70.40 |
||||
|
Dongle-LMG21 |
150 |
ZBMINIR2 |
MQTT |
LQI |
82.67 |
|
Control Latency (ms) |
61.73 |
||||
|
250 |
ZBMINIR2 |
MQTT |
LQI |
68.43 |
|
|
Control Latency (ms) |
63.50 |
||||
|
350 |
ZBMINIR2 |
MQTT |
LQI |
60.53 |
|
|
Control Latency (ms) |
66.37 |
||||
|
Dongle-PMG24 |
150 |
ZBMINIR2 |
MQTT |
LQI |
30.93 |
|
Control Latency (ms) |
59.17 |
||||
|
250 |
ZBMINIR2 |
MQTT |
LQI |
6.00 |
|
|
Control Latency (ms) |
84.18 |
||||
|
350 |
ZBMINIR2 |
MQTT |
LQI |
18.67 |
|
|
Control Latency (ms) |
71.93 |
||||
|
Dongle-M (UART over USB) |
150 |
ZBMINIR2 |
MQTT |
LQI |
41.52 |
|
Control Latency (ms) |
70.41 |
||||
|
250 |
ZBMINIR2 |
MQTT |
LQI |
2.57 |
|
|
Control Latency (ms) |
82.32 |
||||
|
350 |
ZBMINIR2 |
MQTT |
LQI |
4.40 |
|
|
Control Latency (ms) |
79.33 |
||||
|
450 |
ZBMINIR2 |
MQTT |
LQI |
0 |
|
|
Control Latency (ms) |
140.85 |
||||
|
Dongle-M (UART over Ethernet) |
150 |
ZBMINIR2 |
MQTT |
LQI |
40.13 |
|
Control Latency (ms) |
66.23 |
||||
|
250 |
ZBMINIR2 |
MQTT |
LQI |
8.67 |
|
|
Control Latency (ms) |
63.77 |
||||
|
350 |
ZBMINIR2 |
MQTT |
LQI |
14.40 |
|
|
Control Latency (ms) |
63.90 |
||||
|
450 |
ZBMINIR2 |
MQTT |
LQI |
0.00 |
|
|
Control Latency (ms) |
103.43 |
||||
|
Dongle-M (UART over Wi-Fi) |
150 |
ZBMINIR2 |
MQTT |
LQI |
57.27 |
|
Control Latency (ms) |
68.17 |
||||
|
250 |
ZBMINIR2 |
MQTT |
LQI |
4.28 |
|
|
Control Latency (ms) |
77.52 |
||||
|
350 |
ZBMINIR2 |
MQTT |
LQI |
5.73 |
|
|
Control Latency (ms) |
76.20 |
||||
|
450 |
ZBMINIR2 |
MQTT |
LQI |
0.67 |
|
|
Control Latency (ms) |
109.70 |
Communication Distance and Control Stability
Under open, unobstructed test conditions:
- ZBDongle-E, Dongle-M, Dongle-LMG21, and Dongle-PMG24 were all able to successfully pair devices and maintain stable control at distances of 150 m, 250 m, and 350 m.
- Within a 350 m range, the control latency of all SONOFF Zigbee Dongles remained concentrated in the 60–80 ms range, with relatively small fluctuations.
- The Dongle-M was still able to control devices at a distance of 450 m.
Note: Due to site limitations, distances beyond 450 meters (such as 550 m or 650 m) were not tested for the Dongle-M.)
Relationship Between LQI, Distance, and Control Experience
A clear pattern can be observed from the test data:
-
As the test distance increased, LQI values decreased progressively, while the control success rate remained very high, and latency stayed around 70–80 ms.
-
For Dongle-PMG24 and Dongle-M (both based on the Silicon Labs MG24 chipset), device control remained 100% successful even when LQI dropped to single-digit values or even zero, with only a slight increase in latency.
These results indicate that LQI cannot be directly equated with actual control experience or usable communication range.
In practical deployments, LQI is better treated as a reference indicator of link quality, rather than the sole criterion for determining whether a device is usable. Control success rate and latency performance are more representative metrics of user experience.




















































Commenta
Nota che i commenti devono essere approvati prima di essere pubblicati.
Questo sito è protetto da hCaptcha e applica le Norme sulla privacy e i Termini di servizio di hCaptcha.