Yes, I tried the TRADFRI-removal experiment overnight.
The TRADFRI had my Temp5 Sonoff temperature sensor attached to it.
Around 5pm, I removed TRADFRI from the Devices list in Home Assistant, I pulled it out out of the wall outlet.
About 10 minutes after TRADFRI-removal, Temp5 connected back to the Sonoff Zigbee bridge.
No other ZB network changes were seen before bed last night.
This morning, I saw that another Sonoff temperature sensor had paired itself with the Sonoff S31.
Since I didn’t see this additional pairing before bedtime, that means that it took at least 6 hours for the Sonoff S31 to add another temperature sensor to itself.
I just re-added the IKEA TRADFRI to the network (nearest to the Temp2 temperature sensor.
Within 5 minutes, Temp2 was paired with the TRADFRI.
IKEA TRADFRI = FAST
Sonoff S31 = SLOW
Great, really good you did these experiments!
It is clear now that your Temp sensors are able to connect to the Zigbee Bridge via the S31 Light as a repeater. So, the S31 has the repeater feature. That is good. Now we need to figure out what’s causing the long pairing period (sometimes several hrs) between the Temp sensor and the other devices in the network.
You have flashed the Sonoff ZB Bridge with Tasmota and are operating HA (as I understood earlier).
You probably cannot go back, but I would be interested how the set-up would perform with the factory FW of the Sonoff ZB Bridge and the Ewelink App instead of HA. Would be worth knowing if the Temp sensor via the S31 ZB Light would pair smoothly and quick with the ZB Bridge.
Since the Sonoff S31 & Sonoff SNZB-02 both have TI CC2530 chips in them, I’d expect them to like each other and repeater-link very quickly, but they don’t.
Personally I do not think the slow pairing between the devices is primarily related to the HW, Texas CC2530 versus the Silicon Labs chip.
I think that it is more related to the ZB Bridge with Tasmota FW + HA.
To investigate that better, you would need first go back to a ZB Bridge with the factory FW and the Ewelink App, and then check if that works more smoothly from a pairing point of view.
One other remark, the IKEA repeater may have more Zigbee power or a better antenna design which results than in better Zigbee radio coverage around the device.
From other situations I noticed that some of the Sonoff devices have sometimes poor Zigbee radio coverage / pairing resulting from:
a) the small antenna design inside the device —> this can be improved by solding a 5-10 cm wire to the antenna connection inside de Sonoff device.
b) low battery power of the button cell battery in the sensor —> this can either be improved by installing a fresh new battery or connect the sensor to a permanent 3.0V DC power adapter (as I did for some sensors here where I solded 2 wires to the battery contacts in the sensor and conneted that to a 3 V power supply).
Yeah, I think the FW in the various devices in conjunction with HA are likely very significant factors.
In order to perform experiments with the Sonoff Zigbee Bridge OEM firmware, I’d have to order a new unit and rework my HA Zigbee network.
Anyone with the motivation and curiosity could do the same. Feel free.
I do think that the IKEA TRADFRI repeaters have shown better RF performance compared to the Sonfff S31, but I don’t think that RF performance difference accounts for most of the S31’s lack of pairing/re-routing speed.
I had Sonoff temp sensors within 10-15 feet to the Sonoff S31 and it still did no pairing with them over a 24 hour period, while the IKEA TRADFRI paired with a temperature sensor 20-25 feet away, within 5 minutes of pairing it with the Sonoff Zigbee Bridge. And recall, that the Sonoff S31 didn’t do any pairing with Sonoff temp sensors until I added the IKEA TRADFRI to the ZB network.
I can’t explain it, but it appears that the addition of the TRADFRI to the ZB network was a catalyst for the S31 to start functioning as it should have done hours and hours before.
The bottom line: The IKEA TRADFRI repeaters have impressive overall performance while the Sonoff S31 units do not.
Thanks @ tteck, for the pointer to that group of TRADFRI repeaters on ebay! I snapped them up.
Great, many thanks for the useful feedback. Good to know these details!
I think that the interactions of the different FW’s (Tasmota on the Bridge), factory FW on the S31 ZB Lite and HA have caused these issues. By introducing the IKEA TRADFRI in the mix, it very likely helped indeed the S31 to be able to pair to the ZB Bridge.
Btw: Sonoff is very much focussed to miniaturize their devices as much as possible. Probably to make those more appealing for the larger public and perhaps also to minimise shipping costs (most of it they ship outside China).
With that, I believe they compromise battery capacity of the sensors and the effectiveness of the antenna printed on the board.
Did you also open up the IKEA TRADFRI (is it easy to do?)?
If so, could you post then a few pics of it?
Would be very interesting to see the internals and how the antenna design on the print board looks like.
No, I have not opened up my TRADFRI repeater.
There are quite a few internal and external photos on the FCC web page: https://fccid.io/FHO-ICC-1
The antenna section is the part of the module protruding off the side (white coated, rectangle area).
The User Manual posted on the FCC webpage for the IKEA ZB module, calls it out.
OK, great. Thanks for that.
Looking at this, I am pretty sure that the IKEA TRADFRI repeater antenna design is much better and effective than the (minimal) Sonoff S31 ZB Lite antenna design.
It is a more generic concern I have with the Sonoff ZB line of sensors as stated above: their focus is too much on miniaturization of the devices, which leads to compromising range/effectiveness of the antenna in these devices. And if the sensor is also battery powered, their battery life is shorter than of some of the other mature ZIgbee sensor manufactures.
The IKEA ZIgbee line of devices are a little more expensive than the Sonoff Zigbee devices, but in my view a better buy since the performance is better and more stable. Furthermore, the IKEA devices you can have within 24 hrs at home, for the Sonoff’s you usually have to wait 3-6 weeks when ordered from AlieX or BangG.
“Highest measured output power to antenna is 13.1 dBm peak. With -2.5 dBi antenna gain EIRP is 10.6 dBm or 11.5 mW.”
It would be interesting to compare that with the Sonoff Zigbee radio performance (chip RF output + antenna).
Not sure if that info of the Sonoff device is also somewhere available on the web.
By the way, you can read there that the new IKEA Trådfri GU10 RGBW “LED1923R5” that became available in December 2020 in some IKEA stores comes with a newer Silicon Labs EFR32MG21 based module instead of the older EFR32MG12 module.
That new Silicon Labs EFR32MG21 based module both has a much more powerful 20 dBm transceiver (integrated RF power amplifier) and a better antenna design:
.
Now we need to find the published Sonoff S31 ZB Lite performance (RF power + antenna amplification factor) as we already found for the IKEA TRADFRI repeater (see above).
I haven’t found it yet.
I expect that the measured Sonoff ZB RF output power + antenna performance of its Zigbee devices is less than the radio performance of the IKEA line of ZIgbee power plugs, repeater, etc.
But we need to find the Sonoff published data to be able to compare.
The IKEA TRADFRI repeater measured transmitter power (from the antenna into free air) is shown to be a max of 11.5mW.
It looks like the 6.0mW number is the average (duty cycle taken into account) conducted (direct connected to the measuring device, with an attenuator) power level output, not the peak, transmitted-from-antenna power output.
