Zigbee temperature sensors

Hardware
1

Can anyone here give any recommendations for Zigbee temperature sensors, preferably with a display?

I bought 2 of these a while back on a recommendation from a YouTube video:

However, whilst one of them works “ok”, the other was extremely unreliable and would barely connect to the Zigbee network. The one that works doesn’t send updates very frequently though.

These are for driving the heating system though Home Assistant thermostats, so need to do reasonably frequent updates. They could be battery or mains powered (although I’m likely to mod any battery powered devices to run from mains instead). Not ridiculously expensive - I’ll put one in each room. A “current temperature” display on them would be a nice to have, rather than having to look at HA all the time.

2

Those only update about once an hour, I have 2 of them, useless for automation. You might look at the cheap Xiaomi Bluetooth sensors that are flashable to Zigbee

3

These ones: Custom devices (DiY) LYWSD03MMC control via MQTT | Zigbee2MQTT

I have a bunch of them, they’re quite nice and update often. Battery life isn’t exceptional, perhaps 9 to 12 months.

4

I’m running the Sonoff Indoor Temperature Humidity Sensor, model SNZB-02D I’ve been running it for about a year over zigbee2mqtt here’s a link to it on Amazon there about 15 bucks each.

5

In my experience (using 10 of these around the house, flashed to Zigbee) unfortunately the battery lifetime is more like 4 to 6 month. If leaving this model with BT then yes, battery lasts for 9 to 12 month.

6

I’m using the Sonoff SNZB-2D also in a couple of places, my front and back porches. They’re outdoors but protected by the porch roof. Been running a couple of years now with so far just one battery change. I’m using ZHA for my zigbee stuff.
oh… I’m in Central Texas with a temp spread of 10-115F. It updates about every 5 minutes if there is a change.

image
image1812×513 27.2 KB

7

Thanks everyone, I have ordered some LYWSD03MMC devices to try out and if they work ok I’ll get some more.

9

I have modified 10 LYWSD03MMC with the addition of a 1N4148, so that I can power them using rechargable LIR2032 cells instead of disposable CR2032 cells, and just want to report back my findings in case they are of interest to other people.

A fresh CR2032 has a voltage of about 3200mv, whereas the LIR2032 starts off at 4200mv. This voltage is too high and reportedly causes damage to the LYWSD03MMC. Also, running it at a high voltage greatly reduces the LCD contrast. So I modified the sensors by cutting the track from the battery + contact and adding in a 1N4148 diode to drop the voltage by 700mv.

I’ve got 12 LYWSD03MMC sensors running the Zigbee firmware. 10 of them have a 1N4148 diode installed to drop the cell voltage and are powered by LIR2032 cells. The remaining 2 are unmodified and powered by CR2032 cells.

The oldest of the LIR2032 sensors was powered up with a freshly charged cell at 21:00 on 23rd June, with an initial reported voltage of 3500mv (a freshly charged cell is 4200mv, minus the 700mv voltage drop from the diode gives us 3500mv). It is now 7 weeks later and the reported voltage is now 3100mv. Over the first 14 days the reported cell voltage dropped by 200mv, but since then it has dropped by 100mv every 16-17 days.

You can safely discharge the cell to 3000mv, so that means I need to replace the cell when the reported voltage is 2300mv. Assuming the discharge curve stays reasonably linear at 100mv every 16 days, that means that the battery should last another 128 days - a total of 177 days since a fresh battery was installed.

Time will tell whether it actually lasts this long, but I’m calling this a success since this experiment shows that I can go several months between charges.

As a point of comparison, a fresh CR2032 was put into one of the unmodified sensors at 15:00 on June 26th, with a reported voltage of 3200mv. The first 200mv was lost in just 1 day, and then another 100mv 4 days later. Since then the reported voltage has gone back up to 3000mv and stayed at that level, so I don’t really have enough data yet to compare the life of the CR2032 with the LIR2032. However, even if the CR2032 has a longer life, it is disposable, so I’m happy enough recharging an LIR2032 every few months instead of buying a new CR2032.

Note: I have NOT tried any of these experiments with an additional capacitor installed. My gut feeling is that the internal resistance of a LIR2032 is probably low enough for a capacitor not to be beneficial, but I haven’t collected any data to back that up.

A couple of gotchas:

  • if the device moves out of range of the zigbee mesh, it drains the battery fast. I know that @pvvx has said that “zigbee does not like coordinator outages”, but no other battery operated zigbee devices on my network drain the battery like that when they can’t see the network. It would be nice if the device could drop into a low power mode when the network vanishes and leave long intervals before attempting to reconnect.
  • there is no protection to stop the battery being discharged too much. You can monitor it through Home Assistant and manually disconnect the cell once the voltage drops too low, but it would be nice if there was a way to program the device to shut down when the voltage drops below a predefined level.
2 Likes
10

A Zigbee device which lasts months on a single battery is nothing much to celebrate. Normal battery lifetime for Zigbee is usually measured in years, not months.

I’ll admit I don’t have any temp sensors with a display, so not sure how much of a factor the display is on battery drain. However, if I’m having to change the battery twice a year (on top of having to modify the device), then I would think twice about it.

11

That very much depends on the device.

I get about a year out of these (original firmware):

They are powered by a fairly large disposable CR2450 cell, and have no display.

These probably last slightly longer (I don’t think I’ve ever run one until the battery is completely dead):

They are powered by a pair of AAA cells (I use NiMh), send back data very infrequently, are not reliable and are physically quite big.

Conversely, a modified LYWSD03MMC is quite small, can be powered from a small rechargable LIR2032 cell, has a display and (for me at least) very reliably transmits data at a good frequency (better than the CR2450 powered device above).

It would be nice to only have to change the battery once a year. But I would prefer to recharge a device twice a year than to buy a replacement cell once a year. I’ve got 70 devices on my Zigbee network and having to regularly buy disposable batteries every year would start to get expensive, and also not really environmentally sound!

12

I’ve got t&h, motion and door sensors from Aqara. Motion & door sensors are still running on their original coin cell battery 3 years after I put them into production.

The t&h sensors’ batteries died after almost 2 years, and that’s because they were already starting to drain after multiple unsuccessful attempts at pairing them with Vera before I started using HA.

13

A door sensor that only needs to wake up when the door opens/closes is a very different beast to a temperature sensor sending updates every 5 minutes or less.

14

Also looking for recommendations! I bought a Sonoff NZB-02D but didnt like it, since it got stuck after a few nights. It did report visibility to the zigbee network but the temp just got stuck.

Now switching to thirdreality: Amazon.com
It seems to report pretty much the same temp as the Sonoff, so I guess they’re both accurate. But I bought 2 and one didnt even work at all. Going to see how this performs. But if someone is very happy with his temp sensor, I’d love to hear it!

15

I have had this problem with a few sensors, and it was resolved by going to the device’s Reporting tab in Z2M and re-setting up the reporting (you may need to remove some bindings from the Bind tab first if this fails).

16

Ok, I’ve gone a full cycle of a LIR2032 in one of my temperature sensors:

  • 23rd June: Freshly charged 4200mv cell installed, voltage reported to be 3500mv, which is correct as there is a 700mv voltage drop from the installed diode.
  • 24th June: Reported voltage: 3400mv (assumed cell voltage: 4100mv)
  • 8th July: Reported voltage: 3300mv (assumed cell voltage: 4000mv)
  • 25th July: Reported voltage: 3200mv (assumed cell voltage: 3900mv)
  • 11th August: Reported voltage 3100mv (assumed cell voltage: 3800mv)
  • 15th September: Reported voltage: 3000mv (assumed cell voltage: 3700mv)
  • 4th October: Reported voltage: 2900mv (assumed cell voltage: 3600mv)
  • 7th October: Reported voltage: 2800mv (assumed cell voltage: 3500mv)
  • 9th October: Reported voltage: 2700mv (assumed cell voltage: 3400mv)
  • 10th October: Reported voltage 2600mv (assumed cell voltage: 3300mv)
  • 11th October: Reported voltage 2500mv (assumed cell voltage: 3200mv)
  • 12th October: Reported voltage fell rapidly to 2100mv (assumed cell voltage: 3100mv) within around 14 hours. 90 minutes after dropping to 2100mv, the device stopped reporting to the Zigbee network, although the LCD was still working.
  • 13th October: I removed the cell from the device this morning and measured the voltage with a multimeter to be around 2600mv off-load, which is a bit over-discharged.

One of my other sensors which I installed a battery into on 27th June, and is showing a similar drop-off, with the reported voltage currently being 2700mv. Oddly, I have several other sensors started at around the same time, which have not discharged anywhere near as quickly - my guess is that the temperatures in those rooms vary less, so they are not sending zigbee updates so frequently, although I’ve not investigated to see if that’s true.

So my conclusion is that this should be a pretty cost effective way of running these sensors, if you’re willing to recharge the battery around 4 times a year. Its probably worth recharging once the reported voltage drops to 3000mv (which equated to about 14 weeks of battery life for me), given the very rapid drop off in voltage towards the end. I will be setting up HA automations to remind me to recharge the batteries in all of my sensors at about that point.

17

I don’t want to promote any reseller, but I can promote this devices:
SONOFF SNZB-02P

They work like expected and the price is one of the lowest. Next to the temperature you also get the humidty for free. For this low price it would be acceptable to replace some broken ones as one will buy more than one to micromanage all rooms and not the house as one big room.

We have several in the house and all are connected no matter the distance.
I don’t know if this is an error, the battery still shows 100% after 2 years frequently be online.