There's quite a lot of temperature sensors out there on the market obviously. I'm looking to build my own thermostat system within Hubitat, I already have virtual thermostats setup and Shelly switches to control the valves.
What I need is an always on powered temperature sensor. I don't want battery powered, I want it hard wired and forgotten about so it just works 24/7 reliably. It needs to speak to Hubitat locally and not via the cloud for fast reliable updates whilst offline.
I've picked up a Shelly H&T which may tick most of the boxes, though it's £22. There are Tuya USB powered WIFI temperature sensors on Aliexpress. However, can they work locally with Hubitat? I can't find much about these, but they're cheap:
Perhaps flashable? I know HA has a "Localtuya" server you can run. I want to avoid using HE though as Hubitat is my main automation system and the less parts of the puzzle for this the better to improve reliability, which is why I don't want to use Homekit sensors (even though the HomePod mini has a temperature sensor in it, there's no reliable way to get it to communicate in Hubitat) and my HA instant runs on my NAS which is prone to randomly going offline and requiring reboots.
Indeed Mark, you're right - my mistake, let me edit now.
HA, exactly, runs in a container on my NAS, which for reasons unknown to both me and QNAP needs a hard reboot about once a month, so I don't want to rely on it to be honest unless I run HA on a different dedicated stable device.
Yeah, i've mostly had pretty good luck with the odd thing i've got from the, usually by flashing it.
Hmm if these will connect as a generic zigbee sensor to HE that would be amazing for the price and serve me just right. It doesn't even matter if they're not accurate in terms of presenting what the actual temperate is, as long as they are accurate in registering it going up and down - I can always put an offset on what is more accurate to what it is.
I recently setup a Shelly Uni Plus and it supports 3 DS18B20 temperature sensors. I have the Shelly controlling my hot water heater recirculating pump and then have 2 temp probes hooked up so I can monitor the hot water temperature in 2 locations. There is a local community driver for it.
I’ve had this project on the back burner for over a year now. It’s to control my underfloor heating system. I had a Heatmiser system installed with NeoAir stats and remote temperature probes, and it is HORRIBLE. Just totally unreliable, with far too much battery changing. It’s slow, and I just can’t stand it.
So I picked up a Shelly 4PM and connected it directly to one of the UFH actuators. It turns on and off instantly and reliably.
The entire UH8-RF wiring centre can come out. All I need to do is get a couple more Shelly 4PMs and rig them up for each actuator. They can connect to a global neutral bus bar and a global earth bus bar. The only other thing that turns on is the UFH pump, and I can rig it up so that it always comes on when any of the actuators are turned on, without the need for automation.
Then it’s just a temp sensor in each room and a bit of programming to turn the actuators on and off. The Shelly 4PM can be hardwired with Ethernet, so I don’t even need to rely on Wi-Fi. I’ll end up only needing to use Zigbee for the temperature reporting from these devices, which will hopefully be set up once and forgotten about once they’re wired in!
I can even set up Advanced vThermostat and put a virtual thermostat in each room, linking it to a temp sensor and the Shelly switch for a native thermostat-like experience – which I can then share with HomeKit and have Siri control.
I’ll end up replacing 10 Air 2 thermostats, 10 remote temperature probes (and their batteries), the Heatmiser Neo Hub (which provided some smart integration, badly), and the Heatmiser wiring centre, with something that should be both better and more reliable.
I have had this Zigbee temp/humid sensor for several months and it seems to be working well.
It stays connected to Hubitat using @kkossev 's community driver "Tuya temperature humidity illuminance LCD display with a clock" even though the device has no illuminance senor and no LCD display (he suggested I use that driver after providing him data about the device).
To date the only issue is it does not properly report whether it is on USB or Battery power, but I suspect if I asked @kkossev to help with that he could and maybe would...
These are Zigbee end devices, i. e. although they have some kind of USB 5V -> 3V converter inside, they do not report the power source change, and they do not act as a repeater (router).
That would be ideal, though a bit more work - however it looks like as @calinatl as posted below - those are the exact same type i've ordered from Aliexpress.
I've only picked up one so far - which came out at a tenner to test. I'm in the UK so i'm not sure if they're on Amazon UK as i'd have paid a bit more to get one and test (though you never quite know you're going to end up being sent the exact same one)
On a different note. I have a spare 220v feed that was going to a light fitting in one room. It's not being used for anything and it's an ideal location for a hard wired sensor - you can pick up those 220v AC to 5v DC boxes on eBay - anyone used one? More importantly is that going to be safe to be left hooked up 24/7? The wiring in place is fine - it used to run a 50w standard bulb years back, so it'll only ever be pulling a couple of watts to run one of these sensors now.
I had posted about a dht22 a while back. That was going to be one of my options - especially if I could pop them in the ceiling of each room and from the loft somehow run a central unit to them.
Looking now though I think the linked Zigbees might be cheaper.
I have had big success with ESP32 based stuff (The Shelly 4PM is running it and it's honestly one of the most flawless smart home device i've ever run and I don't even have it hardwired yet)
What could be good albeit a more expensive and harder to implement option in the future would be to have a POE feed to an ESP32 with temperature sensor and split it with a POE to DC converter, give it power and data, so it's hardwired for both and will always be 100% reliable. You could also technically very easily reboot them if required then. But it requires a POE switch, POE adapters, the ESP32 boards, plenty of cable and patch leads and a lot of time and effort!
Edit: A one wire solution like the DS18B20 might make more sense especially for price.
You can run one central processing unit - take one ethernet cable out taking power and data from the central unit and connect as many DS18B20's as you require along the cable run. They feed all the data back and if you use a PI you could have it POE powered, although in theory it could just be in your network rack anyway - and have it run an MQTT that Hubitat can tap into to get the data from each sensor.
The sensors are cheap, cheap Cat 5 cable will be more than good enough, so it's just the cost of the central unit.
The bigger problem is what are you going to put the sensor in, in the ceiling to hide it whilst still have it come out into the room enough to sense the temperature properly.
That's true. I mount mine inside a drywall box, with 5v fans to periodically draw air over the sensor. It's all very diy but looks great and works brilliantly.
I do like the idea of using light switch sockets, that's good!
I think an in wall solution will be good as well as a ceiling mount for me.
At the moment i've got a a small vented project box that should work - there's a few flush fit options, like a downlighter type thing but I don't think the sensor would be reading the room temperature very well if it's not actually IN the room.
I'm going to try the one wire solution above and report back, it might be useful for other people trying to get something running with Hubitat - it's also the cheapest solutions and if it works well will never need batteries replacing and all it's data will be hard wired so no fear of wifi or any wireless proticol not working at any point.
I haven't read every post in detail, so apologies if this has already been covered, but you may want to consider / assess the impact of having the sensors on the ceiling in terms of the readings they are likely to detect, particularly in the warmer months. If you plan to use the readings to control any cooling system, then this could result in it running more than is needed.
