Does anyone have any suggestions on monitor up to 20 temperature/humidity sensors for trend aggregation?
We’re building a high performance (passive house). Since there’s very little data on the wall performance at our designed spec of R-63, the builder would like to install a series of sensors to monitor the interior/wall/exterior delta. The goal is to show that it is worth the upgrade costs to install what is 3x the code minimum amount of insulation and a series of water and air barriers not required by code.
My idea would be to install a series of shtc3 sensors connected by 2-pair STP to a central location on each floor. A search of this forum shows that max wire length is 12-15m with STP but drops to 3m with unshielded wire. Reliability and minimal maintenance is the goal so hardwiring is likely the best way of doing this. The number of sensors can be increased by using TCA9548A multiplexers, although ESP32s are cheap enough to have 1 per bank of 8.
Are there other data acquisitions that are as reliable? A bunch of Sonoff TH16 with DS18B20 sensors can work but having sensors in various parts of the way will provide better info since the shtc3 can measure RH. Each Sonoff is also limited to a single sensor.
Victoriametrics seems to have the most efficient data stream compression and storage, so I’ll use that in place of the default HA db.
You might have hard time to get that working. I2C was designed for communication on the same PCB, not for meters long wiring. At least use active terminator like LTC4311. Another problem is that Esp has only two I2C lines and shtc3 has fixed address, so without multiplexer you only get two sensors per Esp32.
You could use RS485 TH sensors solving both issues, but it’s probably cheaper to simply use one Esp for every 1-2 sensors and keep the distances short.
You’d be better with BLE sensors, like Xiaomi LYWSD03MMC. I have a bunch of them, they are cheap, reliable and also accurate.
As a side note regarding this:
sensors are not needed to prove anything, it’s definitely “not worthed” if you look from the money perspective… for double the money you get only a few percent better insulation characteristics. That’s common for all things, not just insulation. But that’s another story (and topic)…
I agree. LYWSD03MMC are cheap enough. Wireless can be a blessing and a curse.
Alternate is deploy DS18B20’s connected to one ESP32 via Tasmota or ESPHome. You can have multiple devices on one GPIO as it uses the two wire bus interface.
Read the docs carefully, use three wires and the single resistor for bias current. Devices with long cables are readily available, but you can deploy a star mesh wiring scenario for your needs.
This says you need senors on the outside and inside, but you probably need a lot more of them to get a better picture, since there will be a variance across the wall both inside and outside.
I have a lot of sensors. For basic temperature the DS18B20 is good. For room temperature I really like the BT ones that also have a display. They are cheap enough I can have multiple in the same room. This is how I know there are typically temperature gradients in most rooms.
But, as others said you probably won’t find any data to back up the claim the extra insulation will pay for itself over a normal lifetime. You don’t need temperature sensors to tell you how long the HVAC system runs or how much it costs. And temperature sensors won’t tell you how much you are saving.
The wiring for all those sensors is going to be a pain and ugly. When I built my house I thought about putting sensors everywhere. I am glad that I didn’t. The technology is much improved since then. The cheap wireless BT or Zigbee senors will probably give you more data than you know what to do with.
Talking abour curses: one wire can also be quite a curse. I’ve had 10 od them on one bus, they worked fine for about a year, then one of them died and of course all 10 stop working. No need to say that i’ve had to dissasemble the whole line to find the culprit… from then on if i use dallas sensors they are one per gpio. Period.
Be aware there are limitations for how many sensors each ESP32 can support out of the tin. This is usually a software limitation.
Having so many sensors may affect frequency of polling, should you want nanosecond response and twelve decimal places in °Kelvin for each sensor
Getting your data into HomeAssistant may be a challenge. Hard wiring via multiplexers, WiFi and BlueTooth channel overload, database storage capacity, sensor accuracy and granularity, sensor placement, etc are all factors that will need addressing.
Aggregating so much data is going to confirm the obvious: the more insulation you have, the less the temperature will vary.
Thermal inertia will be a large overriding factor. A very thick stone wall with a large mass and density is going to change temperature far slower than a single layer brick wall backed by drywall. An igloo is going to heat and melt slower depending on the shape and ice thickness Other factors not related to insulation can have a far greater effect on the overall improved thermal performance of a structure.
Humidity measurement to the level of precision you need may require extensive calibration and expensive sensors.
You may wish to point your builder to university and industry studies that have already solved this argument.
I did something similar.
Initially, there were 22 ds18b20 sensors connected via Arduino Nano. Sensors used moisture-proof
The sensors were divided into 3 branches, respectively 3 GPIOs
Later, the sensors moved to esp32.
The system existed for several years in the basement of an apartment building.
As mentioned earlier, the house is very inertial.
There is no need to take measurements more than once every 6-8 hours.
(sudden changes may occur at the beginning of the heating season).
Your home insulation is great. You can prove the usefulness (costs) of insulation if you have preliminary energy consumption statistics.
