All sensors have some degree of resolution and accuracy. Resolution is essentially the smallest change the sensor can detect (I am talking about digital sensors so not talking about the ADC and signal conditioning of analog sensors) and accuracy is how close its measurement is to reality. With that in mind I set out to find the best sensor(s) I can deploy around my house to take accurate measurements indoors and outdoors. Given I don’t have a calibrated reference to compare to, I am just looking for the sensors to jive between each other (while I test them all in the same location/time). I also a handheld “General Standard EP8710” that appears to provide a fairly accurate temp and humidity reading which I can easily move around to compare readings from HA sensors to a common “standard” (giving it time to acclimatize to the new location of course).
Long story short, in many of my projects I found that proximity to something barely warmer than environment can actually have a large impact on readings. Sensors inside enclosures with an ESP32 in the same box can be way off and once I inspected the build with a thermal camera it was obvious why… Even attaching the sensor to the Outside of the enclosure would skew things considerably due to heat being conducted by the materials in between… while not pretty, letting it hang loose in mid air gave the best results… but I really would prefer something tidier.
Long story short, it is pointless to aim for accurate sensors unless the placement is ideal as well. Below is a chart showing 2 sensors placed on the porch and 2 on the back patio (patio & outdoors) giving me 4 outdoor temperatures and 2 humidity readings.
I am happy to see that all the sensors jive (see between midnight and 4am) but that is a rarity especially if it is warmer outside. During the day there can be significant difference between patio and porch but that is due to the sun’s location. If the sun happens to shine on a sensor (even if I tried to put them as high as possible to prevent it - patio and porch are both covered) it will spike up several degrees.
Based on what I see from my data it appears that the house itself affects readings. In other words, during the summer (most obvious) the outer brick/stone/cement board must be absorbing heat and releasing it affecting readings as the variability between sensors in the same location is significant.
I’d like to hear your strategies to get the most reliable temperature and humidity readings both outdoors and indoors… How do you place a sensor out of the sun, covered, far enough from structure (or other things) that may impact readings without using a weather station? Insulating materials? Ways to affix sensors? Location? Ideal enclosures?
Also, I am working on building a custom measure-it-all ESPhome (ESP32) sensor but the ESP32 itself skews things if in the same enclosure. Is putting the ESP32 to sleep the only way to address this? Any enclosures that minimize this?
I have one outdoor sensor placed on the bottom of alarm siren. It’s in the shade but it will give you different temperature measurement than buderus sensor that is placed on the side of the house that hardly ever get sun. It’s expectable behavior for my point of view. It’s not inaccurate reading but those reading suggest that part of the house that is facing toward sun can gain more heat in the winter time when is sunny than a part of the house that is in the shade all year long. All measurement I done suggest that.
Knowing that it sould be possible to turn off radiator valves when temp on sunny side of the house reach certain degree and heat more in the other parts of the house.
For me problem lies in the fact that for some reason my trv doesn’t close the valve when it reach desired or higher temperature than requested. Its ts0601 trv. People reported problem with this, so maybe I could do something solve this problem.
@ddaniel - I agree that it is “normal and to be expected” but I am trying to figure out ways to lessen the impact if there are any. Placement of sensors is as important as picking the right sensors however it is an uncommon topic in these forums.
Regarding your house heating example… I wish I had a per room heating/cooling ability! While I have 2 independent HVAC systems for upstairs and downstairs, I can’t control the heating/cooling per room as it is a forced air system. There are vents that can open/close but those can mess with the pressure on the system and eventually break the large fan motors (or just wreck havoc with the balancing of the system).
I installed ecobee thermostats that have remote sensors therefore I have 11 temperature measurements to figure out when to heat/cool the house, but it does it using an average of the rooms I pick for that time of day/night so you can imagine what happens… Some rooms are freezing cold or way too hot… I wish I could vary the flow of air based on temperature while keeping the system balanced (total output needs to be the same but where it is output changes).
Even placement of ecobee sensors turned out to be very important as attaching them to an exterior wall (indoors of course) or too close to a window or in direct line of an HVAC vent all skewed the readings causing the HVAC system to operate incorrectly.
Lol, keeping this stuff working right is nearly a full time job…
I almost exclusively use aeotec multi 6 sensors on USB power…
Are they perfect… Absolutely not. There is no such thing as a perfect sensor for all the reasons you specify.
Are they good ENOUGH. ABSOLUTELY.
I also like these sensors because I can set most of the options for the device including report interval and resolution… And a calibration adjustment for the temperature (because of where I have them installed.)
Do I need resolution of a tenth of a degree F? No. I just need to know it’s warmer than I want it to be so something should happen. I’ve stopped looking at graphs of temperatures and things except when I’m trying to troubleshoot an issue. I honestly don’t care as long as the net effect is what I want.
With these sensors on USB power they’re fast enough and have good enough range and resolution to get the job done. What allowed me to do so was that calibration adjustment. I now just put the sensor where it makes sense then take some readings with specific thermocouple (all of my sensors are calibrated to this same thermo, doesn’t matter of its accurate - just that they’re the same) and set the +/- calibration for the sensor.
Below is an example of a case where the “Room Temperature” sensor was attached to the enclosure on the side of the blower that sucks the computer’s hot air exhaust away from my desk. Before installing this system it was getting way too hot under my desk, where the computer is, because the heat could only escape from the front where I sit roasting me (most unpleasant in mid summer).
Right now the room sensor used by this ESP32 based setup is just dangling because attaching it to the enclosure even using thick insulating material was causing the two sensors not to jive when the computer was off. BTW, it is warm in here but not nearly 27C… that is due to the exhaust being released without a duct on the blower so it is warming up the area where the blower is (and the room sensors is within a few CM of the blower). The room is still at about 25C.
@NathanCu I’ve worked for a test and measurement company for 20 years so that is in part why I am fixated with this… but I agree with your take on it. I try to avoid calibrating the sensors unless I can check the calibration on a wide range to ensure that the adjustment makes sense across the full range. Calibrating it in loco may be calibrating out the effect of something that changes with season so you end up with bad readings when in the opposite season (Nathan and I are both in Texas and the joke here is we only have 2 seasons: bit cold and excruciatingly hot).
I use this for your same purpose… just to make things jive:
I do but before I do so, I need to figure out why the reading is off. Below is the contraption I made to measure temp, humidity, pressure and motion (mmWave sensor) on my front porch. The BME280 is in the modified cable gland on the right (sensor and inside of box are separated by air tight thick foam). The measurements are quite good but my current theory is that the house skews the readings in certain conditions. I can’t calibrate that out as it would introduce error when those conditions are not the same. I did turn the IP Camera off for several days (as it gets quite hot) but it did not seem to be the cause.
I know that this is off topic, but I really don’t understand this american way of building houses, heating houses. We don’t do that in europe, at least not in my country. We build houses out of reinforcement cement or reinforcement cement combined with bricks. It’s totally common to have a radiator in every room, at least one, double or triple glasses windows.
Well, that starts to explain a lot of things Alex. And yes I’m currently enjoying the transition from death to just a bit cold - how about you?
Understand on the ‘why’ but is it really knocking the temp that far out? (nice setup on the box btw) Are we talking a degree (F) or a few degrees C? If it’s conditional, then you’re going to have a really hard time filtering it. I have a similar situation with the sensor in my bedroom - it’s actually on the wall behind the range in the Kitchen. (Thanksgiving played havoc with it as the oven was on all day long)
I just deal with the ceiling fan in the bedroom turning on with my oven… My wife hasnt noticed that’s what it’s doing yet.
OT sure - but fun to talk about. One big misconception about ‘the American way’ is that there isn’t just ONE way. You have to remember the state of Texas alone is as big as most of the combined land mass of central Europe. We don’t all live in the same climate. Heck Even IN Texas we don’t have one climate.
I can show you within 50mi (80 KM) of my home no fewer than 6 unique different building styles (and at least 4 of them have no business being built in South Texas…) You also won’t find a basement here - unless you have more money than sense - the water table is too high, and you have to cut through solid limestone to dig the pit.)
Pier and beam is popular in the Southeast (FL, Alabama, Mississippi, Louisiana) because of marshy or sandy unstable soil. South (MS, LA, TX, Ar, OK) - post tension foundation with balloon framing, holds shape when you cant do a deep footing/foundation and both work well where you can’t do a basement. Heat pumps with LARGE cooling capacity is popular here. We also spend a lot of money on radiant barriers. Go to the midwest (Illinois / Chigago, Ohio, Iowa and the middle states for those not in the US) and you will see a lot of brick, still balloon frame but in most cases, with basements where you’ll see large boilers for water-based heating and cooling with radiators. Northeast - concrete and steel. Out west ESPECIALLY in New Mexico and Nevada - arid plains / desert climate you start to see a lot of earth and concrete construction, as well as the disappearance of air conditioners instead swapped out with evaporative coolers. Earth construction with evaporative cooling is enough to keep the building comfortable there where it would create an indoor sauna in South TX.
Each was designed to solve building challenges in the geography its deployed in. You will see heavy use of lumber because white pine grows across the entire country like weeds. But you will NEVER see just one building type. Double pane insulated glass is pretty much standard in all cases tempered glass is required in some cases.
Short version - lots of different kinds of problems to solve.
Challenge accepted.
I’m not familiar with building habits in the us, but I watch some tv, youtube… and all I see is wooden frames for construction of family houses costing a lot of money. Wooden frames are here used for stable.
I never was in us and don’t plan to go, but as I see reinforcement cement, bricks are generally used for government buildings or for some mansions that majority of people cant afford.
Maybe I’m wrong but the thing I wanted to point out that having a solid build house with double glaze windows and radiators in every room is a bare minimum in my country. It’s not something reserved for a few percent ultra rich who can afford it. And that is in most of europe. Ok truth to be told in last years prices of properties sky rocked, but most of our properties are from, that bad and ugly and hard, communist era. Now when I look on it, it was basically a great time for most of the people.
State gave everyone apartments basically for free.
This is not about I’m better than you or something like that. It was just reaction on some things that here is normal for majority of people.
Enjoying it I’ve lived in Texas for about 20 years and have yet to adjust…
While it varies case by case, it actually affects the measurement by several degrees C. Enough for it to clearly not make any sense in the context. And regarding “F”… after 20 years in this country I still don’t get why “y’all” stick with it
I have a similar situation with the sensor in my bedroom - it’s actually on the wall behind the range in the Kitchen. (Thanksgiving played havoc with it as the oven was on all day long)
I just deal with the ceiling fan in the bedroom turning on with my oven… My wife hasnt noticed that’s what it’s doing yet.
[/quote]
@NathanCu I think I just discovered another thing you are expert on! I relocated here from Italy about 20 years ago and have not toured nearly at all as any vacation time I have I go back home to visit. While over the years I saw that some houses are built differently compared to here in Texas, I was definitely not aware of all the variants you listed. That was educational !
My house in Rome has a livable basement underground and 2 floors above. For the basement they used tuff (volcanic soft rock) for outer wall in contact with dirt and “foratini” bricks (picture below) for all inner walls. Between the two there is a walkable passageway to keep moisture at bay. Above ground level, the outer walls are in nicer / denser bricks, while the inner walls are made with “foratini” but the space in between is minimal.
Heating in Italy is done with water radiators in each room, like in @ddaniel 's case. No AC in my house as it is 40 yrs old and designed in a way it would be prohibitively expensive to do so (no separation between 3 floors).
I too was quite shocked at the materials used in this area. An advantage, pre-pandemic, was the low cost of housing allowed higher home ownership compared to Italy, but now I am no longer so sure as my house doubled in value in the past 2 years… and no it is not a good thing as taxes (>$11k) are tied to home value.
You are assuming that a wooden home with drywall walls doesn’t offer the same comfort as a European home, but it does, and in some cases may even exceed it. While I prefer building materials and techniques used back home (Italy), I have to say that well built houses here are supposedly easier to heat/cool than a comparable Italian house (wood is a good insulating material)… and as I say that I remembered that last year I used close to 24MW mostly for cooling… but… that is because I am in Texas .
I am European too, and thought the same when I first moved here. I now see the other side of the coin… but if I win the lottery, I’ll build myself a Euro-style mansion
Anyhow, one good thing, taking this thread back to sensor placement, is that with a wooden home I don’t have to worry about RF signal issues as much.
Another reassuring view… the sensors on patio and porch jive… the porch ones show hotter temperatures as the sun was on that side of the house. This is fine… but I often have significantly different sensor readings from the same location (aka porch 1 and 2, 3C or more degrees apart). The only thing that changed is cooler weather… so maybe it is indeed radiated heat from the structure that heated up during the day…
Perhaps a small fan to circulate the air near your sensor that are picking up radiant heat would help? I know my home weather station has this feature. Could be made “smart” to only turn on when the temp is greater than X…
Interesting… I did not know that weather stations had fans but it makes sense now that you mention it. Air flow is indeed something I tried to improve in the sensor housing but not enough.
The sensor is barely visible behind the net. Behind the sensor is thick dense foam that blocks air flow from the inside of the box where the ESP 8266 is. I wanted to offer some protection to the sensor but maybe it needs to stick out more…
I just wanted to share another case of how sensor placement is important. In my media room, which has no windows and has 3 walls that are exposed to non conditioned attic space or the sun, I have a few sensors providing room temperature. I get wildly different readings and for some it is obvious why given there is networking gear that generates heat but I was surprised at the difference shown in the graph below until I looked at the room with a thermal camera…
The orange trace is a motion sensor with temperature sensor which is typically quite accurate. It was attached to a plastic gang cover for a wall speaker mount I am not using. After looking at the wall space with my thermal camera I noticed that the wall was quite a bit warmer around the box as there is less insulation there due to the box. While it is quite obvious, it just did not dawn on me that it would have such a great effect on the sensor readings. I moved it temporarily on one of my racks (the spike) and then I moved it far from all heat generating sources (but still in the range of the HVAC vent…) and put it next to the room’s ecobee sensor. Great surprise… they now match. The ecobee sensor (blue) is more responsive given its main duty is to provide temperature data to my upstairs thermostat, however overall they are super close.
Just food for thought for those placing sensors and getting readings that don’t seem right…
I see that more people then me is looking in to this topic.
I’m in the middle of a home renovation, and are going from aqara battery temp sensors in all the rooms to esp32 hooked to mains with separate temperature/humidity/air quality sensors to all rooms. (I’m dropping as much battery powered stuff I can).
I’m planning to have sensors in separate housing away from the esp exactly because I don’t want any temperature interfering from esp/ 3.3v transformer to the sensors.
But I’m still not sure what’s the best locations for the sensors…
My idea/best guess so fare is in eye level height, and on interior walls, away from vents, windows and other heat sources.
@tormodnt All I can suggest is to get a good quality temperature meter and investigate the best placement even though you have listed the main considerations in picking a spot. The main issue is that power is not always available in the ideal spot, or having the sensor there is not visually appealing.
In some rooms you may measure different temperatures (this will affect the humidity reading too) but the one that is most important is where you spend most of the time in that room. Therefore, in a bedroom, it may not make much sense to have the temperature sensor in the perfect corner where you can get the most accurate reading when the bed area may be somewhat different. The difference may be small but I am surprised by how often I check the thermostat setpoint while in bed because I feel uncomfortable only to find that the thermostat was set at 24C (in summer at night) instead of my usual 23.5C.
Another aspect, is that if you have more than one sensor in a room, for example your fancy ESP32 multi sensor and an ecobee thermostat sensor (I have a dozen around the house), I generally prefer for the two sensors to match otherwise I always get the nagging impression one is wrong even though they may not if placed in different parts of the room. This is to say, that in your chosen spot you may have a few sensors lined up if you are anything like me
Edit: regarding the hardware pictures I showed above… I have changed how I build them. This is an example of what it looks like now:
The external box with vents was meant to allow more air around the sensor as vents are on all 4 sides. I added insulation between the two boxes and added a tiny mesh on the inside of the small box to prevent insects getting in there. However, the measurement was still skewed by the bad sensor placement (front porch under warmer covered area and close to a camera that heats up considerably).
I moved it to another spot with more ventilation and now this is what I get (Green is the actual temperature from OpenWeatherMap for my area, the other two are front and back of house):
Edit2: @tormodnt - In your list you mentioned to avoid “interior walls” but that should be “exterior walls”. Whether interior walls will affect your measurement depends on the wall’s material. Here in the US it likely doesn’t do much as it is basically just 2 layers of drywall separating you from the other room. In Italy where I am from, the walls are made with bricks, concrete, plaster, etc so they likely do affect the measurement to some extent.
And yes I’m currently enjoying the transition from death to just a bit cold - how about you?
I’ve lived in Texas for about 20 years and have yet to adjust…
