Let's Control It

Hi, do you have any experience which temperature and humidity sensors are 'the best' for ESP8266?
I have a couple of them, they show a different values, of course..

So here is my test, DH11, DHT22, BME280 and HTU21D:




Here are aslo some cheap thermometers, for comparsion (excuse this 'wiring mess', it's just testing ;] ..):




What do you think, what is actual temperature, which sensor should be most accurate?

Hello,

excuse me if i might sound like a teacher, but some basics first.
Every sensor (and every thermometer) has a tolerance. It's unavoidable, we're not living in a perfect world...
If you want to check a sensor, the reference you use must be one class better at least.

What does this say?
If you want to calibrate a sensor to a tolerance of +/- 1 °C you need a reference with an accuracy of 0,1 °C!

If you want to calibrate to a tolerance of 0,1 °C, you need a reference of 0,01°C. And so on.
This is simply mathematics, there is no workaround. Cheating physics and math is senseless.

So which of your sensors and thermoters is correct? I fear none of them.

To make some approximate checks, get a calibrated Thermometer first.
You may get a calibrated glass tube type thermometer at lab equipment sellers for an acceptable price... not cheap, anyways.

A quite good preselected temperature/humidity sensor is the SHT1x made by Sensirion/Switzerland.
The higher the number (SHT10, SHT11, SHT15) the better the accuracy.
Even with the very expensive SHT15 accuracy is +/- 0,3°C and 2%rH....

The best selection you have in your post is the HUT12 (=SI7021). For precise metering I'd prefer to abandon the DHT type sensors.

A word about these sensors:
Precise metering has a (high) price. Don't expect a BME280 from Aliexpress (usually a cheap copy of the genuine Bosch sensor)
to be as accurate as the Bosch type.

So what can you do?
First check: Which accuracy is needed? Is temperature with two decimal digits necessary at all?

Second: Get one reliable reference. A professional thermometer (see above), or a calibrated thermo/hygro-device.
Be aware, that equipment will cost some money.
With a reliable reference you're able to calibrate even cheap sensors to an acceptable accuracy.

How to calibrate? take a look in our Wiki:
https://www.letscontrolit.com/wiki/inde ... d_Accuracy

Hallo, you are right, of course, I just ask wrong question.. Wiki about calibration I know, I just don't have any precise thermometer for calibration (or lab equipment) and I can't also use water with ice to calibrate at 0°C, because none of sensors is waterproof :].
But you answered my question anyway, thanks for that. All sensors are cheap ones, of course, AE/eBay purchase, appr. one year ago. I don't need to have values with two decimal digits accuracy, just 'whole numbers'. In the example that I posted, there is numbers from 24°C to 26°C, so I'll trust HTU21D in this case :].

btw, I'm unable to get from HTU21D any humidity values, just temperature. ANy hint how to force this sensor to measure also humidity? I suppose that I have fault sensor, I tried ot with many different fw's and boards, same results ..

Hello,

i don't have a HTU21 at hand so I can't test.
My SI7021's are working fine. Usually they should (!) be compatible.
Anyways, many sellers say SHT21, HTU21 and SI7021 are same which is definitely not correct.
Try both settings, SI70021 and SHT21, maybe one works if the module isn't broken.

It's not really necessary to go down to 0°C for calibrating a temperature sensor. It just makes things easier.
You may calibrate even with two different temperatures, the bigger the difference, the better.
For example 5° in a fridge and 22°C room temperature. You will have to go back and forth several times
to find the correct parameters then, lower temperature gives the + or minus parameter, higher temperature
gives the multiplication factor.
If your sensor has a bit of cable you may even put it into a plastic sheet and cool it down to 0°C.

Anyways it needs a reliable reference.

Having a look at your Prototyping I would recommend to bring the sensors into clean, circulating air. Means at least a distance of 20cm or more to every concrete object. The way you have placed the different sensors is not suitable to get reliable data.
Before comparing the readings you should give the sensors enough time to adapt the temperature of the surrounding air, an hour or so.

I've made something like that some time ago

I have multiple (about 10 or so DHT21), about 5 HTU21\SI7021, 5 SHT20 and 5 SHT30\31 and one BME280

My personal conclusion (based on few month of test at same location) - htu21\Si7021\SHT20\SH30 are showing the same results (difference usually not more than 0.2 at temperature).

For humidity difference is bigger - about 5%, but also all of them (htu21\Si7021\SHT20\SH30) are accurate. Intresting that I could not get 100% humidity with SHT30\31 (using all 5 that I have) (htu21\SHT20 shows 100% without any problems).

BME280 shows temperature a bit strange. Sometimes it show same as SHTs and HTUs and some time +1. Humidity also not very good (very sensitive, shows 100% when all other sensors only 70-80%).

DHT21 - not bad at temperature, but for humudity - all my sensors died after 1 year (showing 1% humidity all the time).

Conclusion:
Go for Htu21\Si7021\SHT20\21\SHT30\31 - difference between them is very small.

I bought all my sensors on aliexpress

Shardan wrote: Precise metering has a (high) price. Don't expect a BME280 from Aliexpress (usually a cheap copy of the genuine Bosch sensor)
to be as accurate as the Bosch type.

I wouldn't be so sure about someone's ability to copy Bosch's MEMS technology. Even if someone in China could mimic it, I don't see a market at a scale Aliexpress can provide. So I assume those breakout boards contain genuine BME sensors and the price is fair.

chemmex wrote: ↑16 May 2019, 10:22

Shardan wrote: Precise metering has a (high) price. Don't expect a BME280 from Aliexpress (usually a cheap copy of the genuine Bosch sensor)
to be as accurate as the Bosch type.

I wouldn't be so sure about someone's ability to copy Bosch's MEMS technology. Even if someone in China could mimic it, I don't see a market at a scale Aliexpress can provide. So I assume those breakout boards contain genuine BME sensors and the price is fair.

I don't think so. They differ in some points, other markings and such.
I've tested really a lot of these sensors against calibrated reference and
that showed that the deviations mostly are bigger then published in the
Bosch datasheets.
Fair price? I've checked the 1000+ price for genuine BME280. A 10.000 pieces reel
gives a price of 2,68€/piece, tax not included. Would be bad business for the
sellers at Ali.

Another possibility is they grabbed the parts rejected by quality control.
Then it might be genuine and scrap.

Well, Aliexpress isn't the market, I agree in that point.
Just these sensors are widespread, used in a lot of home "Weatherstations" and such.
Most of these devices sold on low supermarket levels are assembled in China so
there is a far bigger market then just AliExpress.

This looks like well documented reports on accuracy :
http://www.kandrsmith.org/RJS/Misc/Hygr ... ature.html
http://www.kandrsmith.org/RJS/Misc/hygrometers.html

For outdoor use an aspirated shelter is a must.

Henk

BME 280 shows higher temp sometimes?

I'm not sure how the plugins handle the SI7021 and the BME280.
Both sensors have an internal heating to dry down the humidity sensor gel pad.
This should be used in some interval to keep the sensor working and precise.
The heating obviously influences temperature gauging.

Up to now I've no problems with my BME280 (genuine Bosch, not bought via Ali) compared to my reference (Peaktech 5160)

I personally use 3 Dallas 1820 I calibrated with Lab thermometer as my probe to help calibrate other temp sensors... It works so far.
You can use them with 12bits resolution mode so 0.06°C. Cheap BMx280 from Aliexpress are giving you an idea and need to be calibrated, accuracy & precision being highly debatable. DHT seemed more accurate but lack precision too.

For Humidity, so far I'm disappointed with everything I tried.

Interesting information.

At this moment looking for a sensor to measure Temperature & Humidity in the soil at - 5 cm (or lower) below the surface in my garden. Have some experience with an encapsulated SHT11, which demonstrates that 'drowning' of such sensor is not a good idea:
measurement of temperature is not a problem, and for humidity pulling out & drying the sensor usually remedies the breakdown.
Regardless of the protected cover/capsule, still in general not a reliable configuration for measurement in a continuously wet/moist environment.

Considering accuracy, a better T&H sensor easily compatible with ESPEasy would be type SHT15 or (even better) type HTU21/SI7021.
The SHT15 can be purchased fitted in protective capsule.

Has anyone seen the HTU21/SI7021 as package-deal in such protective capsule?
Or (as DIY-solution), which protective capsule is compatible with (which) breakout with HTU21/SI7021?

Talking temperature and humidity

Anyone found a simple DIY aspirated psychrometer ?
Never searched before on the net, if someone has a link ?

Not that it is urgent but am planning for “future” new weather station (well project 269 from the list )