i am looking for a kind of accurate humidity sensor which I can trust.
at the moment i have 3 Sensors connected to the same arduino:
BME280
DHT22
HTU21D
I ll get this reading:
Humidity DHT22: 48.30 % BME: 39% HTU21D: 44.97% Temperature DHT22: 29.10 *C BME280: 30.11C HTU21D: 28.98C
All Datasheeds tell that there Sensor has a acuracy of +-2% but there are 10% apart :o
In the datasheed of the BME is a note that it hast du conditioned after soldering at -25°C for 5 day and I doubt the Chinese did that.But still the HTU and the DHT are pretty far apart.
Relative humidity is difficult to measure as it varies by small differences in temperature. You could easily make a test chamber for your sensors using a container of water saturated in table salt (sodium chloride) at room temp. Water saturated with table salt produces approximately 75% relative humidity at room temperature. Search the web for more information about such a test.
I have done the saturated salt water test with a BME280. My test was a fail at first, but I put it away and after a few months I tested it again and it was close to 75%
Why is there a difference of almost 6% between the BME280 and the HTU21D ? Something is wrong. Do you use an Arduino Uno or other 5V Arduino board and are you pushing to much voltage into the BME280 via the I2C bus ?
The humidity and temperature can vary in a room, it might be different in an other corner or at another height. A humidity of 5% accuracy is often enough.
There is no -25°C in the datasheet. I am seeing a tilde, meaning a temperature more or less near 25 degrees: "Ambient Re-Hydration: ~25 °C at >40% rH for >5d". To let the sensor re-hydrate at freezing temperatures would be odd.
The tilde is used a few times in the datasheet. According to Wikipedia a tilde can be "approximately" or "about".
I absolutely love this quote on how to prepare humidity standard solutions for testing purposes:
The best summary I have found of this technique is that of Fossa & Petanga (2003) who describe essentially the same experiment as I perform here for the purpose of testing the hygrometers in CERN's Compact Muon solenoid(CMS). They (sensibly) restrict themselves to using just LiCl (11%), MgCl (35%), NaCl (75%) and KNO3 (92%) instead of the twelve-point calibration I attempt here.
You might very fairly ask why I need a twelve-point calibration for my sensors when they managed to discover the Higgs boson with the CMS and just a four-point calibration.
Thanks I googled for that kind of comperison but didn found it, thanks
Koepel:
I have done the saturated salt water test with a BME280. My test was a fail at first, but I put it away and after a few months I tested it again and it was close to 75%
Why is there a difference of almost 6% between the BME280 and the HTU21D ? Something is wrong. Do you use an Arduino Uno or other 5V Arduino board and are you pushing to much voltage into the BME280 via the I2C bus ?
I definitely need to get a reliable ambient humidity. I ll try the saturated water method today.
I am using an arduino nano and esp8266. With the nano i do experiments, but i want to move everything to esp in the long therm.
I tryed 3.3V and 5V as VCC for the DHT and BME sensor, the HTUs datasheed it tells not to use 5V. I have not thought about the i2c voltage, cloud that be a problem?
Koepel:
The humidity and temperature can vary in a room, it might be different in an other corner or at another height. A humidity of 5% accuracy is often enough.
all the sensors are on the same breadboard.
Koepel:
There is no -25°C in the datasheet. I am seeing a tilde, meaning a temperature more or less near 25 degrees: "Ambient Re-Hydration: ~25 °C at >40% rH for >5d". To let the sensor re-hydrate at freezing temperatures would be odd.
jea my mistake I am thinking about getting finaly glases - its time
If you are serious about accurate humidity then throw this BME280 away and get a new one ! Don't use a 3.3V sensor at 5V. The newer Bosch sensors are sensitive for too high voltages. According to the datasheet the absolute maximum is 4.25V, that means it might be broken when 4.26V is applied.
Also lifting the voltage of a sensor via the SDA and SCL is easy, because the sensors use very little power. It depends on the pullup resistors on the sensor module how much the voltage is lifted above 3.3V. I prefer a I2C level shifter when a 5V Arduino board is used.
With the Nano on a USB cable you are in luck. Suppose the computer has 5.0V, and at the end of the cable the voltage is 4.8V and the Arduino Nano has a diode with extra voltage drop, so the result is that it might run at 4.3V.
Can you read this ? By the way, I was not joking about getting a new BME280 sensor.
Thanks for all your help. I managed to measure 73% humidity over 4 days now. I used an airtight box with table salt saturated water at the button. I suppose I have around 75% humidity in this box.
I ordert a levelshifter but sudenly my arduino nano broke down on every call of bme.begin();
Now i am using a ESP8266 I had laying around which runs purely on 3.3V.
