https://earthobservatory.nasa.gov/global-maps/MOP_CO_M

I currently use this formula, for integrated correction would be for sure far better

• platform: statistics
  name: "min T"
  entity_id: sensor.airq_temperature
  state_characteristic: value_min
  max_age:
  minutes: 2000
  sampling_size: 2000

state: "{{ states('sensor.airq_voc') | float * ( 2 / states('sensor.airq_absolute_humidity') | float ) ** 0.8 - 11 * ( states('sensor.airq_temperature') | float - states('sensor.min_t') | float ) }}"

"The SGP30 features an on-chip humidity compensation for the air quality signals (CO2eq and TVOC) and sensor raw signals
(H2 signal and Ethanol signal). To use the on-chip humidity compensation an absolute humidity value from an external humidity
sensor like the SHTxx is required. Using the “sgp30_set_absolute_humidity” command, a new humidity value can be written to
the SGP30 by sending 2 data bytes (MSB first) and 1 CRC byte."

I have the strong feeling this is not applied, can you please confirm?

I use the correction suggestion from sugo. These values seem much more accurate, when comparing them to the values delivered nativly by airq.

If I think that the Zero-line has shifted and thus a new calibration value is required, I put the airq for around 2 hours outside, and check the values of official measurements. Using them for calibration , it seems.accurate. (anybody without a official station nearby could use https://db.eurad.uni-koeln.de/de/vorhersage/eurad-im.php to get a rough impression for any european location).

Jojo

Option 1 (more likely): The temperature shock has shifted the zeroline permanently. Here you can find how to do calibration manually to correct this shift:

English:
https://support.air-q.com/en/support/manual-calibration-how-do-i-perform-a-manual-calibration/

German:
https://support.air-q.com/support/manuelle-kalibrierung-wie-fuehre-ich-eine-manuelle-kalibrierung-aus/

Here is more info on that:
https://support.air-q.com/en/support/why-is-the-carbon-dioxide-co₂-level-so-high/

If the air-Q is always exposed to those "unusual" conditions like permanently high VOC-value like in a laboratory or medical establishment, it might be useful to deactivate auto calibration like here:
https://support.air-q.com/en/support/how-does-the-automatic-calibration-work/

Option 2: If your measured value does not change over long period of time, it seems it might be blocked or relocated. Sometimes the sensors lose contact to the PCB if the air-Q falls on the ground or is shaked. You can open the air-Q and clean it inside from dust (Tx6 screwdriver needed) and also you could reposition the sensor like described here:
https://support.air-q.com/en/support/a-sensor-has-come-loose-or-is-loose-can-i-reattach-it/

Option 3 (most unlikely): If you get no or stable value that does not change at all from sensor it could be that it might be broken. You can send us diagnostic data from your smartphone or Web App if you think this the case. The best is to do this twice to compare the two measured values with different timestamps from diagnostic data.

After several months, the Air-q shows a very high CO2 value of exactly 5000,0 ppm. This is since around 3 days. When opening windows, or even holding the.airq out of the window, the.value gets for a short time a bit lower, but still much over 4000. This cannot be right. The Airq hasn't changed the position, nor has it been touched when it started that behaviour. The CO2 value cannot get calibrated either. The diagnostics page do also show 5000 for CO2. The description of the sensor from your webpage states, that this sensor has no known cross-sensivity. If there is a failure (software, hardware), it,s still curious why the value gets a bit lower when holding the airq out if the window. It.seems like there would be a calibration in the background active, trying to compensate for the temperaturshock.

What could be the reason why CO2 is at 5000 since 3 days? Is 5000 a special value, like a maximum?

Can you also confirm if and how the CO2 sensor value is compensated? And how to obtain the real value?

The Softwareversions are always kept up to date.

Thank you for your help.
jojo

info@air-q.com

With Firmware Update 1.82 and the new App v1.29.0 many bugs have been fixed and also the auto calibration is now working much better . Just Update to the neweset versions and wait 1 or max. 2 weeks for auto calibration. The more VOCs you have, the longer auto calibration will take because high VOC values will disturb it a little bit.

jojo

docs.air-q.com

for the documentation (enter serial number). There you can see how to read values from all sensors via HTTPS POST, MQTT or in JSON-format. If you want the raw data values from sensors you will find it in the Android- or iOS-App. Go to the Dashboard and open the device options of the particular device you want to read, then select "Diagnostics" and scroll down to "Diagnostic Sensor Data". I think with the API-Access you are also able to read these RAW-values directly out of the device to use them elsewhere or correct them by your own calculations.

In our automatic calibration algorithm there are some cross-sensivities, that will be corrected via values from other sensors of the device via exclusion or difference method, depending on the concentration value/sensor to be measured and the existing cross-sensivities.

You can also create your own dashboard via IFTTT or open HAB with the API-features. Or you take data from HTTPS POST or MQTT to do this.

• Some corrections (such as static offset) can be applied in realtime by the device.
• Other correction data is simply not available at that timepoint (just look for "negative values" ) . These derivations are sometimes severe - my SO2-Values change usually daily between -200 and +400 and take several hours to get reasonable again. Corrections for such things as temperature-shocks can - from my viewpoint - only corrected for the past (depending of the change or shock: some minutes or even several hours). But for this I do require the uncorrected values.
• I've worked with many sensor-types in the past years ; nearly none has a absolut linear response. When the value of the air- e.g. for SO2 should be in the range between 0 and 350ug (1 hour limitation for EU) has a calibration-correction value of currently over +1800 , I expect non-linearity in the response. A real change from 0 to 30 may not result in the same output-value as a change from 1800 to 1830.
• Sensors have always cross-sensitivies. As the effects are sometimes massive (H2S is creates a effect of around 6 for SO2), the calibration is even more important.

Long text short, It would be great if there is option to extract the real sensor values- at least to the science version. This would allow me to be able to create own dashboards, reports and corrections.

since it say 3.8 for calibration

"temperature": {
  "Unit": {
    "humidity": "%",
    "temperature": "degC",
    "dewpt": "degC",
    "humidity_abs": "g/m^3"
  },
  "Part Number": "HDC1080",
  "Manufacturer": "Texas Instruments",
  "Description": "Temperature and Humidity Sensor",
  "Offset": {
    "temperature": -3.8
  },

I did (in home assistant but should be easy to understand)

  - sensor:
    - name: "AirQ Temperature Original"
      state: "{{states('sensor.airq_temperature') | float + 3.8 }}"  
  - sensor:
    - name: "AirQ Internal Saturated Vapour Pressure"
      state: "{{  610.94 * 2.7182818 ** ( 17.625 * states('sensor.airq_temperature_original') | float / ( 243.04 + states('sensor.airq_temperature_original') | float) ) }}"  
  - sensor:
    - name: "AirQ RH original"
      unit_of_measurement: "%"
      state: "{{  0.4615 * ( 273.16 + states('sensor.airq_temperature_original') | float) *  states('sensor.airq_absolute_humidity') | float * 100 / states('sensor.airq_internal_saturated_vapour_pressure') | float }}"

• Well, this is quite important considering the 1:1 sensitivity of O3 to NO2, and since we got a NO2 sensor, looks like obvious to use it, probably better fixing the current stuff than implementing new sensors

"done in the air-Q device"

• after some analisys, see my review on the link above, in the HA forum, I am not sure this is done accurately, it would be nice to have the raw values and/or the formulas you use

"Why do you think so?
without the autocalibration the values ​​would drift over time and could even "jump" permanently when "climatic shocks" occur"

• because a simple correction of the zero value is not enough, I had tried a climatic shock, from outdoor 0 °C to indoor 20 °C and with the compensation I did, it looks everithing works fine, I still have to fix/fine tuning something, but what I mean is: we have a very good device with good sensors, we can and we should achieve better measurement of what we currently receive. Sensors are factory calibrated, which is a damn good stuff, we probably just need a good temperature/humidity/pressure compensation, and a fix of cross sensitivity is a must. Anyway I will go on with my review on HA forum (which BTW is a positive review) providing graph and stuff, at the end I will provide the exact formulas I used for compensation, which seems to work just fine

just my $ .02