October 9, 2018 at 7:03 pm #6121
My CO2 sensor is mounted outside the house and the temperature swings from 30c summer to -30c in the winter.
The CO2 readings from the sensor seem to randomly move up and down with no correlation to time of day, day of the week, wind direction or temperature…not what you would expect in an urban environment.
Reading the specification sheet ….. the utilized indoor ABC calibration below.
Automatic Baseline Correction (ABC logic function)
ABC logic function refers to that sensor itself do zero point judgment and automatic calibration procedure
intelligently after a continuous operation period. The automatic calibration cycle is every 24 hours after powered
on. The zero point of automatic calibration is 400ppm. From July 2015, the default setting is with built-in
automatic calibration function if no special request.
This function is usually suitable for indoor air quality monitor such as offices, schools and homes, not suitable for
greenhouse, farm and refrigeratory where this function should be off. Please do zero calibration timely, such as
manual or commend calibration.
The spec sheet does have a specified temperature range of 0c to 50c https://www.winsen-sensor.com/products/ndir-co2-sensor/mh-z19b.html
It does indicate that the ABC calibration is not suitable for outside use and is meant for inside use.
Could this ABC calibration method for inside use be why we are getting odd swings in the CO2 values if the sensor is installed outside instead of inside the house?
DavidOctober 9, 2018 at 7:32 pm #6122
I have my KIT1 equipped with that MH-Z19B sensor:
When I started using it I also observed these swings. Since I had to add some new routines to the software I thought I did it wrong.
But after several days the swings got lower and lower and now they are quite stable.
My KIT1 is mounted indoors. The slow rising peaks are people going in the room (its a well ventilated office).
Temperature dependence is another topic, though. That needs a good test once my KIT1 goes outdoor.
WolferlOctober 10, 2018 at 1:40 pm #6129
I looked at some of the recent A3 indiegogo units and found some CO2 sensors that looked like they were behaving.
These units looked like they were mounted indoors ( from the temperature swings )
The units mounted outdoors (again from looking at the temperature swings ) had random excursions of the CO2 like mine.
There should be some daily movement outdoors, but it is usually periodic, or related to a weather event.
The movement on my CO2 sensor like others outside appears to be random.
I was wondering if the CO2 sensor in ABC calibration mode was not suited for outdoor installation as indicated by the spec sheet.
DavidOctober 10, 2018 at 1:58 pm #6130
I notice that your data is stopped every day from 7pm to 1 am .
It is not clear whether this is internet connection or the power to the Kit1
This switching is not typical of the A3 unit and could cause different readings especially if the KIT1 lost power periodically.
I also noticed that your CO2 module reads very low with values much lower than CO2 sensors installed indoors in the recent run of A3.
There are however random excursions of your CO2 values on Oct 5 and 7th…
In both our cases it appears that the CO2 sensor behaviour is a mystery.
DavidOctober 10, 2018 at 2:34 pm #6131
That’s my internet connection. It is turned off automatically in the night.
The short excursions are from people entering the room, doing some stuff there, then leaving. I did not do any calibration.
I agree with you the value are a bit of a mystery. If one takes the only 50ppm accuracy into account, its no longer a mystery 🙂 These sensors aren’t accurate at all. I did expect more from them.
WolferlOctober 10, 2018 at 3:20 pm #6132
Even with 50ppm accuracy your CO2 numbers are below the world numbers https://www.co2.earth/daily-co2
I would guess that even at a ski hill in Austria your CO2 would be higher than the mountain top in Hawaii.
If your numbers are correct you have extraordinary good air in your office.
DavidOctober 11, 2018 at 8:09 am #6133
I’d expect higher CO2 values on mountain tops, since there are no plants that collect the CO2.
I only found this about Austria:
(scroll down about midways to “CO2 Trend”). That measurement are taken from a 3000m high mountain, about 410 ppm.
My KIT1 sensor sits in a countryside valley, lots and lots of green around, no bigger cities or highways around.
I didn’t expect accurate CO2 measurements to be so difficult 🙂
There’s definitely more investigation needed how this sensor must be used and calibrated to be useful.
WolferlDecember 25, 2018 at 9:44 pm #6309
The technology of the MH-Z19B instrument is gas selective and precise. It exploits the phenomenon of absorption by CO2 of infrared ray around 15 microns. It certainly has a Fabry Perot interferometer.
But it is not intended to be used outdoors according to the specifications of Winsen the manufacturer. Moreover, to obtain the announced performance, it is still necessary to calibrate it regularly and scientifically.
The daily calibration (mode ABC) should be deactivated before the first 24h of use, otherwise the factory calibration will be lost. This mode of calibration is that of the “poor man” who does not have a source of reference gas. The sensor considers that the lowest value recorded during the last 24 hours corresponds to 400 ppm which is about the concentration of the external atmosphere in the absence of pollution (campaign). This assumes that during this period, the sensor for at least 20 minutes is placed outdoors in a slightly polluted atmosphere. Placed indoors, if the air is never renewed during this period of 24 hours, the sensor will take low values, > 1000 ppm in general and consider that this corresponds to 400 ppm. From there, the values provided will be erroneous.
A more serious calibration approach is to make a true zero with the Z19B placed in a closed container where a maximum of CO2 has been removed. For example by using charcoal filter (foam for extractor hoods) or lime water or soda placed at the bottom of the jar. The concentration can thus be reduced to a few ppm in a few hours. Be careful not to have condensation or water on the sensor. You can also try using nitrogen, argon or bottled helium if you know the specifications of the producer. One can even imagine a vacuum chamber with a primary vacuum. The sensor will calibrate itself to 400 ppm but whatever, in the software, it will be necessary to withdraw this value to that coming from the UART or PWM to obtain the true value. On the other hand, I do not know how temperature compensation is carried out and possibly humidity that can be sources of significant uncertainties. Small chemical reminders to consult:
http://fred.elie.free.fr/soude_et_gaz_carbonique.pdfDecember 30, 2018 at 2:53 pm #6346
Thank you Michel for the valuable input.
You must be logged in to reply to this topic.