CO2 sensor recommendation

Hello,
I just wanted to let you know some information I found over the past few weeks. I asked a few sensors manufacturers about their sensors for indoor air monitoring. I don't have any of the sensors at home yet.

The most reliable sensors of CO2 concentration seem to be the nondispersive infrared sensors (NDIR). An IR light is absorbed by the double bond in CO2 molecules. In a common house environment is practically no cross sensitivity, as other gases like H2O, SO2, NO2 have much lower concentrations. These sensors start at ca. € 100. I only found sensors interfaced via UART, some of them use MODBUS over UART. Powered by 3.3 Volts and have tens of mA consumption during measurement.
The interesting process is a calibration. All sensors have customer calibration possibility, but require proper equipment Since most of the customers don't have sealed chambers with known concentration of some gasses and another measurement device to perform the calibration, the devices are calibrated in factory. However the sensors have some significant drift that requires periodical calibration. That's why the manufacturers came with autocalibration feature. Basically, if the sensor measures CO2 concentration over some period (a week for example), the lowest CO2 concentration measured is considered as a fresh air and all higher values are relative to that.
(I am quite interested to have one in hands to try leave it in a house in the countryside and then place it in a larger city.)

Example: https://www.gassensing.co.uk/products/ambient-air-sensors/cozir-lp-ambient-air-co2-sensor/

There are also some total volatile organic compounds (TVOC) sensors, that are claimed to have the ability to measure CO2 concentration, as well as other pollutants in the air. I googled for trustworthy information and asked one manufacturer of these sensors about the functionality and got quite a surprising answer.

They are suitable for:
1) Environment occupied (and air polluted) only by breathing, nothing else (see below).
2) Detect anything unusual in environment (start of some chemical process, cooking, ...) - either human occupied, or not.

In fact, the sensors have no CO2 sensitivity at all. The correlation between TVOC level and CO2 levels comes from experiments and do not work if the air is polluted by anything not produced by a human (like cleaning agents, cooking, ...) or generally humidity not caused by breathing (ironing clothes, having a shower...). The most import fact is, they produce output that has no unit and meaning - it is just a value and it is up to you to interpret it. You can't determine whether the air is breathable or not.
That could explain the weird measurements that kas.lewis mentioned.

Example: CCS811 / Gas Sensors / Environmental Sensors / Products / Home - ams AG , this is the one features on Thunderboard.

http://uk.farnell.com/silicon-labs/sltb001a/thunderboard-sense/dp/2581458

A useful sensor for home / industrial environment (especially with any combustion device) is a CO sensor. The air can be poisoned by CO and CO2/TVOC sensor won't detect it at all. I guess the sensors can be helpful, however, I would be careful about threshold limit to warn the occupants of the measured environment. For example I found a sensor that outputs a current of 50 ± 20 nA / ppm , which makes it 40 % inaccurate and that could be the difference between a headache and poisoning concentration.

Example: http://www.ddscientific.com/co-sensors.html

Another approach might be measuring the O2 levels. Obviously, O2 is needed by humans to stay alive. The more CO and/or CO2 is in the air, the less oxygen is measured. I found an optical oxygen sensor that outputs partial pressure of O2. If you also measure a barometric air pressure, you can calculate the percentage of O2 in the air.

Example: LuminOx Optical Oxygen SensorsSST Sensing

I will probably order all of the sensors mentioned above and try to make an air quality measurement device.

Note: About two years ago, I bought an SHT21 sensor and use it as thermometer and humidity meter at home. I started watching the humidity carefully and ventilating the apartment at least twice a day. Since that time, I have a much better sleep. According to all the information I've read, it is very probable that humidity, TVOC and CO2 concentration correlate with the human breathing process.

I will definitely be interested in reading about your results.

Been a while since we posted here, as someone noted all NDIR CO2 sensors drift, it is how it is controlled that differentiates manufacturers. At Amphenol we have two types (the only company to have two), one uses Automatic Background calibration, the other has a second internal reference (dual beam or dual wavelength). Read here for more information on which is best for your application.

Farnell and others list our modules. T6613 and T6713 are ABC variants, T6615 is a dual channel.

VOC sensors only offer an equivalent CO2 value, we know because we list one too, the MicS VZ89TE. Ultimately the only real way to monitor CO2 commercially is with NDIR at the moment.

Power is an issue, as the most reliable economic source of IR for sensors remains an incandescent lamp, and they consume pwer, we are doing a lot of work at the moment on reducing consumption. But essentially the common way of reducing it is to measure less frequently, so the downside is reduced response time, which may or may not be an issue. It can also effect any self calibration routine. It should also be noted that in general the sensor modules cannot be turned on and off, as they require a settling time to reach thermal equilibrium.

We also have put up some code to interface to our sensors on github for the Arduino, it is written in C++.

Thank you very much for an additional insight. I noticed there is a RoadTest with T6713-6H CO2 Sensor Module, so I am looking forward to seeing the results. (I didn't apply, because I am just an amateur.)

It took me a few weeks to gather the previously written information and I came to the same conclusion - NDIR is the only usable CO2 measurement technology. By the way, I also found out that Amphenol sensors feature I2C, where CozIR LP from Gas Sensing Solutions only has MODBUS over UART (one more point for Amphenol).

I already got a CCS811 VOC sensor mentioned above and I am afraid it is unusable even in home environment.

Actually the GSS Cozir isn't MODBUS but a simple human friendly protocol so you can easily drive it from Hyperterminal or similar as well as from a micro. This makes it easy to debug systems that use it.

The main difference is power and speed:

Coziir, 3mW average power, 1.2s to first reading no warm up

T6615, 165mW average, 600s to first reading,

This is because the Cozir uses solid state IR LEDs and photodiodes made by GSS in Scotland !

(I must acknowledge an interest here - I do consultancy work for GSS.)

MK

Note for your (and others) clarifiication, our T6713-6HT6713-6H is a T6713 part with a 6 way header on it.

Regards,

WARNING: you might wanna stay away from senseair products!

I was referred to one of their videos/promos  as evidence of CO2 toxicity.

It makes a bunch of absurd misleading devious and false claims. For example, it claims that CO2 levels over 0.5% (5000ppm) are deadly to humans at

https://youtu.be/HutR7V0SVn8 &t=2:36

while displaying stock video showing someone suffocating to death.  Not so truthy.

The concentration needs to be about four times as high to cause any detectable affect in humans. (Source, plus Lots more peer-reviewed toxicity info at https://toxnet.nlm.nih.gov/cgi-bin/sis/search/a?dbs+hsdb:@term+@DOCNO+516  )

Hi Matthew,

The example you give doesn't seem to be a reason to 'stay away' from senseair products. If anything, I appreciate that they made this information available to the layman.

mrlvy  wrote:

 

WARNING: you might wanna stay away from senseair products!

I don't get the reason why you're suggesting such a dramatic warning, when we're all used to recognising the difference between infomercials/marketing content and real test reports. Otherwise we couldn't watch TV!

It's not a medical lecture explaining precisely how to measure that value (is the mentioned 0.5% value an average for instance?). That part of the video which I viewed seems to be an attempt at explaining to a non-expert audience the effects of CO2 at certain levels.

The video just mentions that there is a physical response, and that 'about that' people experience suffocation and other effects, eventually leading to faintness and death.

Given it's not a medical video and their audience could well be engineers who want a value to plug into their CO2 alarms, if I was a customer I'd sleep better knowing the alarm was coded to alert at values well below 0.5%, than anywhere near 2%!

Government information may well use such recommendations too - again, because the website audience won't be people looking for raw medical data based on certain human size and so on.

This website gives the same 0.5% value as a 'time weighted average' as a guideline to work to:

https://www.ccohs.ca/oshanswers/chemicals/chem_profiles/carbon_dioxide.html

If a medic used that website for research it would be silly, but for an engineer to use that as a beginning point toward working out thresholds in their CO2 alarm system, doesn't seem unreasonable on the face of it..

Here are the levels an equipment manufacturer suggests.. (source kane.co.uk). It suggests that some physical effects are noticeable at that value (5000ppm) and below, whereas you're suggesting the level needs to be 20,000ppm for any detectable effect:

Hi Shabaz,

Your table is a bit old - the current atmospheric base level is 406ppm:

https://www.co2.earth/daily-co2

The other numbers are broadly correct.

I did find the video totally weird (doom background music, end of the world tone of narration, people captive in jars, etc etc)- so much so that I wondered if it had anything to do with the CO2 sensor company - but it does seem to be from them.

I don't think it will help them sell sensors !

MK

I stand by my statement :

it claims that CO2 levels over 0.5% (5000ppm) are deadly to humans at

https://youtu.be/HutR7V0SVn8 &t=2:36

while displaying stock video showing someone suffocating to death.

You claim that’s not true. But it is true. and anybody can quickly skip to 2:36 to verify that. so your argument fails.  and your credibility takes a nosedive.

I also stand by my other claim-the claim sourced to NIH, though there is less of an overwhelming consensus behind that claim, and I did use the caveat ‘about’. .. because subtle changes are just that. Reasonable people can disagree over whether a subject is clinically depressed but it’s all a lot harder and rarer to find a situation where they disagree over whether the test subject is alive or dead.

You seem to forget: Correlation does not imply causation. Don't conflate the two.

And MK is right; there’s lots more that’s, questionable in the video.