CO2 sensor for low volumes

It sounds like a very tricky measurement.

The Kemet USEQGSEAC82180 uses I2C, is small and about as fast as I've seen (can be sampled at 1 KHz)

Kemet CO2 sensor

The Kemet sensor needs extensive signal processing by the system controller to get a CO2 reading. The spec on their datasheet is very unclear but it looks as if they claim a response time of 10ms. The usual limitation on this klind of device is noise, so the basic snesor response time tells you very little about how long it takes to get a good reading.

MK

Hi, thanks both for your interest and info!
What the Datasheet of the sensor doesn't really tell me is which range the sensor is active onto. Usually, the other sensors' Datasheets specify if they are active either in sub 5% range or towards the full scale of possible reading, having a tradeoff on accuracy.
By the way i found this other KEMET: https://www.mouser.it/ProductDetail/KEMET/USEQGCDAC82M00?qs=xZ%2FP%252Ba9zWqZhxqSk2f0j7w%3D%3D (italian website).

They don't specify the response time, but is for medical applications so I think it must have an adaptation fast enough, doesn't it?
It also specify low noise, I don't understand if it's about measurement noise or communication noise.

Thanks for any further hint!

The problem with that part is that it is not a complete CO2 sensor.

It needs an IR source, a gas cavity , a measuring system (adcs, micropcontoler etc) and a few man years of development to make a working sensor.

Why don't you just replace the ones you have with new ones ?

MK

I actually wanted to exploit this chance to explore better / cheaper solutions, but yeah in the end I think I will stay on the K30 sensor. I'm working on a master thesis project so the project manager wanted me to look around. To make it short, maybe in the end I will actually end up taking an IR source and 3D printing the cavity.

I gotta ask for a confirmation, on this product I have two readings, where the light lands attenuated according to Lambert-Beers law, on both. Why do I need two readings?

I thank you so much for your time and expertise, even if maybe I'm overcomplicating going too much in depth.

I found it from another manufacturer's site:

"The Active detector is sensitive in the range of absorption of the target gas, while the Reference detector wavelength bands are chosen in a region of the IR spectrum where there is minimal absorbance of the gas of interest.

The differential absorption technique, where the target gas is always monitored with respect to a reference measurement, attenuates the effect of background distortions due to response to other gases, source aging or optical surface contamination. Both Active and Reference channels are equally attenuated when contaminants are present within the IR beam or when the source decays over time."

On page 11 of the data sheet you can see that the different devices offered use channel 1 as a non target gas responsive reference channel.  This is because the IR source will not give the same output over temperature and time.

To make a working sensor with these devices you need to take into account the variations in IR device output, CO2 absorbtion (which varies with pressure, temperature and water content) and the variations in the IR detectors. Then you will need a way of calibrating the device. When I mentioned a few man years of development to go from a working IR reciever to a working CO2 sensor I wasn't joking.

MK

For sure you weren't joking, I'm just looking into the realm of possibilities. The main issue I have is, even if I build a perfect reading system, with the collimator, a perfect source and gas flow, I would still need a reference system for the comparison / calibration.

This isnt an area where I have much knowledge but did you look at the Sensiron STC-31?  https://sensirion.com/media/documents/7B1D0EA7/61652CD0/Sensirion_Thermal_Conductivity_Datasheet_STC31_D1_1.pdf

That's an interesting part. Quite slow, (66ms) and poor accuracy at low CO2 levels. It works by measuring thermal conduction (of the gas I think) and is sensitive to thermal gradients but the data sheet doesn't go into details.

MK