0
\$\begingroup\$

This teardown of a personal weather station shows an LED which is apparently used for detection of carbon dioxide with an accuracy of +/-5% within 0-5000ppm: enter image description here

All the commercially available carbon dioxide sensors I've found retail for above 100USD, so a low-power, low-cost sensor like this is appealing for IOT sensing (eg greenhouse monitoring). However, I haven't been able to find any information on this strange dome-shaped LED.

Another view:

enter image description here

The teardown says the LED is an infrared emitter/receiver pair: in the enclosure, it is pointed at a white surface, and it measures how much IR light is reflected from the white surface to calculate the CO2. This paper implies the sensor is an NDIR LED, and that its accuracy varies linearly with temperature, but doesn't go into specifics, and searching for NDIR LEDs shows expensive modules with waveguides.

Where could I find an LED like this? What specifications should I look for in an LED (eg spectrum, dome)?

Why haven't other companies done this, if it's so cheap and good?

\$\endgroup\$
3
  • \$\begingroup\$ co2meter.com/blogs/news/… \$\endgroup\$
    – D.A.S.
    Commented Aug 3, 2019 at 2:48
  • \$\begingroup\$ how do you know they aren't using this part? \$\endgroup\$
    – Jasen Слава Україні
    Commented Aug 3, 2019 at 11:58
  • \$\begingroup\$ Most (all?) of the modules I've seen on Digikey/Mouser/the web use waveguides, whereas this detector just uses an LED. \$\endgroup\$
    – Zach Morris
    Commented Aug 3, 2019 at 19:10

2 Answers 2

Reset to default
3
\$\begingroup\$

To measure CO2 accurately you need to monitor the spectrum at 4300 nm AND monitor the spectrum at 3900 nm. The first spectral value gives you an indication of how much CO2 attenuates an optical signal that HAS to be reliably broadband in its spectrum. The second spectral value isn’t affected by CO2 hence, it can be used as a reference measurement should the “reliably broadband” light emission lower its intensity (due to temperature or ageing or other contamination in the signal path).

So, ask yourself; does the circuit/module you have uncovered do what I have described above or, does it do it cheaply and with subsequent inaccuracy?

\$\endgroup\$
9
  • 2
    \$\begingroup\$ The CO2 sensor that I worked with in the long ago measured CO2 + H2O at one wavelength, and just H2O at another, and then subtracted out the H2O signal. I don't think I ever knew the wavelengths, though. \$\endgroup\$
    – TimWescott
    Commented Aug 3, 2019 at 1:46
  • \$\begingroup\$ I think the circuit I uncovered measures CO2 cheaply and inaccurately, but that doesn't mean it's not worthwhile. I found a paper that suggests you can get the advertised accuracy (5% between 0-5000ppm) if you measure temperature and apply a correction factor, so I'm interested in finding a suitable LED and using a temperature sensor to get a cheap and accurate result. \$\endgroup\$
    – Zach Morris
    Commented Aug 3, 2019 at 19:14
  • 1
    \$\begingroup\$ I answered why it was cheap and hinted that it probably wasn’t that good. I also told you what the more expensive and accurate sensors use. Product recommendations are an off topic subject on this site. \$\endgroup\$
    – Andy aka
    Commented Aug 3, 2019 at 19:54
  • 1
    \$\begingroup\$ @MrGerber Hamamatsu make an IR LED that covers the 3900 nm and 4300 nm range and that is hence more attractive in terms of power dissipation and speed of modulation. Not cheap though. \$\endgroup\$
    – Andy aka
    Commented Aug 8, 2019 at 10:55
  • 1
    \$\begingroup\$ @MrGerber I’m an engineer on natural gas analysis stuff. There is a guy from Italy using them who sometimes asks questions on stuff. If you don’t find him I’ll try to remember to comment you in if he asks a question or I stumble across him. \$\endgroup\$
    – Andy aka
    Commented Aug 9, 2019 at 16:36
0
\$\begingroup\$

It turns out there was a mistake in the teardown. The "optical CO2 sensor" is in fact an RGB LED. The outdoor weather station does not measure CO2 at all.

The 5% accurate CO2 sensor mentioned in the Netatmo documentation is this chunky module in the internal weather station: enter image description here

I reached out to Fictiv and they said they would update the teardown. I guess the lesson is, if something sounds too good to be true (eg measuring CO2 with an LED), it probably is!

\$\endgroup\$
1
  • \$\begingroup\$ if you find out more information on this I'd be keen to here about it. To me it looks a bit like a gas thermal conductivity sensor if that's any help. Measuring gas thermal conductivity can allow you to estimate CO2 content providing the rest of the content is O2 and N2 i.e. air. If there is methane or propane thrown into the mix it becomes much more difficult (you might guess my company makes gas analysers!) \$\endgroup\$
    – Andy aka
    Commented Aug 8, 2019 at 7:04

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.