Home Creators Posts Import Register Favorites Logout

Content

A closer look at the insides of the NDIR (non-dispersive infrared) sensor. At the top left of the PCB you can see a small tungsten bulb being used as a source of broad spectrum IR. At the right hand side of the PCB you can see the detector with its notch filter to only pass the specific wavelength range that is absorbed by the CO2.

The more carbon dioxide there is in the air, the less infrared reaches the sensor.

The middle part is the mirrored cover for the PCB that acts as part of the sensing cavity, and the section at the right is the mirrored cover that is the main part of the cavity. The curved section at the top left of the cover reflects the infrared down onto the sensor, but also has a port in it, and a light blocking gas permeable membrane on the other side allows ambient air to enter.

I would guess that the components are an amplifier, possibly voltage regulation and a switching component for the lamp. There is a microcontroller on the other side to deal with analysis, communication and self calibration.

Files

Comments

Mike Hughes

I can only assume that the filament lamp is being pulsed and under run to preserve longevity, as filament bulbs were designed to burn for a set time as a household lightbulb, so that they could keep selling them and have a restricted lifespan, but this thing needs to be reliable for a long time. Have you been able to check the bulb? Its voltage etc and whether it's pulsed? Would be interesting to know. Now I've ordered one off ebay, i'm curious to know if it's going to last a while.

Big Clive

It is just pulsed for regular samples to keep power dissipation low. It will also be run at a dull glow for the IR and not for white light. It should last for decades.

Mike Hughes

Oh That's good, a bit like that light bulb in the US that has been burning for over a hundred years, but under-run.

Keith Who

We had a light bulb in the bathroom that ran for close to 20 years. Obviously not on constantly, but by UK standards, that light bulb should have only lasted a year at best. As you say, designed life expectancy.

Keith Who

Very interesting device. I used to use CO2 and O2 monitoring in anaesthetics when I worked as a Nurse in theatre (OR for these in the USA). I wonder what other gases this unit will react to, have you tried butane the most common aerosol propellant to see if it reacts?

Mike Page

I've got a 40W bulb in my downstairs loo that's 16+ years old - from Poundland!

Mike Page

It's a neat trick: one emitter, one detector, two wavelengths and a known physical phenomenon: a ratio measurement.

Big Clive

Traditional light bulbs have a very interesting characteristic. The harder you drive them, the brighter and whiter they get with an improvement in efficiency, but a logarithmic reduction in lifespan. But if you under-run a lamp it will be massively less efficient, but the lifespan increases exponentially. That ancient forever-bulb in the American fire station is more of a heater than a lamp.

Big Clive

The filter on the front of the sensor is optimised for the exact wavelength likely to be absorbed by a specific gas. By using a different filter it would probably allow other individual gases to be detected.

Big Clive

That's exactly why that bulb has lasted so long. It's more of a heater than a light. I dread to think how much the power has cost to run that old lamp for so long.

Ymir the Frost Giant

Other models by the same brand (this one is no longer listed) claim a 10-year operational life. Makes sense for such a product.

Ymir the Frost Giant

Would this have been an off the shelf module or an inherent part of the design of this unit? 2243 could be a date code.

Ymir the Frost Giant

Wasn't there an agreement between rival bulb manufacturers in the 1920s or so to limit bulb lifetime across the board to 1000 hours?

Matthew Marks

This will be a standard module, and yes, that will be a date code.

Matthew Marks

I'm struggling to see where the other side of the ratio is. According to Wikipedia this should be "another chamber with an enclosed reference gas, typically nitrogen." And presumably you'd need two detectors. Wikipedia doesn't spell this out.

Matthew Marks

There might have been, but if the result is that the consumer pays less overall due to the increase in efficiency, then what's not to like? Also on the lifetime of this sensor, that might be as much due to an estimate of the filter getting blocked, or contaminants that do get through affecting the internals, as anything.

Mike Page

The spectral relationships of a tungsten bulb are well known. CO2 doesn't attenuate the visible spectrum. And you have a more or less common optical path. So you use two optical bandpasses (one IR, one visible) addressing two near-identical photodiodes on a single die and a multiplexer going to a single preamp. Most variables cancel out, leaving you to compare the IR with the visible pickup. (To be fair I only heard tidbits some time ago, and guessed the rest. I could be wrong.)

Mike Page

Yes; I should have said matched.