The pages below are from the construction manual for a Heathkit gas tester
I built in the '70s.
Heath stopped offering
kits and test equipment years ago, but several similar, improved units are made by Gunson.
The amazing thing about the CI-1080 is its simplicity. The probe consists
of a pair of matched thermistors that are heated by running current through them. One is in a
sealed chamber filled with air, and the other is exposed to exhaust gasses. The carbon monoxide
in the exhaust conducts heat at a different rate than normal air. One thermistor is cooled more
than the other and the unbalanced resistance is displayed on the meter.
The simple mechanical design of the probe assembly is the tester's biggest
shortcoming. The amount of heat lost by each thermistor is very sensitive to gas flow. (In fact,
that's how fuel injection system "hot wire" type air flow meters work.)
So the gas flow around each of the thermistors must be equal for accuracy. The Heath design
relies on a gas flow that is pretty much zero. That makes the response to changes in gas
composition quite ssllugggishhh... Patience is the order of the day.
I get the impression from Gunson's literature that they pump air over the reference
thermistor at the same rate that exhaust passes over the measurement thermistor, quickening response.
Click on each picture to see the full size version. Each page is about 100-200K.