I'd like to remove the tempco from an MQ-2 sensor. Circuitry is all analog (op-amps), minimal complexity. Perfection is not needed. Attached are two plots
- EPCOS Resistance/temperature characteristics
- MQ-2 Rs/Ro
The plots have somewhat different axis, but I think show essentially the same thing; how resistance varies with temperature.
Looking at the EPCOS NTC B=2000 curve and the MQ2 33% RH curve, and ignoring the small difference in "baseline temperature" (25C at top, 20C at bottom) then it seems below 20C, the curves track well but above the MQ-2 curve is much flatter.
What I would like is that the NTC curve is simply some scaled, shifted version of the MQ2 curve, then I could believe it'd easy to implement compensation.