Have you ever tried compensating tempco? Please just describe how much effort it is to do that.
Hi
Some fairly basic math:
You have a tempco of (say) 20 ppm.
It is measurable / repeatable / predictable to some percentage of that. Let's say 5%
Your math is perfect (it's a formula in a MCU).
Your temperature range is 100C (say -30 to +70)
So:
If you know the temperature to a repeatable 1 C, it is 5X better than you would need for a straight line correction. For a parabolic, you might want 2 or 3C. If you have 1C, it pretty much drops out.
Result would be a part that is compensated to ~1 ppm over a 100 C range.
For under a dollar, you can get a *bunch* of different solid state temp sensors. They have different accuracy ratings, but in general are all very repeatable and stable long term. Resolutions in the 0.06C range are not uncommon. Averaging will get you a bit better than that. Repeatability to 0.1 C is achievable with some care (don't mount the sensor next to the power regulator ...). They are highly linear and hold resolution over the entire range. A NTC plus resistor has great resolution at some central temperature. Not so much at the extremes.
We dumped NTC's almost a decade ago and switched to the solid state gizmos. From what I have seen, there is no reason at all to go back to the thermistors. Our math is about 10X tighter than what I've shown above. We do indeed average the output a bit.
Bob