Quote
uncle_bob
When you fix the thermistor tables, think about some sort of "transparent" fix. There are vendors out there supplying tables for their thermistors, but not flowing them upstream. Why they do this - I have absolutely no idea. How to implement such a thing (or if it's worth it) - that's up to you.
The problem is that the thermistor tables are generated with a Python script. You need to know the number of bits at the time of generation. As long as I am to stick to generating tables, the best I can do is ship separate 10-bit and 12-bit tables, and if the user has a different thermistor, he'll have to regenerate the right table with the right parameters (I'm using the usual NTC model with R_25C and Beta along with constant R1).
On the other hand, I could try to get rid of tables and instead calculate temperature with a formula, in floating point. This wouldn't be so bad because I do the heater power calculations in the main loop anyway. Only problem would be the heater PWM interrupt where I always check for over-temperature, and for that, I'd need to inverse the formula and precompute how the configured temperature limits translate into ADC readings.
But still, this is not a critical problem, a 10-bit ADC really is enough, with proper control you can get pretty close to keeping the temp around +-2C.
Also, I've experimented with generating the tables at compile using template metaprogramming, but it kills the compiler ;)