Best coefficients for extruder thermistor?

[zapta]

The spreadsheet below computes the Steinhart–Hart equation for semitic NT104 using the recommended coefficients for the Duet boards and the error is maximized at the working range of a typical hot end.

Anybody happens to have another set of coefficients that is optimized for that range?

https://www.mouser.com/datasheet/2/362/P18_NT_Thermistor-1535133.pdf

Thermistor function (1).xlsx

[dc42]

@zapta that's probably because Semitech has changed their range of thermistors. You can use the calculator in the online RRF config tool to calculate the new parameters.

[zapta]

Thanks David, found it.

For a hotend thermistor, does it make sense to use for T1, T2, T3, values from the working range 200 - 250C instead of 25C? (Regardless if the values are from the datasheet or actually measured).

[dc42]

@zapta I suggest something like 25C, 150C and 250C. With these values I think you will find that the errors over the usual working range are very small. There will be a significant difference between the nozzle temperature and the thermistor temperature anyway.

[zapta]

Thanks @dc42. That sounds reasonable.

I tried a few functions, one that gave good results across the range (with the datasheet table as a reference) was a spline interpolation using all the datasheet points.

As for the difference between the nozzle temperature and the thermistor, I presume that this is a also a good argument against switching to more accurate sensors such as PT100[0].

[deckingman]

@zapta said in Best coefficients for extruder thermistor?:

As for the difference between the nozzle temperature and the thermistor, I presume that this is a also a good argument against switching to more accurate sensors such as PT100[0].

Yes, pretty much. I replied in your other thread and mentioned that I did a video and posted it on my YT channel. It's focus was primarily on the effect of deflected part cooling air but the difference between the temperature inside the nozzle and the temperature inside the heater block were orders of magnitude greater than the accuracy errors of any sensor. Hardened steel nozzles were especially bad. Which is why, imo the only thing one can do is print temperature towers to find the best setting for the hot end/nozzle/filament combination.