Mk4 reported bad temperature vs real life bad temperature  

RE:

Posted by: @thejiral

I think it was established here that the Mk4 firmware probably needs some changes for temperature control as the results do seem too far off. However, your statement that it all doesn't matter as the temperature on the upper side of the build plate matters fails to acknowledge complexities of reality. That location can not be measured directly and the cooling depends not only on the precise build plate used (there are 3 stock options from Prusa alone) and more importantly on the ambient temperature which can vary a lot, as both scenarios are realistic, inside an enclosure and outside. The corners will always be cooler than the centre as well. 

While I agree that Prusa would need to bring the actual temperature of an examplary stock set-up within less than +/- 10°C, expecting much higher precision is missing the point IMHO. If you really want very high precision and optimal conditions, don't use stock settings (or only use them as a starting point) and calibrate your very specific set up with the filaments you are using. Create your own settings and then it really doesn't matter what the machine shows, as long as the measurements are consistent. 

Everything you said is in place. I would just add the following:

1. Temperature varies across the surface and that is a fact. As long as these variations do not affect a print, they are ok.
2. Temperature is higher on the bottom where the sensor is, than on the top. The two can differ a lot and that could be taken into account by heuristically modifying the resistance vs actual average top temperature formula. That can also very depending on the environment, but with that heuristic correction the error should be lesser than without any.
3. Temperature of the head is much more critical than the temperature of the bed.

Taking all this into account, perhaps there could/should be a disclaimer or a notification somewhere, that the actual temp on the bed can vary +-X from the actual temperature at a given point on the bed.

RE: Mk4 reported bad temperature vs real life bad temperature

Correct me if I am mistaken but the Mk4 firmware does correct the measured thermistor value for the bed temperature, the problem is, if I understand it correctly, that this "correction" would need more finetuning. 

Indeed, the hotend temperature is much more critical. But also there, commonly used thermistors for that job aren't super high precision sensors, the real temperature isn't so critical, the important thing is that these measurements are consistent between printers and that the stock profiles are calibrated against those. The situation at the nozzle is even more complex than at the bed, which is why if you want properly tuned printers, you should do filament calibrations for your very setup. 

RE:

The bad temperature influences the print adhesion. To get the model falling in the middle of a longer print does not make the user happy. Mk3 and MINI do not have those large discrepancies AFAIK.

I looked ať the filamentxx.hpp file and saw some diferenciál. Will double check.

The issue is also that the termistor is used on its lowest sensitivity area. Those NTC thermistors are optimized for room (20°C) temperatures.

RE: Mk4 reported bad temperature vs real life bad temperature

It is well known that those thermistors aren't used at their optimal range. However, this is pretty much industry standard in the desktop printer sector and really not a big deal. I am fairly certain that this is primarily a firmware issue, not a hardware issue. The temperature correction needs to be better tuned. I seriously doubt the hardware setup is significantly different in this regard to that of an Mk3s, different board aside.

This is of course a problem as it makes the print experience for beginners less plug and print and needs to be resolved. However, for proficient printers, which usually do their own calibration at least for each printer but likely also at least for each major filament type they are using as well, it is a non-issue. If the printer shows 10° too low, than the profiles will be simply set to 10°C higher bed temp.