I'm planning to build a temperature controller based on Arduino. I want it for, in conjunction with a k thermocouple, regulating the heat produced by a heating element, in situations from 100ºC to 900°C.
Plenty of moving parts there, I know. But at this momment I have just one specific question. Is there a performance difference, in the sense of accuracy of temperature readings, between Max6675 and Max31855 thermocouple modules?
The MAX6675 outputs 12 bits of data. The MAX31855 outputs 14 bits. That would indicate that the MAX31855 is more accurate. However it has a wider range -250 to +1600 rather than 0 to 1024. So it has about 1 bit more resolution in practice. "1 bit" means it can see changes half the size of what the other can detect.
If you're never going to go outside the 0-1024 range then the MAX6675 is likely to be easier to use. If you also needed to know the temperature of the box where the chip is, then the MAX31855 will give you the "cold junction" temperature (the chip itself) whereas it seems that you can't get this from the MAX6675.
Thank you for your answer. I'm looking into cutting costs wherever possible, however minimum. So that's one reason for hesitating between the two. But also because there might not be a sensible difference between their performances in practice for my choice of temperature interval (which it seems you would agree). And maybe max6675 could have even some untold advantages in practice. Anything in general that one wouldn't necessary grasp from the datasheets.
I'm tending towards Max6675 at the moment. The size of my furnace, latency of my heating element, and specific heat capacities of the materials, make it unlikely that the max discrepancy between temperature of the heating element and the materials will exceed the floor/ceiling measurements of the module.
Hi,
12 or 14 bits does not mean a difference in accuracy, it means a difference in resolution.
Tom.... 
,