Thermocouple Errors at high temperatures but working perfectly under 1400'Fahrenheit

[TinMaan]

I just finished my set up. Armed with understanding about thermocouple issues, I have done my best to eliminate noise issues throughout the system ( shielded thermocouple wire, proper grounding the controller and components, etc). Any suggestions for trouble shooting " thermocouple unknown error " at high temperatures !?

[Saur0o0n]

Hi,
Well, I would start with checking if thermocouple can work in this temperature. It doesn't matter that K type is design to measure up to 1300C, but this single one can be faulty and stop working at higher temps. There is awfully a lot of cheap ones that are K type but doesn't work above 300/400C. I've seen some, that had poor quality joint on the hot end, and it simply stops connecting at higher temps. You can try different thermocouple or connect this one, to some multimeter (that has K type temperature measurements) - and check with a blow torch. If you eliminated probe potential problems, next step is check MAX31855 - most of the boards around have some knock off chips, that doesn't work well at some point anyway. If this is the case - replace it. Pay more - but you will have less problems in the future. One burning failed is more then paying for original single MAX chip.

[TinMaan]

I have bought so many thermocouples I can start a collection. This specific one I tested with is supposedly an OEM for SKUTT kilns with 8 Gauge rods and double shielded thermocouple wire with optional grounding wire over the shields. Also, for this test I ended up swapping two max31855 breakout boards from Adafruit ( they suppose to be the best ) but I am open for suggestions.
Based on my readings, my system generates a lot of noise - especially in higher temperatures. And that noise somehow contaminates the thermocouple signal to max31855 amplifier. I have even put a 10uF capacitor on terminal of max31855 with no effect in filtering noise.
The power supply ( low noise ) and SSR are housed inside a stripped down all metal mini PC computer box. The ESP32 Wrover and ,max31855 housed inside an older all metal MacMini. All power lines are twisted and kept as short as possible.
Perhaps you should know I am having similar crashes inside another Kiln-Controller ( jbruce12000 ). It has been so frustrating over at jbruce12000 that I gave up and came here. I hope I can complete this project and get to work.

[Saur0o0n]

So if you already know your system is generating noise - this is good starting point. The question is if this is EM noise or some power noise on power line.
If it's EM noise, you could start with moving away MAX board (from the noise source) and shield it with some grounded metal casing. The thermocouple is just generating tiny voltage and it's very easy to distort it, but since you have shielded and grounded cable - interferance could come from the thermocouple itself or from the MAX board. Mine themrocouple (the connection point) is also encapsulated with stainless tube.
If the noise is coming over power line (to the MAX chip), you could start with different power source (and common ground), then you could use completely different power source (split ground) or isolated power module and optocoupler board to transfer SPI signal. This will completely separate MAX from ESP32 board.
Good luck! :)

[TinMaan]

Thank you for the thorough suggestions. I have to admit some of the instructions are over my head and I'll have to get help from a skilled electronic tech.
On the same subject, ( max3185x errors above 900'c temperatures only ) there is an interesting thread that hypothesizing about * air conductivity starting at about those temperatures *. That fact combined with open-sensor thermocouples could be the culprit. So, basically the system works as expected to certain temperatures were heat-works start to turn the air to conductive gases. At that point the heating elements ( and some static electricity ) start to " short circuit " to the tip ( or any other exposed parts ) of thermocouple. The fault subroutines of max3185x start deteriorating with all those spikes eventually reading short circuits. Therefore we receive the dreaded " Short to the Ground " or other flavours of the failed reading error messages.
The obvious solution would be, using an ungrounded K-type thermocouple with ceramic-sheathed-sensor ( sealed ). They are relatively expensive but , hey if it solves my problem it's worth it.
Another solution would be to code the " temperature read cycles " out of sync with " heating cycles " in intervals. But this may require a major rewrite of the software.