Comparison of Type 1 thermistor and E3D thermocouple temperature measurements

[printbus]

REVISED COMPARISON TEST RESULTS
The original post has been updated to explain the issue of an imprecise source of supply voltage for the MEGA2560 board affecting the analog-to-digital converter (ADC) measurements of AD597-type thermocouple amp board output. Marlin configures the ATMEGA2560 processor ADC to use the board VCC as the full-scale reference for the ADC. The AD595/AD597 type thermocouple parts output an analog voltage that is scaled to the temperature present at the thermocouple. Those parts, as do most analog parts, strive for a stable output regardless of variations on their supply voltage. So, use of an imprecise source for 5V power on the MEGA2560 board can, and did in my case, add a substantial error into the ADC measurements.

While not a permanent solution, I set the printer up to use a regulated power supply set to 5.00V as the 5V source for the MEGA2560 and reran the thermocouple-thermistor comparison test. I have diode D1 removed on the RAMPS board, which prevents the MEGA2560 on-board regulator from providing 5V derived from the printer 12V power. I also disable the +5/VCC line in the USB cable from my host computer, eliminating USB as a possible 5V source for the board.

TEST SETUP
Except as noted, the test setup is the same as in the original test. The kapton-wrapped assembly of the thermal mass, heater, thermocouple and thermistor was still intact from the prior test. I did have to rewrap the assembly in the fiberglass mesh.

A Fluke 77 multimeter (0.3% accuracy spec, but uncalibrated) was added to the test setup to independently measure the analog output voltage of the E3D thermocouple board at each setpoint. At a scaled output of 10mVDC per degree C, it can be used to validate the accuracy of ATMEGA2560 ADC measurements and the Marlin interpretation of them.

TEST RESULTS
An updated table of test results is available in this zipped PDF file.

The results show a reasonable track between the scaled analog output of the E3D thermocouple board and the thermocouple temperatures provided by Marlin. This somewhat validates that the analog temperature signal is now being correctly measured by the ATMEGA2560 ADC.

The temperatures reported by the thermocouple and thermistor still diverge with increased temperature, but at a lesser extent than they did in the original test. All the test was intended to do was compare the measurement data. The test configuration did have access to a calibrated measurement reference that could be used to independently prove which approach was more accurate.

As stated in the updated original post, based on the new-found concern over accuracy of the ATMEGA2560 processor supply voltage, I do not recommend the use of an AD595/AD597 type thermocouple amp board. I would also discourage any other printer mod that requires an independently sourced analog voltage to be sampled by the ATMEGA2560 ADC and the existing Marlin firmware without understanding the error that the processor VCC can contribute.