Back in December I did a test with the latest T400 hardware revision comparing the measured values against values measured by a Fluke 289 multimeter. This was originally posted in the 24 December 2015 Kickstarter update. The related Junction Temperature Errors issue is on Github.
This first graph shows the offset between the voltage measured by the T400 and the voltage measured by the Fluke 289. The purple line in the graph shows a linear relationship between the offset and the Fluke 289 measured value. We can use the slope of this line to determine a calibrations factor. Applying the calibration factor to the T400 readings would result in the red line.
In this test the calibration factor was implemented in firmware and temperature measurements were taken. The graph below shows the offset between the temperatures measured by the T400 and temperatures measured by the Fluke 289. The blue line shows a small temperature offset of approximately 0.3 C with a slight slope. The slope can be reduced by modifying the calibration factor. The constant vertical offset is likely due to me having hardcoded the junction temperature into the firmware to work around junction temperature compensation.
I've implemented the junction temperature compensation, so there is better accuracy now. I have updated data somewhere. I'll find it and post here. For the moment you can expect a measurement accuracy of +/-0.5 C, but I'm pretty sure the updated numbers are more like +/-0.15 C. There are, however, other factors that can affect temperature measurement including the thermocouple accuracy and how well the thermocouple is connected to the measured object.
I'm in the process of writing the specification sheet and user manual. Should be done in the next week or so.