ok so in a nut shell, i am having what i believe to be temp pin errors.
cpuid hwmonitor is giving read outs of 0.0c min and 0.0c max. this is not unexpected, (im running a modified liquid phase change cooler with plenty of goodies to cool everything )
but for some reason i cant get hwmonitor or core temp or any utility to give negative core temps. i do in fact know i am getting negative temps because my external read out screen, that measures the temp on the evaporator, gives a nice -22 to -34c. and when i throttle back the condenser and compressor, turn off the heaters etc. and oc the temps will rise, but i cannot get the temp pin to drop below 0.0c
(1) first is the limitation imposed by the embedded temp sensor in the Proc. Generally these are designed for a min temp of slightly below ambient. Also at the low end the reported value's accuracy is not always great. Reason, problem is at high end (max designed temp. Temps in the range of 15 ->22 C are just not a concern for the "normal" use. Satellite systems where low temps could posse a problem generally use heaters to ensure the electronics are above a set point, ie -10 -> +20 C. These use an external thermistor to regulate temp. Ground testing is done in a Vac chamber with the walls cooled with Liquid N2.
(2) Most software for temp/voltage reporting for computers will establish "rails" for upper and lower expected values. Any reported values out side this will be limited to the established rails. Good example: +12 V Nominal values are 11.4 to 12.6. The rails could be set to say 5 to -> +16. Any value outside this is then reported as the min/max rail. The two most popular metheds of "getting" the temp/voltage is using a polynominal to convert the "reported/sensor reading" and employing a look-up table. The use of a polynominal is limited to the more linear range as the Upper and lower values can become Very nonlinear. Look-up tables, while limited by the same constraints, are also limited by setting the size of the table to "expected/normal" values.