I have the TMP007 EasyCal kit, which is working out just fantastic. A little rough around the edges, but works great. The only major issue I'm running into is processing speed when performing sweep based calibration on a large data set with a (relatively) high quantity of steps (30 each).
I want to whip up a quick and tasty CUDA (gpu accelerated) program to rip through the calibration, but I want to make sure I fully understand what the sweep-based calibration is doing.
On the TMP007, Tobj is given by Equations 2, 3, 4, and 5 on the calibration guide, which expanded out, gives
Just from inspection, it appears that the TMP007 Easy Cal software uses the default A1, A2, and C2 (C in the above equation) for analysis, and sweeps S0, B0, B1, and B2.
From a high level perspective, the sweep based analysis takes each data point collected (Vsens, Tdie, and Tref), and applies every combination of S0, B0, B1, and B2. Then it compares the output of the above equation (Tobj) to the measured object temperature, and picks the set coefficients that gives the closest Tobj output over the entire data set.
I have 2 questions:
1) Is my expanded equation correct?
2) Does the TMP007 software use the brute force approach (as described above) to calculate the coefficients? Or does it use some black math magic?
