Fujii, Tanioka, Akutsu at Mitaka on Apr. 12, 2019
Did calibration of six LVDT units (#6, 8, 9, 10, 11, 12; #7 was used for TMS-VIS-X) to be used for TMS-VIS-Y. The fittings have been done as well.
- Probably due to our way to tune the gains (the details will be described below), there are tendencies that the calibration factors can be categorized to two groups; around plus 7300 counts/mm and minus 6800 counts/mm, which are comparable with the previous results 5106 and 5046,
- The data used to estimate the fitting lines are shown by filled marks.
- Note that the data of LVDT#9 and #11 are almost overlapped for some reason. Probably I did more efforts than required to set the LVDT#11 to the calibration jig in unnecessariy accuracy.
The steps to do each calibration are as follws.
- Using none-zero outputs, somehow tune the phase of the demodulator.
- Try to find almost-zero position by turning the micrometer on the calibration jig to move the LVDT's outer coil with respect to the inner coil.
- Then move the LVDT's outer coil further more by the micrometer to +3 mm from the zero point, and tune the circuit's gain so that the output would be plus or minus 20000 counts (after some trials, we determined to adopt this way).
- Then did the systematic measurements of the output counts with respect to the moving amount of the LVDT's outer coil using the micrometer.
The step number 3 would be the reason why the calibration factors are different among the two groups according to the signs of the calibration factors. Each sign of the output is just a matter of the direction of current flow, the diffrence however is probably due to the ADC's performance, which would saturate (become detached from linear conversion manner) earlier in the "minus" counts than that in the "plus" counts. (And so, I don't think this is due to the mechanical design of the LVDT coils.) As a result, when we set the gains according to step 3, where we shifted "plus" 3 mm from the assumable zero point and make the position corresponding to about minus 20000 counts for those minus-tilt group, we underestimated the output signal due to the non linearity of ADC, so we would set larger gain than those for plus-tilt group.
