we believe the system is good and reliable now ( the error may about 40 units ). i first checked the signal by changing nothing. i obtains a similar signal strength. thus, i start to change the NMR pulse duration from 0.5µs to 9µs.

the result is good and agree with the fitting curve.

When I depolarize the 1.5 or 1.6us are the 90 degree pulse, the 1st depolarization (the 2nd
measurement) makes the signal go to BG. And 2nd depolarization also BG. Thus, All longitudinal
polarization rotated to traverse polarization.
The last thing is, I fit the curve during experiment and use the fitting curve to predict the FID
area. And the prediction is close to the result. :D