Privacy and Security Notice
Correcting the NA DNL
During the 2002-2003 G0 commissioning run, three sets of data runs were
taken to measure the white spectra. From these white spectra, the width of each TOF bin can be found. After the width of each TOF bin is found, the TOF spectra for asymmetry data runs taken near in time to a white spectra run can then be reconstructed with minimumilization of the DNL problem. To read about the DNL correction procedure, click here
Below in Figure 1 is the corrected(in red) and uncorrected(in black) TOF spectra from asymmetry run 16002 for Octant 1 Detector 7 that was corrected with white spectra run 16026.
Figure 1: Corrected and uncorrected NA TOF spectra.
An important issue to address is for each white spectra run that was taken, which asymmetry runs does that white spectra repair since the DNL changes with time.
An obvious method to try is to fit the TOF spectra and look at the correlation plot between the chi-square and the inelastic dilution factor(which is taken from the fit results of the TOF spectra). This method does not work possibly because the fits to the TOF spectra (3 gaussians and a possible linear background) are not gaussian. In Figure 2 is an example of the correlation plot between the chi-square and inelastic dilution factor.
Figure 2: Correlation plot between chi-square and inelastic dilution factor.
Since there is not a quantitative method to find when the white spectra will no longer fix the NA TOF spectra, this must be done subjectively by looking at the TOF spectra after it has been normalized and variable binning from a white spectra run has been applied. Below in Figure 3 is a comparison between a TOF spectra that has been corrected and it smooths out the spectra and the TOF spectra that has been repaired with the same white noise run but this time the TOF spectra is not nearly as "smooth" and thus not used.
Figure 2: Both TOF spectra have been corrected using the same white noise run. The top TOF spectra is taken to be a "correct" spectra. The bottom TOF spectra is not as smooth as the top TOF spectra and is thrown out assuming that the DNL has changed .
Since there are only three sets of white noise spectra and each white noise spectra only corrects the asymmetry runs taken near in time, not all asymmetry runs could be fixed. Only about 80 runs were able to be corrected. Below in Table 1 are the white noise runs and the asymmetry runs that each corrects.
| White Noise Run | Asymmetry Runs |
| 15712 | 15744-16035 |
| 15905 | 15830-16035 |
| 16026 | 16002-16035 |
Table 1: The first column is the white noise run and the second column is the corresponding asymmetry runs that the white noise spectra corrects.