NaI detector, discrepancy with experimental data

While waiting for your answer two days ago, I made another attempt to find a solution, I tried to adapted the solution you sent me the first time (mgdraw) with my last input.
NaI_eff.inp (2.1 KB)
NaI_eff.flair (2.5 KB)

The results i obtained compared to the last results you sent me (color red : my work / green : result of article / blue : your work / error : between my result and article ).

I want to ask you what is the reason for this simple difference between what I get and the article, and do you have an idea to improve the results? I increased the number of elementary photons to 100 million. ( The error must be <= 5%).
When I change some values, good results are obtained, but the same problem, some other results remain far from what is required : ( change e-e+ ekin form 1e-05 to 0.001)
544
i get this result :
33
mgdraw :
mgdraw_empty.f (9.4 KB)

Also, when the number of cycles is increased, the values change somewhat.
Thank you