The trigger of this search was the pathological behaviour of the fmoderr loop. During the night of the 20 to the 21 we have observed that the offset after the initial fmoderr tuning was increasing, increasing significantly as well, the offset of the error signal that is used for this purpose in Dark Fringe. Knowing that this final loop has a very low control bandwidth, arriving in DF with a big offset means that we can not recover nominal conditions.

Figure 1 shows the initial offset after the Fmoderr tuning was done at the beginning of each lock trial during the night of 20 to 21 (max and min values). It can be seen that there is a trend behaviour towards higher values during the night. There are also four points that show the four succesfull locks, and the initial offset of the fmoderr signal used in Dark Fringe. It can be seen that by the end of the night these offsets become too big to be recovered.

Apart from the fixing of the saturation problem, we have adjusted the tolerance values of the initial fmoderr (reducing them of a factor 10), to try to reduce the offset in Dark Fringe to more reasonable values.