During the last night of the engineering run, we moved the cobras to study the illumination change. We tried to move six cobras, which were selected by being relatively isolated from the other fibers after the dot-roach procedure.
The cobras we tried to move have fiberid [367, 380, 451, 467, 514, 557]. Cobra with fiberid 367 did not move during the experiment. We have taken 2 focused and 3 defocused measurements. First, we conducted one set of defocused and focused measurements. Then we moved these cobras by phi = -2000. We took only defocused data. Finally, we moved cobras by phi=6000. We took both defocused and focused data.

Out of these five fibers, four fibers showed a very large change in the flux after the first move. We believe that this is because they were partially hidden behind the black spots in the first image. As such, the flux increased rapidly after the cobras were moved. Only fiber 380 was not affected in this manner.

Defocused_image1_image2 and Defocused_image3_image2 show the defocused image for a representative position in the detector. This is a relatively central position, giving us a good symmetric view of the illumination. We see that the movement of the cobra in the positive phi direction changes the illumination in a monotonic manner, pushing more flux to the outsides as phi increases.

Defocused_image_1d_with_models shows the 1d flux as a function of distance from the center of the flux, for all three images and for different simple models. The models are:
model_1 - base model, describing relatively well the shape of the image1
model_2 - model in which I increased frd
model_3 - model in which I also added some misalignment
model_4 - full illumination of the pupil

The lower panel shows the change in flux ratio as a function of the distance. The gray bands are areas from which I would not draw serious conclusions, as they have low flux or a lot of obscurations. The brown line shows the illumination change (between image 3 and image 2, the largest possible change). The black lines show the equivalent change in the modeling images. As you can see, I can somewhat describe the seen change. Having said that, I did this by manually adjusting parameters - because I thought it would be simple - I think that I need to do a proper parameter search if I want to do a high quality analysis.

Defocused_images_modelling_illumination shows the 1d illumination of the pupil that I have used for these models. As you can see, I used a super high FRD ratio - starting at 60 mrad, and moving to 80 mrad. That is why this fall-off of flux towards the edges of the pupil. I don't think this is reasonable. However, if true, a lot of flux would be lost outside of the pupil, which might explain the overall discrepancy between the predictions of the flux and the observed flux. I tried to see if the brighter fibers have less loss of flux in the defocused images, but did not get far.

Jim has provided me with a map that describe the pupil illumination on different parts of the focal plane He promised me that he will also supply me with the connection between the angle in the picture and the position on the focal plane (given that I know x,y position of fibers, not the angle). When he supplies me with that info, I will extract the information from the graph.

I have extracted the information from the graph that JEG sent me. I have created a small function that applies the illumination from the JEG plots to the pupil. The fiber for which we have most information (fiberid = 380) is actually at absolute edge of the focal plane (angle=41), i.e., the brown line on the figure above.