Track / SC Matching issue in the forward region

[aaust]

Matching charged tracks to the start counter in the very forward region between 96 and 98cm shows a strong paddle (=phi) dependence, e.g. for run 72180 (Spring 2020):

After the start counter was reinstalled in 2022, this phi dependence is offset by 8 paddles, which matches the change in the geometry file of 96 degrees exactly. See for example run 120365 (Spring 2023):

This strong dependence in phi is not reproduced in any MC sample (example from 2017):

I submit the issue here to collect more information about this effect, whether it is caused by an incorrect description of the start counter in the geometry or a calibration or hardware issue. I don't necessarily believe there is a mistake in the reconstruction.

[aaust]

There is a clear phi dependence of the Delta phi variable, which is used to match start counter hits to extrapolated tracks. This plot is only filled for tracks that project on the start counter with z > 96cm:

[aaust]

This phi dependence is completely absent for MC:

However, a simple shift of the start counter by 2mm in x reproduces the observed distribution nicely:

I will try to suppress this effect in real data reconstruction next by moving the start counter in the geometry.