As reported on slack (https://sumire-pfs.slack.com/archives/C3AU3VCMU/p1673573222493039), we found a boxy (or zigzag) feature in observed spectra taken during the Nov run. This is found mainly in 600-700 nm in red arm, but rarely in blue arm. The boxy feature is found in 3 fibers of 85 FLUXSTD fibers in visit 82596. These features are not always in the same fiberIds in a different visit. They are seen both in pfsMerged and pfsArm. The normalization is smooth in these problematic wavelengths.

Hamano-san found that the problematic fibers have a bright fiber next to them. The boxy features could be leaks from those bright fibers.

There are upturns at the sides in fiberprofile, which could cause this issue.