When T-scrunching a file with pam -T, the resulting pulse profile sometimes has as a spike at a particular pulse phase, even though there is no such spike in any sub-integrations prior to scrunching, and this spike is not present when viewed with pav -DT.
This sounds very similar to issue #446, which was fixed a while back. What version of psrchive are you using?
I'm working on a GBO machine, and it seems the PSRCHIVE version is still back in 2020. I'll ask that it be updated and make sure the bug's not happening anymore. Sorry for the possible duplicate report!
This was fixed by updating the PSRCHIVE version.