The problem seems to be more pervasive than first appeared. I started Mantid again to load in all the datasets with dead time correction applied to start with and for all the files it gives errors: “No dead time correction applied: Unable to load dead times from the specified file: Cannot open group mantid_workspace_1 of class NXentry (trying to open path /mantid_workspace_1)”. It is obvious from the plotted data that no dead time correction has been applied. Analysing the data also gave the same fits as uncorrected data.
I spoke to James Lord to see if I was doing something unusual to get the problem and he tried some HiFi data. This gave some runs displayed with the dead time correction as set, but others were not corrected.
Just had a helpful conversation with Anthony Lim: We tried to do the dead time correction from file on MUSR00062260 in Mantid 3.8.x on his Windows machine and it worked fine. This seems to have broken in the update to Mantid 3.9.0.
I have created an issue for this Reading DeadTime data · Issue #18950 · mantidproject/mantid · GitHub
Should be in the nightly build now