If the subspaces of the two files don't match - e.g. observation A was taken with S5 and that chip wasn't used in observation B - all the events are dropped from the output file (not just those on the chip that's missing).
The bug may also be described as: the tool needs one good time interval (GTI) per chip, otherwise the time for the other CCDs is set to zero.
The *NOM values from the original (unprojected) input file need to be overridden.
The spatial filter is recorded in the file subspace, but it is not reprojected when reproject_events is run. When reproject_events changes the sky tangent point, the region in the subspace is "shifted" in the reference frame.
In many cases, users run into this problem when they do the following order of events:
- Use a spatial filter to select certain ACIS chips (e.g. just the ACIS-I array)
- Run reproject_events on the filtered event file
- Attempt to extract spectra with a source list.
The final output has zero exposure and/or empty GTIs because the source list and the initial, unshifted region filter (stored in the file subspace) don't intersect.
Use the DM subspace-editing capabilities to delete the filter specification written in the file subspace:
unix% dmcopy "reprojected_file.fits[subspace -sky]" reprojected_file_fix.fits
An alternative is to run reproject_events before applying any spatial filtering.