The FWHM of the 1D projection of the zeroth-order "trailed image" (streak) for HETGS observations has been monitored for five-and-a-bit years. The width of the PSF can be broader for sources with harder spectra, so filtering on E<2keV or E<3keV has been applied in these cases in more recent analyses. From the plot here it appears the FWHM is very constant with perhaps a very slight, slow increase over the years - this is hard to conclude given the use of various sources, however.
The 1D FWHM was measured using the "streak" events as described in Measured and Modelled zeroth-order 1D projections.
[ also available as postscript or pdf versions. ]
Software and Data File
See earlier focus trends updates for onfo on software; the latest focus_trends.rdb file is available here.