There are times that the output from mos/pn spectra/back is not optimal. This is usually due to one chip (or one quadrant) being ill-behaved. The multitudinous P-*.qdp files will help you track down which chip or quadrant is causing problems, and will help diagnose the problem with that chip or quadrant. It is also a good idea to look at the P-fovimt.fits files to make ensure that the correct region has been selected and that the point sources have been properly excised. Another important diagnostic is the P-augindiv.qdp which shows the hardness and rate of the observation compared to other observations in the archive, as seen in Figure 15. Observations falling on the edges of the distributions are cause for concern.