A CCD pixel can contain about 
electrons before charge begins to spill
into adjoining pixels. This effect is called ``pixel saturation''. The
charge is not lost but appears in pixels surrounding (preferentially
above) the saturated pixel. Exact saturation level varies from pixel to
pixel but, in general, a signal level of DN=235 (compressed) in a full resolution
image is considered
to be saturated, i.e., the response of the CCD becomes non-linear. For half and
quarter resolution, saturation occurs at DN=255 (compressed). These are
the default levels used in SXT_SATPIX which is called by SXT_PREP. It is
probable that DN=235 (compressed) is about 5% conservative as histograms of
optical images indicate saturation setting in at DN=240 (compressed).
The 12 bit analog-to-digital-converter (ADC) of the CCD camera saturates at
about 
electrons (4095 compressed to DN=255). Charge above this level
is lost. Thus, the signal in saturated half or quarter resolution features
is a lower limit.
The integrated X-ray intensity in a saturated full resolution image is not lost-only the spatial information is lost. The sum of all of the saturated pixels plus one pixel above (north) and one below (south) yields the integrated intensity of the area thus sampled. Only in very extreme cases does charge spill east or west into adjacent columns. NOTE that the signal in the pixel immediately above and below saturated pixels is questionable as it will usually contain spilled charge. We believe that the difference in saturation level between SXT X-ray and optical images is caused by the charge-spill effect which is not important for optical images - which have much lower contrast.
The routine SXT_SATPIX can be used to flag the location of saturated pixels.