This page investigates why the fringe coefficients are bad, which was first described here : [wiki:FringeInvestigations2011]

The GOOD cell example: o5568g0093o.ota06.fits
The BAD cell example:  o5568g0104o.ota06.fits

To investigate, it is necessary to run ppImage by hand. I used these parameters:


{{{
ppImage -file neb://ipp042.0/gpc1/20110107/o5568g0093o/o5568g0093o.ota06.fits -recipe PPIMAGE CHIP -recipe PSPHOT CHIP -recipe PPSTATS CHIPSTATS -stats stats -dbname gpc1 test.fringe.0093.r28950 -trace psModules.detrend 10  >& test.fringe.0093.r28950.txt
ppImage -file neb://ipp042.0/gpc1/20110107/o5568g0104o/o5568g0104o.ota06.fits -recipe PPIMAGE CHIP -recipe PSPHOT CHIP -recipe PPSTATS CHIPSTATS -stats stats -dbname gpc1 test.fringe.0104.r28950 -trace psModules.detrend 10 > & test.fringe.0104.r28950.txt
}}}


This generates output files with extra debugging information.  There are lines that look like this in the output:
{{{
       F -0.000677 -0.051783 0.492753 0
       F -0.000677 -0.052536 0.593636 0
       F -0.000677 -0.059012 0.491503 0
       F -0.000677 -0.108347 0.379989 0
}}}

from pmFringeStats.c, we see that the F # # # # comes from here:

{{{
psTrace("psModules.detrend", 7, "F %f %f %f %d\n",
                    fringe->f->data.F32[j], science->f->data.F32[j],
                    1 / science->df->data.F32[j],(int) mask->data.PS_TYPE_VECTOR_MASK_DATA[j]);
}}}


so it is:

F   (fringe) (science) (science_err) (mask?)

I grabbed all the F lines, and plotted fringe vs science for the good and the bad cell.  

[[Image(fringeplots-0.png)]]

[[Image(fringeplots-1.png)]]