Re: Everett and Howell Paper

[<aah@nofs.navy.mil>]

I downloaded the PASP paper (we don't have our private copies yet;
the USPS usually thinks we are somewhere in South America).
Some brief comments before heading off to the Grand Canyon for
a couple of days.
  TASS won't (and hasn't!) achieve this accuracy for several
reasons, some of which are repeated here.
  First, the KPNO mosaic is a thinned, back-illuminated
set of CCDs, so the overlying gate structure is a small contributor
to the pixel response.  That is not the case for the thick Loral
chips we use, which will show a subpixel QE variation of around 10-20percent.
Second, we are marginally sampled, with ~2pixels fwhm, varying across
the field.  This means a different number of pixels contribute to each
star's profile depending on chip position, and for those stars with tight
profiles, the subpixel response will be a large noise source.
Third, we can't reposition the telescope to the nearest pixel on subsequent
nights, or even within a single 'follow' operation on one night.  This
means the pixel-pixel variation has to be characterized extremely well.
Michael has found a 0.02mag floor to Tom's images; I think we will get
better than that, but a reasonable assumption would be 5-10mmag.  If you
defocus and only consider small regions on the chip, and if the field is
really sparse, you might be able to do better.  We can get higher signal/noise
flats with lightbox systems; I've suggested it before and Tom originally
planned on a lightbox when he described the Mark IV concept.  So that
approach is certainly reasonable and will give us another calibration
direction to see if the 0.02mag floor exists that way as well.
  So my general reading is that Everett/Howell have not done any better than Gilliland,
and I did not see any new techniques in their paper, just a confirmation
of what other observers have been able to obtain.
Arne