[AZ-Observing] Re: Astroman Site

     Actually I'd have to agree with Jeff about the precision possible
with CCDs versus single-channel.  If you're talking about real external
errors adjusted to a standard system through primary standard stars,
about the best you'll see from a CCD is around 0.015 mag rms.  The
systems of standard stars themselves (either UBVRI or Stromgren or
a few others) are established to about 0.001-0.002 mag globally and
about 0.004 mag star-by-star...all done with photomultipliers.  Folks
who do single-channel stuff can reproduce the systems within a few
millimags and the per-star errors are about 7 millimags (0.007 mag)
when a few observations are averaged.
     Differentially, single-channel folks work fairly routinely at
the 1 or 2 millimag level.  What limits CCD precision is usually
flat-fielding errors at various scales across the chip; there are
other sources of error as well that you just don't have with a
photomultiplier (which can be thought of as a single pixel as far as
calibration goes).
     The trade-off of course is in sheer sensitivity---there's lots of
stuff you can't look at with a tube that you can get (with a lot of
noise) using a CCD.  Science tends to work right at the limits of
whatever technology is available, since new stuff is richest there,
so folks go for the glory and works at the faintest possible limits.
But in the regime where real precision is needed, you gain nothing
from a CCD except a lot of calibration headaches, which are often
skipped either through sloth or because people think CCDs are magic
somehow.  huge amounts of garbage in the literature as a result.
(Photomultiplier observers weren't free of error/sloth either of course.)
Jean-Claude Mermilliod has just published a paper dealing with his
open cluster database that includes a couple of embarrassing tables
showing photometric offsets between different observers of the same
star clusters (all recent CCD surveys) where the offsets are 0.2-0.3
mag., all parties of course claiming to have done proper calibration
against Landolt stars etc and with errors of ~0.02 or so.  Just multiply
by _ten_ and you get the real uncertainties in the data!  Details here:
http://simbad.u-strasbg.fr/cgi-bin/cdsbib?2003A%26A...410..511M
...though most of you won't be able to read the full paper without a
subscription.
     As Steve intimates, a lot of it comes down to observing targets
that are most suitable for the hardware you have, and not try to do
something that's inappropriate.

\Brian
--
See message header for info on list archives or unsubscribing, and please 
send personal replies to the author, not the list.

Other related posts: