[vsnet-chat 894] Re: Magnitudes, sequences, etc.

[p.warhurst@auckland.ac.nz (p.warhurst)]

Arne, Brian, Stan, Taichi and all,
                                  much of what has been so far stated in
this thread has been of real interest to me as a novice of photometry. This
post is quite long, but it contains my thoughts on some of the things discussed
I have 2 years experience with CCD phot., using mainly an SBIG ST6 CCD on a
35cm Schmidt Cass. in the CBD of Auckland (pop. 1 million) in New Zealand.
        I agree with Stan Walker (vsnet-chat 881) regarding unfiltered CCD
images going apparently quite deep, but certainly here, with 25 sec
exposures the 13th maybe 14th mag. is as faint as I can go and be happy with
the SDs (ie 0.013 mag over 6 or more hours). 
        Once an even wide passband filter (such as in the BVRI system) is
put in the optical train the Poisson stats don't look too good. I got SDs of
0.013 using a V filter on a V=10.9 cataclysmic binary (TT Ari, Dec 1997),
through ~2 airmass, whereas unfiltered such an accuracy would be for a 13th
mag star, also I am mindful of the fact that this was strictly differential
phot., with the comparison star within the field. At the time all sky phot.
would have been impossible, with thin altocumulus and cirrus across the sky.
I think that the experience of Bessell et al. when designing the BVRI filter
recipes often mentioned, was that an accuracy of 0.01 mag can be attained
with good conditions prevailing and sufficiently bright stars (E Region
standards were used as I recall). 
        Extrapolating from standards at say V=11 down to stars at V=15 seems
unrealistic to me, at least for telescopes with less than ~400 square inches
of mirror, not due to the linearity problems with typical small CCD cameras,
although we are talking a difference in intensity of 40 times, but because
the small number of photons through a BVR or I filter may be swamped by the
noise of a CCD camera with readout noise higher than 15 or 20 electrons, and
dark current of ~60 electrons per sec, when the required exposure times
become as long as required for V=15 stars, and may not be within the
capability of a modest amateur sidereal drive anyway.
        Regarding Taichi Kato's questions about aperture phot software, I
would recommend DAOPHOT II either within ESO MIDAS or the NOAO IRAF, or
MUNIDOS, which uses DAOPHOT II in DOS and is great for time series
differential photometry such as CVs. Careful use of CCDOPS for aperture
phot. will work, and with it it is simple to ensure that the same patch of
sky is used for 'background', but it's far more painful than running DAOPHOT
with a few shell scripts, or the quite automatic MUNIDOS.
        Pretty soon one finds that the best sigmas on the fainter stars are
given by small apertures, maybe only a few FHWM wide, but then there is the
conflict with needing to use say 15 arcsec apertures when observing
standards. Making a growth curve by plotting the fall off in counts with
progressively larger aperture radii for a bright standard, and using the
growth curve function to ascertain corrections for smaller apertures should
be an acceptable way of helping to ensure that whenever standards are much
brighter than program stars, using smaller radii apertures to increase the
faint star S/N ratio does not introduce a systematic error.
        Regarding flatfielding, I find that the best flats I can get are
from using ESO-MIDAS or the NOAO IRAF to median combine a nights CCD frames,
as these compensate well for scattered light (which can be real pain if
trying to make dome flats with a truss tube 'scope for instance), and
importantly use the actual colour of the night sky, not the very blue of
twilight or very deep red of tumgsten filament lamps! When flatfielding with
a screen it is hard to ensure that the effective focal ratio of the
telescope is precisely the same as when recieving parallel rays from the
sky. A pencil of rays emerging from an approximately point source projector,
being refelected by a matt white screen can still suffer an amount of
specular reflection, and such non parallel (and probably off axis) rays
striking the primary will affect the effective focal ratio of the telescope
and thus the size (and location) of the edges of the dust shadow patterns in
the flat. If I take the ratio of appropriately scaled dome and median sky
flats, the dust shadows do not cancel perfectly, when the result should be a
frame devoid of any structure. The result of a dome/sky flat or twilight/sky
flat can also show structure due to the variation in chip sensitivity being
wavelength dependent, so that a red flat is quite different from a blue flat.
       
        Hope this is not too verbose, but I would be glad of any opinions
regarding my comments.

        My thanks to Taichi Kato for this forum,
        cheers to all,

        Paul Warhurst

VSNET Home Page

Return to Daisaku Nogami

---

vsnet-adm@kusastro.kyoto-u.ac.jp