[Aavso-photometry] Ensemble photometry question

[Brad Walter]

 Arne, thanks for several good suggestions and for the reference on
ensemble photometry. 

I always use a V filter when using the refractor and usually a V with
the reflector, but sometimes an R or B, depending on the color of the
target and comp stars. I also try to pick comp stars with B-V color
value listings as close to the target as are available in the "vicinity"
and within 2 mag of the target. If I have to choose between color and
mag, I select the comp stars that mag criteria.  I haven't been
correcting for differences in extinction for differential photometry on
exoplanets since only care about the change in magnitude rather than the
actual magnitude. However I notice that below about 35 degrees altitude,
a systematic error starts creeping in, even with stellar  |B-V|  less
than 1.0. 

With regard to least squares analysis, I am making measurements on the
averaged combination of 5 to 10 images, depending on the frequency of
exposure. I think this is equivalent to a 5 to 10 sample "boxcar"
(evenly weighted) averaging filter. Does the least squares combination
reduce the jitter more than a straight average? If I have done my math
correctly the least squares fit occurs for exposures "a" through "n"
when X =(a+b+.....+n)/n which is what I think you get you get when you
average-combine the same exposures.  Have I got it wrong?

-----Original Message-----
From: arne [mailto:arne at aavso.org] 
Sent: Wednesday, January 11, 2006 8:56 AM
To: Brad Walter
Cc: aavso-photometry at mira.aavso.org
Subject: Re: [Aavso-photometry] Ensemble photometry question

Brad Walter wrote:
> A different but related topic
> I have been using a 106 mm aperture 530 mm F/L refractor to gather V 
> band light curve data on exoplanets. My 250 mm aperture 3000 mm F/L 
> scope doesn't have sufficient FOV to obtain suitable comp stars for 
> these bright HD catalog objects. I don't seem to be able to get better

> than about .006 mag precision even though I have very good (several
> thousand) SNRs. I am defocusing to around 4-5 pixel FWHM. I average 
> between 5 and 10 images depending on exposure duration and take 
> readings from the resulting combined photos. I am using 25-30 each 
> flats darks and flat darks. Exposure durations are constant during a 
> session but range over 5 to 20 seconds depending on the brightness of 
> the target and comp stars for the session. I am not auto guiding 
> because I don't want the re-centering delay. Therefore there is slight

> but steady drift of the stars over a session. I suspect that the 
> culprit is scintillation due to the seemingly random variation, but it

> may also be differences in pixel from movement of the stars over the 
> CCD. Total movement over 4 hours is around 20 pixels. Does anyone have

> any suggestions for improving the precision?
> 
The paper by Castellano et al. in the latest JAAVSO ("Detection of
Transits of Extrasolar Giant Planets With Inexpensive Telescopes and
CCDs", 2004, 33,1 and available on our web site) gives some details
about scintillation and how to work with it.  This is most likely your
biggest source of error.  BTW: there are many typos in that particular
paper; our apologies as this is not at our normal quality level.

If you have a photometric night, you can use your larger telescope and
just sit on the bright object, offsetting periodically to a comp star;
differential photometry in the FOV isn't always necessary.  There are
many ways of handling bright stars that have been suggested on this
list.  The one that might work best, but requires much more care, is the
"neutral density spot" to enable using fainter reference stars in the
FOV.
This can sometimes be made by gluing a spot of exposed film on a glass
plate near the CCD.  There are some stars, such as TrES-1, where the
star is faint enough and has a good reference frame that photometry is
easy; and others that we haven't really covered yet where there is a
reasonably bright comp star in the FOV.  Sometimes you just have to wait
and not do every campaign that arises.

Since you are taking lots of images (~10second duration with transits
that take place over 3hrs), don't always get concerned about the
precision of an individual frame.  Least-squares analysis can find
patterns hidden by random noise pretty easily.  I strongly recommend
using a photometric filter for all transit studies.
Arne