Re: What Do You Want?

[Andrew Bennett <andrew.bennett@ns.sympatico.ca>]

On Thu, 09 Jan 2003 23:04:56 -0600, you wrote:

>Reading all the analysis, I am contemplating what to provide for the next 
>data set.
>
>The new stiffeners are now in place and should cure the camera tilt.
>
>Some thoughts:
>
>A stack of darks for each night.
The last time I looked, the Dark correlated well
with one from months earlier apart from a level shift 
(electronics temperature?). The Dark does not contribute
much noise to the end product so the gain from
having a stack of them is tiny. At most one needs
a very small stack. I would personally be
entirely happy with one good batch to pinpoint
hot pixels (already done for your current cameras)
followed by just one or two a night.

>
>A stack of flats for each night.

Light box flats are an order of magnitude more
efficient than sky flats because you don't need
a huge number to average and get rid of stars.
A single light box flat at say 20,000 adu above
background gets the photon noise contribution
down to under 0.002 mag (star occupies say 12
pixels -> 240,000 adu per star area ->720,000
electrons. Gee; not much over 0.001 mag.)

So again I would be happy with one (or two) per night ...
it had better be easy to get the light box on and
off just for one miserable image!
>
>A seven by seven data set for each night.

A very good idea, I would think. If one gets to run
often enough, you should even get RA overlap from
night to night. Surely one can do something with
that!
>
>Several nights of data so one can look for long term drifts.

Oh dear - now we are getting to science! I'm
not sure I'm ready for that. I have just run a
program on the DS20 data and spent 2 days staring
at the results in comparison with DS24. At this rate
of progress, I should be ready to look at the first
night's DS25 data in just a few months ... unless
something comes up.
>
>Start placing your orders for Data Set 25.  But put your wants in the order.

Um. My first attempt on DS24 gave a scatter around
0.03 mags for the estimated error in a one degree box
on an image. If that were random and I wanted 0.003, I
would need 100 sets to get good enough statistics.
Ouch! Let's see if 4 sets gives a factor of 2 before I
ask for the other 96!

4 nights x (1+1+49 Dark + Flat + 7x7set) x 2 colours = 408 images

Preferably plus spares to cover the nights I throw away -
one of the DS20 nights was a total disaster, I now find.
>
>Note that the pipeline wants the flats to say "object" so they will be made 
>that way.
>
>The darks will say "dark" because the pipeline wants them that way.  Any 
>other peculiarities?

>
>Tom Droege
>

Andrew Bennett, Avondale Vineyard