Re: tn89

[Tom Droege <tdroege2@earthlink.net>]

Arne and all,

I have done (the last week) some of the things Arne suggests below.

First, everything was measured with a relatively long light shield.  The 
shield extends roughly a focal length beyond the end of the lens and has 
several baffles of increasing size.  The opening angle is large enough for 
the light shield that it should not obstruct the light from a star.

I reduced the data using the master sky flat, a flat with a constant value 
of the mean of the sky flat, and a flat composed of 50% constant and 50% of 
the master sky flat.

The main problem with doing this is that the star finding procedure depends 
on the flat fielding operation.  The master sky flat really does reduce the 
background between the stars to a uniform value.  Since the star finding 
operation looks for stars sticking up above the background, stars are lost 
in the corners when a constant is used for the flat.

I compared the three results using a magnitude vs magnitude sigma for the 
multiple measurements.

V Camera

No real difference for the three flats.

I Camera

Worst result is the master sky flat.

No flat is a slight improvement over the master sky flat.

50% constant, 50% master sky flat gave nearly a factor of two improvement 
at low mag values.  At mag 12 the improvement was negligible, at mag 11 
there was about a 50% improvement.

OK, I don't have familiarity with the tools that would give a quantitative 
measure to the indicated result above.  Perhaps some of you with data set 
24 can study this.

Something is clearly making the center of the field too bright.  A point 
source (star) really does give a different response from the sky.  I 
suspect that the inside of the lens barrel is "shiny" at infra red.  So I 
will take a lens and line it with the flocked paper.  I am certainly open 
to other ways to try to make an improvement in the flat.

BTW, I have not studied this recently, but the light box gives a flat field 
very similar to the master sky flat.  I recall one of the data sets 
distributed has a set of light box flats on it.  So some of you have 
everything needed to study this.  One of the reasons for giving away 
telescopes is that I hope that some of you would grab problems like this 
and use your Mark IV to study the problem.  I find that at my age my energy 
level is pretty low.  So it is hard to crank up to go after something like 
this and still keep an upgrade and production line going.

I have also studied (in the past) sky flats made at different zenith 
angles.  They are roughly the same.  Whatever causes a point source (star) 
plot to differ from a sky flat does not seem to be due to zenith 
angle.  Well, I can say the other effect, whatever it is, dominates any 
zenith angle effect.

I think the system is now good enough to make a survey.  It could be 
better.   I still hope that some of you with Mark IVs will set out to solve 
these problems.

Tom Droege

At 10:42 AM 12/31/02 -0700, you wrote:
>Michael and Tom,
>   The main problem with the technical note is the procedure for
>flatfielding.  Here, the data frames were combined to create
>a master sky flat, which was then applied to the data.  Master
>sky flats will typically have an illumination pattern that is
>dependent on local nearby light sources, and also will have a
>gradient across the frame due to the limited sky coverage seen
>by the Mark IV.  In addition, of course, is any scattered light
>that is present in the system.
>   Scattered light is a real problem with unbaffled systems and
>causes insidious effects.  Tom's attempt to clear up any potential
>scattered light problem by using flocked paper is certainly a valid
>trial.
>   However, I have a few other suggestions as to experiments.
>First, try the *same* reduction procedure, but without any
>flatfielding.  You should get a smooth vignetting pattern, but
>a comparison of that pattern to the flatfielded frames should tell
>you whether the flatfielding is working properly (or is being
>contaminated).  Second, Tom (once upon a time) wrote that he had
>built a lightbox for his system; you might try some of those
>flatfields and ratio with the sky flats to see if there are
>residual systematics.  Finally, point the MarkIV to some completely
>different part of the sky (like -30 or +70 dec) and take another
>set of data; ratio the sky flats from this second set and see if
>there are residuals.  I'd really like to see someone build an
>integrating sphere for the Mark IV (we talked about this also)
>so that we can be really sure that the incident light is uniform
>from at least one source.
>   Flatfielding wide-field systems is the most difficult part of
>the process, as I have mentioned before.  For the Mark III, we were
>able to correct the flats after-the-fact since the cameras were
>absolutely stationary; such techniques won't necessarily work here.
>Arne
>