Re: [ATM] Corrector/reducer for a fast Newtonian

[David Whysong <dwhysong@gmail.com>]

First, I'd like to thank everyone for the replies. The people on this
list are an incredible resource, and I'm always impressed with the
collective (and individual!) wisdom.

Originally, I restricted myself to SBIG CCDs because they have the
guiding CCD built-in. Having the guiding CCD is in the same physical
camera simplifies the task of guide star aquisition if the imaging CCD
is reasonably large. The automated control software can use the large
CCD to take a snapshot of the field, select a suitable guide star, and
then move the telescope to center the guide star on the guiding
sensor. That's not too difficult if the two CCDs are at fixed
positions (relative to each other) and never move.

Of course, we could always design an off-axis guide port into the focuser.

I had also ruled out FLI cameras because they lacked Linux drivers for
their USB cameras. That appears to have changed. I am aware of the
sampling issues and the benefits of the microlensed arrays. The Kodak
M chips are amazingly good for front-illuminated arrays, and I may
still prefer the ST-8 to the E2V CCD47-10 simply for
cost-effectiveness: the ST-8 is about 70% the size and has comparable
sensitivity, for less than half the price. Still, the Marconi sensor
is a tempting alternative.

I'm also aware of the potential sampling issues with chips that have
less than 100% fill factor, however I don't think this is likely to be
a problem with any of the cameras I'm considering.

I'm looking at the Paul-Baker design - it's similar to the LSST. There
is some light loss due to the combination of a large obstruction and 3
large reflecting elements, and the secondary and teritary mirrors are
very large! Tempting... if we were experienced at grinding optics, I'd
probably do this. The one concern is field curvature, which will
always be a problem (especially at fast f-ratios) while CCDs are flat!

In addition to examining the Paul-Baker configuration, I'm looking
into a Wynne corrector. I'm not knowledgable enough to design one from
scratch, so I'm trying to adapt other designs (I am looking at the
original paper, 1974MNRAS.167..189W).

I tried adapting the corrector assembly from Peter Smith's web site,
but optimizing it for a 28" f/3.67 primary produces lenses with
negative central thickness and positive edge thickness. This is my
first time using OSLO-EDU, it's fun! Do I need to use different glass
types (higher index?) in order to keep the radii from being too steep?
Could a 4-element corrector have similar correction but looser
construction tolerances?

Thanks again for all the help!

- David

On 10/18/05, fhh@npgcable.com <fhh@npgcable.com> wrote:
> David,
>
> As someone whose day job is building CCD cameras for
> astrometry, I want to comment on one of the assumptions
> in your project, namely the CCDs in the cameras you
> mention:
>
> > I know of one design problem that I don't know how to solve: wide
> > field imaging performance. A coma corrector is necessary of course,
> > but the pixel scale (with 9 micron pixels) is 0.6"/pix with the extra
> > x1.15 factor from a Paracorr. That suggests 2x2 binning, which would
> > render the field of view awfully small with an ST-8XME (never mind an
> > ST-7XME!). There isn't room to add a focal reducer between a Paracorr
> > and the focal plane, and I'm not sure how well a focal reducer would
> > work with an f/4.2 incoming beam anyway.
>
> I suggest that the choice of CCD cameras, containing the
> KAF-0402ME (ST-7XME) or the KAF-1603ME (ST-8XME), should
> be looked at in the context of your wanting to do
> astrometry.  Both of these CCDs contain "E"-series CCDs
> with microlens arrays in front (hence the designation "ME").
>
> A Kodak "E"-series CCD has an image area in which each pixel
> is divided into two regions of markedly different quantum
> efficiencies (QE).  One part of the pixel is covered with
> polysilicon, and has a similar QE response versus wavelength
> to the "old" Kodak CCDs predating the "E" and "ME" types.
> The other part of each pixel is covered with indium-tin oxide
> (ITO), which is much more transparent to shorter-wavelength
> (violet, blue, green) photons.  In a poorly sampled optical
> system, say one in which your typical seeing disk diameter
> (FWHM = full-width half-maximum) covers less than two CCD
> pixels, this variation in intra-pixel QE will potentially
> offset the perceived astrometric position in the vertical
> axis of the CCD.  I do not personally know of any serious
> astrometric endeavor that uses Kodak "E"-series CCDs, unless
> well sampled spatially.
>
> The Kodak "ME"-series CCDs takes the "E"-series CCD and places
> on it the microlenslet array, in an attempt to boost the QE
> by focusing most of the incoming photons onto the portions
> of the pixels covered by the ITO.  I do not know of any
> professional literature indicating that any serious astrometric
> endeavor has been attempted with any CCD utilizing a microlenslet
> array.
>
> When you calculate an image scale of 0.6"/pxl, that would be
> fine sampling for the CCDs in the ST-8XME if your seeing does
> NOT go better than 1.2" FWHM, the 1.2" being 2 pixels wide.
>
> I have proposed at work that we purchase Kodak "E" and "ME"
> CCDs to explicitly test for astrometric and photometric
> accuracy, mostly with the idea of informing the serious
> amateur community regarding what performance they can expect
> from such CCDs.  No go so far.  We are about to host Arne
> Henden of AAVSO as a guest observer, and he intends to try
> some commercial CCD cameras on the scope, but with a 7 meter
> focal length he will not be testing a poorly-sampled optical
> system.  Weather permitting, Arne will have taken his data
> within the next week, and I hope he will report back to the
> community soon.
>
> Rather than assume an SBIG camera and then making the optics
> fit, how about an alternative CCD camera that better suits
> your goals with your existing primary and Paracorr?  Perhaps
> a camera containing an E2V CCD47-10 thinned back-illuminated
> CCD with 13-micron pixels giving 1.03"/pixel, with a field
> that is 17.5 arcmin on a side?  A pricier chip, but one that
> is guaranteed to return good astrometry (and photometry).
> If a custom coma corrector / focal reducer is going to cost
> bucks, maybe spend that money instead on the detector?
>
> Respectfully yours,
>
> Fred Harris
> _______________________________________________
> ATM mailing list http://www.atmlist.net/
>


--
David Whysong
Astrophysics Group
University of California, Santa Barbara
_______________________________________________
ATM mailing list http://www.atmlist.net/