Celestron Edge HD 1100 First Light(s)
The Celestron Edge HD differs from standard SCTs (differs from the older model C11 for example) in that it has a corrector/flattener lens in the baffle tube.
As with other flatteners the backspace between the flattener and the camera chip is critical if you have a large chip camera. If you have a flattener on a
refractor you have to set the flattener to chip distance carefully but as you focus (in a refractor by moving the flattener+camera assembly in and out to reach
the (fixed) focal plane of the objective lens) the focal length of the refractor doesn't change. With an SCT as you focus using the OTAs focuser knob you are
changing the focal length of the scope and it turns out that focal length is critical too.

In setting this up I had data on the ideal backspace between a landmark at the back of the scope and my chip. You would think if you put the camera chip at
that point there would be only one spot where you could put the mirror (using the OTAs focus knob) and therefore only one focal length at which the image
would come to focus but I found this wasn't precise enough. That may be because small differences in FL make a difference in performance (see below),
because my measurement of the chip placement was in error, or because the stipulated backspace was slightly in error. (Note that in calculating backspace
you have to allow for glass in the optical path for example, and that for a large chip camera with a flattner or reducer there is typically only a tolerance in
spacing of about +/- 1mm, at least for a large refractor flatteners.)

In any case I found it easier and more accurate to proceed like this:

My setup has a PDF focuser in the train, so I can adjust the camera backspace (from the corrector and the focal length (moving the mirror with the OTA
focus knob) intedendently.I can measure the FL with a Plate solve but how to measure the camera chip position?

Look at the two images below.  "4CC" is a crop showing the center and extreme corners of the "Full Frame" image. This lets you more easily compare star
shape and how it changes in different areas of the chip.
(at 2858mm FL)
Full Frame
As you can see in "4CC" the stars in this case are round in the center of the chip but elongated in each of the corners. The elongation is "radial" i.e. pointing
towards the center of the optical axis. (This 4CC method is very useful in trying to sort out elongation that is due to camera tilt or astigmatism for example).

That particular image in 4CC above was taken with the camera chip 19mm farther back that recommended and a plate solve of the full frame gives a FL of
2858mm (nominal for the 1100 is 2800mm).

So next I changed the FL and camera backspace:

a) I removed a spacer
b) I left the PDF focuser where it was (centered in the middle of its range)
c) I focused using a Hartman Mask by turning the knob on the OTA
d) Once focused did a plate solve.

The next attempt gave a FL of 2713 to 2715mm (now slightly under nominal) and the 4CC image at that FL looked as below
(at 2713mm FL)
Note that now the stars are still elongated but the elongation is at right angles to the previous example. It is tangential to the circumference ("Tangential"). If
you are having trouble seeing that elongation look closely at the upper right and upper left corners particularly or download and zoom the image. The that
the FL is not 2800mm and the star shape isn't perfect yet.

This pattern of elongation, Radial elongation on one side of the ideal distance and Tangential on the other side is identical to what I found when trying to get
a large refractor reducer to perfect spacing.

Next I split the difference with spacers and got a mirror (OTA focuser) position where the FL was 2786mm. At this FL the camera backspace was 12mm
greater than the  recommended value:
(at 2786mm FL)
The stars in all four corners now look round to me but I'm not done. I've changed the FL, and the slope of a VCurve is dependent on the FL so I now re-ran
my VCurves at this new FL (2786mm) and refocused using the new, more appropriate V Curve:
(at 2786mm FL)
and with VCurve redone at this focuser position/FL

One possibility is that with change in temperature (it took about 45 min to rerun a few VCurves and then refocus) the tube changed length and therefore FL.
However when I do plate solves on these later images I get the same FL as that on the earlier (before rerunning the V Curves) images, i.e. before and after
are both  2786mm +/- 1 mm. If anybody thinks this is due to temperature let me know but I don't think temperature be the culprit if the FL didn't change. (The
temperature could change the camera backspace by changing the thickness of the adapters but remember I have a PDF focuser in the train and that will
accomodate by moving the camera back to the place where "2786mm FL" forms the image).

With the new, more appropriate VCurves I am getting better focus. The first set of images at 2786mm FL have a FWHM of 2.6" (not terribly good) but the
second set have a FHWM of 1.76". So at the moment I think the most likely answer is that with this more accurate focus and smaller FWHM I now see
elongation that was masked at 2.6". (Astrophotography seems to consist constantly of removing one flaw at which point you now have unmasked, another
smaller flaw that was previously hidden).

Next I will a) change the mirror position to make the FL slightly longer. I know it should be longer because the star elongation is tangential.  Then I will recheck
for elongation in either direction. Note that I'm not certain that the 2800mm value will produce the most perfect stars (it might be at 2820 or whatever) but
from the direction of elongation (Radial or Tangential)  I can tell if I have too long or too short a FL and make finer and finer adjustments. (I had to do similar
fine tuning with that refractor flattener).

My final goal is to find the OTA focuser/mirror postion where the stars are round in the corners at the highest resolution I can obtain. Then I'll lock the mirrors
and use the add on focuser to fine focus and adjust for the variation between filters.

The point of all this is to use the star shape in the corners to find the focal length (mirror position) at which the camera backspace (spacer postion) gives
round stars in the corners. You don't necessarily need an add on focuser thereafter (you could use Robofocus) but being able to make small adjustments in
both speeds the process of finding the sweet spot. In determining backspace for that refractor reducer I actually ran Robofocus on the refractor focuser and
the PDF between the reducer and camera chip to speed the process.

I am down to some pretty small quibbles here.  Only with this collimation error (0.9") camera tilt (0.1") and FWHM (1.76") can I see any problem. I'm down to
"Catholic School" levels of fine tuning.