Weird shaped stars, help!

[almcl]

Been making some changes to my imaging set-up over the last few weeks.  

During these a problem has arisen with strange shaped stars (these occur pretty uniformly across all the subs) :

I think focus is OK (Bahtinov image:

Collimation was as near as I could get it.

The now cooled DSLR and its SW Coma corrector is now  being held in place with a Baader clicklock which has replaced  the standard Skywatcher 2" screw-in draw tube adapter.

Guiding is via an OAG and PHD2 is reporting some of the lowest oscillations I ever seen with my EQ6 mount.  

The direction of the elongation doesn't match either axis (star cross test superimposed on sub:

MaxSelector reports 7% tilt, not brilliant, I know, but I have had worse and got better results.

Where's this distortion coming from, I wonder?

[Thalestris24]

Hi

Could just possibly be from thermal currents - had you had a decent cooldown time? Or could be tilt - can you post a whole jpg?

Another possibility could be movement/flex. What exposure time are your subs? Does the effect change between short/longer exposures?

Louise

ps it's a good idea to align your camera along ra/dec

[almcl]

Thanks for the thoughts, Louise.

Scope lives in an unheated part and had been out for 4 hours when sub below was taken so I hope it had cooled, but I'll check when next out.

Subs were 180 seconds but it also happened the previous night with 120 sec ones.  I was hoping that the switch to the OAG from the ST80 guide scope would eliminate flex, but perhaps it hasn't?

Sadly, aligning camera with guiding axis puts the diffraction spikes at what (to me) is a really irritating angle and the demands of balancing the heavy steel tubed SW200P with Canon and cooler attached means I can't rotate the scope to the point where spikes and axis would be aligned.  (Frustrating hobby at times, isn't it?)

Here's one of last night's subs:

[steppenwolf]

Examining the full size image, I am seeing coma but the classic 'StarTrek' effect has perhaps been modified by some tilt. The star shapes are not consistent throughout the FOV hence my belief that this is two issues manifesting themselves. Your focus looks fine.

[Thalestris24]

Seeing the whole pic, I think it's almost certainly tilt - there is a difference in direction between bottom left and bottom right, the stars at the top don't seem to show much distortion. Maybe someone else can chime in - it's a few years since I last used a Newt... I seem to remember that aligning a dslr so it's at right angles with the long side of the tube and with the dovetail inline with the focus tube (i.e. camera underneath) worked ok. That also facilitated balancing the scope in all 3 axes. I did have problem with inherent tilt on the 130pds. I decided life was too short and moved over to a frac

Louise

[Thalestris24]

Just now, steppenwolf said:

Examining the full size image, I am seeing coma but the classic 'StarTrek' effect has perhaps been modified by some tilt. The star shapes are not consistent throughout the FOV hence my belief that this is two issues manifesting themselves. Your focus looks fine.

I was thinking that Steve - I assumed the tilt was enough to affect the spacing on one side.

Louise

[RichLD]

Not sure whether its my monitor but I'm sure I can see a little bit of purple fringing as well, which would most certainly be an issue with the CC. Maybe spacing/tilt? 

[Thalestris24]

2 minutes ago, RichLD said:

Not sure whether its my monitor but I'm sure I can see a little bit of purple fringing as well, which would most certainly be an issue with the CC. Maybe spacing/tilt? 

Yeah, it could be that the camera + cc is tilted - I presume the cc is screwed into the camera's t-ring? But there could be a problem with the cc in the focus tube. Rob (Uranium235) always recommended fitting a third thumbscrew to the focus tube (of his 130pds).

Louise

[RichLD]

4 minutes ago, Thalestris24 said:

Yeah, it could be that the camera + cc is tilted - I presume the cc is screwed into the camera's t-ring? But there could be a problem with the cc in the focus tube. Rob (Uranium235) always recommended fitting a third thumbscrew to the focus tube (of his 130pds).

Louise

The OP is using a Baader Clicklock which I think would lessen or negate tilt in this area. Perhaps the focuser isn't fully orthogonal?

[Thalestris24]

1 minute ago, RichLD said:

The OP is using a Baader Clicklock which I think would lessen or negate tilt in this area. Perhaps the focuser isn't fully orthogonal?

Could be! The focuser tilt was always a problem with my 130pds. I gave up on it in the end even though I knew it was possible to adjust. The focuser tilt was apparent when trying to collimate. My newts stand in the corner gathering dust... I don't recall anyone having similar problems with a 200p though. Perhaps in this case the problem is connected to the oag?

Louise

[almcl]

Thanks for all the thoughts!  

The CC is screwed in to the Canon OAG which doubles as the camera adapater (image below).  Unfortunately, I can't do collimation with the clicklock  as I lack a suitable adapter.

Any suggestions on how to pursue the tilt possibility would be very welcome?

[Thalestris24]

Hi

Does the cc just slide into the focus tube and held by two thumbscrews? I think I'd just try rotating the camera and see if doing that affects the star shapes.

Louise

[RichLD]

3 minutes ago, Thalestris24 said:

Does the cc just slide into the focus tube and held by two thumbscrews?

The Clicklock uses a compression ring, I believe.

[almcl]

 

2 minutes ago, RichLD said:

The Clicklock uses a compression ring, I believe.

That's it.  There's a small lever on the outside so the cc/ camera can be set at any angle, I'll give that try if the clouds roll away.

The clicklock is shown here https://www.firstlightoptics.com/adapters/baader-2-clicklock-m54-clamp.html

[RichLD]

Try the trial of CCDInspector http://www.ccdware.com/products/ccdinspector/

This will allow you to check you tilt with the camera insitu in real time - not sure if it's more effective than MaxSelector, as I've not used the latter.

HTH

Rich

[almcl]

Thanks, Rich.

I'll check that out.

[almcl]

After digesting all the comments (many thanks to all), I removed the clicklock and drilled and tapped a third hole in a spare 2" draw tube screw-in adapter.  

Then took quite a bit of time to get this modified one set up square, clamped, aligned, collimated and focused.  

Downloaded the trial CCDInspector, spent a happy 45 mins trying to get its install script to run without throwing an error message, and then went to image some random star fields.

CCDInspector reports some pretty huge variations in aspect ratio (enlongation) from 54% in one image to 12% in one taken just 30 minutes later.  Total tilt is surprisingly low some of the time at 1% but as high as 21% at others and collimation was good until I tried to use the 'live adjust' feature, when it went pear shaped.

Here are the stats on one of the images

All of this leaves me not much wiser about the cause of the distortion.  It's nowhere near as bad in this sequence as it was a couple of days ago (which points to the clicklock as being part of the problem) but there is still some distortion in the corners...

I wonder if CC spacing (which has not been an issue before) is now the likely culprit?
 

[Thalestris24]

Sorry, can't help with ccd inspector though I think it needs an even star field.

Louise

[Oddsocks]

2 hours ago, almcl said:

All of this leaves me not much wiser about the cause of the distortion

CCD inspector can give quite variable results for many reasons, thermals, pointing, tracking, seeing etc, etc. The trick is to load at least ten single images, exposure time as short as possible, taken a few minutes apart and with small changes in pointing then select them all simultaneously for evaluation, CCD inspector will output the average of the selected images thereby minimizing random variability. When choosing which images to add to the group measurement use those that show the smallest FWHM score,  click on the FWHM column header to sort the loaded images by score, choose the best and remove the worst then select all the remaining for averaging.

Use the 3d view to get a qualitative indication of curvature together with the 2d view for the quantitive report.

I have found Live Adjust can be a bit unstable with big DSLR images, works well with conventional CCD images that don't need RAW conversion, have never had much luck with Live Adjust with a Nikon but QSI and Starlight Xpress FITs images work, maybe worth posting on the dedicated CCDWare forum to see if this is a known issue and what the answer may be.

HTH. 

[almcl]

Many thanks for that, Oddsocks.

Unfortunately I don't have sufficient images of the same area of sky (amazingly when we have just had seven nights in a row where stars were visible) so I'll try to capture a fair number of short ones next time (Friday, on current forecasts).

I'll have a look at the CCDWare forum, although the registration process looked a bit off putting when I glanced at it while battling Windoze scripting issues.

Here are the stats from six images, is a tilt of 9% serious?

  

[Oddsocks]

The 3D image shows the the spacing is not quite right, almost there, the corners are pointing down which shows overcorrection for coma, without a coma corrector the corners would be pointing up. Try a series test removing 0.5mm spacer from the FF distance and a second series test with 0.5mm added to see which way you need to go, if at all.

The result will never be perfectly flat though, as the corners come up the whole flat plane tends to curve so you have to accept a compromise somewhere.

The tilt direction makes a significant difference to the image, if the tilt is across the short axis, as in your image, the resulting stars at top and bottom won't show much distortion, if the same 9% tilt was across the long axis then the stars at the either end of the axis would show more distortion, the tilt direction and it's magnitude work together in the effect you see in the image and reflectors are very difficult to control in this respect as the main mirror will shift slightly in the mirror cell with pointing angle. A square detector is relatively easy to achieve a good result, an oblong sensor is harder since the tilt direction has a big influence across the full width of the image.

Tilt and curvature work hand in hand in the effect they have on the image extremities with both factors almost having a multiplicative relationship.

The best position for measuring and making adjustment is with the OTA pointing up to the zenith so that he mirror is laying flat in the cell, make your measurements in that position and any adjustments to spacing then test at approx 45% elevation either side of a meridian flip. If the 45 degree test shows significant tilt, and you can see the effect in the images, then you will off on the trail of 'Hunt the Sag'. Refractors are much easier to achieve consistent results from one session to the next in this respect. In the end it just comes down to what your expectations are and whether you can really see a problem with your images.

The 'Tilt' measurement is not directly related to mechanical tilt and this causes new users of CCD inspector a lot of unnecessary angst, tilt is a measurement of the distortion of the FWHM profile of all the measured stars in the image that have the distortion lying in a common plane and periodic error, tracking problems, drive vibration etc can all show as 'tilt' when in fact there may be very little, so keep the exposures as short as possible to minimise this effect and ideally the images should only be luminance filtered, no narrow band or RGB and avoid any large areas of nebulae in the image.

Using the lowest overall FWHM scored images will yield the most accurate results, the hard part then is deciding if any residual tilt is from mechanical or optical collimation sources (or both!).

HTH.

[almcl]

Wow!  Thanks for that most detailed analysis, Oddsocks, really appreciated!

Changing the spacing on the Coma Corrector will require a bit of thought - not sure if I've got any spacers to fit for the increasing case and there aren't any to remove for the decreasing one, but I'll see if I can adapt something. 

I'll have another go with the current spacing, taking lots of short subs near the zenith and see what happens.