Resolved SBIG 16803 - image train rigidity

Thanks Colin, I will proceed on this basis.

 

Hi Colin, here is an update on my trials with the SBIG 16803 and C14 Edge HD. Thank you again for your excellent support.

Incidentally, my whole program for commissioning the equipment was put on hold for the best part of 3 months because where I image in Australia is at risk of bushfire and one started about 10kms to the east of my location. After it had been burning for about a week, I decided that the risk was too high, so I pulled all my equipment out of the observatory and relocated it to our house in town some 40kms away. Fortunately the fire which eventually burned out 17,000 hectares, burned away to the north and east. In any case in the ensuing weeks, we had severe dust and/or smoke so it was just as well I had the gear all safely wrapped up. I have recently reinstalled the equipment as the conditions have eased with some nice rain.
However, I had actually completed enough trials prior to pulling the equipment out to reach a preliminary conclusion. Basically the issue was simply one of not having the correct set back distance. As you may recall I calculated that the adaptor theoretically needed to be 51mm to give me the correct set back distance of 146.05mm. The ".05" quoted seems too precise a figure for a manufactured item such as the C14 Edge HD, so basically I was looking at 146mm. As you may recall I made the adaptor 1mm smaller so I could add back spacers to find the optimum. Incidentally, I measured the thickness of the FW and OAG with calipers which were the only parts of the imaging train I could easily measure. They turned out to be 1mm thicker than the specification for whatever reason.

Here are the eccentricity results:


So the best result was when there was no spacer, which is of course at a -1.0mm variation to the spec. This is consistent with the measurement I made of the FW and OAG. Here is a single sub of NGC 253. Actually the stars are really quite good all the way to the edge and beyond the 40mm diameter Celestron guarantee for the Edge HD technology. The eccentricity of 0.565 probably reflects some very slight elongation in the corners, due to the big sensor of the SBIG 16803, but I would be quite happy with this.



The only issue I had was that I couldn't reduce further as I had run out of room and I would have liked to see the eccentricity going up again to prove I was at the optimum. Therefore I bit the bullet and ordered a 49mm adaptor, so a nett -2.0mm versus the spec. Whilst it was coming I decided to try the Celestron 0.7X focal length reducer. I also had a series of spacers made to add to my 1.0mm and 0.5mm ones that I got from Precise Parts. These are 0.1mm and 0.2mm in thickness. I had originally intended to have the spacers next to the adaptor but the flange on the adaptor is internal and it does not mate up to the back of the OTA, so the only place I could find to put spacers was at the camera. These work very well. The cost to get 10 made was basically the same as getting 1 made.



I then carried out a series of trials with the 49mm adaptor and the Celestron 0.7X focal length reducer. In hindsight I should have completed my Prime Focus trials.



Once again the best results were when I had a 1mm spacer, to give a nett -1.0mm versus the spec, although the eccentricity results were poor right across the board. I was never really that happy with the star shapes, although clearly if I crop to a square with a 40mm diagonal, the results are better.

Here is an image of NGC 1365.



Whilst I may comeback to the 0.7X Focal Length reducer I have had advice that I will never get to satisfactory star shapes even when cropping in. This cropped in version to give the equivalent of a 40mm diagonal (i.e. the Celestron spec) has the same eccentricity as the uncropped version and the stars in the corners are still not satisfactory. If I crop in further the field of view advantage versus Prime Focus is lost, although the exposure times are lower although the pixel resolution is worse.



I was about to go on with more trials at Prime Focus and in particular to explore set back distances in the range -1.0mm to -1.5mm versus the spec to hopefully establish the true optimum i.e. to see the eccentricity getting worse again, when the bushfire crisis emerged. However, now that I have reinstalled the equipment, I hope to continue this work.

In summary then:
1) the issue of loose filters was adequately resolved with the spacers made by Astronomik. The trials I ran with and without filters, showed that they were not a factor in star elongation.
2) collimation was not an issue
3) the orthogonality of the sensor was not an issue
4) guiding was not an issue and an Ha image produced over 30 mins showed no issues of star trails
5) The dimensions of the elements in the optical train are about 1mm greater than the specification as measured with callipers.
6) Provisionally the optimal set back distance was 1.0mm shorter than the theoretical spec, consistent with the physical dimensions as indicated above in 5).
7) The images produced at Prime Focus, when the set back distance was correct, gave satisfactory star shapes right across the FOV and in fact beyond the maximum dimension specified by Celestron.
8) The optimal set back distance for the 0.7X focal length reducer was the same as for Prime Focus. However, the star shapes were not satisfactory at the periphery of the image, even when cropped to meet the Celestron specification maximum.

Best regards, Niall

 

Hi Colin,

Thanks for that.

Yes the specification distances were calculated from the back focus table you provided. It specifies that the 'Camera only without 3" accessory ring' is 35.3mm and that the 'Camera + FW5 or FW7 Filter Wheel without accessory ring' is 68.1mm....for no filters.
This means that the FW itself must be = 68.1 - 35.3 = 32.8mm
The optical backfocus for the Guider as specified in its documentation is 27.9mm. The diagram indicates this is the physical width of the unit.
Therefore the specification thickness of the FW7 and the STX Guider is 32.8 + 27.9 = 60.7mm. I measured the two units bolted together at 61.7mm.

The Astronomik specs indicate that their 1mm thick filters reduce backfocus by 1.0mm and yes I did take that into account.

 

Thanks again Colin....

Here is the website where the optical and mechanical thickness are specified:

https://www.astronomik.com/en/available-sizes/mounted-and-unmounted-filters.html

 

Hi Colin, just a further update. I have finally had some opportunities to continue my trials. Frustratingly I am still not able to get consistently round stars.
My previous work had indicated that the optimal adaptor dimension is about 50mm. This is about 1mm shorter than theoretical spec. By the way I did some more measurements of the FW and auto-guider which indicate they are close to the spec dimension, contrary to what I told you previously. It is not easy to measure but I was more diligent this time....it can be difficult to get the callipers square.
Your point about the optical thickness of the Astronomik filters is a good one. If the optical dimension is 0.33mm as you suggested that would tie in with the optimal dimension I determined. In fact on their website it says that the round filters are 3mm thick with an optical thickness of 1mm, but the 50mm square ones are 1mm thick, also with an optical thickness of 1mm. That doesn't make sense to me.
However, when I recently repeated the tests at the nett 50mm adaptor dimension (49mm + 1.0mm spacer) I was getting average eccentricities of 0.59 - 0.72.
I decided to do a series of exposures very close to the South Celestial Pole as I thought I had a problem with auto guiding which I've now resolved.

Here are 3 images of 300 secs each, obviously unguided and the associated Eccentricity plots.












I also took, 30 sec & 60 sec exposures of the same area as this last 300 sec image, just to confirm that the observed eccentricity was not a function of star trailing. The FWHM figures area bit odd, but the Eccentricity numbers are similar to the 300 sec image.



Notwithstanding that Celestron don't guarantee performance beyond an image circle of 42mm, even within this circle the performance is still not satisfactory.

So, I wonder why the performance got worse in the interim. A few things:

- I have had some problems with the Baffle Lock Nut at the back of the OTA coming out in the equipment. It was not properly glued in place during manufacture. It means I was perhaps a bit tentative on how much I tightened the camera so previous variation may have been due to flexure around this joint. I have now applied loctite to the threads for these latest trials and I am confident the camera is now very secure.

- It is colder now so perhaps the imaging train has shrunk somewhat.

I have covered off the following variables:
- set back distance......previously I had determined that an adaptor + spacer dimension of 50mm is the optimum. The latest results suggest that there may be a need to explore each side of this point, especially if the lower temperatures have changed the set back distance. I did some work looking at thermal expansivity that suggested that the image train could change by 0.1mm in dimension for every 10C change in temperature.
- imagine train tilt. Previously I have seen evidence of this, but I am now thinking this may have been due at times to an insecure connection of the adaptor to the baffle lock nut. These last trials where I glued the baffle lock nut and really tightened the connection hard, show that the eccentricity pattern is regular.
- collimation....I ensure the collimation is perfect before commencing trials
- filters....when removed there was little or no difference,
- tracking....the Eccentricity performance is similar for various exposure durations. Where I have used auto-guiding the reported RMS error is less than 0.5 arc secs and often ~ 0.3 arc secs.

I have not covered off:
- mirror flop....if this is an issue I don't have a viable imaging system

Once I further optimise the set back distance, if there is any evidence of tilt in the image train I will address with shims. I think I am getting there, but sometimes it seems 2 steps forward, 1 step back.

Any thoughts or comments on this.

Thanks,

Niall

 

Thanks Colin.
Firstly thanks very much for the support you have given me on this one. I agree that there is no issue in terms of the camera and from that perspective we can close this thread. If you want to go ahead and do that I will send a link via email.
Cheers,
Niall

 

Hi Colin,

I have been doing further experimentation on my star elongation problem and reviewing some of my previous data. These call into question the conclusion that the issue sits with the OTA.

I would really appreciate your continuing help to navigate this seemingly intractable issue.

Firstly, when I reinstalled my equipment after our disastrous summer and commissioned the 49mm adaptor + 1mm spacer, I had to establish the orientation of the camera when the adaptor was fully screwed in. I chose the orientation randomly and as it turned out the camera was at 90 degrees compared to the trials I had run with the 50mm adaptor. I captured some images, but then rotated the camera wrt the adaptor to return the camera to its previous orientation. This is important because I want the bulky side of the FW to be orientated away from the Feather Touch Focuser. Otherwise when I attach the motor to the focuser, it will foul the FW.
In both cases there was elongation across the image in the same direction...i.e. more in the y direction. Now since I changed the orientation at the adaptor to OAG bolted connection, this implies that the issue is downstream of that connection. i.e towards the camera. Had the issue been upstream...i.e from that flange, the adaptor and the OTA, the problem (orthogonality?) would have stayed in the same physical orientation and because I had rotated the camera the orientation of the elongation would have change by 90 degrees and been more in the x-direction in the image.

In fact when I look back at my images across the trials I have run, they nearly always show a y direction elongation. Sure where the set back distance wasn't correct the elongation may have been different in the corners probably due to field curvature, but there is this consistent elongation in the y direction. I believe the work I did to establish the correct set-back distance is still valid as it was mostly about mitigating the field curvature issue in the corners of the image. However, there is this persistent elongation across the image, which has stopped me achieving an eccentricity of < 0.6. I have also generally observed that the stars in the OAG are elongated. I believe it reasonable to expect some elongation because the stars are at the periphery of the FOV and there is an additional 0.7X focal length educe employed. But on the few occasions where I have good stars they have been pretty round on the OAG as well. Whilst the stars in the images from the main camera usually have a y component to their elongation, they are always elongated in the x direction in the OAG.

In some of my early trials, due to issues with the baffle lock nut at the back of my OTA I may not have been tightening the camera assembly as much as I should, frankly for fear that the baffle lock nut would unscrew from the back of the telescope and get lodged in the adaptor. Therefore I may not have had the connection as secure as it might have been. I noticed for example on one occasion that the stars were elongated when the OTA was pointing to the Zenith but were perfect when the OTA was at 45 degrees. I wondered at the time whether some deflection was in fact correcting some issue such as tilt. The fact that I have had some images with good round stars is important because it says that it is possible to achieve. More on this later.

I can definitely say that the issue is not one of elongation due to trailed stars due to poor guiding. Firstly the OAG generally reports a guiding error of less than 0.6 arc secs RMS. I have seen the elongation in the autoguider with 4 sec exposures, also with very short exposures on the main camera. Equally I have taken images very close to the South Celestial Pole and the elongation was present and finally I recently looked at the orientation of the elongation and it does not correspond to either the RA or Dec axes.

Therefore the issue resides with the optics/ imaging train.



I have a C11 Edge HD, which was my first decent scope. I later upgraded to the C14 Edge HD. A friend suggested that I try the camera on the smaller OTA. I did that in the last couple of nights, using the 49mm adaptor and a 1.0mm spacer, which I believe is close to the optimal set back distance. The C11 has exactly the same fittings and set back distance requirements. I got the same elongation in the y direction on the image.

What is interesting however, is that when I looked at the autoguider image the stars were elongated in the x direction. Not only that they showed the appearance of a double star.




This was also apparent in the main camera when focussing. Note the double peak on the histogram here.

In fact I wondered whether the elongation, even on the C14, is due to a double star situation which is normally blurred out to look like an elongated star.

I then thought I would try the red filter. Interestingly, I still got the elongation, but it was in a different direction.



I can think of two reasons why I would be getting this elongation, beyond the ones we have previously discounted:


1) Tilt
In your experience does image train tilt lead to elongated stars right across the image?

I believe it is possible to have differentially elongated stars due to differences in field curvature across the sensor, given one part is further set back than another. Or could also affect the focus from one side to the other. But can it give stars which are elongated right across the image? If this were the case how could the change of filter from Luminance to Red affect the tilt?

2) Non-perpendicular filters.

Could the issues I have been suffering all along be down to the 1mm Astronomik filters not being retained perpendicular to the light path? As you may recall they weren't retained properly by the clamping plate in the filter wheel and I had Astronomik cut spacers to retain them. I really don't know how well this system is working. Further I wondered whether a non-perpendicular filter could produce a double star/ elongated star from internal reflections from the two surfaces of the glass. But.....the OAG is before the filters and yet there appeared to be evidence of double stars there. Then again the changing orientation from one filter to another suggests that there is a difference from filter and is strongly suggestive that the issue may lie here.

Next steps

Go back to the C14.

I propose to remove a filter and see if this resolves the elongation issue. It may introduce a set back distance issue however, but what I will be looking for is round stars at the centre and be less concerned about stars in the corners.

If this shows the filters are an issue, I will probably need to look at buying 3mm thick filters from another manufacturer.

If the filter is not the issue, I am going to set up my Canon DSLR with a guide scope and do some 1 minute exposures close to the South Celestial Pole to see if I get round stars on its large sensor. I can also set up my ZWO ASI 1600 MC with an OAG to verify that I am able to get round stars, albeit across the smaller sensor.

If these prove out, it will verify there is nothing wrong with the OTA and I would assume that there is tilt in the SBIG camera somewhere and start trials using shims to make the sensor orthogonal.

What do you think?

Niall

 

Thanks Colin,

I appreciate the time you've put into this:

A. Tube Rings
I have looked at these and yes they would make a difference to the pointing accuracy and the RMS error in the Point model. I get an RMS error of about 20 arc secs and i could probably get it to 10 arc secs with Tube Rings. i didn't think it worth the $500 for that.

B. Mirror shift
I will carry out the exercise you suggest.

C. Cable Drag
The only cable that doesn't run through the Mount is the power cable to the SBIG. It is actually too large. I run it carefully in a loop and it does drop to the floor, but it cannot foul anything and I've never had an issue.

D. Focuser Draw tube sag
There is no focuser in the imaging train. Just the adaptor which screws directly onto the back of the OTA and which is bolted to the OAG. I focus with the primary mirror.

E. Mount maintenance
The Mount was regreased in December and the internals were reported to be in perfect condition.

F. PEC Train your MX+
The Mount has been PEC trained.

G. Turn on ProTrack
I use ProTrack

H. TPoint model
I regularly perform a Point model with 200 stars and recalibrate frequently. 300 stars is overkill without tube rings.

I. Refine mechanics and optics based on the TPoint model
I haven't done this up to now.

J. Celestial Pole and Equator
OK

K. Single stars looking like doubles
Usual causes:

• Poor Alignment.........the polar alignment is excellent according to the Point model
• Poor PEC / gears rough or sticking........perhaps I will see this with the other cameras
• Wind gusts......inside a dome and not windy when tests were being carried out
• Cable snagging......taken care of
• Guiding software problems........I get star elongation even on short unguided captures.

Thanks again....much food for thought. I didn't expect it to be this difficult, but I will continue to persevere until I sort it out.

Thans & regards,

Niall

 

One other thing you could try... I once had to replace the fan in my STX 16803 camera. It was exhibiting similar issues to what you are describing.

To try that, just turn off the fan in camera settings. Be aware that this will, of course, compromise cooling. But you should be able to tell whether that makes a difference.

Thorsten

 

Thanks Thorsten.

That is definitely worth a try. The trials I have done tell me that the issue is rotating with the camera, so suggesting this is not to do with the OTA and this is the first avenue of investigation that goes to the camera, outside of the filters.

It also reminds me of some advice that I have had from a very experienced astronomer that the primary mechanisms to generate elongated stars in an image are:

• Guiding/ tracking errors
• Field curvature
• Vibration

I have been up and down with the first two, but this is the first time it has been suggested to me that vibration is a potential cause.

Additionally I have a number of options to trial other cameras which will I hope definitively show whether the issues sit with the SBIG or the OTA.

Cheers,

Niall

 

Roger that Doug.
When the OTA is pointing to the zenith and the visual back is at its lowest position the cord will drop to the floor. I didn't think the cord weight would be great enough to materially influence the pointing and tracking, especially as the camera is so rigidly connected to the OTA, but I take your point.
I will therefore ensure that either the cord length is such that it does not touch the floor, from where it leaves the conduit at the top of the pier, but also making sure that the Mount can go through the full range of motion and have enough length to not drag on the camera at the extremities, or employ some sort of hanger to keep it up off the floor..
Thanks for this.

 

Hi Thorsten,
I have been doing research on my issue and there are a number of threads in various forums which identify this as a not uncommon cause of star elongation/ double images.
I am fairly convinced that the elongation has been in the shape of a double stars all along with the C14 and I have had strong hints of it. It is just that I have been trialing with the Luminance filter and it is both more subject to seeing effects than the Red filter, because it incorporates shorter wavelengths and there will be an amount of atmospheric dispersion…..effectively, I believe blurring out the dumbbells to sausages. The above red image shows the dumbbell shape very clearly. So why would vibration cause this shape? If the vibration is sinusoidal, then the camera is going to spend longer at the peaks and troughs on average than in the transition. I just set up a sine wave in Excel with a peak amplitude of 1.0 and measured the amount of time the wave spends from 0.75 to 1.0 or -0.75 to -1.0 (i.e. within 0.25 of the peak) versus from -0.125 to 0.125 (i.e. 0.25 across the centre) and the result is the time at each peak is 3x that in the centre. Therefore one would expect a dumbbell shaped pattern….or double stars. I suspect the ability to judge them as two separate peaks or just an elongated smudge is a function of the seeing. The seeing was much better for the C11 trials and the double nature of the image was clear. Perhaps the amplitude of the vibration is also greater for the smaller OTA.
Can you tell me what exactly was wrong with your fan? I have inspected mine and it looks perfect. Did you replace it with exactly the same model and did the new one resolve the problem for you completely? Did you do anything to try and dampen the vibration. I have heard of a rubber gasket being placed between the fan and the camera housing for example.
It was also suggested that the filters moving in their cells may be the cause, although the double stars are evident in the guider images which are captured before the filters. Nevertheless I have removed one of the filters to put that one to bed.
Hopefully I will get an opportunity to do the trial with the fan off shortly. In the meantime I would appreciate any further rinfmration you can give me on your issue.
Thanks Niall