Resolved SBIG 16803 - image train rigidity

I recently purchased an SBIG 16803 with FW7-STX filter wheel and STX Autoguider. I have a C14 Edge HD OTA. I will focus with the primary mirror using a Feather Touch Autofocuser.
I ordered an adaptor from Precise Parts to give me the correct set back distance of 146mm. At first I specified the bolted connection, but they came back to me pointing out that this would give me no ability to rotate the FOV. So in the end I opted for the dovetail connection, moving the adaptor plate from the front of the camera to the front of the STX-Autoguider. The camera came with a sheet indicating that no shims were required for sensor perpendicularity.
A few days back I achieved first light.
Last night I did a series of commissioning trials, primarily aimed at checking the perpendicularity of the sensor. This entailed imaging at different OTA elevations, starting at the zenith. I had auto guiding working very well with a RMS error of no more than 0.5 arc secs.

I measured the star eccentricity with PixInsight to establish the roundness of stars. I only changed the elevation by moving the OTA around the Declination axis.
OTA elevation (degrees) Eccentricity (% elongation long axis vs short axis)
90 0.5 (15%)
48 0.59 (25%)
22 0.56 (20%)

I would like to target <0.5, so I consider this performance marginal. The worsening performance at lower elevations suggests there is some flexure in the imaging train.

I then slewed to a galaxy at an elevation of ~ 60 degrees. A 10 second calibration image gave an eccentricity of 0.74 (50% elongation). After performing a Meridian flip a similar result was achieved. I then noticed the stars were elongated in the 4 sec exposures of the auto guider. Of course, the slew had involved a change in the RA axis as well as the Declination axis.

Finally I made a series of 5 minute exposures. The measured eccentricity was now 0.9, which means the average long axis of the stars was over twice the short axis.

In all cases the elongation of the stars was consistent across the FOV both in terms of direction and extent.

All this suggests to me that the perpendicularity of the sensor has been compromised by flexure of the imaging train. There is one screw connection (adaptor to the baffle lock nut on the visual back of the OTA), the dovetail with 3 grub screws and the rest of the connections are bolted.

I strongly suspect the dovetail connection is the culprit. I would value anyone's experience in this regard and any perspectives on the experiments I have done and the conclusions I have tentatively drawn.

Thanks Niall MacNeill

 

Thanks to you both for your responses.

Firstly here is an image of my set up.

The C14 is atop a Paramount MX+ Mount. I don't use the guidescope seen here except to allow me to shift the weight backwards and forwards to balance in Dec.




Firstly addressing your points Colin.

I don't lock the C14 mirror, because frankly I will be doing the autofocusing using the primary mirror so I can't have it locked up. I looked at using an inline focuser but there is not enough back focus distance to use the focal length reducer when I get to the point of using it. I do see a small amount of image shift when focusing with the primary mirror, but I have never noticed any issue of out of round stars caused by the mirror flop with other cameras that I have used. I don't think this is a likely cause of the star elongation. I have read in various forums of astro-imagers using this configuration and I know that people do focus with the primary mirror for the reasons I mentioned above. Aside from the odd loss of a sub where mirror flop occurs, I have not seen it reported that mirror shift caused star elongation right across the FOV.

The scope is well collimated (see image). I do a lot of planetary imaging and it was collimated just recently with that. As I'm sure you know, excellent collimation is a pre-requisite for planetary. I can't imaging that changing the camera will affect the collimation of the OTA, so please correct me if that is wrong.



All my cables are run through the Mount, except ironically the SBIG power cable. I tried for 2 hours to also run this through but the connectors are just too big to fit through the Mount. You can see the cable in this image. It hangs down in a nice loop and I am 100% sure it was not fouling anything during the imaging run I described.

Doug, thanks for your additional points.

Yes I am familiar with the differential refraction of the light due to the atmosphere. In fact when I am doing planetary imaging below about 50 degrees I use an ADC to compensate. However, where I had the worst elongation the altitude of NGC 6744 was 58 degrees and at that altitude the differential refraction would not be anywhere enough to account for the stars being twice as long as they are wide. I also made a 30 minute Hydrogen Alpha image of NGC 6744 and the measure eccentricity was 0.72.

I had some instances where the stars were relatively round (pointing to the zenith). In this circumstance the star images on the auto guider sensor were also round, so I suspect the elongation was also due to imperfect perpendicularity. The broader point was the elongation was visible in only a 4 second exposure, so confirming that tracking was not the issue.

In summary, I have had the best star roundness when the OTA was pointing to the zenith, with an eccentricity of ~ 0.5. Even then the elongation is worse in the top right and left hand corners.


When I reduced the elevation to 48 degrees, by rotation of the Declination axis ,when there would have been the maximum torque on the imaging train, the eccentricity got worse to ~ 0.6, although the pattern is similar.



After slewing to the galaxy NGC 6744, which would have involved movement in the RA axis and a change in the orientation of the camera, the elongation got much worse, with the eccentricity ~ 0.9. I presume some insecurity in the imaging train manifested itself here causing the dramatic worsening of the elongation and I presume due to the tilt of the sensor.



Here is a DropBox link to a folder into which I have put the FITS files associated with the above images.

https://www.dropbox.com/sh/v7loei1podtuzw0/AABSsoPrSGlZr0a7M-U6Fz4ea?dl=0

Thanks again for your help and I look forward to your thoughts on the above.

Kind regards,

Niall

 

Thanks Colin,

That is an excellent idea. I will do this whilst I get my filter issue sorted out. It removes that as variable and does not affect the perpendicularity tests.

As I see it, if mirror flop affects image perpendicularity as you are suggesting here, then I do not have a viable camera-OTA combination. Having said that when I have come through the focus from both directions, effectively forcing the mirror to flop, the star elongation/ eccentricity is largely the same. I am not convinced this is an issue but time will tell and I bow to your greater experience in this regard.

Interesting, I hadn't thought of that. This is a bit of "which comes first the chicken or the egg". If I can't be sure the SBIG sensor is perfectly perpendicular then using it to measure collimation could be problematic. However I will do some collimation tests with the 16803. It will be interesting to see.

Agreed. There are a number of cables that connect to usb ports etc that are run internal by Software Bisque. Right now the only additional cable I am running is a 12V power supply for due heaters and to run other lower power cameras.

Exactly my thinking.

No the tracking was absolutely perfect here. I used the SkyX Direct Guide routine. The reported RMS tracking error was no more than 0.5 arc seconds. I could also see the guide star in the cropped image frame through the entire exposure and the deviation was small. I believe the star elongation was due to image tilt and I am convinced that the slew to NGC 6744, which moved the Mount in RA, caused the camera to shift and some insecurity in the dovetail connection manifested itself resulting in the camera's sensor moving significantly out of perpendicularity and thus the elongated stars right across the FOV. Exactly as you would expect for bad tilt.

I concur. Did you notice that the stars at the top right hand corner are elongated to the right? Those at the top left corner are elongated slightly to the left. How can this be? To me it suggests field curvature. I suspect this is because the light rays are in fact convergent/ divergent. Normally where the sensor is perpendicular to the incoming light the curved focal plain does in fact cause elongation/ coma at the periphery of the FOV. All SCTs suffer from this if not corrected. However the Edge HD has lenses in the Baffle Tube to compensate for field curvature. Nevertheless, I imagine this only works when the sensor is exactly perpendicular to the light path. If a part of the sensor is further way due to tilt, then it seems entirely plausible to me that the rays will be again convergent or divergent and this could give stars at the corners which are elongated towards those corners.

This is also the case for the NGC 6744 images….i.e. elongation in different directions in the top corners. However in this latter case the overall sensor tilt was worse and swamps this field curvature effect.

In the case where the OTA was pointing to the zenith the stars are much less elongated overall and the field curvature effect is more apparent.

This suggests that even at prime focus it is going to be very important to have the sensor exactly perpendicular to the light path. I presume it will only come to correct focus at the correct set back distance of 146mm, as achieved by moving the primary mirror.

I agree. I think the guiding was excellent, but to be sure then short exposures will eliminate these variables as you suggest. In fact I took two unguided 10 second images of NGC 6744. The first before I forced a Meridian flip and the second afterwards. Have a look at these files. I measured the eccentricity in both images at ~ 0.7 and consistent right across the FOV. What do you think?

https://www.dropbox.com/sh/jtgskdd61t8v8eo/AABuPL1aDLhiVp4ttxr03tfEa?dl=0

This is the adaptor from Precise Parts. The left side of the part in this image screws onto the baffle lock nut at the visual back of the C14. The right side is a the dovetail connection that attached to the adapter plate which was moved from the front of the SBIG camera to the front of the STX Autoguider. This slips over the dovetail connection on the adapter plate and three grub screws secure the connection.



Thanks again for all your help,

Best regards, Niall

 

Just a thought, but are you sure the filter wheel was adequately tightened after loading the filters? If the wheel is a bit loose it would tilt and cause some problems like this. I seem to remember being unclear on how much to tighten that myself initially and it caused problems.

Kevin Morefield


Sent from my iPad using Tapatalk

 

Thanks again for all your helpful input. I swapped backed to planetary yesterday. As I took hold of the camera I did notice a tiny bit of movement. I am guessing that attaching the camera to the dovetail on the OTA is fraught and despite my best efforts it led to there being a small amount of play within the dovetail connection. When I slewed, moving the RA axis this lack of security in the connection manifested itself and the camera was then tilted and out of perpendicularity. The tracking was fine but the camera sensor was tilted resulting in elongated stars across the FOV. This is consistent with the observations.

So clearly mounting the camera onto the adapter which is on the visual back of the OTA is not an option and I need to attach the adapter on the bench top and then screw the whole assembly onto the baffle lock nut. That is difficult! However I think if I point the OTA downwards, I should be able to manage this.

Next steps therefore are to remove one of the filters to ensure the filter movement is not a factor and then to attach the camera as described above before repeating the perpendicularity tests.

 

Thanks Kevin. The filters are Astronomik ones and are 1.0mm thick. As such they cannot be retained by the press plate. I had it screwed all the way down and there was clearly no contact with the filters, so they are free to jiggle in their holders. However, even though this is not ideal I am guessing the filters will be sitting back in their cradles because the camera is pointing upwards. As such I doubt they are the cause of this issue. It is just not good to have this as a variable. I think I have a solution to this problem. Astronomik will supply me laser cut gaskets to retain the filters.

 

Thanks Doug. I saw that. There was sheet indicating what shims were put in place order the plate adapter. I must have been lucky because there was an O with a line through it on the 4 positions on the piece of paper, which I took to mean that the camera's sensor was perpendicular without the need for shims. Is this a reasonable assumption?

 

OK got it. I will check collimation with the SBIG camera. Thanks!

 

Here is an update based on recent experiments.

August 1st 2019
1) Collimation
It was suggested that the issues of star elongation may have been caused by improper collimation of the scope. I have verified that there is no gross issue here.


Across the wider FOV the stars show that the far edge is chopped off in a circular pattern which I believe is due to vignetting and possibly caused by the dew shield. I don’t believe this will cause star elongation. To be confirmed with a similar test without the dew shield.
Any perspectives on this?



August 5th 2019
The dovetail was attached on the bench to ensure the flange on the accessory plate were flush with the adapter and the grub screws as tight as reasonably possible to give maximum connection security. Previously the dovetail was attached to the OTA on the Mount and it was suspected that this resulted in an insecure connection. Certainly as can be seen below the eccentricities are much better than the 0.7-0.8 figures previously achieved, indicating that there had been movement, or more movement, in the dovetail connection previously.
2) Guiding
There was a question as to whether guiding is an issue. The star eccentricity for short exposures was compared to that for longer exposures
With the OTA pointing to the zenith a 300 sec guided exposure gave an eccentricity of 0.62
With the OTA pointing to the zenith a 10 sec unguided exposure gave an eccentricity of 0.57
These results are sufficiently similar to conclude that guiding is not an issue and confirms what the graphs of guiding performance were telling me and that is that the RMS guiding error was < 0.5”.



The stars in the top left corner are elongated from lower right to top left.
The stars in the bottom left corner are elongated from lower left to top right
The stars in the top and bottom right hand corners are elongated in the vertical direction.
The changing direction of the elongation of the stars across the FOV is inconsistent with guiding errors and suggestive of field curvature.
3) Loose filters
The Astronomik filters are loose in their holders and the filters can be heard rattling around when the camera is moved. This calls into question the validity of tests carried out when the filters are in place. The red filter was removed so that tests can be carried out without this potential variable.
With the OTA pointing to the zenith a 300 sec guided exposure with the Luminance filter gave an eccentricity of 0.58
With the OTA pointing to the zenith a 300 sec guided exposure with no filter gave an eccentricity of 0.62 (as above)
When the test was carried out with the elevation of the OTA at 45 degrees, where the filter may have moved, the eccentricity results with and without the filter were also virtually identical.
I conclude there is little or no difference due to the filters being loose in their holders.
4) Angle of elevation
When the OTA is pointing to the zenith there is no bending moment in the imaging train, however at 45 degrees there will be a significant bending moment due to the weight of the camera. Any difference in the eccentricity is indicative of insecurity or flexure of the imaging train
With the OTA pointing to the zenith a 300 sec guided exposure with the Luminance filter gave an eccentricity of 0.46
With the OTA pointing to the zenith a 300 sec guided exposure with no filter gave an eccentricity of 0.45
This was significantly different and in fact better than when pointing to the zenith. The stars in the top left corner again show an elongation from bottom right to top left, but it is much better than before.



This suggests to me that when the camera was pointing to the zenith the sensor was somewhat out of perpendicularity. The tilting of the camera to 45 degrees clearly resulted in a change in position of the camera and in fact fortuitously brought the sensor closer to perpendicularity. The stars in the centre of the FOV are round but become more elongated towards the edges again suggesting field curvature.
As a final test I slewed to NGC 253 to ensure that there was some rotation in the RA axis, since the altitude tests were all done by rotation in the Dec axis. The altitude was 75 degrees and the measured eccentricity was 0.62….similar to the value obtained at the zenith previously.
August 6th 2019
5) Camera angle
I turned the camera to 90 degrees by rotating the camera at the dovetail connection to see if the pattern of star elongation changed. I also had some concerns as to how tightly the adapter was screwed to the visual back. There was some suggestion of looseness when I removed the camera to change the orientation. This time I tightened the screw connection with more torque to ensure it was solid.
With the OTA pointing to the zenith a 300 sec guided exposure with no filter gave an eccentricity of 0.58.



The eccentricity result and the pattern including the direction of elongation are almost identical to the previous test, where the camera was at a 90 degree different angle.
This implies that the flange connection from the dovetail to the accessory plate is orthogonal and the lack of orthogonality is entirely from the camera dovetail backwards to the camera. Any lack of orthogonality in the dovetail flange, the adapter itself and the screw connection to the visual back would have resulted in rotation of the elongation pattern of the stars.

Conclusions:
Collimation, the looseness of the filters and guiding are not contributing to the elongation of stars.
The dovetail connection to the camera (i.e. the dovetail flange to the adaptor) is orthogonal since a rotation about it produced the same pattern of elongation, when the imaging train had the same orientation (i.e. towards the zenith). However, there is a lack of orthogonality upstream of there. This is most likely the connection of the adaptor plate to the auto-guider, since the camera itself was tested as not needing any shims to achieve a perpendicular sensor. I have of course moved the adaptor plate from the camera to the auto-guider.
However there is some movement/ flexure of the imaging train since the elongation changed at different elevations. The most likely culprit is the dovetail connection, since the only other connections are bolted (camera) or screwed (adapter to OTA visual back).
The star elongation is I believe due to field curvature. At prime focus this field curvature with the Edge HD is most likely due to the coma correcting lenses in the baffle tube and the very large sensor of the SBIG camera exacerbates this issue. It is possible to get away with this, as I have done previously, when using a camera with a smaller sensor such as the ZWO ASI 1600MC.
Any lack of perpendicularity manifests itself with a significant worsening of the field curvature and gives differentially elongated stars across the FOV.
The sensitivity of the camera to set back distance is going to require the ability to vary it in order to fine tune it. Further any lack of perpendicularity is likely to require adjustment with shims.
Next steps:
Move away from the dovetail connection to a bolted connection and put up with the inability to rotate the FOV. The sensor is square so that is not big deal.
Design an adapter to give a set back distance 1mm shorter than the design of 146mm. The best place to add a spacer will be where the Precise Parts adapter bolts to the auto-guider, which is bolted to the filter wheel which is bolted to the camera. These should be made from 0.5mm, 1mm and 2mm brass sheet and cut to fit, with 4 holes for the bolts.
The only question here is whether to design 0.5mm short rather than 1.0mm short. Any views on that?
If shims are used they can be made from brass washers to fit over the bolts. Any issue of light ingress due to the tilted connection will be dealt with by duct tape. That will not be necessary hopefully.
Measure as accurately as possible, the actual thickness of the filter wheel and auto-guider with a set of callipers to establish as closely as possible the size of the adaptor.
Once everything is sorted for Prime Focus, repeat for the Celestron 0.7X focal length reducer.
Any thoughts on this?
Thanks again for your help.

 

Hi Colin et al,

Here is an update. As indicated I bought the bolted adaptor from Precise Parts, in order to remove any doubt about the security of the dovetail connection, although I have had feedback from others that this should not cause a problem. In any case the bolted connection is secure. Naturally I had to modify the dimensions to account for the fact that there was no accessory ring dovetail connection.

The set back distance specified by Celestron is 146.05mm.



From the back focus table supplied with the camera, I deduced that the following optical back focus dimensions apply for an STX camera.
Camera only without 3" accessory ring: 35.3mm
Camera + FW7 Filter Wheel without accessory ring: 68.1 mm
The STX Guider User's Manual specifies that it has an optical back focus distance of 1.1" (27.9mm)
The Astronomik 1mm filters give an equivalent optical set back distance of -1.0mm, effectively increasing the required adaptor length by 1mm.

I also designed the adaptor to be 1.0mm short so that I could add spacers with flexibility to go up and down in dimension.

Here are the calculations:



So I bought an adaptor with a set back distance of 50.0mm. I also bought 1.0mm and 0.5mm spacers from Precise Parts, which were supplied as large annular rings. Whereas my original intention was to put them between the adaptor and the visual back of the OTA, I didn't realise that the adaptor has an internal flange to give the correct dimension such that when screwed all the way in (to that flange) there was still a gap to the OTA. In the end I cut slots into the spacers so I could fit them between the adaptor and the STX Guider and account for the bolts. This worked perfectly.

I ran trials for each of the following scenarios:
1) no spacer (equivalent to -1.0mm versus the theoretical set back distance)
2) 0.5mm spacer (equivalent to -0.5mm versus the theoretical set back distance)
3) 1.0mm spacer (equivalent to the theoretical set back distance)
4) 1.0mm + 0.5mm spacers (equivalent to +0.5mm versus the theoretical set back distance)

I imaged through the Luminance Filter towards the zenith, or near to an elevation of 45 degrees, to check that the bending moment on the imaging train wasn't causing an issue of flexure.

In all cases there was an unacceptable level of star elongation.

Here is a summary for each trial"










One other check I ran was to image NGC 253 towards the zenith with my Canon EOS 5D mkII for which I have the required spacer from Celestron. It has a very large sensor too. I can't easily auto guide in this set up so I just made a series of 30 sec captures. There was much less star elongation with an eccentricity of just 0.6. In fact the slight eccentricity was in the East-West direction so it is probably due to tracking.

I am perplexed as to why I can't seem to get round stars at the set back distances I've tried.
The scope is well collimated and the auto guiding has an RMS error of < 0.5 arc secs typically.
I have tried coming both ways through the point of focus to force mirror shift and the results are largely the same. Also if mirror shift were an issue, I would expect to have seen it in the DSLR image.

The only thing I can think of from here is that the sensor is not perfectly perpendicular to the light path and I should now start some trials using shims.

Of course I would be open and would in fact welcome any other suggestions you may have.

Thanks & kind regards,

Niall MacNeill