Canon Sensor Tilt - Sanity Check Required Please

[richyrich_one]

Hi All

It's starting to look like my sensor is tilted on my 600D since coming back from being astromodified.

I would really appreciate a sanity check that my method for confirming this is correct before I raise the issue with the modder.

I set up my light panel and placed the camera with coma corrector nose directly on the panel through 3 sheets of white A4.

 

Using APT I took a flat in a dark room as usual and used PI to perform a contour plot on it.

 

Can anyone see any problem with my method?

If not, any advice would be very much appreciated.

 

[ronin]

Cannot comment on the method, sounds sort of reasonable but I cannot think why the people doing the modification would want to remove then replace the sensor. I would have expected them to avoid the sensor as much as possible since any damage to that in effect makes the camera scrap.

If they replaced the filter with another, likely, then maybe the filter is not sitting flat to the sensor and so the new filter is tilted, that may produce the same, and is more likely, I would suggest.

The problem I see is that having in effect changed the nature of the optical path then what appears at the sensor will also have altered in some way.

[RayD]

Did you do the same check before you sent it away for modding?

[Dr_Ju_ju]

One point with your method, is that you're relying on the machining of the nosepiece is parallel to the sensor, which may or may not be true. Also I have found paper to be very variable on densities between and across sheets, so to your eye, everything looks evenly illuminated, it may not be to more sensitive instruments.

Have you tried the same test but using a stock lens ?

I've had 2 cameras modified by Juan @Cheap Astrophotography, and he performs a full test\setup on the cameras before returning them...

[yelsac]

Don't know if it helps but I had a sensor tilt in a 450d which could be clearly seen when you take a 15 second shot of a bright star through a reflector telescope.

You can clearly see the reflection is out near the star (see below) its down & to the right a bit. This also effected the focus too. My scope was not the problem as when tried with another camera the reflection was dead center.

In my case I found out that the chap who astro modified it accidentally allowed a couple of shims to fall out during the process.

I took the camera apart & removed the remaining 4 shims so the sensor lie's flat & it sorted the problem.

hope you sort the problem

[alacant]

Hi. PM sent and maybe this will help. HTH.

[alacant]

4 hours ago, richyrich_one said:

Can anyone see any problem with my method?

Everyone will say it's the coma-corrector. If you're worried about squareness or stray light, do you have any 2" extension tubes you could use or perhaps an old full frame camera lens set at infinity? I took everything from the camera. Try holding it up to a wall. In my case it didn't matter how square it was. No matter what I took, you could see the vignett along one edge and the off centre diffraction rings when stretched. HTH.

[richyrich_one]

3 hours ago, ronin said:

Cannot comment on the method, sounds sort of reasonable but I cannot think why the people doing the modification would want to remove then replace the sensor. I would have expected them to avoid the sensor as much as possible since any damage to that in effect makes the camera scrap.

If they replaced the filter with another, likely, then maybe the filter is not sitting flat to the sensor and so the new filter is tilted, that may produce the same, and is more likely, I would suggest.

The problem I see is that having in effect changed the nature of the optical path then what appears at the sensor will also have altered in some way.

I believe the whole sensor assembly is removed from the camera in the modification process.

[richyrich_one]

3 hours ago, Dr_Ju_ju said:

One point with your method, is that you're relying on the machining of the nosepiece is parallel to the sensor, which may or may not be true. Also I have found paper to be very variable on densities between and across sheets, so to your eye, everything looks evenly illuminated, it may not be to more sensitive instruments.

Have you tried the same test but using a stock lens ?

I've had 2 cameras modified by Juan @Cheap Astrophotography, and he performs a full test\setup on the cameras before returning them...

I thought the same so I even tried holding the camera at quite an angle to the light panel and it was at near 45 deg before even the slightest difference could be noticed, so a slight machining error in the nosepiece would seem to have neglible effect.

[richyrich_one]

6 minutes ago, alacant said:

Everyone will say it's the coma-corrector. If you're worried about squareness or stray light, do you have any 2" extension tubes you could use or perhaps an old full frame camera lens set at infinity? I took everything from the camera. Try holding it up to a wall. In my case it didn't matter how square it was. No matter what I took, you could see the vignett along one edge and the off centre diffraction rings when stretched. HTH.

I have found the same results. I removed everything from the body and held it just above the panel with the same effect.

[richyrich_one]

55 minutes ago, alacant said:

Hi. PM sent and maybe this will help. HTH.

VERY helpful, thanks.

[richyrich_one]

Thanks everyone.

It seems it isn't the first time this has happened and I will be raising the issue with the modifer asap.

[sharkmelley]

15 hours ago, richyrich_one said:

 

Can anyone see any problem with my method?

 

I don't see how sensor tilt can be measured using this method of an evenly lit panel.  For instance the bottom part of the sensor will always have increased vignetting compared to the top part because the parked mirror causes shadowing.  Note that image formation is upside down, so the mirror parked at the top of the mirror box causes shadowing that appears at the bottom of the image.  

As an aside, mirror shadow is why I removed the mirror from my astro-imaging cameras: http://www.markshelley.co.uk/Astronomy/Projects/Mirrorless/canon550mirrorless.html

So I'm afraid this method is flawed as a way to determine tilt.  The only way to determine sensor tilt (without specialist test equipment) is to examine star images taken with actual optics.  But even then it is impossible to eliminate tilt caused by slight mis-collimations in the optics themselves.  The easiest amateur method to detect tilt in your 600D is to compare its images against a known good Canon camera (i.e. one with no tilt) using the same adapters, optics etc.

Mark

[richyrich_one]

3 hours ago, sharkmelley said:

I don't see how sensor tilt can be measured using this method of an evenly lit panel.  For instance the bottom part of the sensor will always have increased vignetting compared to the top part because the parked mirror causes shadowing.  Note that image formation is upside down, so the mirror parked at the top of the mirror box causes shadowing that appears at the bottom of the image.  

As an aside, mirror shadow is why I removed the mirror from my astro-imaging cameras: http://www.markshelley.co.uk/Astronomy/Projects/Mirrorless/canon550mirrorless.html

So I'm afraid this method is flawed as a way to determine tilt.  The only way to determine sensor tilt (without specialist test equipment) is to examine star images taken with actual optics.  But even then it is impossible to eliminate tilt caused by slight mis-collimations in the optics themselves.  The easiest amateur method to detect tilt in your 600D is to compare its images against a known good Canon camera (i.e. one with no tilt) using the same adapters, optics etc.

Mark

Thank you Mark.

You have given me a little bit of hope. I really didn't want there to be a problem with the camera and all the associated hassle getting it sorted.

So I took a set of flats at 0,90,180 and 270 deg rotations using the same method but with the scope attached.

I would appreciate anyone educating me what this is revealing. I cannot get my head around what would or should be happening as I rotate the camera.

 

0 deg

90 Deg

180 Deg

270 Deg

 

 

[sharkmelley]

4 hours ago, richyrich_one said:

 

So I took a set of flats at 0,90,180 and 270 deg rotations using the same method but with the scope attached.

I would appreciate anyone educating me what this is revealing. I cannot get my head around what would or should be happening as I rotate the camera.

 

Let me first clear up another point.  When a camera is modified it is impossible for the sensor to replaced off centre.  An off centre sensor was suggested in this thread (https://stargazerslounge.com/topic/283403-sensor-tilt/) but it cannot happen because the sensor assembly can only be screwed back in one location.

Now to the interpretation of what you are seeing with those on-scope flats.  Again it is easy to jump to incorrect conclusions.  What you are essentially doing here is taking a flat frame.  Taking good flats is fraught with problems and there are pages and pages of discussion on the subject on SGL and various other forums.  Your flats are all clearly quite different and this illustrates the point quite well.  Part of the reason for this might be the way the experiment was designed.  To even begin to understand what we are seeing, further information is required.

1) What kind of scope is it and what is its native F-ratio?  Some scopes have asymmetrical illumination patterns.  What other lens elements and adapters do you have attached?

2) Is the scope mounted vertically looking down on the panel?  If not, what is the orientation used and which position is the 0 deg position?

3) Do you rotate the scope and camera as one unit or are you rotating the camera on the scope?

4) Is the camera orientation sensor switched on?  This is what detects the orientation of the camera and rotates your images accordingly.  For astro-imaging I leave it switched off.  Is it possible that one or more or your images are rotated?  A related question is this: are you using RAW or JPG?  JPGs might be rotated but RAWs might not - depending on the raw converter being used and the options that were set in the software.

5) What is the "flicker frequency" of the illuminated panel you are using?  What is your exposure length? Is your exposure long enough to integrate over a sufficient number of cycles?

6) What gives you the confidence that your light panel is evenly illuminated?

7) Does the 600D give you the choice between a physical shutter and an electronic (silent) shutter?  If so, which are you using?  Electronic shutters can result in horizontal banding with flickery light sources but it's probably not an issue here.

Mark

 

[richyrich_one]

 

Thanks for your comprehensive response.

1 hour ago, sharkmelley said:

Let me first clear up another point.  When a camera is modified it is impossible for the sensor to replaced off centre.  An off centre sensor was suggested in this thread (https://stargazerslounge.com/topic/283403-sensor-tilt/) but it cannot happen because the sensor assembly can only be screwed back in one location.

I don't believe anyone suggested it could be. It was a question of whether when the sensor assembly was replaced it was correctly and evenly shimmed in order to maintain auto focus.

1 hour ago, sharkmelley said:

Your flats are all clearly quite different and this illustrates the point quite well.

They are only different because I am rotating the camera, if i don't, they are consistant.

Information as requested:

1) Skywatcher 130P-DS reflector F5

2) Yes

3) Rotating the camera on the scope

4) The images are not being rotated by the camera and are taken in RAW

5) I don't know but it's 12v LED. 1/2 sec exposure and there is no obvious banding which I would have thought would be picked up in the contour plot if it was present.

6) When I move the panel around the flats are consistant, let's just assume it is

7) It only has a physical shutter

 

[RayD]

To eliminate all the queries relating to the optical train and flat panel can you not take some flats with your QHY camera to see whether you have similar results?  If they are centred then you are definitely looking at the DSLR rather than anything external.

[alacant]

1 hour ago, sharkmelley said:

Some scopes have asymmetrical illumination patterns.

Hi. The OP has eliminated a telescope effect from his tests. He can reproduce the issue with or without camera attachments. HTH.

[RayD]

Forget that actually, you've taken them with nothing at all attached so I suppose you've pretty much done that anyway.

I feel your frustration 

[sharkmelley]

3 hours ago, richyrich_one said:

Information as requested:

1) Skywatcher 130P-DS reflector F5

2) Yes

3) Rotating the camera on the scope

4) The images are not being rotated by the camera and are taken in RAW

5) I don't know but it's 12v LED. 1/2 sec exposure and there is no obvious banding which I would have thought would be picked up in the contour plot if it was present.

6) When I move the panel around the flats are consistant, let's just assume it is

7) It only has a physical shutter

 

Thanks for your replies.  It appears to me that those flats are telling us that for some reason the scope is illuminating the sensor slightly off centre.  So as you rotate the camera on the scope, the bright area of the illumination moves relative to the sensor.   This may or may not be a problem.  Take an image of a star field and see if the centre of the pattern of star elongations is consistent with the centre of illumination.  However, the good news is that for this scope, the mirror and the mirror box are casting very little shadow.  Mirror shadow tends to be a problem only with fast scopes e.g. my Tak Epsilon.

The flats are telling us nothing about possible sensor tilt.  If the scope were illuminating the very centre of the sensor and if it were possible to detect sensor tilt using this method (which I strongly doubt) then we would expect to see the same pattern of brightness on the sensor, whatever the orientation of the camera because the part of the sensor tilted away from the scope would always be the darkest area.

Mark

[michael8554]

Sorry everyone, but I don't see illumination being the main effect of sensor tilt.

IMHO if the sensor is slightly tilted, due say to a missing shim from under one of the locating lugs, then the predominant effect will be differing focus across the frame.

Michael