[popo]

I wonder if some of the more theoretically knowledgeable members can give me a hand here. I'm not quite up to speed on working out depth of focus (NOT depth of field) and possibly using the amount of de-focus to estimate the shift distance from ideal.



As background to this, I'm still dabbling with astrophotography. Using my modified 450D with the 135mm f/2 at f/2, I find that the focus is not consistent across the frame. It has a characteristic that a diagonal line is in focus, and deviating from that line the defocus grows. As such I suspect the lens axis is not perpendicular to the sensor, and this is not field curvature which would be a circular effect wouldn't it? The question then is how much deviation do I have?



For example, using this calculator and a CoC value of 10.3 microns (double the pixel pitch of the 450D) I get a depth of focus of about 21 microns. The difference in size of a nominal point source (faint distant star) is about 4x greater in the worst defocus zone. Is the difference in distance between that and the edge of the depth of focus then (4-1)*0.5*depth_of_focus? I had to draw that out to visualise it! But that is in the ball park of 30 microns... I'm not sure I can hand adjust the sensor placement to that sort of accuracy!



If anyone hasn't fallen asleep or just gave up already, some sanity checking would be welcome.

[wim]

Hi popo,

[quote name='popo' timestamp='1326998992' post='15034']

I wonder if some of the more theoretically knowledgeable members can give me a hand here. I'm not quite up to speed on working out depth of focus (NOT depth of field) and possibly using the amount of de-focus to estimate the shift distance from ideal.



As background to this, I'm still dabbling with astrophotography. Using my modified 450D with the 135mm f/2 at f/2, I find that the focus is not consistent across the frame. It has a characteristic that a diagonal line is in focus, and deviating from that line the defocus grows. As such I suspect the lens axis is not perpendicular to the sensor, and this is not field curvature which would be a circular effect wouldn't it?

Yes.



However, this could be caused by a variety of issiues. It could be a decentered lens (slight tilt), be it very nicely centered, so not likely, or the lens mount, or the mount on the camera, or the sensor itself. All of the latter are possible.

The question then is how much deviation do I have?



For example, using [url="http://www.tawbaware.com/maxlyons/calc.htm"]this[/url] calculator and a CoC value of 10.3 microns (double the pixel pitch of the 450D) I get a depth of focus of about 21 microns. The difference in size of a nominal point source (faint distant star) is about 4x greater in the worst defocus zone. Is the difference in distance between that and the edge of the depth of focus then (4-1)*0.5*depth_of_focus? I had to draw that out to visualise it! But that is in the ball park of 30 microns... I'm not sure I can hand adjust the sensor placement to that sort of accuracy!



If anyone hasn't fallen asleep or just gave up already, some sanity checking would be welcome.

[/quote]

cosine(5.52 degrees)*((4-1)*10.3 microns * 0.5) I would think, or 15.4 microns if I calculated this correctly. This based on a point source being 10.3 microns... <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='' />. This is one way of course. It would be similar but in opposite direction in the opposite corner. 5.52 degrees because that is half the AoV on APS-C of a 135 mm lens; I think you may have used the entire AoV, which is not correct, IMO.



In order to check whether it is the lens, or the camera mount or sensor, check another lens, I would say. Same result would indicate it is the camera, either sensor or mount (which can be adjusted by mere humans, apparently, but is a very time consuming process), even if it is the sensor in that case. Underneath the mount ring, there are shims of different thicknesses and inclinations. There are also shims behind the lens mount itself, to position it correctly. Of course only change those if it is the lens rather than the body.



Considering the minimal amount of tilt required, it is going to be tricky to do...



Kind regards, Wim

[popo]

Hmm... just had a look at similar shots with the same lens on a 600D, and I'm not seeing the variation. So I think we can say it is not the lens, but the body. This is further confirmed looking at some 50mm f/2 shots on the 450D where I get a similar effect.



Bear in mind I had the 450D modified for astro use, I wonder if that could be a contributing factor. If I get bored on the weekend I might open it up and check it. Maybe there's a big speck of dust in there causing the offset!

[wim]

[quote name='popo' timestamp='1327005359' post='15037']

Hmm... just had a look at similar shots with the same lens on a 600D, and I'm not seeing the variation. So I think we can say it is not the lens, but the body. This is further confirmed looking at some 50mm f/2 shots on the 450D where I get a similar effect.



Bear in mind I had the 450D modified for astro use, I wonder if that could be a contributing factor. If I get bored on the weekend I might open it up and check it. Maybe there's a big speck of dust in there causing the offset!

[/quote]

If that means removing/replacing the AA-filter, yes, that would well be possible <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='' />.



However, where would you expect this dust speck to be, in that case? Between filter and sensor? Under the mount?



Kind regards, Wim

[popo]

Look at this guide on taking apart a 450D. I assume the person who did my filter swap (remove IR block, fit clear glass) had to go through a similar procedure. One possible candidate is between the module which holds the sensor assembly and the rest of the internals. While I'm there I also want to have a go cleaning between the filter and sensor as there is one annoying piece of dust in there.

[wim]

Interesting. I don't think I would want to do that ... <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='' />.



Success, and the best of luck with this, popo!



Kind regards, Wim

[popo]

I had a little practice on my earlier 300D IR mod. Cheap ebay IR filter, cut myself. Held in place by tape! Hence me getting someone else to do a better job of it on the 450D mod. With a guide at least I should know where they all are...

[frank]

Hi Wim, may I also ask a theoritical/technical question here. In an ideal case, the optical axis of the lens should be orthogonal to the image censor and pass through the center of the sensor. But in reality errors always exist: if the optical axis of a lens is not perfectly orthogonal to the censor, then there is a tilt; if the optical axis does not pass through the center of the censor, then there is a shift. How large this kind of errors are considered to be within the acceptible limit?



Regards,

Frank

[Guest]

[quote name='popo' timestamp='1327006811' post='15040']

Look at this guide on taking apart a 450D. I assume the person who did my filter swap (remove IR block, fit clear glass) had to go through a similar procedure. One possible candidate is between the module which holds the sensor assembly and the rest of the internals. While I'm there I also want to have a go cleaning between the filter and sensor as there is one annoying piece of dust in there.

[/quote]



Would a very crude solution work? I.e. take a piece of thin paper, aluminum foil, teflon tape or something similar, put it between body and lens mount on one side as you mount the lens. I never tried anything like this, but since you need so little correction, putting the shim between lens and body could be the least amount of work and would allow you to tinker around with thickness. You might have to seal for light leaks tough.

[wim]

[quote name='popo' timestamp='1327018538' post='15047']

I had a little practice on my earlier 300D IR mod. Cheap ebay IR filter, cut myself. Held in place by tape! Hence me getting someone else to do a better job of it on the 450D mod. With a guide at least I should know where they all are...





[/quote]

Wow! Hat off to you!



Kind regards, WIm