Hi Han,
I've been trying to figure out whether my imaging train has a tilt in it and I'm having some problems with my interpretation of what ASTAP is telling me. I think in cloudynights, you said that the setting "ignore stars less than [HFD]" should be set to greater than 0.8. I also did not select Image Inspection -> Extra Stars for the tests that I show below.
My camera is the STF8300C which is an OSC and it is not as sensitive as a CMOS camera. The integration time for each frame is 120 s. I pointed my telescope, Stowaway, at M44 to get enough stars for ASTAP to analyze.
For "ignore stars less than [HFD] 0.8", (file: hfd0.8_17%tilt.jpg) I get 17% tilt.
For "ignore stars less then [HFD] 1.5", (file: hfd1.5_16%tilt.jpg) I get 16% tilt
For "ignore stars less than [HFD] 2", (file: hfd2_4%tilt.jpg) I get 4% tilt
Since I get different results for each value of "ignore stars less than [HFD]", I'm wondering whether I actually have a tilt > 10% or not?
That is an interesting image. Looks like your having the excellent seeing. Much better then most locations. The faintest stars are around one pixel large. First I tought these where hot pixels but after solving and star annotation these are real stars. See attachment.
The brighter stars are imaged much larger. That indicates that your optics are not working so well. Either due to the optics itself or your system is sensitive for UV or IR light and out of focus for these wavelengths. Do you use a UV/IR block filter?
Since that stars can be so small, i think it is better to set " ignore stars" to 0.0. Manual inspection indiates that south-east side is worse. That is confirmed by the inspector. But it is difficult to measure because brighter stars are larger then fainter stars. Normally all imaged stars have about the same diameter.
What I would do:
- Check and fix UV/Ir block filter.
- Try to correct for tilt.
The brighter stars are imaged much larger. That indicates that your optics are not working so well and for UV or IR
light are out of focus for these wavelength. Do you use a UV/IR block filter?
Not in this case. Brighter stars will look larger than fainter stars because more rings around the Airy disk (aka Airy pattern) are
visible. Given a perfect telescope, the minimum size of the Airy disk (in microns) is directly related to telescope focal-ratio.
-Ray
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I'll use the formula (from Wikipedia) where the first minimum occurs is approximately:
theta = 1.22* lambda/d
where lambda is the wavelength of the light in m, d is the diameter in m.
If I assume lambda = 550e-9 m, and d = 92e-3 m, I get
theta = 7.2 urad = 1.4 arcsec. If 1.4 arcsec is the radius, then the diameter of the Airy disk is 2.8 arc which is just larger than the pixel < 2.3 arcsec/pixel size
I want it to be bigger, so I'll use IR 750 nm, which is about 1.4x larger, so
theta = 1.4 * 1.4 = 1.9 arcsec. If 1.9 arcsec is the radius, then the diameter of the Airy disk is 3.8 arcsec which is a lot larger than the pixel (2.3 arcsec/pixel).
Maybe the above formula is not the right one to use because Ray says that it is directly related to f/D which the wikipedia formula does not have.
cytan
Last edit: cytan 2022-04-18
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Pixel scale is 2.3" arcseconds. Will the airy disk be visible? Normally you need some magnification to see them.
That requires a calculation.
The Airy disc and rings can move around because of scintillation. Since the accumulated signal is recorded in an image, the ability
to capture the Airy disk and rings depends on pixel scale, seeing conditions, and exposure length. But my point was that brighter
stars in an image look larger because the accumulated sum of more Airy rings rises above the noise floor.
-Ray
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hi Han,
It's really interesting that you thought my seeing was excellent because I thought my seeing wasn't that great :) My rig sits right on the sidewalk right next to a street light. And according to meteoblue, my seeing that night was > 2.3 arcsec. Maybe I was saved by having a camera with large pixels: 5.4 um which gave a resolution of 2.27 arcsec/pixel and it is an OSC as well. In this photo, M44 was not that high in the sky, alt=65 deg.
I'm also surprised that hot pixels wasn't an issue because my STF8300C because it certainly has about 0.2% hot pixels.
I didn't use an UV/IR filter because I thought that the STF8300C already had one. See this link: https://www.cloudynights.com/topic/525878-stf-8300c-ir-filter/
Thus, I didn't add any filters in the train because I wanted to eliminate any effects from the filter. However, I can insert a LPS-P2 filter the next time I try this out to see if the bloat that you see is eliminated.
As for the optics, the image came from a brand new Astro-Physics Stowaway which I received just after X'mas last year (2021). So, I seriously doubt that there is an optical problem.
I had double checked that all the screws were tight in the imaging train before taking this photo (I did find screws loose between the filter wheel and the camera which I tightened before this test). I do have a focal reducer inserted in the train which uses DoveLoc rather than being screwed on. I made sure that the focal reducer was properly inserted into the train by using gravity, i.e it was inserted with the scope set to vertical.
The next test with a tilt corrector will be a few weeks from now because it is coming from Germany. And I will also need a precise parts adapter as well. It's more $$$$ :(
Thanks for the analysis. If you have more thoughts, please post. I really appreciate your help.
In the mean time, attached is M44 (Beehive cluster, 23 subframes at 120 s for each subframe) processed with StarTools. The light from the street lamp is at the top of the photo which I didn't get rid of.
Yes it is puzzling. Lets see what the LPS-P2 filter brings.
Nice image. The light at the top can be removed in pixel math 1, "Equalise background tool". But do that only with unstretched 16 or 32 bit images, preferably FITS.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
The Gaussian waist size is 4.65 micrometer. See attached. So a little less then one pixel. The airy rings will contain much less flux then the central spot. It doesn't look like that. But it is rare to see such a sharp image.
Hi Han,
I also got more data that night. Here's a comparison between M44 and some other point in the sky. The M44 had the telescope pointing to the West and the other one had the telescope pointing to the East.
It looks like the larger stars are on the Southeast corner for the M44 image, while for the other one, it's on the Northwest corner. This looks like flexure to me, right?
For both screenshots the worst corner is south-east. That is indicated by the north arrow after solving. One image is 180 degrees rotated due to meridain flip. So it still look like a fixed tilt to me.
Since you ordered a tilt adjuster this could easily be fixed in future.
With my APO I had a lot of tilt problems using filter wheel with noise piece connection. The behavior was unpredictable. After making everything screwed except the camera noise piece the tilt is stable. But still a tilt adjuster was necessary even with my not so large ASI1600 camera.
Han
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hi Han,
Maybe I'm insane (well, maybe crazy :) ) , but since the bad corner actually moved from SE to NW when the camera rotated by 180 deg, doesn't this effect show that the tilt is caused by a sag rather than a fixed tilt. See image that I drew in chicken scratchings :)
Here's my description of the chicken scratchings:
1. Figure 1 (top left figure) shows the physical orientation of the camera at 0 deg. ABCD marks the corners of the camera. The image on the camera is the up arrow. The bad corner is at C and is circled in red, i.e. bloated stars should be at the SE.
2. Figure 1a (top right figure) shows the image on a computer. The corners are again ABCD with the bad corner at C circled in red.
3. Figure 2 (bottom left) Now, I physically flip the camera by 180 deg. The arrow is still pointing up because of circular symmetry of the lens system. The bad corner C on the camera is now at the NW corner.
4. Figure 2a (bottom right) shows the image on the computer. This time the arrow is pointing down as expected. The bad corner is still at C. So since the bad corner of the image is still at SE, I should still see bloated stars at the SE, i.e. the bad corner shouldn't move even when I rotate the camera by 180 deg.
Therefore, since I see the bloated stars change from corner C to corner A, I would think that it is actually a sag, i.e. flexure rather than a fixed bad corner.
Hi Han,
I've been trying to figure out whether my imaging train has a tilt in it and I'm having some problems with my interpretation of what ASTAP is telling me. I think in cloudynights, you said that the setting "ignore stars less than [HFD]" should be set to greater than 0.8. I also did not select Image Inspection -> Extra Stars for the tests that I show below.
My camera is the STF8300C which is an OSC and it is not as sensitive as a CMOS camera. The integration time for each frame is 120 s. I pointed my telescope, Stowaway, at M44 to get enough stars for ASTAP to analyze.
For "ignore stars less than [HFD] 0.8", (file: hfd0.8_17%tilt.jpg) I get 17% tilt.
For "ignore stars less then [HFD] 1.5", (file: hfd1.5_16%tilt.jpg) I get 16% tilt
For "ignore stars less than [HFD] 2", (file: hfd2_4%tilt.jpg) I get 4% tilt
Since I get different results for each value of "ignore stars less than [HFD]", I'm wondering whether I actually have a tilt > 10% or not?
The fit file can be downloaded here:
https://1drv.ms/u/s!ApsRFMpkd9cwiS_GdIOV9FVFYOf0
Thanks!
cytan
Last edit: cytan 2022-04-17
Hello Cytan,
That is an interesting image. Looks like your having the excellent seeing. Much better then most locations. The faintest stars are around one pixel large. First I tought these where hot pixels but after solving and star annotation these are real stars. See attachment.
The brighter stars are imaged much larger. That indicates that your optics are not working so well. Either due to the optics itself or your system is sensitive for UV or IR light and out of focus for these wavelengths. Do you use a UV/IR block filter?
Since that stars can be so small, i think it is better to set " ignore stars" to 0.0. Manual inspection indiates that south-east side is worse. That is confirmed by the inspector. But it is difficult to measure because brighter stars are larger then fainter stars. Normally all imaged stars have about the same diameter.
What I would do:
- Check and fix UV/Ir block filter.
- Try to correct for tilt.
Han
I will the second image after posting.
Last edit: han.k 2022-04-18
Not in this case. Brighter stars will look larger than fainter stars because more rings around the Airy disk (aka Airy pattern) are
visible. Given a perfect telescope, the minimum size of the Airy disk (in microns) is directly related to telescope focal-ratio.
-Ray
Pixel scale is 2.3" arcseconds. Will the airy disk be visible? Normally you need some magnification to see them. That requires a calculation.
Han
I'll use the formula (from Wikipedia) where the first minimum occurs is approximately:
theta = 1.22* lambda/d
where lambda is the wavelength of the light in m, d is the diameter in m.
If I assume lambda = 550e-9 m, and d = 92e-3 m, I get
theta = 7.2 urad = 1.4 arcsec. If 1.4 arcsec is the radius, then the diameter of the Airy disk is 2.8 arc which is just larger than the pixel < 2.3 arcsec/pixel size
I want it to be bigger, so I'll use IR 750 nm, which is about 1.4x larger, so
theta = 1.4 * 1.4 = 1.9 arcsec. If 1.9 arcsec is the radius, then the diameter of the Airy disk is 3.8 arcsec which is a lot larger than the pixel (2.3 arcsec/pixel).
Maybe the above formula is not the right one to use because Ray says that it is directly related to f/D which the wikipedia formula does not have.
cytan
Last edit: cytan 2022-04-18
The Airy disc and rings can move around because of scintillation. Since the accumulated signal is recorded in an image, the ability
to capture the Airy disk and rings depends on pixel scale, seeing conditions, and exposure length. But my point was that brighter
stars in an image look larger because the accumulated sum of more Airy rings rises above the noise floor.
-Ray
Hi Han,
It's really interesting that you thought my seeing was excellent because I thought my seeing wasn't that great :) My rig sits right on the sidewalk right next to a street light. And according to meteoblue, my seeing that night was > 2.3 arcsec. Maybe I was saved by having a camera with large pixels: 5.4 um which gave a resolution of 2.27 arcsec/pixel and it is an OSC as well. In this photo, M44 was not that high in the sky, alt=65 deg.
I'm also surprised that hot pixels wasn't an issue because my STF8300C because it certainly has about 0.2% hot pixels.
I didn't use an UV/IR filter because I thought that the STF8300C already had one. See this link:
https://www.cloudynights.com/topic/525878-stf-8300c-ir-filter/
Thus, I didn't add any filters in the train because I wanted to eliminate any effects from the filter. However, I can insert a LPS-P2 filter the next time I try this out to see if the bloat that you see is eliminated.
As for the optics, the image came from a brand new Astro-Physics Stowaway which I received just after X'mas last year (2021). So, I seriously doubt that there is an optical problem.
I had double checked that all the screws were tight in the imaging train before taking this photo (I did find screws loose between the filter wheel and the camera which I tightened before this test). I do have a focal reducer inserted in the train which uses DoveLoc rather than being screwed on. I made sure that the focal reducer was properly inserted into the train by using gravity, i.e it was inserted with the scope set to vertical.
The next test with a tilt corrector will be a few weeks from now because it is coming from Germany. And I will also need a precise parts adapter as well. It's more $$$$ :(
Thanks for the analysis. If you have more thoughts, please post. I really appreciate your help.
In the mean time, attached is M44 (Beehive cluster, 23 subframes at 120 s for each subframe) processed with StarTools. The light from the street lamp is at the top of the photo which I didn't get rid of.
cytan
Yes it is puzzling. Lets see what the LPS-P2 filter brings.
Nice image. The light at the top can be removed in pixel math 1, "Equalise background tool". But do that only with unstretched 16 or 32 bit images, preferably FITS.
The Gaussian waist size is 4.65 micrometer. See attached. So a little less then one pixel. The airy rings will contain much less flux then the central spot. It doesn't look like that. But it is rare to see such a sharp image.
More calculation are here, but most flux will end up in one pixel. Not spread out over a diameter of 10 pixels.
https://www.cloudynights.com/documents/resolution.pdf
Last edit: han.k 2022-04-18
Hi Han,
I also got more data that night. Here's a comparison between M44 and some other point in the sky. The M44 had the telescope pointing to the West and the other one had the telescope pointing to the East.
It looks like the larger stars are on the Southeast corner for the M44 image, while for the other one, it's on the Northwest corner. This looks like flexure to me, right?
The NW.jpg fits file can be downloaded here:
https://1drv.ms/u/s!ApsRFMpkd9cwiTAGv1G90OxHD4C1?e=xjmxJL
cytan
Last edit: cytan 2022-04-19
For both screenshots the worst corner is south-east. That is indicated by the north arrow after solving. One image is 180 degrees rotated due to meridain flip. So it still look like a fixed tilt to me.
Since you ordered a tilt adjuster this could easily be fixed in future.
With my APO I had a lot of tilt problems using filter wheel with noise piece connection. The behavior was unpredictable. After making everything screwed except the camera noise piece the tilt is stable. But still a tilt adjuster was necessary even with my not so large ASI1600 camera.
Han
Hi Han,
Maybe I'm insane (well, maybe crazy :) ) , but since the bad corner actually moved from SE to NW when the camera rotated by 180 deg, doesn't this effect show that the tilt is caused by a sag rather than a fixed tilt. See image that I drew in chicken scratchings :)
Here's my description of the chicken scratchings:
1. Figure 1 (top left figure) shows the physical orientation of the camera at 0 deg. ABCD marks the corners of the camera. The image on the camera is the up arrow. The bad corner is at C and is circled in red, i.e. bloated stars should be at the SE.
2. Figure 1a (top right figure) shows the image on a computer. The corners are again ABCD with the bad corner at C circled in red.
3. Figure 2 (bottom left) Now, I physically flip the camera by 180 deg. The arrow is still pointing up because of circular symmetry of the lens system. The bad corner C on the camera is now at the NW corner.
4. Figure 2a (bottom right) shows the image on the computer. This time the arrow is pointing down as expected. The bad corner is still at C. So since the bad corner of the image is still at SE, I should still see bloated stars at the SE, i.e. the bad corner shouldn't move even when I rotate the camera by 180 deg.
Therefore, since I see the bloated stars change from corner C to corner A, I would think that it is actually a sag, i.e. flexure rather than a fixed bad corner.
What do you think?
cytan
Last edit: cytan 2022-04-19
Yes that makes sense. If the bad corner moves 180 degrees after the meridan flip it must be flexture.