Switching systems, trying to understand the pros/cons.

[FiveByEagle]

Hello all!

Added a 8" astrograph to the fleet this week. Changing the focuser, and modding it out for my usage and I am having FUN! With my wide field setup I am using a ASI183MM and having an OK time but I do genuinely believe my patience and desire for monochrome has waned.. however I really do not want to waste time and money on going to a OSC that won't really work out for me.

I am concerned using the 2.4um pixels on an 800mm scope. The math tells me ideal sampling pixel size for 800mm and my area and band is 3.56, and closest to that is 533MC for me. I know the 2600 is great.. I do not want that sensor size. Personal preference, I shoot a lot of medium format and square formats look right to me. I had a 533 for a scope at my old site and loved it, so I am confident in its capabilities. The 294 is cool too, but not a fan of the sensor size and resulting FOV.

However, it'll be a minute before I can make this transition. Might even have some clear sky time to give the 183MM a shot on this scope and I was honestly wondering... what's the worst that can happen? Being sampled at .62 is really oversampled but will my guiding have to be perfect? I use an OAG and get .4-.6 on average, regardless of proximity to meridian. With a larger OTA, I am not going to expect those numbers, but I do expect to stay below 1". Is that "good enough" for round stars at that scale? I get perfectly round pin-points currently, and would hate to settle for anything worse.

 

Either way, me going back to OSC is gunna happen soon, its more or less an argument of what is a great fit for what I want to do, and unfortunately a lot of my responses on facebook groups and social media has really been driven by what influencers and youtube channels are saying versus some real math and explanations.

[vlaiv]

I'm not sure I understood half of what is bugging you , but I'll try to help. I'm going to make couple of points, and you can take from that what ever you like.

1. With 8" astrograph it is very likely that you are oversampling at anything below 1.2"/px if your guiding is in 0.4-0.6" RMS range - so bin your pixels accordingly. Depending on seeing - even 1.8"/px will be oversampling at some nights.

2. Not sure what math are you using but 3.56µm at 800mm is 0.92"/px - again most definitively over sampling for 8" of aperture even with premium mounts. You should really bin x2 ASI183MM to get 4.8µm effective pixel size

3. Larger sensors are better sensors. Sensor real estate translates into speed when paired with appropriate scope. You can pair larger sensor with longer FL scope and for same F/ratio longer FL scope - means larger aperture. If you adjust your sampling rate to be the same using pixel size / binning - you have more aperture at target resolution and that equals more speed. Don't like that sensor is not square? You can always crop away what you don't need (it is a waste of sensor area - but if you really like square images ...).

4. Consider lowering your sampling rate further. I know that 1.2"/px - sounds great - you are right there - face to face with galaxies and all - but most of the nights - you won't be able to achieve that. You'll need 1.18" FWHM seeing during the night in order to get 1.92" FWHM stars in your image with 0.6" RMS guiding and 8" aperture - and that is sharp enough for 1.2"/px. In fact - take some of your old subs and measure star FWHM in arc seconds - divide that with 1.6 and that will give you optimum sampling rate for that image / night. I'm convinced that most of your subs will be properly sampled at 1.5" or above

 

[FiveByEagle]

15 minutes ago, vlaiv said:

I'm not sure I understood half of what is bugging you , but I'll try to help. I'm going to make couple of points, and you can take from that what ever you like.

1. With 8" astrograph it is very likely that you are oversampling at anything below 1.2"/px if your guiding is in 0.4-0.6" RMS range - so bin your pixels accordingly. Depending on seeing - even 1.8"/px will be oversampling at some nights.

2. Not sure what math are you using but 3.56µm at 800mm is 0.92"/px - again most definitively over sampling for 8" of aperture even with premium mounts. You should really bin x2 ASI183MM to get 4.8µm effective pixel size

3. Larger sensors are better sensors. Sensor real estate translates into speed when paired with appropriate scope. You can pair larger sensor with longer FL scope and for same F/ratio longer FL scope - means larger aperture. If you adjust your sampling rate to be the same using pixel size / binning - you have more aperture at target resolution and that equals more speed. Don't like that sensor is not square? You can always crop away what you don't need (it is a waste of sensor area - but if you really like square images ...).

4. Consider lowering your sampling rate further. I know that 1.2"/px - sounds great - you are right there - face to face with galaxies and all - but most of the nights - you won't be able to achieve that. You'll need 1.18" FWHM seeing during the night in order to get 1.92" FWHM stars in your image with 0.6" RMS guiding and 8" aperture - and that is sharp enough for 1.2"/px. In fact - take some of your old subs and measure star FWHM in arc seconds - divide that with 1.6 and that will give you optimum sampling rate for that image / night. I'm convinced that most of your subs will be properly sampled at 1.5" or above

 

Thank you!

Sorry, I am aware my posts are word-vomit. I've considered binning the 183MM.. but going down to like 5MP is cutting it way too close for someone who doesn't own a non-4K display anymore, its fairly low for me, But I understand the benefits!! 

I should have added. I also do plan on never buying into this hobby again, as this will be my last setup. I am moving and changing careers, and going from 5 scopes to 1, which is the 800mm scope. Figured it is the average FL of all my stuff so it makes the sell off easier for me, plus at F4 I can pull in some images quicker. The others are already gone, so that is a constant in this equation.

I know since the sensor doesn't exist, the camera obviously won't but man I wish there was a mono camera with the 533/2600 specs but 1" size. 

http://www.astro-imaging.com/Tutorial/MatchingCCD.html This what the math I used - 

[FiveByEagle]

Also - please excuse and forgive my ignorance on this. Guiding numbers relating to end PQ.

I have CHASED the lowest RMS arc-second numbers since entering this hobby (lowest was .28, 3 weeks ago, almost fainted) just because I know it results in sharper images with pleasant star shapes. That is it. I have those numbers with an OAG.

Can you elaborate a but on this? Do my numbers at that FL indicate what I can expect with this new OTA?

[vlaiv]

10 hours ago, FiveByEagle said:

I've considered binning the 183MM.. but going down to like 5MP is cutting it way too close for someone who doesn't own a non-4K display anymore, its fairly low for me, But I understand the benefits!! 

I understand that you want higher megapixel count for your image, but shooting native oversampled will give you same the same result as binning your data and then enlarging image when you are done - except binned version will have better SNR. When you are oversampled - you are recording "empty" resolution - it is the same thing as when you enlarge image - image size will be there but detail wont.

Resolution in astronomical images does not come from number of pixels - but rather how good seeing is, how good mount tracking (guiding) is and what is the size of your aperture.

10 hours ago, FiveByEagle said:

http://www.astro-imaging.com/Tutorial/MatchingCCD.html This what the math I used - 

I just briefly had a look at the math on that site, and I'm sorry to say - it is faulty. I'll just mention few things that are wrong with it:

1. Nyquist criterion is always x2 max frequency component - in 1d and in 2d for square sampling. Stating that it is x3.3 is wrong. You can find numerous sources online that explain why is that.

2. Nyquist criterion holds for band limited signal and x2 relates to maximum frequency. Relating it to anything in spatial domain (like FWHM of star or Rayleigh criterion) is wrong use of sampling theorem.

For long exposure imaging - there is no well defined cutoff frequency. If we approximate star shape with Gaussian profile - then we can use some sort of convention and put cutoff at some sensible frequency. For example, since Fourier Transform of Gaussian is Gaussian - we can easily do the math and put cutoff frequency at place where all higher frequencies are attenuated to more than 90% (their value is less than 10% of original). If we do that, we get that sampling rate is roughly FWHM of Gaussian profile divided with 1.6.

3. Seeing is not the only thing that affects the star image. For perfect aperture, we can devise formula that goes like this:

sigma_profile = sqrt( sigma_seeing^2 + sigma_guiding^2 + sigma_aperture^2)

Where sigma_seeing is FWHM / 2.355 - or regular seeing divided with 2.355 (relationship between Gaussian sigma and FWHM is with factor of x2.355)

https://en.wikipedia.org/wiki/Full_width_at_half_maximum

sigma_guiding is guiding RMS value that you get

and sigma_aperture is Gaussian approximation to Airy pattern - and its sigma, given by this expression:

https://en.wikipedia.org/wiki/Airy_disk#Approximation_using_a_Gaussian_profile

This is of course for perfect aperture, but most fast astrograph telescopes are not diffraction limited over whole field. This means that actual FWHM of stars will be lower.

You can get info on expected FWHM seeing value / forecast from this website:

https://www.meteoblue.com/en/weather/outdoorsports/seeing/indianapolis_united-states-of-america_4259418

(this is for Indianapolis - but do select your place)

Fourth column - astronomical seeing is value that you want, it is given in arc seconds. Compare that with FWHM values in arc seconds in your images - and you'll see that FWHM values in images are higher than seeing figures. This is because there is guiding component and there is aperture size component that has not been taken into account.

As far as guiding goes - don't relate guiding to imaging focal length. Although general rule of thumb is that your guide RMS should be at least half of imaging resolution (and I agree with this rule in sensible range of imaging resolutions) - but I have another rule - make your RMS as low as you can regardless of your working sampling rate

Above rule is very good to show you what sort of resolution you should not be using. Say you are guiding at 1" RMS - well, that means that you really should not go much below 2"/px. You can, but you really should not if you want sharp images.

Only thing that I've found important for guiding is to get your guiding resolution fine enough to be able to reliably determine RMS and guide star position. Here math goes like this - you need to have guiding "/px lower than x6 your target guide RMS.

Say you want to reliably guide at 0.4" RMS (or feel that you can with your mount) - then your guider resolution should be 2.4"/px or higher (higher as in resolution - lower in the number so 2"/px is good but 3"/px is bad). This has to do with centroid precision and all. Other than that - mount stability, mount mechanical soundness and shielding from wind (or not being undermounted) - is the key for good guiding performance.

Hope this helps

 

[FiveByEagle]

3 hours ago, vlaiv said:

I understand that you want higher megapixel count for your image, but shooting native oversampled will give you same the same result as binning your data and then enlarging image when you are done - except binned version will have better SNR. When you are oversampled - you are recording "empty" resolution - it is the same thing as when you enlarge image - image size will be there but detail wont.

Resolution in astronomical images does not come from number of pixels - but rather how good seeing is, how good mount tracking (guiding) is and what is the size of your aperture.

I just briefly had a look at the math on that site, and I'm sorry to say - it is faulty. I'll just mention few things that are wrong with it:

1. Nyquist criterion is always x2 max frequency component - in 1d and in 2d for square sampling. Stating that it is x3.3 is wrong. You can find numerous sources online that explain why is that.

2. Nyquist criterion holds for band limited signal and x2 relates to maximum frequency. Relating it to anything in spatial domain (like FWHM of star or Rayleigh criterion) is wrong use of sampling theorem.

For long exposure imaging - there is no well defined cutoff frequency. If we approximate star shape with Gaussian profile - then we can use some sort of convention and put cutoff at some sensible frequency. For example, since Fourier Transform of Gaussian is Gaussian - we can easily do the math and put cutoff frequency at place where all higher frequencies are attenuated to more than 90% (their value is less than 10% of original). If we do that, we get that sampling rate is roughly FWHM of Gaussian profile divided with 1.6.

3. Seeing is not the only thing that affects the star image. For perfect aperture, we can devise formula that goes like this:

sigma_profile = sqrt( sigma_seeing^2 + sigma_guiding^2 + sigma_aperture^2)

Where sigma_seeing is FWHM / 2.355 - or regular seeing divided with 2.355 (relationship between Gaussian sigma and FWHM is with factor of x2.355)

https://en.wikipedia.org/wiki/Full_width_at_half_maximum

sigma_guiding is guiding RMS value that you get

and sigma_aperture is Gaussian approximation to Airy pattern - and its sigma, given by this expression:

https://en.wikipedia.org/wiki/Airy_disk#Approximation_using_a_Gaussian_profile

This is of course for perfect aperture, but most fast astrograph telescopes are not diffraction limited over whole field. This means that actual FWHM of stars will be lower.

You can get info on expected FWHM seeing value / forecast from this website:

https://www.meteoblue.com/en/weather/outdoorsports/seeing/indianapolis_united-states-of-america_4259418

(this is for Indianapolis - but do select your place)

Fourth column - astronomical seeing is value that you want, it is given in arc seconds. Compare that with FWHM values in arc seconds in your images - and you'll see that FWHM values in images are higher than seeing figures. This is because there is guiding component and there is aperture size component that has not been taken into account.

As far as guiding goes - don't relate guiding to imaging focal length. Although general rule of thumb is that your guide RMS should be at least half of imaging resolution (and I agree with this rule in sensible range of imaging resolutions) - but I have another rule - make your RMS as low as you can regardless of your working sampling rate

Above rule is very good to show you what sort of resolution you should not be using. Say you are guiding at 1" RMS - well, that means that you really should not go much below 2"/px. You can, but you really should not if you want sharp images.

Only thing that I've found important for guiding is to get your guiding resolution fine enough to be able to reliably determine RMS and guide star position. Here math goes like this - you need to have guiding "/px lower than x6 your target guide RMS.

Say you want to reliably guide at 0.4" RMS (or feel that you can with your mount) - then your guider resolution should be 2.4"/px or higher (higher as in resolution - lower in the number so 2"/px is good but 3"/px is bad). This has to do with centroid precision and all. Other than that - mount stability, mount mechanical soundness and shielding from wind (or not being undermounted) - is the key for good guiding performance.

Hope this helps

 

Really do appreciate the time and this response! Unfortunately I did not understand about 85% of it, but that is OK! I can just wing it. I'll bin 2x2 because it sounds like the right thing to do.

Went out last night and guided this beast at .5-.7 with its normal payload for 4 hours until the focuser slipped (its a WIP) and I was too tired to watch the monitor anymore.

Looking at that chart, it says it was 2.39" last night and an index of 4. It was really really mushy with the Moon however. Definitely not the best conditions.

[FiveByEagle]

My biggest issue last night, because I did start to use Bin2 and noticed it was no real loss in resolution, especially with drizzle and deconvolution in post.

Biggest issue was stray light! I never had to deal with that on my 81mm refractor. but my goodness every stray light beam somehow found its way into the tube!!

[vlaiv]

26 minutes ago, FiveByEagle said:

especially with drizzle and deconvolution in post.

Don't drizzle either

It is maybe useful for very under sampled data - odds are - you'll never have such data.

[FiveByEagle]

Fair enough!  

Seems there is too much of a compromise in any direction. I won't go with a larger sensor, as I only use 1.25" filters and enjoy this FOV.. but I also hate low resolution. a 5MP sensor would be the lowest resolution camera I own by a factor of 6.

 

Edited April 23, 2021 by FiveByEagle

[FiveByEagle]

Okay - so I am almost entirely set up and running. So far so good!

So my main question is now how to take flats. I use a smaller tracing pad for my 80mm refractor but it does not fit over the new tube.

With it being an 8", should I just go out and find a larger one? A4 sized should cover it but not sure on reflections and possible other problems. 

[vlaiv]

I use flat box like this one:

https://www.teleskop-express.de/shop/product_info.php/info/p8241_Lacerta-LED-Flatfield-Box-with-240-mm-usable-Diameter.html

except mine is older model.

You can DIY flat box with LED strip and couple of pieces of matte plexiglass panel to act as diffuser (in fact if you google it - there are quite a few options for light diffusion).