Is long focal length a waste for imaging?

[Padraic M]

I'm trying to get my head around the relationships between pixel size, focal length and seeing, and trying to understand whether there is a benefit in having a second imaging setup with a longer focal length. I can just about follow other conversations on this topic, so my logic may be completely out - looking for corrections and suggestions.

My current main imaging setup is ASI1600MM Pro on an Esprit 80/400, NEQ6 (or HEQ5) guiding at ~0.6-0.8". HFR on good nights is 2.6 which implies seeing of 1.6". I also have a C8 SCT @ 2032mm f/10 or 1280mm f/6.3 with a flattener/reducer. Intuitively, there seems to be a logic that says that the two rigs can work together, with the overall wide-ish field being captured by the Esprit, and narrow-field detail being added by the C8. Think, as an example, capturing Orion with the Esprit, and getting close-in detail of the trapezium with the C8 at shorter exposures. But I'm wondering if (given the seeing) the two setups are actually equivalent and the C8 will give no benefit whatsoever. As it is, the optical quality of the Esprit is much better than the C8, so unless there's a tangible benefit that would warrant investment in the C8 setup, I may be better off not going down that road.

Pixel scale for the Esprit/1600MM setup is 1.9". This is pretty much spot on given the seeing of down to 1.6", and the guiding is well within this limit.

Pixel scale for the C8/1600MM with 0.63x reducer is 0.61" which is way oversampled. If I understand this correctly, I get no additional detail with any pixel scale below the seeing (obviously without using AO); I would need to bin 2x2 or even 3x3 to sample correctly, which will improve capture speed but not resolution.

Even changing to a different camera with larger pixels still hits the same limit - the pixel scale needs to be >1.6" to capture detail un-binned, and that puts it right back into the Esprit resolution territory. The only difference is that the C8 has a much narrower FOV than the Esprit, but I can achieve that at the same resolution and with better image quality by cropping an Esprit image. What am I missing here?

[vlaiv]

I'm talking from diffraction limited point of view here - some scopes are not diffraction limited over their imaging field so things will be a bit different for them.

Detail in the image that you capture depends on several components - seeing, guiding precision and aperture size - to name 3 main ones.

They work together and no single component is main one, but important thing to note is that - aperture size plays a part.

For example - with 80mm scope, you simply can't hope to realistically go below say 1.8"/px in long exposure astrophotography.

In my view - you have rather good match with your current setup - Esprit 80/400 and ASI1600 if it samples at 1.9"/px.

I don't really understand this part:

HFR on good nights is 2.6 which implies seeing of 1.6".

There is relationship between FWHM (which is the same for perfect Gaussian as HFR) of stars in the image and above mentioned 3 components.

If you have FWHM of 2.6" in your image and you guide at 0.6" RMS, then seeing must be around 1.7" FWHM. If you guide with 0.8" then seeing needs to be 1.1" FWHM.

In any case - larger scope will give you both a bit of edge on possible resolution and speed of capture. You can almost never achieve 1.5"/px image with 80mm scope - but you'll be able to do it with 6"-8" of aperture. On very good nights you might be able to go down to 1.2"/px (but this requires good skies, good mount and good focus - and of course diffraction limited telescope).

 

[Padraic M]

@vlaiv, OK, I got you. The larger aperture of the C8 (200mm) gives a smaller Dawes limit of 0.67". So if seeing supports this, or this binned x2 to 1.34", it can provide better detail than the Esprit, design limitations apart.

1 hour ago, vlaiv said:

I don't really understand this part:

HFR on good nights is 2.6 which implies seeing of 1.6".

Sorry, my confusion. I misremembered one of your previous posts relating FWHM to seeing by a factor of 1.6.

I've checked my image stats in APP for my most recent session, and I got FWHM between 2.5 and 3.0 consistently, with average guiding RMS from Phd2 log viewer of 0.6" over 2hrs 15mins. Can you tell me where the relationship below comes from?

2 hours ago, vlaiv said:

If you have FWHM of 2.6" in your image and you guide at 0.6" RMS, then seeing must be around 1.7" FWHM. If you guide with 0.8" then seeing needs to be 1.1" FWHM.

Thanks!

[powerlord]

Not sure I really get FWHM to be honest. I get 2.5 or so if I'm shooting wider field (400mm), but only 4.5 on a good night with 1300mm. So it's related to FL it seems, and maths says my asi1600 with 1300mm (c925x6.7) is massive oversampling, but I can get this, which is miles better than widefield results.

[vlaiv]

9 hours ago, Padraic M said:

Sorry, my confusion. I misremembered one of your previous posts relating FWHM to seeing by a factor of 1.6.

I've checked my image stats in APP for my most recent session, and I got FWHM between 2.5 and 3.0 consistently, with average guiding RMS from Phd2 log viewer of 0.6" over 2hrs 15mins. Can you tell me where the relationship below comes from?

Ah ok.

So, factor of x1.6 relates to FWHM of final image and "optimum" sampling rate (optimum is quoted because it is really approximation so we can't really say optimum value).

If you have 2.6" FWHM then you need to sample at 2.6" / 1.6 = 1.625"/px

Since you have between 2.5 and 3.0 - then your sampling rate of 1.9"/px is spot on, as above would give sampling between ~1.56"/px and 1.875"/px and you are on upper bound of that.

As far as getting expected FWHM, it goes like this:

We take three major sources of blur and we approximate each one of those with Gaussian. These three convolve one another and produce final blur - also Gaussian (convolution of Gaussian by Gaussian is another Gaussian). Their variances add to produce final one - which means that standard deviations add in quadrature.

- seeing FWHM - we divide FWHM value with 2.355 to get sigma of Gaussian

- Airy disk - Radius is given by 1.22 x lambda, while sigma of Gaussian approximation is given by 0.42 x lambda (source: https://en.wikipedia.org/wiki/Airy_disk#Approximation_using_a_Gaussian_profile)

- guide RMS is already sigma of the spread

You take those three values and you take square root of sum of their squares and then you multiply resulting value with 2.355 to get FWHM.

[vlaiv]

2 hours ago, powerlord said:

Not sure I really get FWHM to be honest. I get 2.5 or so if I'm shooting wider field (400mm), but only 4.5 on a good night with 1300mm.

Are those values in pixels or in arc seconds?

They need to be in arc seconds in order to be comparable.

[powerlord]

2 hours ago, vlaiv said:

Are those values in pixels or in arc seconds?

They need to be in arc seconds in order to be comparable.

I don't know, they appear in asiair when focusing, and in AAP when registering. I assumed pixels for it to make sense.

[Padraic M]

@powerlord that's a nice Triangulum! Is it the non-Edge HD 9.25, and what flattener/reducer are you using?

[Padraic M]

@vlaiv, great, thanks for that answer, I think I understand now. There is a benefit in a second setup with a longer focal length. The question now is whether the C8 XLT is up to the job optically.  I need more experiments!

[vlaiv]

1 hour ago, Padraic M said:

@vlaiv, great, thanks for that answer, I think I understand now. There is a benefit in a second setup with a longer focal length. The question now is whether the C8 XLT is up to the job optically.  I need more experiments!

I don't think regular C8 will be up for the job. From what I've seen - C8s always produce bloated stars for some reason.

Could be the fact that SCTs have spherical aberration unless their mirrors are at exact prescribed distance? Since focusing is done with moving primary - there is actually only one position where telescope will behave the best - as soon as you remove diagonal and/or add elements to optical train and refocus - you'll change distance between mirrors and introduce some spherical in the mix.

I've seen some good images produced with EdgeHD optics, so that is better bet - but also check spot diagrams on those as well, as some models are not quite diffraction limited.

[Padraic M]

Yes I'm inclined to agree. I've spent quite a bit of time adjusting back distance etc. to try to get round stars with the C8 and haven't succeeded so far. Was considering trading up to an EdgeHD, or possibly using the Starizona flattener which gets good reviews. For much the same cost of upgrading, the EHD may be the better solution.

[barbulo]

Sorry for jumping in but, how can I measure the FWHM? I’ve searched in imageJ and astroimageJ with no success. 
I believe you’re taking about an average of the FWHM of all the stars in the image, am I right? Thanks in advance for the clarification. 

[vlaiv]

With astroimageJ it is rather easy - you alt + click on any particular star and it will give you FWHM profile in new window.

It won't do average on the image, but if you do that on few stars - you'll get pretty good idea of what average FWHM is.

Another way is - just select and configure photometry tool and select multiple stars with it - you'll get results table containing FWHM - you can then copy/paste that in spread sheet and do average

[barbulo]

Thank you very much @vlaiv. I had in front of my eyes 🙈

[powerlord]

3 hours ago, Padraic M said:

@powerlord that's a nice Triangulum! Is it the non-Edge HD 9.25, and what flattener/reducer are you using?

standard non HD XLT job. with standard celestron flattener/reducer.

And before that, I had an old 1980s C8 with the same flattener, and took this with it, which i was reasonably happy with.

[Padraic M]

23 hours ago, vlaiv said:

With astroimageJ it is rather easy - you alt + click on any particular star and it will give you FWHM profile in new window.

It won't do average on the image, but if you do that on few stars - you'll get pretty good idea of what average FWHM is.

Another way is - just select and configure photometry tool and select multiple stars with it - you'll get results table containing FWHM - you can then copy/paste that in spread sheet and do average

 

This is puzzling me now - when I load a fits file I can alt+click, or do multi-aperture photometry, but the measurements table doesn't contain FWHM. I can see in the Aperture Photometry Settings a list of fits file header keywords to include in the measurements table and FWHM is not one of them.

I don't see any way to ask APP to include FWHM in the fits header, so maybe this is why it's not appearing in the table??? 

Also, AstroImageJ seems to approximate a FWHM based on user-defined constants (i.e. the 13/19/26 radii above, and/or the FWHM factor pf 1.4). Do you have a different way of calculating FWHM?

[vlaiv]

13 minutes ago, Padraic M said:

This is puzzling me now - when I load a fits file I can alt+click, or do multi-aperture photometry, but the measurements table doesn't contain FWHM. I can see in the Aperture Photometry Settings a list of fits file header keywords to include in the measurements table and FWHM is not one of them.

@barbulo had similar issue, don't know if it is solved now, but this is what I suggested. In above dialog - Aperture Photometry Settings - there is More options button on the bottom. This takes you to another screen where you can select FWHM measure to be included:

As far as alt-click goes - you should get profile for a single star looking like this:

In top part it says FWHM in pixels (in this case since image is not plate solved and I did not enter plate scale factor).

16 minutes ago, Padraic M said:

Also, AstroImageJ seems to approximate a FWHM based on user-defined constants (i.e. the 13/19/26 radii above, and/or the FWHM factor pf 1.4). Do you have a different way of calculating FWHM?

I don't think it approximates anything - it does Gaussian fitting and calculates FWHM (and other parameters).

Above radii are essential part of photometry process.

You need to tell how much of signal is considered star in the image - first radius / first aperture

You also need to tell software how to calculate background - this is where second and third radius come in - anything between second and third radius will be considered "Sky"

And yes, you need to set those sensibly for photometry calculations to work.

[barbulo]

20 minutes ago, Padraic M said:

the measurements table doesn't contain FWHM.

Same here! When I alt+click I have the seeing profile of the single star, as depicted by @vlaiv in the previous post. But in the FWHM column is missing in the tote view. 

Trying to solve the issue, I updated the AtroimageJ app to its latest version (3.40). Now I can't open any image file. 🤦‍♂️

Tonight is the fist clear in weeks so I will try to troubleshoot any other night. 

[alacant]

40 minutes ago, Padraic M said:

different way of calculating FWHM

There's a good method described here.

Open source download here.

HTH

[Padraic M]

Thanks Vlaiv,

57 minutes ago, vlaiv said:

you can select FWHM measure to be included

Yes, I have that selected. Still no sign. I'll keep reading and trying - I don't want to hog your time! I can get FWHM in the seeing profile. Also in pixels, although I've plate-solved and set the plate scale. 

This is how my radii are set. 

57 minutes ago, barbulo said:

Now I can't open any image file. 🤦‍♂️

Oops! I'm on 3.2 so maybe it's worth the risk??

@alacant thanks for that link. I have Siril installed so will take a look. 

 

[vlaiv]

11 minutes ago, Padraic M said:

es, I have that selected. Still no sign. I'll keep reading and trying - I don't want to hog your time! I can get FWHM in the seeing profile. Also in pixels, although I've plate-solved and set the plate scale. 

That profile looks like clipped, so be careful - use stars that are not saturated in their cores (less bright ones) for FWHM measurement.

[Padraic M]

Yes, I noticed. It may be showing that a shorter exposure/lower gain might be appropriate as quite a few stars are saturated at that level.