The Importance of Aperture in Astrophotography?

[Whistlin Bob]

I was having a friendly virtual chat with @Ken82 last night and we reached a question that I'm hoping some of the gurus of SGL may help with- especially as it may have consequences for my bank account!

Ken has a few imaging scopes and has been comparing results from 106mm and 280mm scopes- both very fine instruments- and hasn't really been able to see much difference between them on individual subs when looking at individual objects. The reason for this would appear to be that his guiding setup is working well and is essentially seeing limited, so he's getting a resolution you would expect from around 0.7" RMS depending on conditions regardless of what scope he's using.

This has given me pause for thought. I use 2 different scopes for imaging- a 130pds for wider stuff and a 200p for smaller targets (both at f4.5 with Coma Corrector). I've been trying to improve my guiding and I've found that my belt modded HEQ5 with a finder-guider and the 130 on it typically ranges between 0.7 to 0.9 RMS- which is fine for that scope and I'm very pleased with results. For the 200p I've never really been able to get it below 1" (and usually a bit higher) with the same mount and guider. I'm still pleased with the output, but also aware that the scope is probably a little undermounted and I could get improved resolution with a higher capacity mount. I was pondering an upgrade, and my thinking was that it would be worth it for the less wobbly guiding combined with the effect on SNR that comes from the 200p catching 2+ times as many photons. In my view I can see the effect of aperture in the images I'm getting- but I don't know if I'm just biased because I like big Newts, and it's hard to experiment on this given changing conditions between different sessions.

What do people think? 

[vlaiv]

You are right - larger aperture means more photons captured.

Given two scopes - a smaller one and larger one, both paired with cameras (binning or otherwise) that result in same sampling rate / resolution and cameras being otherwise equal (QE, noise and all) - larger scope will be faster.

Why don't people see this sometimes?

Because larger scope means larger focal length and that means larger sampling rate and most people don't bother to bin or otherwise adjust sampling rate.

Another thing is that small changes in SNR might not be obvious to human eye - no two images are processed the same, but they are measurable quantities. If you want to visually assert the difference in SNR - it is best to form single image while data is still linear - by using half of one image and other half from second image. That way processing is guaranteed to be the same as you are processing only one image. Of course, you must take care to "equalize/normalize" signal levels so that they match when you form single image.

 

[Whistlin Bob]

Thanks @vlaiv- really helpful!

[Ken82]

Yea thanks Vlaiv always very helpful👍

Does the guiding accuracy only play a part in the resolution then ? 
 

say guiding at 0.7”-0.9” with a 130mm scope. But then guiding at 1.5” with the 200mm ? Potentially your going to get better resolution with the 130 ??? 

Does this not also effect SNR? 

[vlaiv]

9 minutes ago, Ken82 said:

Does the guiding accuracy only play a part in the resolution then ? 

Depends what you mean by resolution.

That word has many uses that can sometimes get mixed up.

If by resolution you mean sampling rate / number of pixels per part of sky or number of pixels of the image in general - guiding accuracy does not play any direct part in that (it is related thru other use of the word if you want good results).

On the other hand, if by resolution you mean actual detail captured / resolved (hence resolution) - guiding precision does play significant part. Better guiding means less blurring due to poor tracking - in turn that means sharper image. Sometimes it can depend a lot on guiding precision and sometimes there won't be much difference between good and bad guiding - atmosphere/seeing can play significant part as well - so much so that in some cases good guiding simply won't help. Sometimes poor seeing is preventing you from guiding good - guide star just jumps around too much.

14 minutes ago, Ken82 said:

say guiding at 0.7”-0.9” with a 130mm scope. But then guiding at 1.5” with the 200mm ? Potentially your going to get better resolution with the 130 ??? 

This is true. Larger scope does not necessarily mean better resolution - in sense of resolving things. In particular this is true for undermounted large scopes. There are other intricate details of how atmospheric seeing effects depend on aperture size and so on - not easy topic to both understand and predict correctly.

In general - larger aperture has the possibility of resolving more, but host of factors - seeing, guiding and such, can interfere and prevent it from delivering.

16 minutes ago, Ken82 said:

Does this not also effect SNR? 

Yes and no. It does affect highly contrasting features, but such features often have enough signal for impact to be minimal. If something is high contrast (like star against black space) - one component has to have strong signal, and strong signal means good SNR.

On the other hand if you have low contrast feature - like uniform nebulosity, then blurring that won't do pretty much anything to disturb signal and hence snr (blur of uniform intensity is again that same uniform intensity. only place where you see blur is where there is contrast).

In most practical cases - a bit more blurring due to poor guiding or poor seeing won't affect SNR much. There will be cases, for example - trying to catch very faint star, or similar where it will reduce SNR and can mean difference between detection and having no detection, but for regular imaging - it won't make a difference.

[ollypenrice]

We've been here before but it seems to me that the useful unit would be one we never see: area of aperture per pixel.  This unit would clear up the confusion sowed by the term F ratio and would work nicely with the useful term arcseconds per pixel.

Olly

[andrew s]

52 minutes ago, ollypenrice said:

We've been here before but it seems to me that the useful unit would be one we never see: area of aperture per pixel.  This unit would clear up the confusion sowed by the term F ratio and would work nicely with the useful term arcseconds per pixel.

Olly

I like the idea  but I am not sure that the right unit as it depends on the focal length and pixel size.  Not sure I can think of anything better but might go for aperture per FWHM . However, I know that has issues too.

Regards Andrew 

How about aperture per arc sec.?

Edited May 23, 2020 by andrew s

[ollypenrice]

54 minutes ago, andrew s said:

I like the idea  but I am not sure that the right unit as it depends on the focal length and pixel size.  Not sure I can think of anything better but might go for aperture per FWHM . However, I know that has issues too.

Regards Andrew 

How about aperture per arc sec.?

Yes!

Olly

[wimvb]

Isn't that the Rayleigh criterion?

Rayleigh: angular resolution =1.22 L/D, where L is wavelength, D is aperture.

[andrew s]

15 minutes ago, wimvb said:

Isn't that the Rayleigh criterion?

Rayleigh: angular resolution =1.22 L/D, where L is wavelength, D is aperture.

For me "aperture" is D*D so that it measures the energy collected and hence how much gets into one arc second .  Then you have to decide what limits the resolution. Scope seeing, guiding . Need to think it through. Regards Andrew 

Edited May 23, 2020 by andrew s

[wimvb]

18 minutes ago, andrew s said:

For me "aperture" is D*D so that it measures the energy collected and hence how much gets into one arc second . 

I'm sorry, I don't follow.

[Captain Scarlet]

My understanding is that a fundamental unit is etendue, or area of aperture _multiplied_ by Solid-arcseconds (subtended from location of aperture to pixel) and that etendue multiplied by exposure time (or square root of) is what determines SNR

[andrew s]

43 minutes ago, Captain Magenta said:

My understanding is that a fundamental unit is etendue, or area of aperture _multiplied_ by Solid-arcseconds (subtended from location of aperture to pixel) and that etendue multiplied by exposure time (or square root of) is what determines SNR

Quite right, I had forgotten about that. Although I am not sure it quite mirrors the same idea. If I have remembered correctly it measures across the filed of view not per arc second.

Regards Andrew 

Edited May 23, 2020 by andrew s
Add more.

[Whistlin Bob]

Thanks for the input. I can see there's a maths answer brewing here, which should give me something firmer to use. One for the morning I think! 

[Captain Scarlet]

1 hour ago, andrew s said:

Quite right, I had forgotten about that. Although I am not sure it quite mirrors the same idea. If I have remembered correctly it measures across the filed of view not per arc second.

Regards Andrew 

Yes I think it’s per pixel subtending the (solid) angle, and hence across the field of view. But per arcsecond is a simple extra step. Getting dangerously close to the edge of my understanding here

Edited May 23, 2020 by Captain Magenta

[vlaiv]

1 hour ago, Captain Magenta said:

My understanding is that a fundamental unit is etendue, or area of aperture _multiplied_ by Solid-arcseconds (subtended from location of aperture to pixel) and that etendue multiplied by exposure time (or square root of) is what determines SNR

Not quite as easy as that.

SNR depends on signal - which is what is discussed here, but also on all noise sources - read, thermal, shot and LP noise all play a part. In recent times, LP noise has become quite dominant noise component since set point cooling has become widely available in amateur setups and more and more man made light is present at night.

Defining speed of telescope in light gathering capability only is very useful for comparing two scopes - but can be really misleading to some.

We can conclude that particular scope and camera is very fast system - yet people in heavy LP will struggle to get decent images in multiple nights of exposing. For example - I calculated (or better term would be estimated?) that exposure difference between my current and future site - mag18.5 vs mag20.8 will be about x6 in exposure time. Same scope - x6 times faster in darker skies!

[wimvb]

https://arxiv.org/pdf/1401.5473.pdf (eqs. 3 and 4, mainly)

A crude approximation:

the light gathering ability of an imaging system is proportional to the square of pixel size divided by focal ratio, (p/F)^2

This is equivalent to:

the light gathering ability of an imaging system is proportional to the square of the pixel scale multiplied by the aperture (diameter), (rD)^2

This is under the assumption that the system isn't undersampled, and any light smearing effects, such as poor seeing, poor focus, or poor guiding, are not accounted for.

Oh, and nothing about signal to noise ratio.

[vlaiv]

How about if we take pixels out of equation and just concentrate on aperture per arc second squared as "speed" of the scope.

I know that many will say - but you can't take pixels out of the equation, and I say - we can match resolution by using fractional binning or other resampling methods.

[wimvb]

7 minutes ago, vlaiv said:

We can conclude that particular scope and camera is very fast system - yet people in heavy LP will struggle to get decent images in multiple nights of exposing. For example - I calculated (or better term would be estimated?) that exposure difference between my current and future site - mag18.5 vs mag20.8 will be about x6 in exposure time. Same scope - x6 times faster in darker skies!

Do you mean like this:

https://skyandtelescope.org/astronomy-blogs/astrophotography-benefits-dark-skies/

[wimvb]

4 minutes ago, vlaiv said:

How about if we take pixels out of equation and just concentrate on aperture per arc second squared as "speed" of the scope.

I know that many will say - but you can't take pixels out of the equation, and I say - we can match resolution by using fractional binning or other resampling methods.

That changes the outcome of the image, but not the capability of the hardware.

[vlaiv]

3 minutes ago, wimvb said:

Do you mean like this:

https://skyandtelescope.org/astronomy-blogs/astrophotography-benefits-dark-skies/

Something similar. I would not agree that all targets and all conditions have same rule - x2.51 for mag off difference.

It depends on brightness of the target - I did my calculation based on mag26 targets, and also on my particular setup (sampling rate affects both target and LP signal in the same way - spread over more pixels, however signal grows linearly and LP noise as square root of LP signal with time).

[vlaiv]

5 minutes ago, wimvb said:

That changes the outcome of the image, but not the capability of the hardware.

It shows something else - speed depends on sensor size as well as aperture size. Not something that we consider often.

Take for example ASI1600 + 8" RC scope vs ASI6200 + 16" RC scope.

Roughly speaking, since ASI1600 has diagonal of about 22mm and ASI1600 has diagonal of 45mm - we can say that it is twice as large sensor (in height and width) - or x4 in surface area.

16" RC scope will have twice the focal length of 8" RC scope.

Thus both will capture the exact same FOV (five or take a bit). With use of binning - you can make them sample at same sampling rate.

16" RC  + ASI6200 will be clear winner as it grabs x4 more light and illuminates same FOV. In that sense we can say that with scope it is aperture per FOV and basic unit of FOV is arc second squared.

In the end - we conclude what visual observers new all along - with scope, aperture is king .

As imagers, we need to add our own saying, and I propose: With sensors - sensor size is king.

[tooth_dr]

Are these images I took useful as examples?  Same camera, same night, one with 250mm and one with 80mm scope?

 

 

[Captain Scarlet]

9 minutes ago, vlaiv said:

It shows something else - speed depends on sensor size as well as aperture size. Not something that we consider often.

Take for example ASI1600 + 8" RC scope vs ASI6200 + 16" RC scope.

Roughly speaking, since ASI1600 has diagonal of about 22mm and ASI1600 has diagonal of 45mm - we can say that it is twice as large sensor (in height and width) - or x4 in surface area.

16" RC scope will have twice the focal length of 8" RC scope.

Thus both will capture the exact same FOV (five or take a bit). With use of binning - you can make them sample at same sampling rate.

16" RC  + ASI6200 will be clear winner as it grabs x4 more light and illuminates same FOV. In that sense we can say that with scope it is aperture per FOV and basic unit of FOV is arc second squared.

In the end - we conclude what visual observers new all along - with scope, aperture is king .

As imagers, we need to add our own saying, and I propose: With sensors - sensor size is king.

Is this not all suggested, from both sides, imager and visual, in the etendue expression? Except the other “king” for imagers is exposure time...

[andrew s]

9 hours ago, Captain Magenta said:

Is this not all suggested, from both sides, imager and visual, in the etendue expression? Except the other “king” for imagers is exposure time...

Yes we have reinvented Etendue. What it says in broad terms is the bigger the aperture (to capture more light) and the larger the FOV (to capture more sky) the  more efficient the system. However, if your target is significantly smaller than the FOV the it's just aperture that counts.

As @vlaiv points out noise sources also play a role in detecterbility .

Regards Andrew