Focal Length for Imaging

[Clarkey]

Currently I have an RC8 which I use for imaging which gives a pixel scale  just under 0.5"/pixel when used with my ASI1600. I usually bin 2x2 to give a pixel scale just under 1"/pixel and the scope is at F8.

I am considering an F4 -F5 newtonian for imaging which would be at a FL of around 800 - 1000mm. At this FL is gives the right pixel scale for 99% of seeing conditions in the UK but the additional benefit of a faster scope. (I am talking about DSO imaging - obviously planetary / lucky imaging is different)

My question is simple really. Is there any benefit in using a FL of more than 1000mm with modern small pixel CMOS cameras if rescaling on a computer will ultimately give the same result?

Any thoughts would be appreciated.

[vlaiv]

Yes there is benefit as with increased focal length often comes increased aperture.

You are now using 200mm diameter 1600mm FL scope - and you are happy to bin x2 (while in reality you should be binning x3 most of the time and sometimes even x4).

Say you get 300mm F/5.3 imaging newtonian. It will also have ~1600mm, so nothing will change in the way you capture and process your data (you'll still need to bin with certain bin factor due to small pixels) - but 300mm will collect more photons than 200mm.

Take this to another extreme - take 16" F/8 RC and pair it with ASI6200 - which again has small pixels at 3.76µm (almost the same as ASI1600 -3.8µm).

It will give you the same FOV - you'll need to bin differently (maybe x6 or x7), but it will be x4 as fast as your current setup (with longer FL). It will also be x4 as expensive .

Don't think that small pixels are advantage - rather think that small pixels are nuisance for astro imaging and that larger pixels actually better suit telescopes. Luckily CMOS sensors have low read noise and it makes sense to bin them in software. Small pixel size has another advantage - that no one is using at the moment (as far as I know).

They are closer to true point sampling and introduce less pixel blur if used properly (split bin instead of regular binning).

 

[newbie alert]

As an option you could buy a ccd47  as a x.67 focal reducer and you're be imaging at 1072mm or f5.36 for the f ratio freaks... 

[vlaiv]

17 minutes ago, newbie alert said:

As an option you could buy a ccd47  as a x.67 focal reducer and you're be imaging at 1072mm or f5.36 for the f ratio freaks... 

Only on smaller sensors.  CCD47 reduces FL by 0.67 - but also shrinks usable field. If RC8 can utilize about 27-28mm - that also gets multiplied with 0.67 - so it ends up being about 19mm. Most people don't use CCD47 / CCDT67 at prescribed x0.67 - but rather x0.72 - x0.75 as that is much better match for 4/3 sensor size like ASI1600, ASI294 or KAF8300 (28mm x 0.72 = 20.16mm).

I'd rather go with Riccardi x0.75 FF/FR as that both reduces FL and corrects field curvature. I think that way up to APS-C becomes usable when reduced (that would be equivalent of almost 38mm not reduced - that sort of field is affected by field curvature quite a bit).

 

[ollypenrice]

I don't think we can assume that binning will 'just work.'  I've had two CCD cameras on which it produced too many artifacts to be of any use at all and others on which it worked perfectly. This includes two cameras of the same make and model. I'm also finding with a CMOS camera (ASI 2600) that software binning in AstroArt is giving me poor stars on which the colours are not properly mixed. I get different coloured stellar edges from one side to the other. Perhaps other software will work better but I mention it as a 'problem in progress' which we're still exploring.

Olly

[vlaiv]

1 hour ago, ollypenrice said:

I don't think we can assume that binning will 'just work.' 

It's a bit like saying that we can't just assume that stacking works.

Both things work - mathematically they are quite similar things. It is implementations of each that can be better or worse. Even hardware binning can fail if sensor manufacturing process failed - or you simply got defective units. Software binning can also "fail" - if it's not implemented properly.

Binning can also "fail" if one uses it improperly. If we say take single sub and "stack" it with another 8 exact copies of itself - we might conclude that stacking simply does not work. Problem is that we did not get statistically independent samples - but rather copy of one sub and of course it won't work properly.

Similarly, binning can also "fail" - If one bins debayered single sub - it won't produce expected results as debayering already contained interpolation and added pixels are not statistically independent. Similarly, binning stack of debayered subs is not the same as binning mono sensor data or binning stack of differently debayered subs (for example if one uses super pixel mode to debayer instead of interpolation).

If all these things are taken care of - then binning works as expected every time.

[ollypenrice]

22 minutes ago, vlaiv said:

It's a bit like saying that we can't just assume that stacking works.

Both things work - mathematically they are quite similar things. It is implementations of each that can be better or worse. Even hardware binning can fail if sensor manufacturing process failed - or you simply got defective units. Software binning can also "fail" - if it's not implemented properly.

Binning can also "fail" if one uses it improperly. If we say take single sub and "stack" it with another 8 exact copies of itself - we might conclude that stacking simply does not work. Problem is that we did not get statistically independent samples - but rather copy of one sub and of course it won't work properly.

Similarly, binning can also "fail" - If one bins debayered single sub - it won't produce expected results as debayering already contained interpolation and added pixels are not statistically independent. Similarly, binning stack of debayered subs is not the same as binning mono sensor data or binning stack of differently debayered subs (for example if one uses super pixel mode to debayer instead of interpolation).

If all these things are taken care of - then binning works as expected every time.

But this is of no consolation if you've bought equipment intending to bin and find that you can't do so...

Olly

Edited September 12, 2021 by ollypenrice

[vlaiv]

7 minutes ago, ollypenrice said:

But this is of no consolation if you've bought equipment intending to bin and find that you can't do so...

Olly

You can bin any data in software. It won't have benefit of hardware binning - if your gear is capable of it, but it will bin just fine in software.

If you purchased gear with intent of hardware binning and it does not work - well then gear is defective and you can return it, can't you? (unless you bought it second hand).

[newbie alert]

3 hours ago, vlaiv said:

Only on smaller sensors.  CCD47 reduces FL by 0.67 - but also shrinks usable field. If RC8 can utilize about 27-28mm - that also gets multiplied with 0.67 - so it ends up being about 19mm. Most people don't use CCD47 / CCDT67 at prescribed x0.67 - but rather x0.72 - x0.75 as that is much better match for 4/3 sensor size like ASI1600, ASI294 or KAF8300 (28mm x 0.72 = 20.16mm).

I'd rather go with Riccardi x0.75 FF/FR as that both reduces FL and corrects field curvature. I think that way up to APS-C becomes usable when reduced (that would be equivalent of almost 38mm not reduced - that sort of field is affected by field curvature quite a bit).

 

Thanks for that info Vlaiv 

[Clarkey]

4 hours ago, vlaiv said:

You are now using 200mm diameter 1600mm FL scope - and you are happy to bin x2 (while in reality you should be binning x3 most of the time and sometimes even x4).

I probably should have been more specific with my question to state that it would be aperture limited. With my current set up I could push the aperture a bit more, but I would not want to go above 250mm as I don't want to go sailing🤣

 

3 hours ago, ollypenrice said:

I don't think we can assume that binning will 'just work.'  I've had two CCD cameras on which it produced too many artifacts to be of any use at all and others on which it worked perfectly. This includes two cameras of the same make and model. I'm also finding with a CMOS camera (ASI 2600) that software binning in AstroArt is giving me poor stars on which the colours are not properly mixed. I get different coloured stellar edges from one side to the other. Perhaps other software will work better but I mention it as a 'problem in progress' which we're still exploring.

Interesting thought. If software (or hardware) binning does not actually represent the limitations linked to seeing and pixel ratio then it adds another variable to considered.

 

4 hours ago, newbie alert said:

As an option you could buy a ccd47  as a x.67 focal reducer and you're be imaging at 1072mm or f5.36 for the f ratio freaks...

I have got a ccd47 reducer which I use to about 0.73 which works OK with the ASI1600. However, I decided that although the results were pretty good - I managed to get round stars across the frame - it was one more piece of glass to add optical aberrations / reflections etc, so I only use it if I need the extra FOV. But then I could use the ED80 instead......

[Xilman]

FWIW, my SX814 has a native pixel size of 0.29" which matches nicely the Airy disk of ~0.31" of 0.4m f/6.5 (i.e. 2614mm focal length) optics. Almost all the time I use hardware 2x binning which spreads the typical seeing-limited 2.5 FWHM stellar image over a 4 pixel FWHM. As my interest is primarily stellar photometry that is close to ideal. The photons are spread over a good number of pixels, thereby mitigating pixel-to-pixel variations in sensitivity, and the binning allows very nearly the full 16-bit digitization range to be used without saturating the (binned) pixels.

As the minimum exposure time is 1ms, very occasionally I image unbinned as fast as possible and hope to get Lucky on bright objects. Managed Sirius B that way but have yet to image Procyon B. Must try for some bright and close (sub-1") double stars some time on nights of good seeing. Sirius has a very wide separation at around 8"; the only difficulty observing it is the 10-magnitude difference in brightness between primary and secondary.

YMMV.

Edited September 12, 2021 by Xilman
Minor clarification

[Xilman]

5 hours ago, vlaiv said:

Small pixel size has another advantage - that no one is using at the moment (as far as I know).

They are closer to true point sampling and introduce less pixel blur if used properly (split bin instead of regular binning).

 

This is rather important, IMO, but perhaps beyond the interests and/or capabilities of many amateurs at image processing. If you wish to perform super-resolution of images beyond the scale imposed by the seeing, you need to be able to sample the point spread function adequately. Ideally you should be sampling at half the size of the Airy disk or better for your telescope. For most people here that means a pixel size of under 0.2 arcseconds.

Martin Lewis, to mention just one, has presented some superb images from a 0.46m telescope (Airy disk of 0.3") at 0.11" per pixel. He resolves features on the disk of Ganymede which tally perfectly with the images returned from Voyager and subsequent Jovian orbital probes.

[Xiga]

On 12/09/2021 at 16:01, ollypenrice said:

I don't think we can assume that binning will 'just work.'  I've had two CCD cameras on which it produced too many artifacts to be of any use at all and others on which it worked perfectly. This includes two cameras of the same make and model. I'm also finding with a CMOS camera (ASI 2600) that software binning in AstroArt is giving me poor stars on which the colours are not properly mixed. I get different coloured stellar edges from one side to the other. Perhaps other software will work better but I mention it as a 'problem in progress' which we're still exploring.

Olly

@ollypenrice how are you stacking your ASI 2600 subs? Are you using Bayer Drizzle? When you have well-dithered subs, and lots of them (which is pretty much always the case with OSC CMOS) then Bayer Drizzle basically bypasses the need to do interpolation. You might find that fine-scale colour details are a bit more accurate, and it could make a difference when then binning. Worth a shot i would think. Unless you are already using Bayer Drizzle, in which case ignore me! 😄

[ollypenrice]

30 minutes ago, Xiga said:

@ollypenrice how are you stacking your ASI 2600 subs? Are you using Bayer Drizzle? When you have well-dithered subs, and lots of them (which is pretty much always the case with OSC CMOS) then Bayer Drizzle basically bypasses the need to do interpolation. You might find that fine-scale colour details are a bit more accurate, and it could make a difference when then binning. Worth a shot i would think. Unless you are already using Bayer Drizzle, in which case ignore me! 😄

Thanks Ciarán. (I had to copy and paste to get the accent! 😁) I'll see if I can give this a try. We do have APP but I haven't yet tried it.

Olly

[Newforestgimp]

Sorry to resurrect this but my head is scrambled trying to work out back focus and reductions. I'm yet to test my new set up but i wondered if someone could give me a yay or nay with what i have put together and if im going to be able to achieve focus ?

RC8"  >  50mm extension ring  >  focuser  >  CCD47  >  ZWO OAG 16.5mm  >  ALTAIR Filter Drawer 17mm  >  WO M48 T Mount 11mm  >  550D DSLR 44mm

I have pushed the CCD47 as far into the focuser as possible so that the OAG butts up against the focuser, & by my measurements & component specs i get 88.5mm from focuser to the sensor plane on the 550D APS-C sensor. I can see TSO recommend 85mm, will the extra 3.5mm hurt me or should i look to ditch something like the T Mount for a slimmer version ?

I'm getting a bit lost when people say they are using the 0.67 reducer at 0.72 and how this is achieved ?

Do I need the 50mm Extender Ring or should i attached the focuser direct to the OTA ?

Any advice appreciated, even better if its in proper layman terms. Id like to know before i spend hours trying to achieve the impossible if one of the clever people on here could say Yes that'll work or don't be silly you'll never get focus with that lot 🤣

Regards

Andy

 

[vlaiv]

1 minute ago, Newforestgimp said:

will the extra 3.5mm hurt me or should i look to ditch something like the T Mount for a slimmer version ?

It will hurt a bit. Further away you move sensor - larger reduction factor. You are already trying to illuminate and correct APS-C sensor.

There is formula for reduction that goes something like this

reduction_factor = 1 - distance / focal_length

Focal length of x0.67 reductor is 303mm. This means that you need to place sensor at 101mm (from the center of the lens) to get x0.67 reduction as

1 - 101 / 303 = 0.66666 = ~0.67

You actually get 101mm away from center of the lens when you use 85mm of extensions. Which means that center of the lens is roughly 16mm away from the thread (this is important to know)

Say you want 0.72 reduction - what distance do you need?

0.72 = 1 - x/303

x = 303 * (1-0.72) = 84.84mm

This is from center of the lens, but since lens to thread distance is 16mm - actual spacing is 84.84 - 16mm = ~ 68.84mm = ~69mm

RC8 will illuminate at max 30mm without too much distortion - but if you add x0.67 reducer - it will "shrink" that field to 30 * 0.67 = 21mm.  That means that beyond central ~21mm you can expect some rather poor star shapes on your 28mm sensor.

I'd use APS-C sensor without reductor on RC8 - only with flattener if you have too much curvature at the edges.

[Newforestgimp]

Thanks Vlaiv,

i “think” I now understand the relationship between the reducer back focus and sensor, your explanation really helps.

So in theory if I dropped the filter drawer that would get me somewhere closer to ideal. Or remove the reducer from the equation and the filter drawer could stay for example ?

how about the extender rings between OTA and focuser ? Do I need 50mm or can I reduce that ?

[vlaiv]

1 minute ago, Newforestgimp said:

So in theory if I dropped the filter drawer that would get me somewhere closer to ideal.

Yes, that would make it 71.5mm or x0.72 reduction factor - which is "ideal" for 4/3 size chip (about 21-22mm diagonal).

22mm / 0.72 = 30.55mm - you can just about squeeze 30.55 onto 22mm using 0.72 reduction.

3 minutes ago, Newforestgimp said:

Or remove the reducer from the equation and the filter drawer could stay for example ?

I'd probably opt for that version. That + x3 software binning to get you at ~1.6"/px.

That would be a good place to be with DSLR.

Just keep in mind that you'll benefit from longer exposures since you are natively at only 0.55"/px (which means read noise will dominate as sky flux will be low), so keep exposures as long as feasible.

When you get your images - stack them as you normally would and then before you start processing - bin result x3.

6 minutes ago, Newforestgimp said:

how about the extender rings between OTA and focuser ? Do I need 50mm or can I reduce that ?

It changes nothing with respect to focal reducer - but you might want to loose it when using focal reducer (or at least use 25mm version). This is because focal reducer moves focus point inward / shortens back focus, but its really best checked - if you can reach focus with extension - leave it, if not - remove it. Simple as that.

Without reductor - you'll probably need extension to comfortably reach focus.

[Newforestgimp]

Perfect Vlaiv, thank you.

As you suggest I think I will remove the reducer and see how I get on with exposure times. I will also try the binning on the ASI AIR too see if that helps.

Andy

[vlaiv]

Just now, Newforestgimp said:

I will also try the binning on the ASI AIR too see if that helps.

Do look into that, but I always prefer to bin later in software - gives a bit more options and with CMOS sensors - makes no difference.