Jump to content

  •  

CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.

Photo

Extreme Focal Ratios and Narrowband Filter

  • Please log in to reply
39 replies to this topic

#26 AtmosFearIC

AtmosFearIC

    Apollo

  • -----
  • Posts: 1005
  • Joined: 10 Dec 2015
  • Loc: Melbourne

Posted 24 August 2019 - 06:25 PM

Hi AtmosFearlC,

 

Where are you finding the data on the amount of intensity drop at different focal ratios?

 

Thanks,

 

Jerry

I did a quick Google search but cannot remember the website. A number of years ago someone did some testing with 3nm and 5nm filters and a lens to test in a controlled setting the change in illumination as the aperture becomes wider.

From memory 3nm shows no change until you drop below F/3.5 and by F/3 has a 14% drop in intensity. 5nm is good to F/2.5-2.8 before it shows any change.

 

I believe this is different to the pass band shift that also occurs but I'm not entirely sure.



#27 freestar8n

freestar8n

    Vendor - MetaGuide

  • *****
  • Vendors
  • Posts: 8828
  • Joined: 12 Oct 2007

Posted 24 August 2019 - 06:31 PM

Hi Frank,

 

Well, I don't know where the exit pupil is for the 105mm, but I did some tests and my suspicions were confirmed.

 

A front filter aperture stop increases the vignetting. See the animated GIF below.

 

This was with a 52mm stop that made the scope f/2 roughly.

 

Jerry

 

Hi Jerry-

 

Yes - I expect some vignetting - but that should flatten out fine.  For a sub aperture in front of the lens my main concern would be aberrations - but I doubt they are bad and the field may even be improved over wide open.

 

Although you are losing light at the edge, you are losing equal amounts of emission signal and sky background as you move out to the edge.  But if you put a narrow filter on the sensor side you would be losing a good amount of emission signal everywhere in the sensor - and the sky background would be at full power everywhere on the sensor.  So I think that after flattening the SNR would be much higher with the filter in front.

 

But yes - it would be an expensive way to do it.  I am not aware of anyone who has tried it.  Maybe I will sometime - but with a 1.25" filter on the 135 f/2.

 

Some people really don't like vignetting and they don't like using flats.  If that is the case here, then filter in front is not an option.

 

Frank



#28 Jerry Lodriguss

Jerry Lodriguss

    Vendor

  • *****
  • Vendors
  • topic starter
  • Posts: 7254
  • Joined: 19 Jul 2008
  • Loc: Voorhees, NJ

Posted 24 August 2019 - 08:56 PM

Yes - I expect some vignetting - but that should flatten out fine.  For a sub aperture in front of the lens my main concern would be aberrations - but I doubt they are bad and the field may even be improved over wide open.

 

Although you are losing light at the edge, you are losing equal amounts of emission signal and sky background as you move out to the edge.  But if you put a narrow filter on the sensor side you would be losing a good amount of emission signal everywhere in the sensor - and the sky background would be at full power everywhere on the sensor.  So I think that after flattening the SNR would be much higher with the filter in front.

 

But yes - it would be an expensive way to do it.  I am not aware of anyone who has tried it.  Maybe I will sometime - but with a 1.25" filter on the 135 f/2.

 

Some people really don't like vignetting and they don't like using flats.  If that is the case here, then filter in front is not an option.

Hi Frank,

 

The problem isn't shooting flats, the problem is that flats don't recover the lost s/n in the corners.

 

Jerry



#29 Jerry Lodriguss

Jerry Lodriguss

    Vendor

  • *****
  • Vendors
  • topic starter
  • Posts: 7254
  • Joined: 19 Jul 2008
  • Loc: Voorhees, NJ

Posted 24 August 2019 - 09:11 PM

Frank,  maybe you can answer this question for me.

 

The argument for using the filter in front of the lens is that the light from the sky is parallel so you don't get a bandpass shift with inteference filters, but after the lens, it is a cone and you do get a shift. 

 

But if you try to use in interference filter in front of a wide-angle lens, it is unusable because of the radial bandpass shift off-axis. But it's in front of the lens. Why doesn't it work if the light from the sky is parallel? And, for whatever reason this is, why doesn't the same thing apply to a narrowband Ha filter in front of the lens?

 

However, if you try to use an interference filter in between the same wide-angle f/2 lens and the sensor, it works fine. I have done both.

 

What is different conceptually here between the wide lens and the 105mm?

 

If the argument is the light from the sky is parallel, the interference filter should work in front of a wide-angle lens too.

 

The interference filters were broad band LPR filters.  Is it the narrowness of the bandpass with a narrowband filter that is the reason?

 

When I see how far the rear optical element is from the sensor with the 105mm, I have to think it's mostly parallel light coming out of the back of the lens at that point. 

 

What would be interesting would be to use a 3nm Ha filter in front of the lens and then compare it when used in between the lens and sensor with the same step-down ring so the focal ratio is the same and the only variable is filter location.

 

Jerry



#30 freestar8n

freestar8n

    Vendor - MetaGuide

  • *****
  • Vendors
  • Posts: 8828
  • Joined: 12 Oct 2007

Posted 24 August 2019 - 09:12 PM

Hi Frank,

 

The problem isn't shooting flats, the problem is that flats don't recover the lost s/n in the corners.

 

Jerry

It's true the transmission and SNR is slightly reduced in the corners - but I expect even there it will be much higher than even in the center if you had the filter on the sensor side.

 

If the filter is on the sensor side let's say it reduces the light cone for emission to f/2.8 - while the sky background is still coming in at full f/1.4.  That means your signal is 1/4 and the noise is full - so SNR is 1/4 what it could have been - and that's everywhere in the image.

 

But if you put it in front and reduce the f/ratio to 2.1, in the center both the sky and the emission will be at f/2.1.  In the corners it may be equivalent to f/3 if the vignetting reduces the transmission by about 50% - and that is probably still higher snr than the other way, in the center.

 

And I doubt you would notice any increased noise after flattening and stacking.

 

The main alternative is to use wide filters on the sensor side.  That might allow full f/1.4 but I'm not sure.  But it requires wider passband and lower snr.

 

Note that narrowband filters with fast systems is a fundamental problem in professional work.  For fast systems you can't use narrow filters - and in that case you just can't isolate the Ha line from Nii.  But in your case you actually could - if you put a 3nm filter in front.


  • kingjamez likes this

#31 freestar8n

freestar8n

    Vendor - MetaGuide

  • *****
  • Vendors
  • Posts: 8828
  • Joined: 12 Oct 2007

Posted 24 August 2019 - 09:23 PM

Frank,  maybe you can answer this question for me.

 

The argument for using the filter in front of the lens is that the light from the sky is parallel so you don't get a bandpass shift with inteference filters, but after the lens, it is a cone and you do get a shift. 

 

But if you try to use in interference filter in front of a wide-angle lens, it is unusable because of the radial bandpass shift off-axis. But it's in front of the lens. Why doesn't it work if the light from the sky is parallel? And, for whatever reason this is, why doesn't the same thing apply to a narrowband Ha filter in front of the lens?

 

However, if you try to use an interference filter in between the same wide-angle f/2 lens and the sensor, it works fine. I have done both.

 

What is different conceptually here between the wide lens and the 105mm?

 

If the argument is the light from the sky is parallel, the interference filter should work in front of a wide-angle lens too.

 

The interference filters were broad band LPR filters.  Is it the narrowness of the bandpass with a narrowband filter that is the reason?

 

When I see how far the rear optical element is from the sensor with the 105mm, I have to think it's mostly parallel light coming out of the back of the lens at that point. 

 

What would be interesting would be to use a 3nm Ha filter in front of the lens and then compare it when used in between the lens and sensor with the same step-down ring so the focal ratio is the same and the only variable is filter location.

 

Jerry

Which exact wide angle lens are you referring to?  This is where the exit pupil distance is important - because even a fisheye lens may have little change in chief ray angle across the sensor if the exit pupil is pushed far away.

 

The thing that kills narrow band filters on the sensor side is the f/ratio and the steep angles of the cone.  There is a separate effect of the changing chief ray angle out from the center - but that depends on the lens design and the pupil location.  A wide angle lens may have less impact from this than a medium lens - which is very contrary to expectations.

 

For the filter in front everything is more straightforward.  If you have a wide angle lens then you know immediately that as you go off axis you will hit a point where emission no longer makes it through the filter.  And that will show as vignetting of the emission - but no vignetting of the sky background.  If you have a nebula in the center of the view, you may not even notice that you are losing emission as you go out from the center - because the sky background is constant.

 

But for a 105mm lens and medium sized sensor the field angle at the edge isn't very big compared to the steepness of the cone at f/1.4.  So it would be ok to put the filter in front.  But with a wide angle lens it wouldn't work at all.

 

Frank


  • kingjamez likes this

#32 Jerry Lodriguss

Jerry Lodriguss

    Vendor

  • *****
  • Vendors
  • topic starter
  • Posts: 7254
  • Joined: 19 Jul 2008
  • Loc: Voorhees, NJ

Posted 24 August 2019 - 10:09 PM

It's true the transmission and SNR is slightly reduced in the corners - but I expect even there it will be much higher than even in the center if you had the filter on the sensor side.

 

If the filter is on the sensor side let's say it reduces the light cone for emission to f/2.8 - while the sky background is still coming in at full f/1.4.  That means your signal is 1/4 and the noise is full - so SNR is 1/4 what it could have been - and that's everywhere in the image.

 

But if you put it in front and reduce the f/ratio to 2.1, in the center both the sky and the emission will be at f/2.1.  In the corners it may be equivalent to f/3 if the vignetting reduces the transmission by about 50% - and that is probably still higher snr than the other way, in the center.

 

And I doubt you would notice any increased noise after flattening and stacking.

 

The main alternative is to use wide filters on the sensor side.  That might allow full f/1.4 but I'm not sure.  But it requires wider passband and lower snr.

 

Note that narrowband filters with fast systems is a fundamental problem in professional work.  For fast systems you can't use narrow filters - and in that case you just can't isolate the Ha line from Nii.  But in your case you actually could - if you put a 3nm filter in front.

Hi Frank,

 

This is a really interesting way to look at it, I hadn't considered this. I can really appreciate what this means and thanks to you and everyone else here for helping me to understand.

 

I measured a raw 16-bit linear file at:

F/#   Corner     Center

1.4    837          1596  with no aperture stop (48%)

2.0    667          1546  with aperture stop in front of lens (57%) 

 

Lenstips says the vignetting in full-frame on the 105mm f/1.4 Sigma Art is

47% (−1.85 EV) at f/1.4 with no aperture stop

25% (−0.82 EV) at f/2 with the in-camera diaphragm

11% (−0.32 EV) at f/2.8 with the in-camera diaphragm

8% (−0.25 EV) at f/4 with the in-camera diaphragm

 

So, stopping down with the in-camera aperture to get to f/2 improves the corners to only 25% light loss, but stopping down in-front of the lens makes it much worse at 57%.

 

Jerry


  • Jon Rista likes this

#33 Jerry Lodriguss

Jerry Lodriguss

    Vendor

  • *****
  • Vendors
  • topic starter
  • Posts: 7254
  • Joined: 19 Jul 2008
  • Loc: Voorhees, NJ

Posted 24 August 2019 - 10:10 PM

Which exact wide angle lens are you referring to?  This is where the exit pupil distance is important - because even a fisheye lens may have little change in chief ray angle across the sensor if the exit pupil is pushed far away.

 

The thing that kills narrow band filters on the sensor side is the f/ratio and the steep angles of the cone.  There is a separate effect of the changing chief ray angle out from the center - but that depends on the lens design and the pupil location.  A wide angle lens may have less impact from this than a medium lens - which is very contrary to expectations.

 

For the filter in front everything is more straightforward.  If you have a wide angle lens then you know immediately that as you go off axis you will hit a point where emission no longer makes it through the filter.  And that will show as vignetting of the emission - but no vignetting of the sky background.  If you have a nebula in the center of the view, you may not even notice that you are losing emission as you go out from the center - because the sky background is constant.

 

But for a 105mm lens and medium sized sensor the field angle at the edge isn't very big compared to the steepness of the cone at f/1.4.  So it would be ok to put the filter in front.  But with a wide angle lens it wouldn't work at all.

Hi Frank,

 

Thanks for the explanation, I always wondered about that. (Not that I really know what a chief ray angle is, but that is simple ignorance on my part).

 

The lens was a 16mm f/2 Rokinon used with an Astronomik CLS-CCD clip-in filter on an APS-C Canon.

 

Jerry



#34 freestar8n

freestar8n

    Vendor - MetaGuide

  • *****
  • Vendors
  • Posts: 8828
  • Joined: 12 Oct 2007

Posted 24 August 2019 - 10:29 PM

Hi Jerry-

 

Yes - stopping down in front is not optimal at all for vignetting or aberrations - but it's what you have to do if you want to have a fast cone arriving at the sensor.  If you had a lens with 50mm front aperture and an exit pupil far from the sensor then there would be no downside at all to putting the filter in front - regardless of how fast it is.

 

But for any lens, if it is fast you will lose emission signal with a narrow filter on the sensor side.

 

Unfortunately you can't stick the filter inside where the true aperture stop (iris) is - because then the rays would be coming in at an angle and you would lose emission signal.

 

I guess one way to look at this is to look for a smaller f/1.4 lens with 50mm front aperture - and combine it with a sensor that gives the field of view you want.  Then you could create the exact picture you want with a 3nm filter - and no losses or disadvantages - as long as the field of view isn't too wide so you lose emission toward the edges.

 

The complication in your situation is that the front lens is larger than 50mm, so you would be stopping it down a bit.  But even then I think it is the best way to go in terms of the final result and SNR.

 

Frank 



#35 Jerry Lodriguss

Jerry Lodriguss

    Vendor

  • *****
  • Vendors
  • topic starter
  • Posts: 7254
  • Joined: 19 Jul 2008
  • Loc: Voorhees, NJ

Posted 24 August 2019 - 10:47 PM

Hi Jerry-

 

Yes - stopping down in front is not optimal at all for vignetting or aberrations - but it's what you have to do if you want to have a fast cone arriving at the sensor.  If you had a lens with 50mm front aperture and an exit pupil far from the sensor then there would be no downside at all to putting the filter in front - regardless of how fast it is.

 

But for any lens, if it is fast you will lose emission signal with a narrow filter on the sensor side.

 

Unfortunately you can't stick the filter inside where the true aperture stop (iris) is - because then the rays would be coming in at an angle and you would lose emission signal.

 

I guess one way to look at this is to look for a smaller f/1.4 lens with 50mm front aperture - and combine it with a sensor that gives the field of view you want.  Then you could create the exact picture you want with a 3nm filter - and no losses or disadvantages - as long as the field of view isn't too wide so you lose emission toward the edges.

 

The complication in your situation is that the front lens is larger than 50mm, so you would be stopping it down a bit.  But even then I think it is the best way to go in terms of the final result and SNR.

 

Hi Frank, this is just a classic case of me having confirmation bias combined with the Dunning-Kruger effect.

 

I only listen to the facts that support my theory that I'm going to use this thing wide-open no matter what. :-)

 

If someone one day makes a Z to F mount filter holder (it seems so simple if you don't need electronics in it)  I will probably end up getting a 50mm narrowband filter, and trying it both ways.

 

Theory and facts are nice, but reality can surprise you sometimes with something completely different.  grin.gif

 

I guess I'm going to have to start researching narrowband Ha filters and how much the bandpass shifts by focal ratio just to get a ballpark feel for what to expect. 

 

Luckily I have all the time in the world. I just got the camera and don't plan to modify it just yet.

 

Jerry



#36 freestar8n

freestar8n

    Vendor - MetaGuide

  • *****
  • Vendors
  • Posts: 8828
  • Joined: 12 Oct 2007

Posted 24 August 2019 - 11:33 PM

Sounds good.

 

Well you can test things out pretty well by choosing a bright nebula and selecting one patch on it.  Then image that patch in the center and on the side of the frame in the two different ways: with filter on sensor side and in front of lens.

 

Also measure sky background in center and on side.

 

Convert everything to electrons.

 

SNR then goes as EmissionSignal/sqrt(SkyBackground)

 

That should answer everything concretely.  Make sure to measure everything with the same patch of nebulosity and move it around.

 

I think there is a good chance SNR will be higher for both center and edge if the filter is in front - and it will be a meaningful test and not just theory.

 

I also hate the idea of truncating the cone of a nice fast lens.  But unfortunately the filter will be doing that to some extent anyway if it is on the sensor side.

Frank


  • kingjamez likes this

#37 Jerry Lodriguss

Jerry Lodriguss

    Vendor

  • *****
  • Vendors
  • topic starter
  • Posts: 7254
  • Joined: 19 Jul 2008
  • Loc: Voorhees, NJ

Posted 25 August 2019 - 01:38 AM

Hi Frank,

 

That's what I will do.

 

It might also be interesting to see how stopping down the lens with the built-in diaphragm to f/2 affects the test with the filter in between the lens and camera and no aperture stop in front of the lens. (I know, it violates my prime directive of using the thing wide open).

 

Thanks!

 

Jerry



#38 TxStars

TxStars

    Surveyor 1

  • *****
  • Posts: 1870
  • Joined: 01 Oct 2005
  • Loc: Lost In Space

Posted 25 August 2019 - 01:40 AM

If you had enough cash you could get a custom full aperture filter.

https://www.chroma.c...ation/astronomy



#39 Jerry Lodriguss

Jerry Lodriguss

    Vendor

  • *****
  • Vendors
  • topic starter
  • Posts: 7254
  • Joined: 19 Jul 2008
  • Loc: Voorhees, NJ

Posted 25 August 2019 - 01:52 AM

If you had enough cash you could get a custom full aperture filter.

https://www.chroma.c...ation/astronomy

I definitely have to request a custom quote.

 

That should be interesting. 

 

Jerry



#40 freestar8n

freestar8n

    Vendor - MetaGuide

  • *****
  • Vendors
  • Posts: 8828
  • Joined: 12 Oct 2007

Posted 25 August 2019 - 03:36 AM

Hi Frank,

 

That's what I will do.

 

It might also be interesting to see how stopping down the lens with the built-in diaphragm to f/2 affects the test with the filter in between the lens and camera and no aperture stop in front of the lens. (I know, it violates my prime directive of using the thing wide open).

 

Thanks!

 

Jerry

Great!  That would be really good if you can take the measurements.

 

One other basic thing is to take data with no filter at all.  At a really dark site you don't need the filter in the first place - except to isolate the particular emission line.  It depends on what you're trying to image and how much you want to isolate a specific emission line - but at a dark site you might do well with just R G B imaging and no narrowband - in which case you can use the full aperture.

 

But if there is any sky background - that's when the narrowband filter could have an SNR win.

 

Frank




CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.


Recent Topics






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics