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

Spider and Secondary Diffraction: what to do, what to avoid

  • Please log in to reply
370 replies to this topic

#1 Oberon

Oberon

    Mercury-Atlas

  • -----
  • topic starter
  • Posts: 2831
  • Joined: 24 Feb 2013
  • Loc: Hunter Valley NSW Australia

Posted 27 March 2015 - 10:22 PM

This might be helpful. From Maskulator...

 

gallery_217007_4746_176849.jpg

 


  • Nils Olof Carlin, ed_turco and Duesing6 like this

#2 Jeff Morgan

Jeff Morgan

    Voyager 1

  • *****
  • Posts: 10641
  • Joined: 28 Sep 2003
  • Loc: Prescott, AZ

Posted 27 March 2015 - 11:14 PM

Interesting simulation. I'm curious as to the differences in spiders in column three. The central obstructions are the only obvious differences (to me).

 

The four vanes actually look pretty good. The offset vane pattern was mentioned by Sidgewick nearly 60 years ago, what interested me most about it was his claim that tube-deforming vane tensioning was not required for mechanical stability. Given the current prevalence of Sonotube, that should be a prized quality - but rarely seen today.


  • gene 4181 likes this

#3 Oberon

Oberon

    Mercury-Atlas

  • -----
  • topic starter
  • Posts: 2831
  • Joined: 24 Feb 2013
  • Loc: Hunter Valley NSW Australia

Posted 27 March 2015 - 11:56 PM

The top one has thin vanes...I wanted to compare it directly to the 3rd down to see what the effect of thicker vanes are. Looking at a few of these shows that the thickness of the vanes makes a greater impact than the size of the secondary.



#4 Oberon

Oberon

    Mercury-Atlas

  • -----
  • topic starter
  • Posts: 2831
  • Joined: 24 Feb 2013
  • Loc: Hunter Valley NSW Australia

Posted 27 March 2015 - 11:59 PM

The bottom right has 4 x 180 degree vanes; it would be better in every way with only 45 degree vanes. 



#5 Oberon

Oberon

    Mercury-Atlas

  • -----
  • topic starter
  • Posts: 2831
  • Joined: 24 Feb 2013
  • Loc: Hunter Valley NSW Australia

Posted 28 March 2015 - 12:02 AM

The offset vanes are definitely the way to go for straight vanes. Most rigid.



#6 Oberon

Oberon

    Mercury-Atlas

  • -----
  • topic starter
  • Posts: 2831
  • Joined: 24 Feb 2013
  • Loc: Hunter Valley NSW Australia

Posted 28 March 2015 - 01:36 AM

Another simulation. This time I have a 125mm central obstruction (previous was 100mm) to reflect what I plan to use on my binoscope. The most interesting thing to come out of this one is how good the single curved stalk is despite being very thick. The other thing is how bad the 6 vane 'triangular' results were, no doubt due to the length of vane exposed.

gallery_217007_4746_101263.jpg


  • Nils Olof Carlin likes this

#7 Nils Olof Carlin

Nils Olof Carlin

    Vanguard

  • *****
  • In Memoriam
  • Posts: 2227
  • Joined: 26 Jul 2004

Posted 28 March 2015 - 03:51 AM

Harold Suiter and Bill Zmek wrote a "Dialogue on Spider Diffraction"in issue 11 of the ATMJ, reprinted by Willmann-Bell in "The best of Amateur Telescope Making Journal vol 1". Their discussion is worth reading if you can find the book. They do simulations of much the same patterns as you do here (but also of the "rolling pin" pattern and others by Couder, as mentioned in a parallel thread). One thing that struck me was that the offset 4vane spider (3rd in top row of #6) shows somewhat widened spikes, but thinking again, I believe this may be the effect of a 2-slit diffraction pattern across the spike - I doubt that this matters in practice. Even the slightest loss of parallellism of ordinary crossing vanes would widen the spikes some.
The width of the spike is ideally the width of the Airy disk, despite the apparent size of the (bloated, overexposed) star image. Useful for astrometry, but if that is what you want, you could add accurately straight dummy vanes.

Nils Olof

#8 Nils Olof Carlin

Nils Olof Carlin

    Vanguard

  • *****
  • In Memoriam
  • Posts: 2227
  • Joined: 26 Jul 2004

Posted 28 March 2015 - 06:50 AM

The top one has thin vanes...I wanted to compare it directly to the 3rd down to see what the effect of thicker vanes are. Looking at a few of these shows that the thickness of the vanes makes a greater impact than the size of the secondary.

I don't think it is unexpected. The influence of the secondary obstruction is mostly confined to the size of the Airy disk (Babinet's principle) of the obstruction itself. Your simulations are grossly "overexposed" in this area. In order to show the effects of the obstruction, showing the central part magnified, with diffraction rings separated, would no doubt have shown the effects of secondary obstructions better. The obstruction will add in phase to the innermost ring increasing it, in opposite phase for the second ring decreasing it, etc, but further out I believe there is little difference.
I've used Aberrator with enhanced contrast to show the idea: from left to right 0, 25 and 40% obstruction (the contrast is the best I could make).

Nils OlofDiff2a.jpg
  • Jon Isaacs, Oberon and Earthbound1 like this

#9 hottr6

hottr6

    Skylab

  • *****
  • Posts: 4171
  • Joined: 28 Jun 2009
  • Loc: 7,257', Magdalena Mtns, NM

Posted 28 March 2015 - 11:00 AM

Could someone clarify what Maskulator is doing?  I don't see any diffraction rings, and field contrast seems to be constant regardless of size of central obstruction.



#10 raal

raal

    Viking 1

  • *****
  • Posts: 761
  • Joined: 10 Nov 2009
  • Loc: Serbia

Posted 28 March 2015 - 11:17 AM

No matter how high-res mask I draw, Maskulator will work only up to 2048x2048 pixels, it seems.

That leaves artifacts oriented at X and Y axis in the result that are greater than PSF distribution, so it won't show the Airy disk and rings around it.

The closest I got was with lowest "brightness" setting for one single wavelength.

 

Anyway, FWIW, it is telling me that tangential incidence of vanes to CO, or aperture stop, is to be avoided, as in this example:

 

 

 

 

Attached Thumbnails

  • normal incidence.jpg
  • normal incidence  of vane to CO.jpg
  • tangential incidence.jpg
  • tangential incidence  of vane to CO.jpg


#11 mark cowan

mark cowan

    Vendor (Veritas Optics)

  • *****
  • Vendors
  • Posts: 9385
  • Joined: 03 Jun 2005
  • Loc: salem, OR

Posted 28 March 2015 - 11:20 AM

The simulations for the simple circular obstructions are incorrect as well - those are spurious patterns.   :ohmy:


  • DesertRat likes this

#12 raal

raal

    Viking 1

  • *****
  • Posts: 761
  • Joined: 10 Nov 2009
  • Loc: Serbia

Posted 28 March 2015 - 11:22 AM

The smoothest polychromatic "halo" that I ever got is with 3 x 63deg. vanes, oriented to be as normal as possible to both the aperture stop as well as to CO, so I'll do it that way.

Alex

Attached Thumbnails

  • 3 x 07mm 63deg normalized.jpg
  • 3 vanes 63 n.jpg

  • Oberon and Duesing6 like this

#13 pstarr

pstarr

    Skylab

  • *****
  • Posts: 4318
  • Joined: 17 Sep 2004
  • Loc: NE Ohio

Posted 28 March 2015 - 11:27 AM

The smoothest polychromatic "halo" that I ever got is with 3 x 63deg. vanes, oriented to be as normal as possible to both the aperture stop as well as to CO, so I'll do it that way.

Alex

See post #80 I made in reflectors. http://www.cloudynig...e-spikes/page-4



#14 raal

raal

    Viking 1

  • *****
  • Posts: 761
  • Joined: 10 Nov 2009
  • Loc: Serbia

Posted 28 March 2015 - 11:53 AM

Paul, that's it!



#15 Nils Olof Carlin

Nils Olof Carlin

    Vanguard

  • *****
  • In Memoriam
  • Posts: 2227
  • Joined: 26 Jul 2004

Posted 28 March 2015 - 12:57 PM

No matter how high-res mask I draw, Maskulator will work only up to 2048x2048 pixels, it seems.

That leaves artifacts oriented at X and Y axis in the result that are greater than PSF distribution, so it won't show the Airy disk and rings around it.

The closest I got was with lowest "brightness" setting for one single wavelength.

 

Anyway, FWIW, it is telling me that tangential incidence of vanes to CO, or aperture stop, is to be avoided, as in this example:

Those coloured artifacts could not likely be from the insertions on the CO, but more likely from the (short) purely vertical sections where the vanes leave the aperture.Try increasing the diameter by a fraction. I don't believe you could reproduce them in real life.

 

Nils Olof



#16 raal

raal

    Viking 1

  • *****
  • Posts: 761
  • Joined: 10 Nov 2009
  • Loc: Serbia

Posted 28 March 2015 - 01:40 PM

Right now I'm not at the computer where I have Maskulator and Corel installed to try that, but you're right purely from the point of limited resolution in Maskulator, that would pixelize the arc segments tangential to XY grid into bars. 

 

Other than that, everything there is a part of a circle, there are no straight lines and that's true for any orientation. If the mask would be rotated, nothing would appear to be vertical any more.

 

My point about avoiding tangential incidence was about increased darkening between the diffraction "fans'',while the "fan'' adges are brightening at the same time, thus making greater and more obvious separation between them, even though in both cases, the total sum of arc is exactly 180 degrees.

 

I tent to think that this is because of small "wedge" shaped areas of aperture where vanes meet the CO, as this happens with any vane shape and distribution. Most probably, those wedges are not the best shape that one should let the light pass through.



#17 Oberon

Oberon

    Mercury-Atlas

  • -----
  • topic starter
  • Posts: 2831
  • Joined: 24 Feb 2013
  • Loc: Hunter Valley NSW Australia

Posted 29 March 2015 - 07:00 AM

OK so...in for a penny...I kept going. Here are are variety of 4 curved vanes.

 

gallery_217007_4746_73829.jpg



#18 Oberon

Oberon

    Mercury-Atlas

  • -----
  • topic starter
  • Posts: 2831
  • Joined: 24 Feb 2013
  • Loc: Hunter Valley NSW Australia

Posted 29 March 2015 - 07:03 AM

And here is a really odd variety of curvaceous possibilities, some of them surprisingly good...

gallery_217007_4746_323629.jpg



#19 raal

raal

    Viking 1

  • *****
  • Posts: 761
  • Joined: 10 Nov 2009
  • Loc: Serbia

Posted 29 March 2015 - 08:44 AM

That was a lot of work!

 

I'm intrigued by the possibility of masking a single stalk like you showed in 3rd and 4th in the top row.

Also, the 1st one in the second row is equally intriguing.

 

My layman point of view is:

 

I know that the thinner the vanes are, the further they will throw the spike and vice versa.

 

So, the question is, If we make the stalk as thick as CO, then wouldn't the diffraction be confined to just a few first rings around Airy disk and get something that shows no halo or spikes at all?

 

Can that case be regarded only as reduced aperture in one direction? Or a bigger CO in one direction and smaller in the other?

 

If no spikes and halo,couldn't that be equal to having an optical window,only with somewhat reduced resolution in one direction?

 

Given the aperture size advantage of a bigger reflector to a smaller refractor,shouln't that truly emulate the crisp contrasty view in a refractor?

 

I'd sure love to see some PSF distrubition for that case and how it translates to MTF.



#20 DesertRat

DesertRat

    Fly Me to the Moon

  • *****
  • Posts: 6261
  • Joined: 18 Jun 2006
  • Loc: Valley of the Sun

Posted 29 March 2015 - 11:42 AM

There are lots of artifacts seen in these simulations, especially evident when the image is overexposed or stretched.

 

To control them it is recommended to do the following:

 

1) Use as large a working space as possible.

 

2) Ensure that the pupil or object covers no more than ~20% of the working space. There is a tradeoff between precision and reality, because the FFT assumes the pupil in the above simulations is repeated forever in all directions.

 

3) Sometimes its useful to 'apodize' the pupil with a small blur, so that the staircase or pixelation of features which are not purely horizontal or vertical is controlled. This helps avoid the high frequency artifacts seen from fourier techniques applied to binary images.

 

Glenn

 


  • Dave O likes this

#21 Nils Olof Carlin

Nils Olof Carlin

    Vanguard

  • *****
  • In Memoriam
  • Posts: 2227
  • Joined: 26 Jul 2004

Posted 29 March 2015 - 03:06 PM

My layman point of view is:
 
I know that the thinner the vanes are, the further they will throw the spike and vice versa.

“It ain't what you don't know that gets you into trouble. It's what you know for sure that just ain't so.” Mark Twain might never have said so.
But as Harold Suiter has shown:
http://home.digitale.../TM/Spiders.htm
The spikes are, on average (they have peaks and valleys) equally long and taper out equally fast away from the star, the difference is near the star (or even within a planet image), where the spikes are clearly brighter for thicker vanes.
This said, with thin vanes, spikes from a not so bright star may not be obvious with thin vanes but visible close to the star/planet with thick vanes.

Nils Olof

#22 ChristianG

ChristianG

    Surveyor 1

  • -----
  • Posts: 1540
  • Joined: 18 Oct 2012
  • Loc: Ottawa, Canada

Posted 29 March 2015 - 04:12 PM

Guys,

 

Glenn is right. None of the patterns that appear above are correct. If you use Fourier transform, you need to understand a few things. If your mask or aperture takes up a significant portion of the whole 2048 x 2048 image, the transform will be characteristic of the mask and of the square image, and the whole thing will be swamped by four-fold symmetric artifacts. The reason is that Fourier transforms of objects of similar sizes are similar sizes.

 

One way to avoid this problem, beside apodization, is to have the desired Fourier transform much bigger than the Fourier transform of the square image frame. To do that, the aperture of interest must be much smaller than the whole image. So the aperture one wishes to use needs to cover only a small part of the central portion of the 2048 x 2048 image. Then its transform will be larger than the artifacts.

 

When I post diffraction patterns, they are only the central portions. Below is a transform pair for a centrally obstructed circular aperture, showing many diffraction rings:

 

c20_diff.jpg

 

But the image I started with is this 4096 x 4096 one:

 

c20.jpg

 

And its Fourier transform was in fact:

 

c20f.jpg

 

Only this way can I be sure that whatever artifacts that appear when I add a spider indeed are caused by the spider. Here's an example below:

 

c20so2.jpg

 

Have fun!

 

--Christian


Edited by ChristianG, 29 March 2015 - 04:16 PM.


#23 mark cowan

mark cowan

    Vendor (Veritas Optics)

  • *****
  • Vendors
  • Posts: 9385
  • Joined: 03 Jun 2005
  • Loc: salem, OR

Posted 29 March 2015 - 04:59 PM

Right, and that diffraction pattern illustrates what actually happens with offset but parallel vanes. :waytogo:



#24 dag55

dag55

    Viking 1

  • *****
  • Posts: 656
  • Joined: 13 Apr 2011
  • Loc: Hamburg, IL. U.S.

Posted 29 March 2015 - 05:20 PM

Now this seems to confirm what I saw at the EP when I conducted experiments with the thin wires stretched across the UTA of my Dob. and then inserted a 3/16" dia. rod to only half the aperture. I did this at full aperture 17.5" and off axis mask at 6.75" both experiments showed the single post as less diffraction spikes, with the single post as a slightly wider spike across the field of view as compared to the four thin but brighter spikes from the wires. What surprised me was the spikes from the wires were dimmer than from the four vane spider of my Dob. but not to the extent that I expected. This has caused me to stop with the wire spider and pursue the single post design albeit with a 5/32" carbide rod instead of the 3/16" I was going to use.
Dane

#25 Oberon

Oberon

    Mercury-Atlas

  • -----
  • topic starter
  • Posts: 2831
  • Joined: 24 Feb 2013
  • Loc: Hunter Valley NSW Australia

Posted 29 March 2015 - 06:37 PM

Right...thanks for the feedback, I'll rerun some of them when I get another opportunity.

Meanwhile...from what I've done so far it seems to me:-

 

1. that the "vanes must be as thin as possible" rule only applies to parallel lines, whether straight or curved.

 

2. that eliminating parallel lines is highly beneficial.

I imagine this is why the second down on left is so good despite the large obstruction. If these conclusions are true (and I will follow up with masks on my telescope when the weather fines up) then it seems we can build extremely stiff curved support structures that produce minimum diffraction, albeit at the cost of a larger obstruction. And that is exactly what I am trying to do for my binoscope.

Keep the feedback coming guys... :waytogo:

 


Edited by Oberon, 29 March 2015 - 06:53 PM.



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