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

Unmasked Foucault testing and related topics

  • Please log in to reply
61 replies to this topic

#1 mark cowan

mark cowan

    Vendor (Veritas Optics)

  • *****
  • Vendors
  • topic starter
  • Posts: 7679
  • Joined: 03 Jun 2005
  • Loc: salem, OR

Posted 25 May 2016 - 03:17 AM

To start off with here's a few more screenshots of both Jim Burrow's SixTests (archived here) and FigureXP for a unmasked mirror test I did some time ago on a 14.4" f/4.58 mirror that produced this comparison that I've shown recently in another thread:

 

compares manual w digital foucault.gif

 

The digital acquisition on that test was over an arbitrary range that yielded 8 zones when inverted (description to come), trimming the mirror diameter to 320mm to better reflect the actual measured diameter (not the actual mirror diameter, just closer to the measured area) yields the following setup data in SixTests and FigXP, respectively:

 

sixtests setup orig.png

 

figXP setup orig.png

 

This results in the following analyses:

 

sixtests surface orig.png

 

figXP surface orig.png

 

Anybody will of course notice the error figures shown here are very low, to say the least.  The first image above shows the correspondence between this data and careful manual testing.  

 

It is NOT PART OF THIS THREAD to exclaim publicly about how "impossible" that is - all of the factors involved will be discussed in some detail eventually.  The sole purpose of this thread to characterize this particular method of unmasked digital Foucault by providing a worked example (shown here) and then apply it more thoroughly over the next few days to an even more stringent test on an 18" f/3.57 mirror, and possibly to more as time allows...

 

We'll look more at that later, but the next step in this thread is to post the digital data that was taken of the 14.4" mirror for inspection, and then explain the technique, so anybody interested can follow along.

 

(continued)

 

 

 

 


  • gregj888 and jdupton like this

#2 RAC

RAC

    Viking 1

  • -----
  • Posts: 678
  • Joined: 30 Sep 2011
  • Loc: New Zealand

Posted 25 May 2016 - 04:01 AM

I like it.

 

I'm not sure how your system works but it would be neat to have some software that can do the flip and diff live from a video output so with a mask on all you need to do is move the KE and take down the numbers when the lines hits the halfway make in each zone. It would be so fast and super accurate.



#3 Pinbout

Pinbout

    Hubble

  • *****
  • Posts: 17069
  • Joined: 22 Feb 2010
  • Loc: Montclair, NJ

Posted 25 May 2016 - 05:29 AM

I would use cadtools for illustrator to measure the radius of the zone that's nulled.



#4 jdupton

jdupton

    Viking 1

  • -----
  • Posts: 636
  • Joined: 21 Nov 2010
  • Loc: Central Texas, USA

Posted 25 May 2016 - 10:27 AM

Mark,

 

   I too, have used this method ever since Michael Peck described it back in 1999 on the old ATM Listserv. I find it very easy to perform and it yields a permanent record of your Foucault readings. In my implementation, I use a pocket digital camera which is mounted on my tester rather than a video camera. (Although tiny video surveillance cameras are plentiful and very cheap these days.)

 

   The only difference for me is that in processing, I also add a step to Michael Peck's processing procedure. After the basic Flip-and-Diff, I do a Solarize operation. On slower (more normal for us rank amateurs) mirrors, the nulled zone from the Flip-and-Diff operation can still look rather broad. The Solarization operation, finds the exact zero difference area and you can get a one pixel wide band to measure the nulled zone even with slower mirrors. Fast mirrors like your's here will not need the Solarization step

 

   In operation, I usually decide how many zones (n) I want to measure up front and then divide the expected zonal length (r^2/2R) at the ROC by n+1 to get an offset for each reading. Since I use a digital still camera, I take three images at a zone, step by the calculated increment and rinse and repeat until I have covered everything from the center zone (nearly impossible to read) to beyond the edge of the mirror. Due to rather poorly controlled thermal air currents in my garage or house during testing, the stack and average the three images as part of my processing routine.

 

   For others here not familiar with the basics of Michael Peck's method, it is described at the following Web page.

 

http://home.earthlin...autof/autof.htm

 

   I look forward to your experiences and thoughts on this testing method, particularly any systematic problem areas your have worked through.

 

 

John



#5 Redstone2

Redstone2

    Explorer 1

  • -----
  • Posts: 77
  • Joined: 28 Feb 2009
  • Loc: AL, US

Posted 25 May 2016 - 02:03 PM

Doesn't James Lerch's Robofoucault automate Michael Peck's method?

His opening page gives credit to Michael's work.

http://lerch.no-ip.c...m/Robo_V_Inter/



#6 Pinbout

Pinbout

    Hubble

  • *****
  • Posts: 17069
  • Joined: 22 Feb 2010
  • Loc: Montclair, NJ

Posted 25 May 2016 - 02:39 PM

here's your CAT scan in motion

 

marks-14in.gif


Edited by Pinbout, 25 May 2016 - 02:39 PM.


#7 jdupton

jdupton

    Viking 1

  • -----
  • Posts: 636
  • Joined: 21 Nov 2010
  • Loc: Central Texas, USA

Posted 25 May 2016 - 03:39 PM

Tom,

 

   If I recall correctly, James' RoboFoucault testing was inspired by Michael's work but used a different image analysis method. (I think Figure 6 in Michael's write-up is what prompted James to come up with the RoboFoucualt methodology.)

 

   Subject to my sometimes iffy memory, James used a program to scan across the horizontal diameter of the image to record brightness readings  and then found the radii (left and right) on the image where they were equal. (See Michael's Figure 6.) It was sort of an "electronic eye" methodology where a program did the work of matching zones of rising and falling slope on each side of the mirror. I don't recall James' software doing the Flip-and-Diff portion of Michael's method although I could be wrong.

 

   I used to use PaintShopPro to perform the Flip-and-Diff methods but have since found that PixInsight (Image processing software) has all the tools to do this pretty well although I still have to measure the width of the one pixel wide null zone myself. The methodology does lend itself to scripting and I have scripts / process containers that do some of the work. Still, all in all, it doesn't take too long to go through this process and feed the resulting in SixTests or FigureXP (the two reduction programs I use most). Best of all, it takes my eyes out of the equation regarding determining when two zones are nulled.

 

 

John


  • Redstone2 likes this

#8 mark cowan

mark cowan

    Vendor (Veritas Optics)

  • *****
  • Vendors
  • topic starter
  • Posts: 7679
  • Joined: 03 Jun 2005
  • Loc: salem, OR

Posted 26 May 2016 - 07:55 PM

Ok, so moving along here.  Today I'll step through the procedure in as much detail as I can stand.  First the basics:

 

The mirror is unmasked obviously, but the setup otherwise is exactly the same as for Foucault testing with a Couder mask, only in place of your eyeball you use a camera.  The camera is NOT attached to the stage but wants to be secured in space behind the KE, focused on the mirror, and attached to a tripod or similar support so it doesn't move at all.

 

It doesn't matter if the tester is fixed or moving source, but for correct analysis in FigureXP a fixed source tester MUST have the KE where the center of the mirror shows a null aligned longitudinally with the light source or the analysis can show significant error for fast mirrors.  For a moving source (usually slitless ;) ) this condition will always be met.  If you use SixTests for analysis (you might want to because it doesn't appear to have any limit on the number of zones that can be entered, whereas FigXP tops out at 15 zones) you can have a misalignment longitudinally between the source position on KE, but you'll need to know what that is.  Easier just to line it all up in the first place.

 

For this test to work correctly you need a mirror with a good diffraction edge, smooth, with no defects of figure of revolution.  But you're wasting your time if those conditions aren't met so I won't mention them again.

 

Setup is to have the KE aligned as above on the center null of the mirror, the gauge ideally reading zero at the same position, and the axis of the tester aligned with the axis of the mirror.  I accomplish the latter by pushing the tester forward a bit and then running the stage back and forth while adjusting its axis until the position of the KE on the mirror center doesn't drift left or right - then dragging the tester straight back while keeping the same center position.  Then, at least the way I test, I run the micrometer back until the very edge of the mirror is nulled.

 

This takes the amount of travel that FigXP reports as the "ideal knife edge value" for the outer zone.  But because the zones tested this way can be very narrow - and you don't know what they are until after doing the test - just get it to where the edge is illuminated and note that value as the starting point.

 

Next you take your eye away and put the camera in its place.  It doesn't have to be close to the KE, so long as it captures the entire face of the mirror (or at least all the way across the center). 

 

Using the camera viewfinder (or computer screen or whatever you have) adjust the setup until you see good contrast and the diffraction ring on the mirror.  This is the first exposure, so capture it.

 

Now, iteratively, you want to step through the caustic depth of the mirror with the shadows changing (like in posts 2 and 3) from edge balanced to the central null.

 

You'll want to parcel out the expected travel of the KE longitudinally (the value of the outer zone) into the number of "zones" you'd like to capture, whether 10, or 100, for analysis.  Just divide the value of the outer zone by the number of images desired.  That will give you the incremental travel between each exposure, and due to the way that correction increases on a paraboloid as you move out, it won't be an even traverse across the FACE of the mirror, but it will be an even traverse across the SLOPE of the mirror, which matters much more for nailing down the surface correction.

 

So you just calculate out those steps, move the stage by the correct (always the same) amount to the next "zone", and capture another exposure.  Save them or rename them with some scheme that makes sense.

 

Having done that, analysis comes next.  To do this you need to know a couple things - the # pixels across the diameter of the mirror in each image (which will remain the same because the camera doesn't move at all) and the # pixels across the "zone" for each exposure.  You already know the actual physical optical diameter of the mirror (let's hope).  In some sort of paintshop program, open each of these images.  I'll demonstrate it with the second exposure from the series above, test02.jpg:

 

test02.jpg

 

When I measure the diffraction ring in pixels I get 428.  You can of course increase the contrast to make this easier to read.

 

Determining the "zonal" value is a little more complicated.  It's measured between the spots across the diameter where the gray values are identical.  For a monotonically increasing curve like a paraboloid should be there's exactly three of these spots - left, right, and center.

 

Fortunately an easy visual method exists to make this measurement.  

 

First open the image capture you're working with, and (important!) crop the image to the exact edge of the mirror on both sides:

 

flip and diff 1.jpg

 

Make a copy of this and mirror reverse it:

 

flip and diff 2.jpg

 

Then you take the difference between them - because the two (three) locations across the mirror have the same value, the difference there becomes zero, whereas it's positive everywhere else:

 

flip and diff 3.jpg

 

OK, well there may be zero values in there but they're hard to see. :lol: The final step clears that up.  Using "color corrections" in IrfanView crank the gamma up and then adjust the contrast (or use any image manipulations program to do the same thing):

 

flip and diff 4.jpg

 

Here's the result:

 

flip and diff 5.jpg  

 

Now you can measure the pixels across the mirror, since the only values that remain zero after that last step are the ones that were already zero.  Alternatively, of course, all of this can be determined in software from the image files, which is how you'd do it in a fully automated tester.

 

flip and diff 6.jpg

 

You can see that there are 388 pixels across that diameter, so the diameter of that "zone" is 320mm (the mirror diameter) divided by 428 (the pixel diameter of the diffraction ring) times 388, or 290(.1) mm, and the radius is 145(.0) mm.

 

(continued)


Edited by mark cowan, 26 May 2016 - 08:43 PM.

  • jdupton likes this

#9 mark cowan

mark cowan

    Vendor (Veritas Optics)

  • *****
  • Vendors
  • topic starter
  • Posts: 7679
  • Joined: 03 Jun 2005
  • Loc: salem, OR

Posted 26 May 2016 - 07:58 PM

You can see that there are 388 pixels across that diameter, so the diameter of that "zone" is 320mm (the mirror diameter) divided by 428 (the pixel diameter of the diffraction ring) times 388, or 290(.1) mm, and the radius is 145(.0) mm.

 

(Continued)

 

What this number is, exactly, in terms of what FigXP uses, is the "Effective Radius" in the setup page:

 

temp.png

 

These are the "zonal" radii, the second part of the input is the longitudinal micrometer readings (call them "Y").  

 

(to be continued with new data, because the numbers I get don't match the original analysis for obscure reasons.  But I also notice in looking at the other sets of images that the mirror hasn't got any center issues, it was just something in the camera setup.  But I can't edit those first two posts - also for obscure reasons...)

 

 

 

 

 

 


Edited by mark cowan, 26 May 2016 - 09:57 PM.


#10 mark cowan

mark cowan

    Vendor (Veritas Optics)

  • *****
  • Vendors
  • topic starter
  • Posts: 7679
  • Joined: 03 Jun 2005
  • Loc: salem, OR

Posted 26 May 2016 - 10:06 PM

I like it.
 
I'm not sure how your system works but it would be neat to have some software that can do the flip and diff live from a video output so with a mask on all you need to do is move the KE and take down the numbers when the lines hits the halfway make in each zone. It would be so fast and super accurate.

 
That would indeed be nice.  But I'll explain later on a bit why you don't really want to be moving the KE to match arbitrary zones, as the other way round is a lot easier to implement.
First I have to do live tests on another mirror to continue.
 

The only difference for me is that in processing, I also add a step to Michael Peck's processing procedure. After the basic Flip-and-Diff, I do a Solarize operation. On slower (more normal for us rank amateurs) mirrors, the nulled zone from the Flip-and-Diff operation can still look rather broad. The Solarization operation, finds the exact zero difference area and you can get a one pixel wide band to measure the nulled zone even with slower mirrors. Fast mirrors like your's here will not need the Solarization step

 
Yes, but even fast mirrors have slow zones near the center and there the pixels get spread out again.  Software that's looking at the actual data can just locate the zeros and numerically extract a center by averaging, which is the approach I'm taking in the robotic system I haven't discussed yet. ;)

 

 

 


Doesn't James Lerch's Robofoucault automate Michael Peck's method?
His opening page gives credit to Michael's work.
http://lerch.no-ip.c...m/Robo_V_Inter/

 

Mostly the reverse of this unmasked method, where you try to hunt down the zero crossing by moving the KE robotically.  A lot more trouble, since there's no need to have prespecified zones.  The only people I know who took this approach at all were Dale Eason and me, at least that's what he said when we discussed it.   :shrug:

 

It doesn't hunt for the zone centers.  It just assumes they are there (because they are!) and steps along the caustic, divided into as many intervals as you want, and then extracts the appropriate "zone" radii from the images it obtains.  Automation to do that gets the job done in mere seconds.

 

I will throw this out there now - you don't really want to take the flip-and-diff images and fit to the whole ring to get the radii of the zero crossing.  Work I've done with this so far shows that the image is very sensitive to test stand aberration, and that plus some other factors possibly best left unexplored means the only accurate data you get here is along the axis (same as Foucault with mask).  It reduces the error contributors to just look at that.


Edited by mark cowan, 26 May 2016 - 10:27 PM.


#11 mark cowan

mark cowan

    Vendor (Veritas Optics)

  • *****
  • Vendors
  • topic starter
  • Posts: 7679
  • Joined: 03 Jun 2005
  • Loc: salem, OR

Posted 26 May 2016 - 10:22 PM

here's your CAT scan in motion
 
attachicon.gifmarks-14in.gif



Had a feeling that was coming. :waytogo:

#12 Chriske

Chriske

    Kijkerbouw Urania

  • *****
  • Posts: 3113
  • Joined: 15 Aug 2004
  • Loc: Boechout, Belgium

Posted 27 May 2016 - 05:28 AM

 

here's your CAT scan in motion
 
attachicon.gifmarks-14in.gif



Had a feeling that was coming. :waytogo:

 

 

Here's another one: large file(sorry)



#13 mark cowan

mark cowan

    Vendor (Veritas Optics)

  • *****
  • Vendors
  • topic starter
  • Posts: 7679
  • Joined: 03 Jun 2005
  • Loc: salem, OR

Posted 27 May 2016 - 03:45 PM

 

(to be continued with new data, because the numbers I get don't match the original analysis for obscure reasons.  But I also notice in looking at the other sets of images that the mirror hasn't got any center issues, it was just something in the camera setup.  But I can't edit those first two posts - also for obscure reasons...)

 

 

Yes, well, I'm missing an entire set of data - namely the images that were on the automation computer from which I derived the final profile shown in the first post.   And that computer's been reformated a couple times because it went from W2K to XP and back to W2K on account of driver incompatibility with the automation system's dedicated firewire camera.  

 

So just on to new data today.  What I do recall distinctly about the original final set is that it showed a tiny bump on the mirror - tiny meaning like 40th wave or something - that I'd missed in my original manual testing.  But it was actually there.  This impressed the heck out of me at the time.

 

What I'm planning to do is take 30 exposures across each of 3 axes on the 18" f/3.57 mirror (or as many as I can stand to do manually).  I'll use half of those to fit into the 15 zone limit on FigXP, and the full 30 for SixTests, to compare in terms of resolution.  

 

Then I'll do manual testing on the mirror - which SFAIK is finished - and see how that goes (with an 11 zone mask).  The mirror has a distinct profile around 20th wave thanks to the way I finished it, at least that's what my testing while working it revealed.  

 

So if the two approaches yield up the same profiles I'm happy.  If the unmasked reveals more work to do on it, well I'm happy too.  That's what it's intended for - a very quick (when automated) approach to single axis profiling in order to control the figuring robot in near real-time.

 

But first (in a bit) I'll illustrate the potential accuracy of the flip-and-diff with synthetic Foucaultgram from Diffract.


Edited by mark cowan, 27 May 2016 - 03:49 PM.


#14 jdupton

jdupton

    Viking 1

  • -----
  • Posts: 636
  • Joined: 21 Nov 2010
  • Loc: Central Texas, USA

Posted 27 May 2016 - 04:10 PM

Mark,

 

   When you do your write-up with the fresh data, please include comments about methods you may use to combine or correlate the data from three runs on different axes.  That is something I have never found a satisfying way to do. The best I have been able to do is treat each as a totally separate run and simply visually compare the profiles given in SixTests or FigureXP.

 

   I really like this test and have used it on a meager number of mirrors since 2001. I am a big fan of the ease of use and repeatability of the testing. I also look forward to more of your commentary on the best-practice methods and results you have found.

 

   Your automated analysis would be a welcome relief to the image processing / manual analysis methods I use. It only takes me about 10 minutes to run 15+ zones on a mirror but processing the image data in PixInsight takes at least 45 minutes before I can load up the final data into SixTests.

 

 

Best regards,

John



#15 mark cowan

mark cowan

    Vendor (Veritas Optics)

  • *****
  • Vendors
  • topic starter
  • Posts: 7679
  • Joined: 03 Jun 2005
  • Loc: salem, OR

Posted 28 May 2016 - 01:33 AM

As promised, here's a synthetic look at the process. That'll be followed by the real thing now that I've got the camera running again on its computer.  These images replace those from posts 2 and 3 (now deleted) because they weren't what I actually used to produce the first comparison graph, so new data is required.  First though here's what it looks like in theory, all worked out.

 

What I'm going to do is model the original mirror in Diffract at each of the effective radii of a 7 zone Couder mask (according to FigXP's numbers) then analyze those and see if it maps back to what should be a perfect mirror.  The FigXP radii, along with DIFFRACT values added:

 

figXP radii for Diffract.png

 

Although I entered Diffract's zone inner and outer radii into FigXP it doesn't produce the same values for expected null position or effective radii (though the latter is very close).  I don't seen an obvious explanation.  For the first test I'm using the Diffract values as that should work back correctly.  Then we'll see about why they might be different.

 

Diffract settings for the first of 7 "exposures":

 

Diffract settings.png

 

And in turn, each of the simulated Foucaultgrams, with their flip-and-diff versions and the measured zero crossings.  The diameter of the mirror is 512 pixels.

 

zone7.png  zone7 b.png

 

zone6.png  zone6 b.png

 

(continued)


Edited by mark cowan, 28 May 2016 - 03:16 AM.


#16 mark cowan

mark cowan

    Vendor (Veritas Optics)

  • *****
  • Vendors
  • topic starter
  • Posts: 7679
  • Joined: 03 Jun 2005
  • Loc: salem, OR

Posted 28 May 2016 - 01:33 AM

zone5.png  zone5 b.png

 

zone4.png  zone4 b.png

 

zone3.png  zone3 b.png

 

(continued)


Edited by mark cowan, 28 May 2016 - 02:01 AM.


#17 mark cowan

mark cowan

    Vendor (Veritas Optics)

  • *****
  • Vendors
  • topic starter
  • Posts: 7679
  • Joined: 03 Jun 2005
  • Loc: salem, OR

Posted 28 May 2016 - 01:38 AM

zone2.png  zone2 b.png

 

zone1.png  zone1 b.png

 

Looks good so far, including the avocado at the end.  (The reason I've attached all these files is in case anybody wants to repeat or check the measurements for themselves, and to give a feeling for what they look like as I'm going to delete posts 2 and 3 since the data doesn't actually apply correctly.)

 

So, if these numbers, produced from DIFFRACT, are entered directly in FigXP, what is the result?

 

Here's how it's done.  Recall this set of numbers (the DIFFRACT ones off to the side):

 

figXP radii for Diffract.png

 

What I've done is used the right hand numbers (red circle) for the synthetic Foucaultgram KE position (as if the knife was advancing to these exact positions), and then entered, for the effective radii, the measured mm from the analysis with flip-and-diff (blue circle), as so:

 

figXP analysis.png

 

And how does that turn out?  Not bad:

 

figXP analysis 2.png

 

figXP analysis 3.png

And I noticed now that if I entered the "ideal" radii from DIFFRACT into FigXP, as expected it now produces the correct "ideal" KE positions (the two columns of longitudinal match well).  So they both are on the same page, but FigXP determines the average radii differently than DIFFRACT.

 

figXP analysis 4.png

 

 

(to be continued)


Edited by mark cowan, 28 May 2016 - 02:12 AM.

  • PrestonE likes this

#18 MKV

MKV

    Fly Me to the Moon

  • *****
  • Posts: 5858
  • Joined: 20 Jan 2011

Posted 28 May 2016 - 10:34 AM

Nice work, Mark. A couple of things -- just my observation, not critcism. I'm wondering how long does it take to measure 7 zones (I suppose several times for a good average), then data processing. It seems that this would be a very tedious process during the final stage of figuring -- when it counts the most -- as figuring spells get shorter, cooling periods longer and longer.

 

In comparison, an autocollimation null test takes but a few minutes. You take a look at the Ronchi bands, inside/outside the focus, pan them left and right, notice if there's any bending, or if there's any astigmatism (bands change angle), and you're back to another figuring session.Nothing to measure.

 

Likewise, using an interferometer,  such a the Bath, you take several images at 0, 90, 180, 275 degrees, feed them into a program, average the wavefronts, than average the average (2 clicks) and -- bingo, you get not only the quantitative RMS, conic, and Strehl results, but a 3D image of the surface (which can be rotated, tilted, etc), a profile graph through different sections of the wavefront (which shows you which sections of the whole wavefront are high or low), Zernike's coeffcients (so you can know exactly where the problem is, if there is any), and you can -- if you so choose -- look at the MTF, PSF and simulated star test (based on your mirror's results) with just a click of a button. FInally, you can get an artificial null (as in autcollimation), plus both Foucault and Ronchi simulations of your mirror on a neat all-inclusive printable report. These simulations, when compared to actual tests, agree very, very closely and there's plenty of documentation for that. It all takes but 10 minutes. And you're back to figuring. No measuring involved.

 

Please, don't get me wrong, your method is fascinating, and I love your presentations, but it seems terribly time-consuming and tedious to me, as  I fail to see where the advantage is -- especially if it's applied where in counts the most -- in the final phases of figuring a mirror, when figuring spells last a couple of minutes at most. Hopefully you can see my point and not take this comment the wrong way. Apparently, the method seems to work for you, and that's all that counts in the final analysis.

 

All the best,

Mladen

 



#19 brucesdad13

brucesdad13

    Apollo

  • *****
  • Posts: 1034
  • Joined: 25 Jan 2015
  • Loc: Rutland, MA

Posted 28 May 2016 - 11:20 AM

:confused:



#20 Chriske

Chriske

    Kijkerbouw Urania

  • *****
  • Posts: 3113
  • Joined: 15 Aug 2004
  • Loc: Boechout, Belgium

Posted 28 May 2016 - 12:58 PM

In my opinion the FT is the most logic test there is. Like I said before in a glans you see everything there is to see.

But more(most) important to me is the Foucaulttest is rather easy to learn by novice mirrorgrinders. You learn these guys how they should handle the darkening of the zones and the figures that go with it and your done. For all other tests you need far more experience.



#21 mark cowan

mark cowan

    Vendor (Veritas Optics)

  • *****
  • Vendors
  • topic starter
  • Posts: 7679
  • Joined: 03 Jun 2005
  • Loc: salem, OR

Posted 28 May 2016 - 06:10 PM

Progress - here's the first image of the 18" f/3.57, and a sample flip-and-diff from it.  The mirror needs another cleaning and it's not aligned quite right, but I think this is the correct exposure - the bright zones should be overexposed to favor resolution in the mid grays.  The 25mm Fujinon lens I have on there is too short so I'm going to substitute the simple 50mm achromat I've shown before, that'll give 4 times the area and almost fill the frame.  Like every other commercial c-mount lens I've used internal dust is a big factor in getting a clean Foucaultgram.  If you rotate the lens the spots rotate with it.

 

18 first 25mm.jpg

 

18 first f&d.jpg



#22 mark cowan

mark cowan

    Vendor (Veritas Optics)

  • *****
  • Vendors
  • topic starter
  • Posts: 7679
  • Joined: 03 Jun 2005
  • Loc: salem, OR

Posted 28 May 2016 - 08:05 PM

Another mirror I have, a 16" f/3.75, that's just undergone final testing is a lot cleaner, so I'm using that one instead.  Will see how well the unmasked matches the manual soon.

 

Here's a test Foucaultgram, the raw flip-and-diff, and one stretched twice with gamma. The focus with the achromat isn't as sharp for a couple reasons (some of which are fixable - alignment) but at least the image scale is a lot better.

 

16 inch 50mm 1.jpg

 

16 f&d raw.jpg

 

(continued)


Edited by mark cowan, 28 May 2016 - 08:09 PM.


#23 mark cowan

mark cowan

    Vendor (Veritas Optics)

  • *****
  • Vendors
  • topic starter
  • Posts: 7679
  • Joined: 03 Jun 2005
  • Loc: salem, OR

Posted 28 May 2016 - 08:08 PM

Here's the stretched version:

 

16 f&d stretched twice.jpg

 

That one's easy to measure even though there are multiple pixels at the equal zones.

 

I'm going to do some experimentation to find if there's a better capture setting in terms of the output result...


Edited by mark cowan, 28 May 2016 - 08:09 PM.


#24 mark cowan

mark cowan

    Vendor (Veritas Optics)

  • *****
  • Vendors
  • topic starter
  • Posts: 7679
  • Joined: 03 Jun 2005
  • Loc: salem, OR

Posted 29 May 2016 - 04:03 AM

Turns out the best setting is what "looks" the best, focusing the lens better is helpful.  Have completed full set of 30 images on one axis:

 

 30 sliced and diced.jpg

 

Next step is to measure and fit the results.  Very time consuming process done this way...   :imawake:

 

 


  • philipdo likes this

#25 brucesdad13

brucesdad13

    Apollo

  • *****
  • Posts: 1034
  • Joined: 25 Jan 2015
  • Loc: Rutland, MA

Posted 29 May 2016 - 06:29 PM

You need a software engineer ;-)




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.







Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics