Jump to content


Photo

Ross (et al) Null testing parabolas

  • Please log in to reply
82 replies to this topic

#51 MKV

MKV

    Mercury-Atlas

  • *****
  • Posts: 2610
  • Joined: 20 Jan 2011

Posted 12 January 2013 - 03:41 AM

Now that the construction and other issues of the Ross are handled, one of my favorite bench tests to analyze a null, is the bench star test, and that should be useful in the Ross. Anyone tried it?

Ever read the whole thread? Well, the short answer is yes it has been mentioned and illustrated on this thread already.

regards,
Mladen

#52 MKV

MKV

    Mercury-Atlas

  • *****
  • Posts: 2610
  • Joined: 20 Jan 2011

Posted 12 January 2013 - 05:10 AM

re: "The point source I've most often used is a lensless laser pointer diode" [ ccaissie]

Indeed, a nonlasing diode will have a tiny point but a true artificial star has to be equal to, or smaller than the Airy disc of a particular optical configuration.

The Airy disc diameter is given by d = 2.44*w*F#, where w = wavelength, and F# is the focal ratio (F/D). For an f/5 and w = 0.000633 mm (red laser), the Airy disk will be 0.0077 mm or 8 microns. I seriously doubt a lensless laser diode meets that requirement.

Texereau's method of using a steel ball can easily produce an artificial star source of accpetable size. The problem with Texereau's method is the quality of the steel ball, which would have to be optically perfect smooth and zone free, keeping in mind that on reflection any imperfection on the ball surface will be doubled. Perhaps unused precision ball bearings are smooth enough and meet astronomical surface requirements, but I doubt it.

Keep in mind that you'll get some kind of diffraciton pattern no matter what, it's just that it's not going to be equivalent to a star test because you will never know if the imperfections are due to the mirror tested or the light source.

An artificial star source must be optically corrected, and the best way to achieve that is by the use of quality microscope objectives and applying the lensmaker's principle in the setup. Microscope objective are specifically corrected in such a way that makes them ideal for artificial star source.

By placing the light source at the objective's focus a microscope objective will form a fully corrected image of that source some distance in front of it.

Say that you have an objective that was configured to examine specimens at a distance of 2 mm, and forms an image of the specimen at a distance of 200 mm. You have a 200/2 = 100X microscope objective. If your lensless laser source is 0.3 mm in diameter, the resulting artificial star will be 0.3/100 or 0.003 mm (3 microns), which will satisfy requirements for any system f/2 or slower.

regards,
Mladen

Attached Files



#53 MKV

MKV

    Mercury-Atlas

  • *****
  • Posts: 2610
  • Joined: 20 Jan 2011

Posted 12 January 2013 - 06:05 AM

Brian, you can measure the lens to mirror distance with sufficient accuracy to have a reliable Ross null test, but in your case the problem is the f/4 mirror. That configuration leaves you almost no room for error and makes it practically impossible to do a reliable test when your lens-to-mirror spacing must be within 0.001 inch with your Jaegers lens!

A slower lens mirror would work just fine. The working diameter would be only For example even an f/4.5 would leave you with +/- 2.8 mm margin, using the same lens lens, and that's plenty. A BK7 lens with R1 = 250 and f = 484 mm would give you almost +/- 5 mm of tolerance

Now, the real problem is the working diameter (WD) required. for a 20-inch mirror. In your case, this would not be eased even if you decided to go with a slower focal ratio. The WD of your lens would be a whopping 78 mm, which means you'll need a lens whose correction is better than 1/8 wave over at least that much - a toll order indeed.

Now, depending on how important this is to you, you may consider buying a larger lens, say a 6-inch PCX, and rework it to the radius and finish required for a 20-inch test.

Surplus Shed often gets new lenses and it's only a matter meticulous checking (along perhaps with eBay and similar sites) before you find a suitable lens. They sell large 6-inch lenses from time to time at bargain prices and it would not be too much work to have them reground, repolished and refigured to required surface standard over the WD area, keepong in mind that lens requirements are more relaxed than those for a mirror.

That's probably not what you were hoping to hear but there is a lot of good in it. For one, it will be a valuable learning experience with lenses, and, two, once you have a Ross lens that size it will always be useful. And don't forget that with the Ross you can NULL elliptical, as well as hyperbolpodial surfaces for Cassegrains and exotic telescopes of all kinds.

A null test is a dream compared to other tests. And, as you said, it gives you an instant idea where you stand correciton-wise, no tedious and numerous measurements needed, not squinting or guessing.

Mladen

#54 Ajohn

Ajohn

    Mariner 2

  • *****
  • Posts: 283
  • Joined: 03 Dec 2007

Posted 12 January 2013 - 06:27 AM

There is probably a very simple answer to all of this based round Dall's instructions. In fact given the vagaries of figuring it's probably the only way of doing it as the radius of curvature is very likely to change as the mirror is figured eg going too far in one zone and then playing with the middle.

Dall's instructions effectively mean placing the knife edge at the centre of curvature - easily done when the mirror is a sphere - and then adjusting the tester so that the image of the pinhole falls as close as possible to the knife edge. Not so easy when the mirror is not parabolic.

What is happening is that the base sphere of the mirror is being used as a reference sphere for the tester. It will always be possible to find the radius of that with the aid of a mask and the knife edge so that can always be placed in the correct position. As the parabolic form takes shape it will get easier and easier to position the tester so that the image of the pin hole falls in the same place. When that happens and there is a null it has to be a parabola providing the distance from the lens to pin hole is correct. Dall takes care of that aspect because that distance is fixed as part of the test and can be multiplied to obtain nulls on different conics. He doesn't mention correcting that distance according to changes in the central curvature of the mirror but if some one for instance wanted to form the conic by deepening the centre which starts getting attractive on very fast mirror that would have to be accounted for.

The question really is just what reference sphere the Ross test uses. I would say it's bound to be the same sphere. Then adapting this procedure to suite the Ross test.

When I used the Dall test I did this without really thinking about it as the mirror was already parabolic. Rather than making in a tubular holder for the lens and pin hole I put both in perspex holders that slid along a perspex bar. Much easier to do as suitable tube may not be easy to get hold of. Also easier to set up. I set the distance with vernier callipers - analogue ones. It was a long time ago :o

Perspex was a bad idea. Sufficient light came from the pin hole to illuminate the lens holder and give the mirror a sort square profile mostly removed by painting the front of the holder.

The pin hole was a slit using a rather small hole. Dall suggest no more than 0.030in long which is a size that is fairly easy to drill. The biggest problem with slits is the need to square them up to the knife edge.

On Ofner I'm fairly sure I just read something famous suggesting that a mirror can be figured to 1/20 wave with 1/12 wave error in the measuring system.

On suitable lenses I have seen mention of making a lap and repolishing them to a sphere without actually being able to check that. I can believe this can work out. I once had the spherical surface of a commercial lens tested on a Johansson coordinate measuring machine. The radii was all over the place and varied markedly across the surface of the lens. Dall's advice on lens focal length was to measure it as accurately as possible via a clear image within the aberration halo. I have seen mention of using this test with the convex side of the lens facing the mirror. My guess is that as he was a very bright cookie there will be very good reasons why he chose to use it the other way round.

John
-

#55 Ajohn

Ajohn

    Mariner 2

  • *****
  • Posts: 283
  • Joined: 03 Dec 2007

Posted 12 January 2013 - 06:47 AM

Texerau mentions lapping a ball bearing to improve it's figure. Also rotating the ball to see if that is producing astigmatism. I also assumed from his comments that the set up is intended to give a "star" that is well below the diffraction limit of the test set up. Much like how certain lenses are tested with a pin hole that may or may not be round that is under 1/2 the size of the Airey disc.

I am heavily into microscope and I would be very cautious about exact calculations based round their parameters. They are all fully corrected with additional optics even if it is only the manufacturers eyepieces that are intended to be used with them. An additional factor of 2 wouldn't be a bad idea given how the NA seems to be specified. When checked they seldom live up to it. It would also be a good idea to use what is normally referred to as a metallurgical objective. These are intended to be used without a cover slip and even at an NA of 0.5 the absence of one on normal objectives has a noticeable effect. :( Some manufacturers now don't bother making 2 versions of those.

John

#56 MKV

MKV

    Mercury-Atlas

  • *****
  • Posts: 2610
  • Joined: 20 Jan 2011

Posted 12 January 2013 - 08:35 AM

re: "Texerau mentions lapping a ball bearing to improve it's figure. Also rotating the ball to see if that is producing astigmatism. I also assumed from his comments that the set up is intended to give a "star" that is well below the diffraction limit of the test set up. Much like how certain lenses are tested with a pin hole that may or may not be round that is under 1/2 the size of the Airey disc"

How do you lap ball bearings? And why would a ball bearing produce astigmatism?

A ball bearing can easily produce an artificial star that's below the optical resolution of the optic being tested even if your light source is relatively large.

Let's say you use a 3 mm red LED, and a 4 mm diameter precision ball bearing. The ball bearing radius will be 2 mm and its focal length -1 mm. If you set the light source at a distance of 300 mm (~ one foot) from the ball bearing the image will be formed "inside" the ball bearing at a distance d2 = 1/f - 1/d2 or ~ 1 mm. The magnification ratio then will be d2/d1 = 0.00333*. Multiplying that with the original LED diameter of 3 mm you get an artificial star image that is 10 microns. Increase the distance d1 = 500 mm and the artificial star will be 6 microns, etc.

You can also create a more compact artificial star tester by using several short focus lenses and shorter seperations.

re: "I am heavily into microscope and I would be very cautious about exact calculations based round their parameters. They are all fully corrected with additional optics even if it is only the manufacturers eyepieces that are intended to be used with them"

That's a good point, John, especially for newer microscopes with ED lenses and complimentary proprietary eyepieces. Older microscopes objectives wre designed to give a corrected image at the focal plane and then magnify it withn an ordinary Huygenian and Kellner eyepiece. For artificial star images yu can be pretty safe with older microscope objectives.

They form textbook Airy discs with monochroamtic light source (usually a simple superbright LED and a narrow band filter of about 10 nm width).

#57 ccaissie

ccaissie

    Viking 1

  • -----
  • Posts: 634
  • Joined: 13 Sep 2010
  • Loc: Whitefield, Maine

Posted 12 January 2013 - 09:19 AM

Exactly that. I found the ball bearing had some microdefects on the surface so buffed it up, and much of the scatter etc. was cured.

LD chips can be pretty small and rectangular, like .3 x 5 microns.

Indeed, at high power bench star testing, like 800x, the image at focus does suggest the actual chip shape, but what is sought is a difference in the appearance of the out of focus rings, etc., which shows up well after the KE test gets difficult. Y'all must have seen that yourselves. I'm just touting that if we are looking at a correctly nulled point image, the star test protocols are useful in complement to KE and Ronchi.

#58 MKV

MKV

    Mercury-Atlas

  • *****
  • Posts: 2610
  • Joined: 20 Jan 2011

Posted 12 January 2013 - 09:26 AM

I was hoping to find a way "around" it by fixing the KE to lens distance and then move the whole stage back and forth to get the best null I can as I work the mirror (just like if it is was a sphere).

Actually, even the lens thickness is a variable in figuring out the correct postions, and I am talking a fraction of a mm of lens thickness! So you have to be absolutely sure about the lens parameters, glass type, distances, mirror radius of curvature, and clear apertures (i.e. lens minus the bevel, etc.)

You can't change anything in the Ross null solution without changing all other variables proportionally. But I can totally relate to your thought process. I thought of the exact same thing because it seemed logical at the time, so I actually put it to a test! As they say, live and learn. When it became obvious I wasn't getting the expected results, I sat down to learn more about it, which is not easy because very little is written on how the equations were derived.

#59 Ajohn

Ajohn

    Mariner 2

  • *****
  • Posts: 283
  • Joined: 03 Dec 2007

Posted 12 January 2013 - 10:15 AM

Texereau mentions polishing a 1/4in ball bearing up in a lathe. An optician would do that by centring it in rosin the same way as lenses used to be mounted for centring. Me I would mount it on the end of stick and rotate by hand. Hopefully that could be arranged to cause the bearing to centre itself with a bit of thought.

He does mention rotating the ball bearing as a check that astigmatism if present is due to the mirror. He uses that as a check for the mirror and later uses the same idea to check the flat using the main mirror with a mask over it just leaving the central portion clear to approximate to a good sphere. The mirror ideally needs to be aluminised though.

On Huygenian eyepieces and microscopes it fairly generally accepted that these do not follow the usual prescription in order to compensate for objective aberrations.In my experience when pushing things to the limit it pays to get the correct eyepieces. The capabilities of Huygen types tend to vary over the years. Very early on they would be often used for all types of objectives, later additional variations may be available for plan eyepieces in particular or apo's etc. Nikon were the 1st to completely compensate for colour purely in their objectives. They come for 160 and 210mm tube lengths and at one stage where relatively cheap for what they are used but people started using them for direct projection colour photography so prices have rocketed. The 210 mm tube length ones are metallurgical types. With 160mm tube length objectives it's best to avoid Zeiss as there is no way of knowing how recent they are and they are known to have delamination problems up to some point in time. Leitz 170mm have to be rather old. Old japanese objectives are rather short. The 160mm tube ones aren't and I would say are good bet for a decent objective followed by 160mm Leitz. At the cheap end Vickers objectives are amazingly consistent and last well providing they don't have DIN marked on them but know one is entirely sure what tube length they used. The optical tube length on 160mm objectives is to the DIN standard which is 10mm down the mechanical tube so is 150mm. Vickers used a 160mm mechanical tube and people suspect the optical tube length was 17mm down that. Modern infinite tube length objectives are a different kettle of fish - amongst other problems the opticians have yet another piece of glass to play with in the microscopes tube. It is probably worth trying to project a near parallel beam of light down them but I would have doubts about them meeting their specs. The 160mm and green light would be the best bet as compensating eyepieces usually show an orange tint round the edge when just held up to the light. That can apply to Huygens types too.

Hope this all reads ok - I've been out of the house twice and eaten dinner while typing it.

John
-

#60 Mike Lockwood

Mike Lockwood

    Vendor, Lockwood Custom Optics

  • *****
  • Vendors
  • Posts: 478
  • Joined: 01 Oct 2007
  • Loc: Usually in my optical shop

Posted 17 January 2013 - 06:23 PM

Well, coming from someone who professionally specializes in under-f/4 mirrors of up to 60-inch diameters (and won't even touch anything smaller than 14-inch disks), I'd say that for your experience and tooling it's no problem for sure! :)

Just for clarity, I refigure mirrors of nearly any size, and occasionally make 10" and 12" mirrors from blanks that I have on hand. I just don't take orders for them. I also make smaller Cassegrain secondaries and elliptical flats.

So, yes I will touch smaller than a 14" disk.

On the subject of the Foucault, I am curious, since you deal with much larger optics than commonly encountered among amateurs, why do you think the professional community abandoned the Foucault way back in the 30's for their large observatory mirrors and bothered to devise alternative testing methods?

Simple - they didn't, and your assertion is inaccurate.

Though you are complimentary of my work in later messages, and even if it wasn't your intention, the message above implies that I use outdated methods, which I assert is not true.

The 200" Hale telescope was mostly tested with Foucault, then a little bit of caustic testing, and finally Hartman mask testing in the telescope itself. Yes, a 200" f/3.3 mirror was made mostly with Foucault testing.

While the occasional parabola might have been done with autocollimation during the era you describe (extremely large flats just don't exist, as it turns out), as I have heard and it has been described to me by those who did serious optical work during those times and at major companies, most tests were done at the center of curvature.

I should point out that for Ritchey-Chretien telescopes, autocollimation is not a null test for the primary. Testing the system in autocollimation is possible, but tricky even with one mirror complete and coated, and thus two reflections off of an uncoated mirror. One has to work match wavefront intensities and get good fringes with an interferometer. Also, the larger the system, the more vibration becomes an issue due the physical size of the test setup.

Lasers (and thus interferometry) weren't even available to most shops until the mid to late 1960s, well after the date you mention.

So, the methods that I use are time-tested and have been used with great success for very large projects. They are still used today by other respected professionals, arguably producing better end products than other methods.

However, I must say that I highly recommend use of the more modern interferometer for checking figure of revolution of large mirrors. While other tests can show when there is a problem, the interferometer will quantify and locate it so that it can be remedied most efficiently and effectively.

I would completely trust Mike ability/skill to Foucault test fast mirrors.... comes with extensive experience. The thing is, I doubt *mine* :-). Can I tell the difference in a zone null with the steep slopes of a f/4, within 1-2 thousandths?.... not so sure... :-).

Brian, give it a try. You might surprise yourself. One thing is for certain - if you never try it, you will never learn it.

While I'll try to keep reading this thread, clearly others who have been posting here have far more time to spend on posting than I do, and I don't have time to debate test methods.

#61 MKV

MKV

    Mercury-Atlas

  • *****
  • Posts: 2610
  • Joined: 20 Jan 2011

Posted 17 January 2013 - 07:29 PM

Mike, the caustic test was devised precisely because the professional community felt there was a need for a more accurate quantitative method. Since the 1930's the community has moved away form reliance on the Foucault test. In the process, it has also abandoned the caustic test because it is so time-consuming. Intrerferometry pretty much excludes the human bias factor, to which the Foucuat is not
immune. But that doesn't mean an experienced and well tooled individual can not produce superb mirrors with the Foucualt. So, yes the method is old, but it's not the tester that makes the mirror, the mirror maker does.

Mladen

#62 Ed Jones

Ed Jones

    Mercury-Atlas

  • *****
  • Posts: 2716
  • Joined: 06 Apr 2004
  • Loc: Sin-sin-atti

Posted 29 April 2013 - 01:48 PM

This is an old thread but here is a video I did of a cheap, easy null test that might be of interest.

#63 Pinbout

Pinbout

    Cosmos

  • *****
  • Posts: 8039
  • Joined: 22 Feb 2010
  • Loc: nj

Posted 29 April 2013 - 02:30 PM

Great vids Ed!

#64 Ajohn

Ajohn

    Mariner 2

  • *****
  • Posts: 283
  • Joined: 03 Dec 2007

Posted 30 April 2013 - 04:47 AM

Thanks Ed. Another option.

I wonder if the Ross test can be re arranged to be set up like the Dall test. That is set with the knife on the centre ROC, easy to do to a fraction of a mm. The tester is then moved to bring the slit image to the same place - ie to null out the centre again or use an optical aid to achieve both settings. That then leaves the problem of measuring the distance from the mirror to knife edge and setting the test lens accordingly. To me this is a better option than a measuring stick. All depends on how accurately the mirror vertex rad is measured anyway.

Walland used Foucault to test fast mirrors having the advantage of knowing what he would see. I get the impression that this method was common augmented with a subsequent Hartman test as a final check. During final titivating he used a Ross type lens positioned to completely or partially null out SA, no measurements at all. No super precision lens either. He puts the method down to Dall and suggests a lens twice as fast as the mirror being tested.

There isn't much information about on modern methods but they seem to use nulling lenses and an interferometer. I wonder if that is a more sensible method of using Bath for figuring leaving the software for a last final check. The eye is much better at gauging flickering fringes than a camera.

On Bath I recently bought a double convex 4mm FL lens. While in theory it should be easily capable of putting out an F3+ beam from an ordinary laser diode set up it falls well short. All I can put this down to is a radial error right on the vertex of the lens. This wouldn't matter in normal use as it contributes little to the image. It looks to be a hot pressed lens and clearly shows variations in RI.

No one seems to come clean on exactly what a precision lens is for any of these uses, other than one Ross type which is too expensive in my view. Precision lenses seem to be shaped to 1/4 wave in red light but are these sufficiently precision? Dall for instance talks about standard ophthalmic crown lenses and measuring the FL so just how important is the precision of the lens for each type of testing.

John
-

#65 Ed Jones

Ed Jones

    Mercury-Atlas

  • *****
  • Posts: 2716
  • Joined: 06 Apr 2004
  • Loc: Sin-sin-atti

Posted 30 April 2013 - 06:58 AM

John,
I'm not sure I follow you, it sounds like the Dall set up.
I'm advocating this test for beginners or anyone because it's easy to set up, inexpensive ($40), insensitive to spacing and you can test any parabola even down in the F/3 speeds where the Ross starts to fail. I've looked at a number of Newport and OptoSigma lenses and find the middle area is reliably good enough. If you are really doubtful you could buy a plano concave lens and fringe test the convex curve and test the flat on a good flat. You only need to set the KE/lens space with only a little accuracy and the light source to the nearest foot.

Ed

PS It's also perfect for testing a purchased telescope without having to remove the mirror.

#66 MKV

MKV

    Mercury-Atlas

  • *****
  • Posts: 2610
  • Joined: 20 Jan 2011

Posted 30 April 2013 - 08:49 AM

John, you can get a decent laser-quality lens by cannibalizing medium to higher price laser pointers.

Other than that, like Ed noted, I got a little lost as far as what you were trying to say, but be mindful of the fact that you can obtain an optical null even if the optic under test is not fully corrected (the "Hubble") unless your distance is within the required enevlope.

In the Ross null test, the essential parameter is the mirror to lens distance. Depending on how much wavefront error you're willing to tolerate that distance can vary as little as a fraction of a mm to dozen or so millimieters!

The conjugate null test Ed described is a litte more forgiving in that respect but it's down side is the requirement for the light source to be at a distance that makes the test subject to air currents.

Also, let's not forget that the conjugate null test is a single-pass test and not that much more sensitive than than Ronchi null-testing your scope using star light -- which is even simpler and cheaper to accomplish.

#67 Ed Jones

Ed Jones

    Mercury-Atlas

  • *****
  • Posts: 2716
  • Joined: 06 Apr 2004
  • Loc: Sin-sin-atti

Posted 30 April 2013 - 10:17 AM

Mladen,

The big problem with the Ross is the cost of a qualified null lens, not something a beginner would use. BTW the critical airspace on the Ross is the KE/lens spacing.

The Dall conjugate is affected by air currents but there's a big difference between 100 feet in your back yard and looking through miles of atmosphere.

#68 DAVIDG

DAVIDG

    Aurora

  • *****
  • Posts: 4887
  • Joined: 02 Dec 2004
  • Loc: Hockessin, De

Posted 30 April 2013 - 10:47 AM

Ed,
Actually the critical spacing on the Ross is the Lens to Mirror spacing and you just move the knife edge or Ronchi screen behind the Ross lens to find focus just like your doing a type Foucault or Ronchi test. Just like say in your video about your test many times only a small portion of the lens is used in the Ross Null so a typical 50mm off the self lens can also work well. As you say in your video many ATM's have problems measuring zones and a null test is better alternative. With a little bit of testing of the lens like you say in your video you can qualify the accuracy of a Ross lens as well. Places like Surplus Shed have "precision" lenses that are inexpensive so one can purchase a couple and test them to find the best one.
Having help teach mirror making classes for over 20 years now, using Null test methods ( double pass autocollimation and Ross Null) ATM first get hang up with accuracy of the test methods but soon find out that isn't the real issue. What is, is getting the mirror to show a clean null. A true 1/8 wave mirror that shows a clean null will perform better then 98% of the optics out there and one can have the confidence that they have achieved that accuracy with a little homework to prove out their test methods by checking the quality of the nulling lens they use and the accuracy in the setup for their spacing. I alway recommend double checking the results with other test methods as well. If they don't agree, then one needs to take the time to find out why and not simple ignore the results of one in favor of another. So depending on the diameter of the lens one could do a Ross Null and then your test as well using the same lens and then also a Foucault test and finally a star test. Then compare the results and see how well they agree.
Nice video of what I viewed so far. YouTube for some reason is only downloading about 3/4 of the full movie. So I don't know if you shows this or not but if you haven't it would be educational to show what a 1/4 wave of error looks like in the amount of bowing of the Ronchi lines to get a feel for the sensitity of your test under real world conditions. Maybe test a 6" f/8 sphere which is just 1/4 wave and typical of the size and F-ratio many ATM's make the as a first mirror to see how much bowing is visible compared to a parabolized 6" f/8 of known wavefront.

All the Best,
- Dave

#69 Ed Jones

Ed Jones

    Mercury-Atlas

  • *****
  • Posts: 2716
  • Joined: 06 Apr 2004
  • Loc: Sin-sin-atti

Posted 30 April 2013 - 12:22 PM

Hi Dave,

No the KE/lens air space is the more critical airspace in the Ross. For example a 20 inch f/4 set up with a Ceravolo null lens has .015 wave nominal wavefront test error. If you would make a 0.05 inch error in the mirror/lens airspace and find focus with the KE then the wavefront error increases to .23 waves. However a 0.05 inch error in the KE/lens airspace results in a .54 wave front error. Not sure how you can test the convex side of a Ross lens reliably without a test plate or inteferometer.

Not sure why you're not getting the whole download, I've downloaded it on 2 computers here? Yea it would be nice to have shown an uncorrected mirror. I didn't have many Newts to pick from (mostly Chiefs) and of course it had a good mirror :lol:. Later I can do one on parabolizing perhaps.

#70 MKV

MKV

    Mercury-Atlas

  • *****
  • Posts: 2610
  • Joined: 20 Jan 2011

Posted 30 April 2013 - 12:43 PM

No the KE/lens air space is the more critical airspace in the Ross. For example a 20 inch f/4 set up with a Ceravolo null lens has .015 wave nominal wavefront test error. If you would make a 0.05 inch error in the mirror/lens airspace and find focus with the KE then the wavefront error increases to .23 waves. However a 0.05 inch error in the KE/lens airspace results in a .54 wave front error.

I made a setup with precise spacing (using radius bars)

Posted Image

for both mirror-lens and lens-ke setting, and when I looked at the mirror I saw -- nothing! Of course, because the image was in focus! Moving the Ronchi screen in and out a little produced characteristic Ronchi bands, which in this case were straight. So, I am not exactly sure how can lens to k-e distance be critical, considering that in order to see the bands you may have to move your screen more than 0.05 inches in and out of focus.

Mladen

#71 Ed Jones

Ed Jones

    Mercury-Atlas

  • *****
  • Posts: 2716
  • Joined: 06 Apr 2004
  • Loc: Sin-sin-atti

Posted 30 April 2013 - 02:17 PM

Well yes they both should both match. I don't think I made a mistake in my Zemax set up. I did make 2 typos. In the nominal test error, .15 not .015 waves error or about 1/6.5 waves and for an F/3 not an F/4 mirror. :foreheadslap:

#72 Ajohn

Ajohn

    Mariner 2

  • *****
  • Posts: 283
  • Joined: 03 Dec 2007

Posted 30 April 2013 - 06:12 PM

Sorry if confusing - another go. Not sure which part caused confusion.

Ross test. I assume that it's essentially the same as the Dall null that uses and offset test unit and knife or could be except that the lens is on axis in Ross's test. On this basis rather than using a measuring stick a knife edge could be set to the centre of curvature of the mirror and the lens then introduced at the correct distance from the knife edge. As far as I am aware the only feasible way for most people to measure the radius of curvature is via a knife edge anyway. The distance is needed for the calculations and it's likely to vary as the mirror is figured. The distance from the lens to the knife edge is short and more easily measured. If the radius changes that needs changing as well. One way of taking knife edge to mirror distances quickly might be to build a steel tape into the set up.

Waland. Waland used a moving source Foucault tester. He gives detailed information on figuring 2 60in mirrors. A precision F4 and a F2.5 for IR work. He figured via shadow position using tooth picks at 1in intervals. To finally titivate a precision mirror he introduced a plano convex lens into the beam from the tester in much the same way as it's used in the Ross test and moved it around to control shadow density and spot minor zonal problems etc. Used this way the precise characteristics of the lens aren't important. I would hazard a guess that the Foucault tester would be set at the vertex radius of curvature. At some distance from the tester the lens will achieve as complete a null as it's capable of producing so it's essentially a sensitive test. As he suggest a lens that's twice as fast as the mirror under test like most he seems to be relying on the centre of the lens being good. Maybe some work with a hard pitch lap would ensure that it is.

I like shadows especially via a mask. I suspect most peoples problems relate to too coarse a control of knife edge into the beam. Maybe too narrow a slit too. I've struggled with the usual M6 tilt bolt with it's normal head. It really needs a 50mm dia head at least to give sufficient control. That way the way the shadows behave as the knife cuts into the beam can be observed. I'm not a fan of fuzzy ronchi bands I'm afraid.

Lens precision. I'm in the UK and can only buy easily from sources like this one. Shudder to think what research grade would cost

http://www.galvoptic...o.uk/lenses.htm

Other sources give similar tolerances and state things like grade A glass and max surface errors of 1/4 wave.

There are a couple of other tests that deserve some interest. One uses a camera to read the shadows. There are 3 separate software packages available for this. The other is the caustic wire test with a loupe some way behind the wire. The wire can be centred in the fringes very quickly but the X readings need to be taken with a high degree of precision. A lashed up a mod to my tester to try it. It's basically 2 Texereau type sliding tables one on top of the other. My next version will be a more refined version of that. Once I have made it any of the usual testers can be place on it. The Bath people have finally produced a more complete write up on what is needed so if I ever build one the same x-y stage will be useful. 1/20 wave beam splitter if possible though. I am inclined to conclude that this test gets more and more complicated as one gets into it. It looks quick, cheap and simple to do but ........ Using it to figure means processing photo's at each stage or a very stable environment. I do wonder if adding a null lenses and working for straight projected fringes might be a good way of taking a tester as far as it can reasonably go. Then using the software for a final analysis.

:bawling: My tester has been in a sad place for several years. Needs a rebuild anyway really. It had a sliding stop to allow the mic head to be zero'd when the knife is set to the mirror. It was basically as per Texereau except I tilted the source to set it parallel to the knife edge. Less bits to make. It was made out of odd off cuts that were available for free at the time.

I can now add the conjugate test to the list. :grin: I'm likely to need a number of options on testing.

John
-

Attached Files



#73 MKV

MKV

    Mercury-Atlas

  • *****
  • Posts: 2610
  • Joined: 20 Jan 2011

Posted 30 April 2013 - 09:33 PM

Ed, when you do a Ross null test and your lens to mirror distance is correct, your focus and your light source should be equal and you should observe a clean null with a k-e. With a Rnchi grating you don;t observe anything at the focus! To the contrary you will see the bands (straight if the resultatnt wavefront error is nulled) as you move your Ronchi screen in and out of focus. The shape of the bands will not change indicating the wavefront is n ot shifting; only the number of Ronchi bands will change as you move the screen along the axis.

#74 MKV

MKV

    Mercury-Atlas

  • *****
  • Posts: 2610
  • Joined: 20 Jan 2011

Posted 30 April 2013 - 10:00 PM

As far as I am aware the only feasible way for most people to measure the radius of curvature is via a knife edge anyway

In my experience a spherometer and a radius bar provide for accuratee and repeatable setups.

For the Ross test, you need a couple of clarifications: (1) set the mirror lens distance accuratelty, (2) move your light source and ke tester away or towards the lens until you see the bands. This will happen at the correct lens to source distance.


The Bath is not something you want to use during figuring. For aspheric mirrors, the Bath should be used with some type of null nulling test.

I highly recommend an optical flat. Either buy or make one slightly larger than the largest scope you think you'll ever own.

If that's not an option, then stick to the good old Foucault and Ronchi tests with Ed's conjugate null or a simple Polaris null with a Ronchi eyepiece.

#75 Ajohn

Ajohn

    Mariner 2

  • *****
  • Posts: 283
  • Joined: 03 Dec 2007

Posted 01 May 2013 - 07:08 AM

I tried to put some one off on another group from using a ronchi screen for figuring via matching patterns against screen movement. He had read it in a book so it must be good as I often say. Matched 6 patterns that I assume are shown in the book, why 6 pass. He then took the mirror for coating and asked the coater to check. Seems it's at least 90% corrected, we hope a lot more. If he has managed 95% or greater which is doubtful he would probably have a mirror just within Rayleigh's limit. If it's close to 90% figured it's way way short of that.

Out of interest the knife edge on my rusty tester can turned through 90 degrees so that the proper mic head can be used in the X direction. It reads to 0.0001in via a vernier and there were indications that finer readings than that could be taken. Biggest doubt is the Y measurements. Rather than following the method outlined in ATM III it may be better to note Y readings and then use trig to work back to the normal Foucault readings.

I do have a coated flat but have no idea how flat it actually is. Once I have a sphere it should be easy to find out. As I understand it smoothness is important and slight dish etc of the order of a 1/4 wave or so doesn't really matter. It's like testing against a star that is several miles away. This made flat backed achromats popular. The tool can be polished up and used as a flat to finish the lens the argument being that dishes of this order can just be detected with a home made spherometer zero'd on one and readings taken on the other surface. Mentioned in case some one fancies making a flat. Some people have simply ignored errors due to none figures of revolution errors in the flat.

I believe a few people have figured mirrors with Bath plus software but I suspect it's only really convenient with a web cam style set up. It all boils down to how many shots need to be taken, mirror rotated because it sags etc and how quickly they can be got into a PC for analysis.

I fancy giving Waland's method a try. I'm fairly convinced that the main problem with Foucault is the need for fine knife control both ways. I'm told deep shadows are a problem at faster ratio's and his use of the lens gets round that aspect without using an excessively wide slit. I don't need a Ronchi screen to test for a sphere as fringes can be generated with the knife edge as Texereau outlines. That allows it to be checked right up to the edge. :) Not disregarding advice. How well a test can perform is one of the interesting aspect to me and there is no harm in trying several. Figuring F3 is bound to be difficult. It seems spin polishing and a harder than usual lap might help with that. Something else to try as are semi flexible laps and ring laps plus anything else I can think of.

John
-






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics