Click here if you are having trouble logging into the forums
Privacy Policy |
Please read our Terms
of Service | Signup and
Troubleshooting FAQ | Problems? PM a Red or a Green Gu… uh, User
Nils Olof Carlin
Pooh-Bah
Reged: 07/26/04
|
Re: I have read all the posts about tests
[Re: MKV]
#5309347 - 07/09/12 07:18 AM
|
Edit
|
Reply
|
Quote |
Quick Reply
|
|
|
Quote:
"...and its inadequacy is therefore more evident."
Let it be clear that Linfoot refers not to the Foucault test itself here, but only to the inadequacy of ray theory to explain what is observed. (Unfortunately, there seems to be indeed a wide-spread belief among amateurs that Foucault is a purely geometric test.)
In the paragraph preceding the sentence you quoted, Linfoot says "Because of its simplicity, the zonal Foucault test seems likely to remain one of the most widely used methods of testing the figures of medium-sized astronomical mirrors, although for large mirrors, the 'caustic' method of Platzeck and Gaviola is probably superior in accuracy." Discussing their 1939 paper, he goes on: "[they] tried to estimate the limits of accuracy of the zonal Foucault test ... and of their own method of testing on the caustic."
And in the following sentence: "Not until a diffraction theory of each of these tests is available will a reliable theoretical estimation be possible of their limits of accuracy and of their systematic errors".
Thus, Linfoot does not contradict my belief that zonal Foucault is not clearly inferior to the Caustic test for small to medium size mirrors (suitable for inexperienced amateurs). Indeed, the zonal Foucault is not very practical for large and/or fast mirrors, but this is the realm of experienced mirror makers.
In the closing paragraph, (p. 445) Linfoot gives this take home message: "The zonal settings are most accurately determined, not by comparing the brightness of the apertures with the knife-edge stationary, but by observing which darkens in advance of the other as the knife-edge is brought across[...] The accuracy of each zonal focus determination is therefore limited in a practical case only by the observer's limited power to perceive that one aperture is darkening a little ahead of the other."
But there remains a clear and perhaps puzzling discrepancy between what critics of Foucault testing estimate the limits of accuracy to be, and what skilful professionals actually do achieve using it.
The beginner will not master optical testing, nor any of the other skills needed to make a first-class mirror, on a first try. But beginners have made satisfactory mirrors before, and I still think it is fair to recommend Brad (and his likes) to use the zonal Foucault test for figuring, and the star test to decide the acceptability of figure.
|
MKV
Carpal Tunnel
Reged: 01/20/11
|
Re: I have read all the posts about tests
[Re: Nils Olof Carlin]
#5309762 - 07/09/12 12:51 PM
|
|
Edit
|
Reply
|
Quote |
Quick Reply
|
|
|
Nils, I agree with much with what you wrote. Also, select quotes can never present the whole picture, but they can set the general mood and intent of the author, and the general mood of the professional community, in this case, was: the Faucult Test needs to be replaced with something better as soon as possible. And it was. And the professional community never went back to it. That's a fact.
There are professional tools and homeowner's tools, there are cheap lathes and top quality lathes, there are bargain cars and luxury cars, etc. In the hands of experienced and talented users, cheap tools can produce better results than expensive tools in the hands of a newbie.
So it is not surprising that a few talented and experienced opticians can turn out better mirrors with nothing more than a Foucalt Test. That speaks more of their talent and experience than of the quality of the test itself (But, I seriously doubt that people who make quality optics for a living only use the FT.)
Freebie software and mythical quality of the FT resurrected in recent years helped create an apparent belief among some amateurs that the FT can turn out 1/100th wave optics (recently, I have seen even six times higher claims)!
This kind of "wave inflation" is leading ATMs to believe they have top-quality optics simply because the computer says so, and because they trust what they measured is accurate without having calibrated their measuring devices (because no one told them to).
Instead, they are told to get a pulley wheel from a hardware, add three commercial ball bearings to it, stick a $15 dial indicator in the center, and they will have a "precision" spherometer that will give them exact radii of curvature.
Or, they are told to get some plywood, a few pressed out 1/4-20 screws, a plumbers brass pipe, stick a $15 dial indicator on one end and, with a simple knife-edge blade, they have a "precision" insturment for making top quality optics, even to 1/600th a wave!
(Makes you wonder what was the professional community thinking when it decided to drop such a 'perfect' test!)
Then they are told to get some freebie software, put into it the numbers they read with their 'precision' instruments, and it will spit out results to an accuracy of four decimal places! It's time for a wakeup call. We need to stop this deception.
Mladen
Edited by MKV (07/09/12 03:30 PM)
|
MKV
Carpal Tunnel
Reged: 01/20/11
|
Re: I have read all the posts about tests
[Re: Nils Olof Carlin]
#5310081 - 07/09/12 04:15 PM
|
|
Edit
|
Reply
|
Quote |
Quick Reply
|
|
|
Quote:
The beginner will not master optical testing, nor any of the other skills needed to make a first-class mirror, on a first try. But beginners have made satisfactory mirrors before, and I still think it is fair to recommend Brad (and his likes) to use the zonal Foucault test for figuring, and the star test to decide the acceptability of figure.
Yes, beginners have made satisfactory optics with Foucualt and Ronchi tests. That doesn't mean Brad and other shouldn't be encouraged to try easier and better null tests available to amateurs.
What is satisfactory? If your telescope delivers 1/4 wave or better on the wavefront as the light enters your eye, you have a decent telescope - considering all the obstacles that wait in the optical path.
That's why the final test of quality is, as some - including myself - have suggested: under the nighttimes sky, and under acceptable observing conditions.
But this must be done photographically, as Texereau did (see page 313, Figure A, of his How to make a Telescope) to get an idea of the average wavefront quality delivered at the focus and not, as some have suggested, using a Ronchi eyepiece or a knife-edge visually, looking at scintilating air from one instant to another.
Also, it should be borne in mind that the success in making quality optics with the FT/RT test at COC diminishes exponentially with faster focal ratios, which is why it is much easier to make a quality f/8 then a quality f/4 in moderate diameters. In larger diameters, slower mirrors are simply not practical and fast focal ratios dominate.
Unlike the COC tests, null tests are simply easier to interpret and work much better with faster focal ratios and larger apertures. There are no numbers to manipulate, nor reading errors to commit, no curve estimates, etc. - just simple straight bands or clean k-e darkening of the disk.
Null tests must be set up properly (like ALL tests, including the FT) with known limits of accuracy and quality of auxiliary optics used.
Null tests are not magic. They also DON't produce 1/20 wave P-V optics! Contrary to so many claims, evidence suggests that null tests produce optics with wavefront error between 1/5 and 1/10 wave P-V, depending if the test is a single or double pass.
At the lower end, single pass tests will not produce optics better by signifcant margin than those produced by traditional COC method, but they are much easier, especially in the final stages of figuring then the traditional methods.
Most null tests are very simple, and most of them don't require an expensive, large flat. Their auxilliary optics are very affordable and easy to find (they are single element, usually planoconvex lenses), and the only critical aspect of these tests is to measure two distances with reasonable amount of accuracy, just as with traditional tests.
The rest is the same as when testing a sphere at the COC. So, why so much resistance, confusion, anger and angst in the community over these tests except they were misinformed? Remember that professionals and experienced amateurs routinely used them.
Mladen
|
Mike I. Jones
Carpal Tunnel
Reged: 07/02/06
Loc: Fort Worth TX
|
Re: I have read all the posts about tests
[Re: Nils Olof Carlin]
#5310292 - 07/09/12 06:38 PM
|
|
Edit
|
Reply
|
Quote |
Quick Reply
|
|
|
I am NOT taking ANY sides here, one way or 'tuther. It ain't my style. But I do want to describe the best mirrors I've made over the last 40 years, how they were tested, and how they tested by different methods.
1. 16" f/6: Me and my buddy Mike Marcario worked that mirror back in 1981. We used my Foucault tester, and called it quits after 7-zone testing across several diameters, and hours of equalization. Data reduction said the mirror was 1/18 wave P-V, RMS somewhere in the 1/40 wave range as I recall (very smooth). Star testing on it, again after plenty of equalization, and on several different nights of very good seeing, showed absolutely Suiter-good textbook diffraction behavior, with the secondary shadow coming in perfectly symmetric either side of focus. Caustic testing on this mirror gave essentially the same figure quality as KE readings. I also took the mirror into my optics lab at work and null-tested it using the method I put up in this post a few days ago. The null test showed a flat, smooth sphere with only very faint zonal shadows, almost imperceptible.
2. 8" f/10 fused silica mirror: Again, I Foucault tested this mirror with 4 and 5-zone masks that overlapped each other. Fused silica works very slowly, and I was able to creep up on the figure very controllably. At the end, KE data reduction showed the mirror to be an unbelievable and repeatable 1/80 wave. As you can imagine, we were skeptical, and tested it independently on a Zygo interferometer in double pass. The best-focus residual wavefront error was below 1/50 wave, really at the very limit of the Zygo optics and digital camera (after cavity error measurement and subtraction), and air stability in the test path (using an insulating tunnel).
3. 12.5" f/7.5: Again, a matter of using the right strokes and pulling the lap off at the right time. This mirror rated at 1/50 wave P-V, again with a wonderfully smooth curve. It is now mounted in a nice long Newtonian tube, and word is it gives the best, sharpest imagery of any telescope in Lubbock, Texas. Same star test behavior - perfectly symmetric extra-focal diagonal shadow, perfectly circular Airy disk at focus. I would love to test that one in interferometry some time.
4. 6" f/8 on Zerodur: My last figuring stroke was just right, and the mirror tested at some value below 1/35 wave. Its owner sent it to Ed Cortez, a superb and fastidious mirror maker in Corpus Christi known for his own beautiful mirrors. Ed sent us a note saying, "What is this, a joke? The worst I could make that mirror was no worse than 1/100 wave!".
5. 6" f/4 RFT primary: KE said about 1/10 wave P-V, interferometer said right at 1/10 wave P-V. Very nice agreement, very nice imagery even at 200X (after GOOD collimation!).
6. 10" f/6: My buddy, Msgr. Ron Royer figured this one back in 1972 to better than 1/35 wave P-V. Again, tested in double-pass on a Zygo, its figure accuracy was so good that it pushed the limits of the optics, cavity subtraction and camera resolution on that Zygo.
I agree that credibility in Foucault readings below around 1/20 wave become questionable in beginner hands. But I also believe in this important point: that after you make enough mirrors, you develop a "feel" for reading Foucault zones that enables you to credibly push below that 1/20 wave barrier. I can usually repeat edge zone readings on a 10" f/6 mirror to within 0.002", and often to within 0.001". Augmenting the visual KE test with a 10 or 12-bit video camera can improve this accuracy considerably. I'm going to upgrade my Foucault/caustic tester to video very soon, but not all the way up to the RoboTester level - I still want to turn the knobs.
I will reiterate my personal experience:
(a) Direct surface measurement (Foucault, caustic) for absolute accuracy, but only subjective for smoothness
(b) Null testing for smoothness evaluation, but not so much for absolute accuracy,
(c) Interferometry to cross-validate the accuracy and smoothness of finished or nearly finished surfaces, and
(d) Your personal accuracy with all these tests will improve as you make more optics, and get the feedback on their performance in observational testing.
Just my 20 millidollars, but with very real foundation,
Mike
PS: I could post my bibliography as well, but it would go on for three pages (Applied Optics, SPIE, IEEE, etc.) and I don't see it as necessary to establish credibility here. This is a fun ATM forum, not a CV.
|
mark cowan
Vendor (Veritas Optics)
Reged: 06/03/05
Loc: salem, OR
|
Re: I have read all the posts about tests
[Re: Mike I. Jones]
#5310389 - 07/09/12 07:53 PM
|
|
Edit
|
Reply
|
Quote |
Quick Reply
|
|
|
Just a general comment here, as I did reexamine the original sources Mladen was discussing and now quoting. As I thought before, Schroader got some things wrong while getting some things right, but the underlying assumptions that Schroader used to support his conclusions are unsupported.
I'll not take sides either, because I'll always use whatever test is most appropriate for whatever task it's being asked to do. For new production that will be unmasked computerized Foucault to monitor the process and interferometry to verify.
Have a look at this comparison for a 14.4" f/4.6, it shows results for both careful manual Foucault testing and unmasked digital Foucault (over a smaller range of zonal radii). Note that the overall rating for this mirror using either method is around 1/60th wave PVW. But I don't guarantee test results, just performance.
Manual Foucault combined with other tests already discussed and applied conscientiously IN THE RIGHT HANDS is perfectly capable of producing reasonable sized and speed mirrors (up to, oh, around 20" and f/4) that show textbook patterns under careful star testing. Some makers are going faster with it, and since I'm currently making a 14.7" f/3 via manual testing will share that experience soon enough. I could cite more examples of my own using other testing regimes. But this isn't speculation or argument-from-authority, it's observation from commercial production of many mirrors.
Bottom line as always is the testing tools are only as good as the person using them.
Best,
Mark
|
MKV
Carpal Tunnel
Reged: 01/20/11
|
Re: I have read all the posts about tests
[Re: Mike I. Jones]
#5310714 - 07/09/12 11:50 PM
|
|
Edit
|
Reply
|
Quote |
Quick Reply
|
|
|
Mike and Mark, after reading your posts I believe it's fair to ask Mike: Does your company use the Foucualt test, when and how often? And what about the rest of the professional community in general?
If this test is capable of such accuracy, then why spend money on more expenisve equipment? Apparently the Zygos were outdone by the good ole' knife-edge - every time!
I must say this is quite encouraging, Mike. I am throwing all my flats and Ross lenses out, and tomorrow I am going to build my plywood Foucualt tester with a $15 dial indicator and brass plumbing pipes. No doubt, after two dozen mirrors I too could be making 1/80 wave or better mirrors with my Foucualt tester. Thanks.
Mladen
Edited by MKV (07/10/12 12:07 AM)
|
MKV
Carpal Tunnel
Reged: 01/20/11
|
Re: I have read all the posts about tests
[Re: mark cowan]
#5310735 - 07/10/12 12:05 AM
|
|
Edit
|
Reply
|
Quote |
Quick Reply
|
|
|
Mark, here is a question for you: what good is 1/50 PVW mirror if it's secondary is only 1/10 wave and the f/3 mirror has (at least) 30% lienar obstruction? A 30% CO is enough to throw even the best mirror back to the 1/4 PVW, never mind the secondary!
What if the collimation is off by a smll amount at f/3? Why don't you tell us by how much can the collimaiton be off before the PVW is thrown out of the 1/4 wave range at f/3? What's the point of chasing shadows to 1/100 WRMS if the wavefront reaching the eye will be no better than a 1/4 wave?
Again, I love anecdotal stories and software simulation, but what Texereau did - published actual photographs of a finished telescope's wavefront reaching the eye under the nighttime sky - is worth a thousands stories. And, gee, it wasn't fantastic but 1/8 PVW - which is actually quite realistic and darn good, all things considered.
Thanks.
Mladen
|
cheapersleeper
Carpal Tunnel
Reged: 01/22/10
Loc: Sachse TX
|
Re: I have read all the posts about tests
[Re: MKV]
#5310835 - 07/10/12 02:25 AM
|
|
Edit
|
Reply
|
Quote |
Quick Reply
|
|
|
Quote:
Mike and Mark, after reading your posts I believe it's fair to ask Mike: Does your company use the Foucualt test, when and how often? And what about the rest of the professional community in general?
If this test is capable of such accuracy, then why spend money on more expenisve equipment? Apparently the Zygos were outdone by the good ole' knife-edge - every time!
I must say this is quite encouraging, Mike. I am throwing all my flats and Ross lenses out, and tomorrow I am going to build my plywood Foucualt tester with a $15 dial indicator and brass plumbing pipes. No doubt, after two dozen mirrors I too could be making 1/80 wave or better mirrors with my Foucualt tester. Thanks.
Mladen
Mladen, much of what you posted is immaterial. The difference between the professional environment and what an amatuer can do is volume, which equates to time, which equates to money. Under those conditions investment in expensive testing equipment makes perfect sense. The same logic applied the other way, is that a hobbyist likely has more time than money to spend on a project and will over time do relatively few pieces of optical work and therefore economically it would be unreasonable to spend huge amounts of money on equipment to test optics.
|
Mike I. Jones
Carpal Tunnel
Reged: 07/02/06
Loc: Fort Worth TX
|
Re: I have read all the posts about tests
[Re: MKV]
#5310843 - 07/10/12 02:43 AM
|
|
Edit
|
Reply
|
Quote |
Quick Reply
|
|
|
Hey Mladen, don't throw those flats out, send them to me! And if you can read zones repeatedly to 0.001" with your $15 tester, I'm throwing out my 2-axis precision tester in the pile! 
I'm assuming (I hope) your comments had smiley faces, too?
And I still need my phase-shifting interferometer, cuz it's much easier to test big plane-parallel optical windows in transmission than with a KE tester and 24" aperture 200" ROC sphere. And it's pretty much the only way to sensibly test 300mm aperture f/1.5 conics in DPAC.
That being said, the answer is yes, if the focal ratio of the optics system under test is above about f/4 or f/5, it is reassuring to cross-check the interferometer with a direct surface survey test, using both Foucault and caustic testing. These tests still have a place in the lab, just not for all optics.
Oh, and probably less than half of the optics I work on in my day job function like telescopes!
Mike
|
MKV
Carpal Tunnel
Reged: 01/20/11
|
Re: I have read all the posts about tests
[Re: cheapersleeper]
#5310848 - 07/10/12 02:59 AM
|
|
Edit
|
Reply
|
Quote |
Quick Reply
|
|
|
Quote:
<snip>The difference between the professional environment and what an amatuer can do is volume, which equates to time, which equates to money
That is true to some extent but not always. My point was that, according to Mike, his Foucualt Test beat Zygos hands down - every time!
The most logical question then is: Why not invest in automated Foucalt testers instead? Think how many Foucualt testers you can build for the price of one Zygo.
My historical approach was meant to show that the professional community didn't have as much confidence in the Foucualt Test and sought to replace it as soon as possible.
That's why it was decided to figure the 200-inch Palomar with a different test. Likewise the professional community chose not to use the Foucualt Tests or the Hubble Space Telescope primary. The fact that the HST primary turned out bad was due to the operator's error and not the test they used.
But one can't escape to question why did they not use the good ole' Foucault test, given that it pushes Zygos to their limits?
It also makes you wonder how come Texereau was happy that his 76-inch reflector delivered "only" 1/8 wave at the focus. Does that mean he should have made more mirrors to perfect the process? 
Regards,
Mladen
|
MKV
Carpal Tunnel
Reged: 01/20/11
|
Re: I have read all the posts about tests
[Re: Mike I. Jones]
#5310852 - 07/10/12 03:13 AM
|
|
Edit
|
Reply
|
Quote |
Quick Reply
|
|
|
Quote:
Hey Mladen, don't throw those flats out, send them to me! And if you can read zones repeatedly to 0.001" with your $15 tester, I'm throwing out my 2-axis precision tester in the pile! I'm assuming (I hope) your comments had smiley faces, too?
Sure, Mike. I am not as serious as people seem to take me. This is a hobby for all of us - but discussions should be rigorous on controversial issues. There is a lot of confusing information out there and only a good an open grownup discussion can hopefully sort things out - for the good of all. Of course, without attaching egos to issues or making it personal.
After all, ins't that what forums are all about - exchange of ideas?
Mladen
PS Oh, I was only kidding about throwing out my flats (I think I will use them as giant paper weights instead!), or about making a $20 Foucault tester (you know, splurging on the dial indicator and brass plumbing). Just adding a little drama to all this humor, that's all.
Edited by MKV (07/10/12 03:17 AM)
|
MKV
Carpal Tunnel
Reged: 01/20/11
|
Re: I have read all the posts about tests
[Re: Mike I. Jones]
#5310858 - 07/10/12 03:38 AM
|
|
Edit
|
Reply
|
Quote |
Quick Reply
|
|
|
Quote:
is reassuring to cross-check the interferometer with a direct surface survey test, using both Foucault and caustic testing. These tests still have a place in the lab, just not for all optics. Mike
Mike, I am surprised you didn't try figuring a mirror with a null test and then subjecting it to the rigors of interferometry and finally to the Foucualt test. This would be a way of determining how well null tests work, and what one can epexct from them.
Mladen
|
kfrederick
Carpal Tunnel
Reged: 02/01/08
|
Re: I have read all the posts about tests
[Re: MKV]
#5310965 - 07/10/12 07:36 AM
|
|
Edit
|
Reply
|
Quote |
Quick Reply
|
|
|
With a ronchi to check for astigmatism get one band and move it through RC and see it it rotates .A bad edge is very easy to see as well as zones and smoothness For me a ATM I use the ronchi first. Then if by the stroke of God it looks good . I Foucault check center to edge This should be twice the sag .[Fixed light] Then if that is close do knife edge readings . Then have a pro take a look .I do not like a fine grating I like 65 line .I would like to hear from the" Old Salts" how they do knife edge readings.
|
Mike I. Jones
Carpal Tunnel
Reged: 07/02/06
Loc: Fort Worth TX
|
Re: I have read all the posts about tests
[Re: kfrederick]
#5311007 - 07/10/12 08:22 AM
|
|
Edit
|
Reply
|
Quote |
Quick Reply
|
|
|
More comments on the Foucault test, all from my perspective:
1. As in musical instruments or any other learned skill, I do think you acquire an improved ability with the Foucault as you test more and more mirrors with it. There are VERY subtle shadings in the zone subapertures, and behavior that these shadows exhibit as the knife is cut into and out of the light cone, that may escape a beginner's eye. Within these slight nuances in subaperture shading and relative movement lies the ability to push the Foucault beyond what most literature will tell you. A beginner might overlook these subtleties at first, but then they begin to be very apparent, and the tester's accuracy improves accordingly.
2. It's odd that those of us that have sweated making enough mirrors to acquire that "brown belt" or "black belt" ability in Foucault testing, have to keep our data analysis hushed for fear of getting scolded by the general populace as lacking credibility. Unfortunately, the one device that can prove your test results to be accurate, a phase-shifting interferometer, really is too expensive for most individuals, or even clubs, to purchase. Sometimes you get lucky, like Ed Jones, and can find a used Zygo at a fantastic price, but that's very rare. Mastering the Foucault test is physically tiring, too. Hours spent hunting for the return spot and walking from there to Mars aligning the mirror and tester, hunched over a test bench, eyes getting tired, neck getting a crick, that &%^$% lamp getting hot (ahh the good old pre-LED days), arms getting tired while super-gently nudging the knife edge axes, probing and probing again into the light cone and watching the tiniest of details in the zone mask apertures. Then finishing a set of numbers, scooting the tester around on the table to try to remove bias from the numbers, and doing it all again, maybe five times or more. Wears a guy out!
3. The Foucault is by its nature a Fourier spatial filter (man were those French guys smart!). Thus, it must have access to a far field image, and a little room either side of it for defocus, to work properly. Ya can't Foucault test collimated light! A standard Fizeau (see, there's another smart French guy!) or Twyman-Green interferometer cavity is just the opposite: it can only test in collimated light. If an optical system forms a point source image at its far field, you have to have an auxiliary optic such as a return sphere to test it.
4. That's the beauty of the two tests: they offer highly precise testing in both focused and collimated light, two entirely different electromagnetic domains, and yet can agree to within centiwaves on a wavefront profile. The difference is that an interferometer analysis is but a mouse click away, while the Foucault or caustic test takes a long time to perform. The interferometer may be more expensive, but the time labor it saves is big.
Just my 20 millidollars,
Mike
|
Benach
Pooh-Bah
Reged: 01/24/08
|
Re: I have read all the posts about tests
[Re: Mike I. Jones]
#5311032 - 07/10/12 08:41 AM
|
|
Edit
|
Reply
|
Quote |
Quick Reply
|
|
|
Uhm, Mike, I assume you don't count time to build up your test bench setup for the Zygo? Especially if you have to wait for the time till the air is steady... Oh boy, I doubt it is much quicker than a Foucault.
|
DAVIDG
Post Laureate
Reged: 12/02/04
Loc: Hockessin, De
|
Re: I have read all the posts about tests
[Re: Benach]
#5311256 - 07/10/12 11:33 AM
|
|
Edit
|
Reply
|
Quote |
Quick Reply
|
|
|
Mike,
After reading your post what I find critcial is that you cross checked your Foucault results with other test method to show that your doing the test correctly. I've been doing scientific research of 27 years now and doing measurements of all types is what we rely on. We need highly accurate and highly precise data to draw conclusions from. Poor data and you can easily go down the wrong path. To have any confidence in the data we run standards both internal and external ones and cross check the results with other methods. That is what makes your result creditable as well, you have independent checks on your complete method of doing knife edge readings. That states YOUR doing the method correctly. It was refreshing to read your statement on doing the Foucault readings were you said that you took a set of reading then redid the test from start which including repositioning tester and starting all over. That shows that you have a really understand of precision and accuracy of the complete PROCEDURE in doing the test.
The lack of validating the results is what is lacking in most ATM made mirrors. Also one needs to understand that one cross check on one mirror doesn't mean that all the rest of them made will be of the same quality. The results for each mirror needs to be double checked to keep the Gremlins out. As you pointed out a Star Test is a great second test and requires no addition equipment just an understanding of how to do the test and read the results. What again is critical is that both methods need to agree if not one needs to find out why they don't.
All the Best,
- Dave
|
MKV
Carpal Tunnel
Reged: 01/20/11
|
Re: I have read all the posts about tests
[Re: DAVIDG]
#5311398 - 07/10/12 02:02 PM
|
|
Edit
|
Reply
|
Quote |
Quick Reply
|
|
|
What I find fascinating is that a technique that looks at four sample points produces a surface accuracy greater than a machine sampling 50,000+ points can establish.
Come on, that's astonishing! Especially in view of the fact that a spherometer machined and reading to a precision of 0.001 inch can still give results that are inches off on the radius if curvature and that we rarely if ever state the calibrated Foucualt test reading error.
But, then, all you need are three points to define a curve, and perhaps by the law of averages the more you work on something with random motion, the more the surface reaches an average smoothness across the whole area.
Mladen
|
Mike I. Jones
Carpal Tunnel
Reged: 07/02/06
Loc: Fort Worth TX
|
Re: I have read all the posts about tests
[Re: MKV]
#5311457 - 07/10/12 02:29 PM
|
|
Edit
|
Reply
|
Quote |
Quick Reply
|
|
|
Surface / wavefront profile errors on a smooth mirror are of very low aperture spatial frequency, which is why a 4, 5, etc. zone mask can do such a good job of representing the surface. Errors of higher aperture spatial frequencies, above 3-4 cycles/pupil, will be undersampled for sure, but then, you don't want those on a telescope mirror curve. A mirror with a TDE, multiple annular zones, and a deep central hole has a much higher aperture spatial frequency spectrum than a nicely blended, smooth conic that's slightly off in magnitude. The high-frequency zonal errors are blatantly obvious, drive up the RMS error relative to P-V, and of course shred the contrast in the image. You have to adjust your polishing strokes to blend out these high-frequency errors and achieve a nice smooth blended conic. Thus a zonal mask with only a few openings is enough to adequately sample the surface, provided (1) it is a smooth figure of revolution and (2) different zone masks are used that stagger zonal coverage and leave no zone untested.
Our 4D PSI has a 1.4 MPix fringe camera in it, which means that we might have more like a quarter million pixels spanning an interferogram aperture. But, if the surface errors are very broad, such as the case with a smooth conic, the curvatures in the fringes are thus very slight, and the fringes are highly oversampled. If, on the other hand, I was making and testing a strong hyperboloid with dozens of waves of spherical aberration, the fringes would be crowded together at very high aperture spatial frequency, and the full resolution of the camera would be needed to sample the tightly packed fringes without aliasing.
Helpful?
Mike
|
MKV
Carpal Tunnel
Reged: 01/20/11
|
Re: I have read all the posts about tests
[Re: Mike I. Jones]
#5311544 - 07/10/12 03:26 PM
|
|
Edit
|
Reply
|
Quote |
Quick Reply
|
|
|
Thanks, Mike, as always. I never realized a surface can be described with too many points!
On a related subject, what is the limiting size of wavefront error detectable in a Foucualt test? It seems to me that the successful production of very accurate mirrors is determined by the law of averages more than by any objective measure.
Then, of course, one must ask why strive for such perfection, given that most commercially available secondary mirror surfaces are not much better than 1/10 PVW (1/5 waves on the wavefront), and that central obstruction, decollimation and other wavefront-degrading obstacles in a telescope have their effect in a completed OTA (not to mentione atmospheric conditions). In other words, other than bragging rights , what other advantage does one get from 1/80 PVW mirror?
Mladen
|
Starman1
Vendor (EyepiecesEtc.com)
Reged: 06/24/03
Loc: Los Angeles
|
Re: I have read all the posts about tests
[Re: MKV]
#5311842 - 07/10/12 07:13 PM
|
|
Edit
|
Reply
|
Quote |
Quick Reply
|
|
|
Let me emphasize the star test as an arbiter (NOT the only one because eyepiece optics are seldom accurate enough to reveal ALL the truth) of good optics.
I find that 1/8 wave on the wavefront (both mirrors icluded) is easy to see in good seeing, and even smaller amounts of astigmatism are visible.
It is a VERY rare mirror that doesn't show ANY astigmatism in the field. I have read all the explanations of why astigmatism may be subtracted in the lab interferometer testing, and it is reasonable. But I wonder if star testing may not be MORE accurate at exposing astigmatism than tests in the lab?
Also, very small amounts of surface roughness (micro-ripple) are easily visible in the star test.
And, after gaining more experience with it, very small amounts of over/under correction.
But, I'll bring up a point, which is: What is the purpose of testing in the lab? The answer is to produce an optic that works well in the field. And there are very few lab technicians that can produce an optic in the lab that yields near-perfect field results in the star test. Fortunately, perfection isn't necessary to produce an astoundingly good image in the field under good conditions.
One mirror manufacturer I know (he's a friend) relies heavily on the star test to tell him when the mirror is done. He sometimes see things in the star test that weren't revealed completely in the lab. The difficulty, of course, is that seeing and cooling play a big role in what is revealed, not to mention mechanical stresses caused by the structure.
At Parks, I got to see a KE test done on an 8" mirror that revealed a superior optic and VERY smooth surface. Yet, in the field, the star test showed a slight astigmatism. Rotating the mirror also rotated the astigmatism. It went back to the bench.
I have found this thread illuminating (pun intended), but it seems to me the easiest test for the end user to use and master is the star test. And that, I believe, is the point of all the forms of testing.
One last thing: I hear so many people mispronounce Foucault, I'll bring up the fact it is pronounced "Foo-koh' ", with the l and t silent.
|
|
9 registered and 14 anonymous users are browsing this forum.
Moderator: ausastronomer, richard7
Print Thread
|
Forum Permissions
You cannot start new topics
You cannot reply to topics
HTML is disabled
UBBCode is enabled
|
Thread views: 8807
|
|
|
|
|
|
|
Previous
Index
Next
Threaded
