You are not logged in. [Login] Entrance · Main Index · Search · New user · Who's Online FAQ · Calendar

Equipment Discussions >> ATM, Optics and DIY Forum

Pages: 1 | 2 | 3 | 4 | 5 | 6 | 7 | (show all)
MKV
Carpal Tunnel

Reged: 01/20/11

Re: Herrig [Re: DAVIDG]
#5566204 - 12/11/12 04:10 PM

Dave, yes of course your concic constant must be huge to be perceptible on a very long focus surface. The problem is that when the surface appears spherical it may be quite a ways off anything close to spherical. The reason is the limiting resolution of the test method.

Take, for example, an 8-inch mirror with a radius of curvature of 480 inches, and a focal length of 240. It's an f/30 mirror. If it is a parabola (cc = -1), its sagitta is 1/60 of an inch or 0.01666667 inches. If it is a perfect sphere, its sagitta is 0.016666957. The difference between a parabola and a spehere of that diameter and focal ratio is 0.0000003 inches (3^-7), or 1/74 wave, which means that the resulting wavefront error would be 1/37 wave.

Do you really think you can see that difference, given the air currents, and numbers of glass surfaces which themselves could have additional errors? Do you have a k-e stage that can record such difference?

Assuming autocollimation and testing at the focus (240 in) instead of RoC, you are still talking about a difference well beyond common mechanical measuring devices.

Now, let's assume the mirror is not really a paraboloid but a slight prolate ellipse (say, cc = -0.25). The sagittal difference between the two wouldbe even smaller, 0.0000001 (1^-7) or about 1/200 wave. Then on relfectionn the wavefront error difference would be 0.0000002 or 1/99 of a wave.

There is just simply no way I can see how such surface can be reliably figured to a sphere no matter what method you use Maybe I am missing something.

The good thing is that OSLO tells us that, even if the Herrig mirrors depart from sphericity by substantial amount, the system would still perform well.

 Post Extras:
MKV
Carpal Tunnel

Reged: 01/20/11

Re: Herrig [Re: Mark Harry]

Quote:

MVK, it almost looks like if you deformed one or both of the optics somewhere between (less) than what you did above, it could make it better than an all-spherical scope....?
M.

Not really, Mark. Here is one with the primary deformed to an oblate ellipsoid (cc = +0.2), and the secondary to a prolate ellipsoid (cc = -0.3)

The performance is the same as if they were spherical. For some reason, the Herrig seems to be resistant to any conic deviation beyond any reasonable expectation.

 Post Extras:
MKV
Carpal Tunnel

Reged: 01/20/11

Re: Herrig [Re: MKV]

And here is the same configation except both mirrors are extreme hyperboloids (cc = -3.0), and then refocused!

 Post Extras:
DAVIDG
Post Laureate

Reged: 12/02/04

Loc: Hockessin, De
Re: Herrig [Re: DAVIDG]

I worked out a way to test the convex primary mirror for the 166mm design I posted. Using this test method one can get a direct look at the surface of the convex primary. One would need to make an addition 8" f/7 concave mirror and then figure it to an ellipse that has a conic of -0.76. One be could off by +/-0.05 in the conic. You then set up the mirror like your doing a double pass autocollimation test but instead of a flat you use the convex primary mirror from the Herrig as the "flat". The spacing between the two mirrors is set to 80" but it can be off by a few inches and it doesn't make much difference.
One uses a small Newtonian diagonal placed at 45 degree that isn't shown in the drawing of the test layout. So the optical test layout looks like a type newtonian facing the convex primary. The Foucault/Ronchi tester is off to one side. So light from the tester is directed at the 45 degree mirror, then to the 8" f/7 concave mirror then to the long radius convex mirror, back to the 8" f/7 and back to the tester. Both the distance from the tester and too the 8" f/7 mirror and back from it the mirror to focus point are the same. So you use a tester were the light source and knife edge/Ronchi screen are coplanar. This forces them to always be the same and at the correct spacing. So you don't have to measure this. You just move the tester until you find the focal point. Now you figure the convex mirror until the system nulls like a sphere being tested and shows straight Ronchi lines. If the 8" f/7 is smooth, the conic is close on it's surface and one achieves a null condition by figuring the convex surface, the convex surface will be smooth and actual figure will be no worse then mildly aspheric with a conic value of 100 or less. If you plug a conic of 100 into OLSO for the convex primary one gets a PV value of 1/10 wave for the complete Herrig system. Too bad and now one has chance to make to optics so the actual telescope performs like the theorical one.
- Dave

 Post Extras:
DAVIDG
Post Laureate

Reged: 12/02/04

Loc: Hockessin, De
Re: Herrig [Re: DAVIDG]

Here is the OLSO file.

- Dave

 Post Extras:
MKV
Carpal Tunnel

Reged: 01/20/11

Re: Herrig [Re: DAVIDG]

DaveG, I love the null test, except that it is a null no matter what conic constant you assign to the f/51.4 primary. Here is your test except with the primary deformed to a hyperbola whose conic constant is -12! The off-axis performance actually improves the more hyperboloidal the mirror becomes! You don't even have to refocus.

I think this only confirms my skpeticism about knowing when your very very long radius mirror is actually a sphere.

Trying to get it to be spherical simply doesn't change anything, even if you could determine when such a mirror actually is spherical.

Added: Therefore, I'd say polish the convex mirror, center over center, until polished out - its final figure will surely fit the super generous conic tolerance envelope. I don't think any fancy tests are really necessary. What this boils down to is that the convex mirror really doesn't have to be spherical at all.

Regards,

Edited by MKV (12/12/12 01:59 PM)

 Post Extras:
Mark Harry
Vendor

Reged: 09/05/05

Loc: Northeast USA
Re: Herrig [Re: MKV]
#5567035 - 12/12/12 06:29 AM

Mladen, I think I see why there is a problem you are having understanding how to detect such a long sphere.
You figured the sag of the paraboloid as being .0167"- which also represents the movement of th KE stage with moving source. (rounded off) Fixed source would be .0334"
A perfect spherical surface would be 0.000" on the KE stage. (~1/32" difference, or about 3/4mm)

I have a good (fixed source) KE rig made with a lathe cross-slide, dial indicator, and KE with no slop or backlash. At 200" ROC, I can detect/repeat measurements to .002" on average, and often hit measurements well under that. A bit less than 480" is the limit of the length of my cellar, but occasionally I get contorted and make a reading or two at the major part of that length depending on what I'm doing. Never at any time has this sensitivity exceed + - .005"; within .010" when that far out there; depends on exactly what I'm testing, and how large it is. That puts the mirror you use above, being tested with around 1/100th wave sensitivity. These are conservative parameters. last mirror I tested that had a 200" radius, I was hitting the measurements to .001".
*******
I just made a rather large 8" convex surface, and used a 4.5" dummy-shined TP to check the regularity. The radius was 39.125" and is the front surface of a positive meniscus. The TP had a ripped edge, but the central half was a really close slightly concave match to the meniscus surface in question. I used this area to screen the surface. It took about 20 minutes to make the TP, and about 80-90 minutes to pitch polish the meniscus surface smooth and with no edge problem. (-ZERO- tde) The central area of the TP remained a pleasing and accurately round bullseye, anywhere on the meniscus, from center to edge. If there was any edge problem left, the round bullseye would have adopted the shape of a lightbulb for a moderate edge problem, and a horseshoe, or 'omega' symbol for a bad edge. The rest of the meniscus surface is beautiful.
The TP? I don't really care about being all that fussy with it. Later on, it will be turned into something else; so I'm not about to waste a bunch of time to dream up all kinds of tests, or spend a lot of time worrying about the TPs edge. I put the "horsepower" where it belongs- on the meniscus. This makes perfect sense considering I'm not new to contact IF testing- saves me a ton of needless work, most certainly!
M.

 Post Extras:
MKV
Carpal Tunnel

Reged: 01/20/11

Re: Herrig [Re: Mark Harry]
#5567538 - 12/12/12 12:52 PM

Mark Harry, my understanding is that a real difficulty with Foucualt testing a mirror at f/30 or slower, such as used in the Herrig configurations, is in the lack of intensity of the shadows. As you undoubtedly know, the intensity of the shadows increases with faster focal ratios, and decreaes with slower. This is proportional to the optical path difference (OPD) of the wavefront. The smaller the OPD the fainter the shadows.

In my example, I used an 8-inch mirror of RoC = 480 inches (40 ft), f30, and David G's configurations calls for an 8-inch mirrors of RoC ~ 823 inches (68.6 ft), f/51. Both mirros have extremely small OPD values even if parabolic in profile, and therefore would present very faint shadows, if not indsicernible shadows, even if we assume all other conditions, such as air currents, etc) were controlled during testing.

At conic constants closer to the sphere, say cc = -0.25 instead of -1 (parabola), the shawods would become even more difficult to discern. The only rational conclusion from this is that it would be difficult to know if your mirror is truly spherical or not. But, the good news is that in a Herrig it really doesn't matter. The configuration has a very generous tolerance range when it comes to that. Any well polished Herrig convex mirror would probably fall well inside the tolerance envelope.

You describe much faster surfaces, whose shadows could still be clearly discerned. At f/10, or f/12, the shadows can easily be seen under the k-e. At f/30 with great deal of diffculty, if at all, and at f/51 I seriously doubt it.

Regards,

Edited by MKV (12/12/12 01:47 PM)

 Post Extras:
Mark Harry
Vendor

Reged: 09/05/05

Loc: Northeast USA
Re: Herrig [Re: MKV]
#5567656 - 12/12/12 02:03 PM

All I can say, change your test, or modify it to get the -REPEATABILITY- down to the range where you can be sure what it's telling you about error and what type to ascertain with any reliability what you're saying or assuming is correct. I've told you what a reasonably done measurement in relation to ROC can reveal.
******
In the past you've advocated accepting 1/4-1/2" wide error as being acceptable for edge error. In IFgrams, you've "verified" reductions of .33 wave by XP and ZEMAX as 2 fringes (direct observed) error. Do you deny this????? Do we have to discuss this further????
I think that most comments by you should be taken in context with your actual capability and assessment of what you -THINK- is the "real world" so to speak.
*********
If you cannot get your mesurements reasonably close to what I state, then I think you should refrain from pontifying anything about test procedures, accuracies or conclusions; for they certainly are not legitimate, or correct in any sense.
Maybe you should stick to selling aspirin? Just a thought...
At any rate, I view your past posts as minimizing the accuracy necessary to obtain a respectable performing optic, or to minimize the accuracy required to be considered legitimate. What -IS- the case???
M.
**********
If I could get one of you guys to get me a -TOLERANCE- on the 5% I requested (now 3 times!) I am sure if you supplied me the glass and that tolerance is acceptable, I could get you an optical set within 1 week, and I could enjoy my cup of 'joe' in the morning, and not feel rushed, one bit; and it would be a shining example of what this scope design would offer. End of story.
********

The A36-(forerunner to the P51 Mustang) was designed, built and flown in 117 days. The rate this discussion is going, there would be no hope of that ocurring at all- a shame, and a bow to EGO- an expensive mistress!!!!
Cordially,
M.

 Post Extras:
mark cowan
Vendor (Veritas Optics)

Reged: 06/03/05

Loc: salem, OR
Re: Herrig [Re: Mark Harry]
#5567673 - 12/12/12 02:13 PM

I've got some glass sitting around. What version of this would you like to do? Somebody get him the 5% please!

I like the TP approach - was trying to figure out the easiest way to test one of these suckers but that didn't occur to me.

Best,
Mark

 Post Extras:
Mark Harry
Vendor

Reged: 09/05/05

Loc: Northeast USA
Re: Herrig [Re: mark cowan]
#5567735 - 12/12/12 02:34 PM

TP verification has been in practice for over 100 years or so!
Hi there, Mark!
M.

 Post Extras:
Mark Harry
Vendor

Reged: 09/05/05

Loc: Northeast USA
Re: Herrig [Re: Mark Harry]
#5567787 - 12/12/12 03:03 PM

Heaven forbid I ever think of a shortcut to lessen the work load with TP mfr, or testing procedure. I guess I should understand, most folks that haven't had to ever think of this stuff, might be a bit slow on the uptake on what I'm suggesting.
M.

 Post Extras:
Gary Fuchs
Pooh-Bah

Reged: 05/22/06

Loc: Easton, PA, USA
Re: Herrig [Re: Mark Harry]
#5567895 - 12/12/12 04:18 PM

Mark (H),

Didn't Dave O. answer your tolerance questions in posts #5565470 & #5565928?

My take on what MKV said about testing the really long radius spheres needed for this design was that it was a matter of not being able to resolve the shadows sufficiently to tell if you really had a null or not (or I guess Ronchi line straightness); not anything to do with repeatability. I'm currently working on a 313" or so radius convex surface and it's not easy (for me) to accurately determine the null/line straightness at that distance--at least compared to something in the under 100" range.

(Darn, now I'm thinking about accuracy vs precision and wondering if I got it right?)

As a side note, I'm surprised to see you using the term "pontifying". I would have thought you would be more sensitive about that sort of thing. I'm assuming you mean 'pontificating' but if not and "pontifying" is actually a compliment I withdraw my comment...

Now the "aspirin" remark sure sounded like the dreaded snark to me...

Gary

 Post Extras:
MKV
Carpal Tunnel

Reged: 01/20/11

Re: Herrig [Re: Mark Harry]
#5568580 - 12/13/12 01:43 AM

Re: this is about Mark Harry's passionate (putting it mildly!) response to my skepticism about being able to test an f/30 or f/51 Herrig-type mirror using the Foucualt knife-edge method and obtaining "repeatable" results.

Mark Harry: As I sad to you privately, I am the old school, and the old school taught me that, the slower the mirror, the fainter the shadows. This is ATM 101.

Note: 200 mm diameter paraboloid (cc = -1), tested at RoC, compared to 200 mm sphere also tested at RoC.

Mark, you can talk the talk, but I believe the pictures, because there is good science behind them. Unfortunately, based on your post, you don't seem to have that science down. I find that odd.

Even under ideal conditions, no air currents, etc., at about f/20, the parabolic shadows are so faint they are difficult to discern. At f/30 and f/51 they are indistinguishable from a sphere.

So, then, Mark, what does your "repeatability" have to do with this?

Better yet, please explain, what measurements are there to "repeat" in a clean null? Your precision x-y table is not going to help you measure something you can't see!

This is so basic, so fundamental that it's actually embarrassing to me to have to teach you this, because you advertise yourself as an optical expert. Strange, isn't it?

Just how basic is this? Well, it can't get any more basic than this. Please, open Albert G. Ingalls' Amateur Telescope Making, Book One, the 1978 reprint edition (small red hardcover book), and turn to page 10 (yes page 10, it's that basic!).

There is a drawing of four shadowgrams. If you look at shadowgram F, listed as "Long Paraboloid", you seen barely perceptible shadows! Bingo! Anyone who has read ATM Book One knows this. So, then, please explain why you don't?

As I've already mentioned to you, the intensity of shadows is proportional to the optical path difference (OPD) between real (deformed) and ideal (perfectly spherical) wavefronts.

From this we know that the slower the focal ratio the smaller the OPD and therefore the fainter the shadows will appear. The actual and ideal wavefronts will approach a point of no distinction.

At this point the sphere and the parabola will look the same in a K-E test, i.e. a "flat" disk disk, with longitudinal travel of exactly 0.0000". There is NOTHING TO REPEAT, Mark!

So, based on this fact, I ask you, with regard to DAVIDG's f/51 Herrig convex mirror: Given the extremely small OPD of the surface at that focal ratio, how will you ever know the surface is spherical, oblate, prolate, parabolidal or even hyerbolodial? Simple: You won't (and can't)!

 Post Extras:
mark cowan
Vendor (Veritas Optics)

Reged: 06/03/05

Loc: salem, OR
Re: Herrig [Re: MKV]

The slowest mirror I've tested so far was a 10" CA f/18 mirror spec'd as a "paraboloid". I did a companion sphere at f/9 which was easy to test with a Foucault derived Strehl of 0.993 - and P/S IF supplied real Strehl of 0.994. The longer one was not so easy, but quite doable. The client (ITT Space Systems) kind of laughed over the paraboloid spec; indeed...IIRC the figuring time was about 3 minutes total.

In any case...interestingly enough these simulations show the shadows would be resolvable with imaging - I stretched the contrast to bring out the real detail in the bottom row.

Best,
Mark

 Post Extras:
MKV
Carpal Tunnel

Reged: 01/20/11

Re: Herrig [Re: mark cowan]

Mark Cowan: Interesting indeed. Nice. I suppose a suitable camera could be mounted behind the K-E and push the contrast/brightness to maximum. Unfortunately, an 8-inch f/51 mirror has a radius of curvature of 68 feet. That represents a serious problem with air currents and even finding such a testing facility.

If the bottom row were resolvable for a parabolid, anything with a smaller conic constant then -1 would reach a null way before f/51. Say you have a 54% paraboloid, cc = -0.2916. IN which case you probably couldn't reoslve the anything even with a higher contrast. The picture below shows an 8-inch f/51 mirror cc = -0.2916, tested at RoC, "raw" and enhanced constrast and brightness.

It would also be necessary to make a Foucaugram of an actual f/51 or so mirror and see if the photograph could be enhanced the same was as the simulated image. In other words, the program may by itself form a pattern (artefact).

DAVIDG's solution is actually ideal as far as the long radius testing is concerned and all the problems with air currents are concerned. It cuts the whole test down to a neat 80-inch package, plus double precision, and a clean null.

Unfortunately, for a Herrig there really isn't any good reason to go through that trouble because the performance is not at all affected even if the mirror's figure is way off. That mirror would more likely then not be within the conic constant tolerance envelope without figuring, as soon as it was polished out.

Regards,

Edited by MKV (12/13/12 04:06 AM)

 Post Extras:
Dave O
sage

Reged: 12/21/11

Loc: Sri Lanka
Re: Herrig [Re: MKV]
#5568663 - 12/13/12 04:28 AM

Quote:

Unfortunately, for a Herrig there really isn't any good reason to go through that trouble because the performance is not at all affected even if the mirror's figure is way off. That mirror would more likely then not be within the conic constant tolerance envelope without figuring, as soon as it was polished out.

Not sure that I'd call it 'unfortunate' ... perhaps I'm missing something?

Any rate, the 'problem' is still the one about "testing' that long ROC CX primary -- it does need to be smooth, it should be somewhat 'spherical', and it needs to be in the 'ball park' of the desired radius.

As to Mark H's question(s) on "tolerance"; for the 820" ROC primary of Dave G's 6.5" f/18, anything +/- 5% (41") is 'good enough' -- just plug in the high and low end of the range into OSLO, refocus, and you'll see very little changes -- even w/o re-optimizing the secondary ROC or tilts and separation.

Edited by Dave O (12/13/12 04:39 AM)

 Post Extras:
MKV
Carpal Tunnel

Reged: 01/20/11

Re: Herrig [Re: Dave O]
#5568715 - 12/13/12 06:23 AM

Quote:

Not sure that I'd call it 'unfortunate' ... perhaps I'm missing something? Any rate, the 'problem' is ... it does need to be smooth, it should be somewhat 'spherical', and it needs to be in the 'ball park' of the desired radius.

You got me there, Dave O, perhaps 'unfortunate' wasn't the best choice but I was trying to be sarcastic (considering how much fuss has arisen over a surface with extremely "fat" tolerances).

Generally speaking, proven polishing techniques should result in a smooth surface and if figuring is not critical then the mirror could be ready as soon as it polishes out.

The radius should not be problem, given that its sagitta is well within micrometric range (~0.01 inch).

I have also shown in the spot diagram above that even at cc = -12, the performance doesn't suffer, so I wonder what is meant by: the mirror needs to be "somewhat spherical". A mirror with cc = -12 is nowhere near the sphere, and yet the sport actually look better off axis then with a shpere.

Regards,

Edited by MKV (12/13/12 06:29 AM)

 Post Extras:
Dave O
sage

Reged: 12/21/11

Loc: Sri Lanka
Re: Herrig [Re: MKV]
#5568916 - 12/13/12 10:06 AM

Quote:

I have also shown in the spot diagram above that even at cc = -12, the performance doesn't suffer, so I wonder what is meant by: the mirror needs to be "somewhat spherical". A mirror with cc = -12 is nowhere near the sphere, and yet the sport actually look better off axis then with a shpere.

Actually at the given radii and aperture, a CC of -12 is still pretty close to a sphere ... you could increase the CC of the primary on Dave G's 6.5" f/18 to -100 and still get diffraction limited spots .... By "somewhat spherical", I was simply implying a surface of revolution that is close enough to a sphere to produce a diffraction limited image -- whatever that CC tolerance happens to be for the particular design.

 Post Extras:
MKV
Carpal Tunnel

Reged: 01/20/11

Re: Herrig [Re: Dave O]
#5569190 - 12/13/12 12:43 PM

Quote:

By "somewhat spherical", I was simply implying a surface of revolution that is close enough to a sphere to produce a diffraction limited image -- whatever that CC tolerance happens to be for the particular design.

In this particular case that is absolutely true, Dave O.

 Post Extras:
Pages: 1 | 2 | 3 | 4 | 5 | 6 | 7 | (show all)

Extra information
10 registered and 22 anonymous users are browsing this forum.

Moderator:  ausastronomer, richard7

Forum Permissions
You cannot start new topics