44 replies to this topic

[ngc4910]

Hello!

 

I had long planned to optically test my star diagonals -- for no other reason than that I believe this could be something which I would have fun doing. I have been very curiously reading the threads on star testing and DPAC that I could find here, and in this thread I plan to document my progress, and ask questions along the way.

 

As a very first step, I'd like to attempt contact testing with an optical flat, simply by placing a few of the old, cheap prisms I own on a clean optical flat, wait till fringes form under monochromatic light, and see how different the patterns are between the different diagonal prisms.

 

For this task, I bought and old and somewhat abused optical flat, 4 inch in diameter, labelled "accuracy +- .0000010" , which I think translates to 1/20 wave error. The flat has some scratches and a chip, the figure may hence not be accurate any more. But, I am just trying to learn here, and this flat should provide an affordable attempt for me to find out if I do enjoy indoor optics testing at all.

 

 

The first obstacle is a source for monochromatic light. A low-pressure sodium lamp is not an option given the cost, and (at least locally) I only found LED energy-saving light bulbs but no CLF blubs, though I could order a CLF bulb online somewhere.

 

I wanted to ask, has someone tried the combination of a green LED and a Baader "Solar Continuum" filter?  The filter has a band width of 10nm ; the green led likely around 20-30nm ;  if those are slightly offset, that could make for an even narrower band width, perhaps sufficient to see the fringes? Did anyone try this by any chance already? (I do have the filter, I'd just need to obtain a bright-enough green LED).

 

Cheers

ngc

 

 

[Arjan]

"The first obstacle is a source for monochromatic light."

What about a laser diode?

[walt r]

I use a Green Laser pointer that is defocused (wide spot) and a ping pong ball diffuser (hole in ball to stick the Laser into).

[BGRE]

A green LED and a 10nm bandwidth filter result in a coherence length of around 50 waves or about 25 microns which corresponds to an airgap of 12.5 microns and you would want it to be somewhat smaller to ensure adequate fringe contrast. This is likely impractically small.

 

A laser diode can be used as even multimode laser diodes have coherence lengths of around 300 microns or so, but an efficient illumination system will be required or a high power diode will be required to achieve the same illumination level that can be achieved with a compact fluorescent or similar source.

 

In contrast it's much easier to use a laser diode source to view Haidinger fringes (fringes of equal inclination) from the airgap bounded by the two nominally flat surfaces being compared.

[ngc4910]

What about a laser diode?

 

 

I use a Green Laser pointer that is defocused (wide spot) and a ping pong ball diffuser (hole in ball to stick the Laser into).

 

 

A laser diode can be used as even multimode laser diodes have coherence lengths of around 300 microns or so, but an efficient illumination system will be required or a high power diode will be required to achieve the same illumination level that can be achieved with a compact fluorescent or similar source.

 

 

 

Excellent, thanks for the feedback. I had read about the table tennis ball + green laser. Would any green 1 mW (Class 2) laser do, or is a more powerful diode necessary when diffusing? If you have a specific product that works for you, I'd appreciate a message with your recommendation.

Cheers

[BGRE]

The laser power captured by the eye somewhat below 1 microwatt is required.

 

Stationary diffusers can generate speckle.

 

Without a large area secondary diffuser how exactly does a small area diffuser like a table tennis ball ensure that light from the entire flat enters the eye/camera?

 

The problem can be seen by raytracing from the eye to the flat and back to the source.

Without any illumination optics rays from the edge of the flats will miss the source if it isn't large enough (at least twice the size of the flat depending on the geometry of the illumination and viewing setup.

[Mike I. Jones]

Here's an old optics shop lesson:  Buy a box of Kim Wipes or other lens cleaning tissue.  Place a Kim Wipe over the flat, then put the test part on the Kim Wipe.  Holding onto the part, gently pull the Kim Wipe out from between the part and flat.  That will allow them to settle into contact with danger of scratching.

 

Thorlabs sells narrow-band LEDs that work nicely.  The Thorlabs LED560L emits at 562nm with 11nm FWHM.

 

https://www.thorlabs...ctgroup_id=2814

Edited by Mike I. Jones, 03 December 2024 - 06:48 PM.

[BGRE]

That LED is likely a Marktech Optoelectronics MTE2056N-UYG.

Mouser's list price is significantly lower than that of Thorlabs.

 

With a small source like that simple illumination optics will likely be required.

[DAVIDG]

 There is an easy way to make a large monochrome light source for testing.  You need a  BL type fluorescent  bulb ( NOT BLB) which is an  uncoated fluorescent black light blub,  and a  piece of green plastic. The BL bulbs have a strong green emission from  Mercury at 546nm and the green plastic will filter all but that source out.

   This how my commercial Unilamp monochrome source works.  You can see it in the upper part of the attached picture. It uses two small BL fluorescent tubes a piece of white diffusing plastic and a piece of green plastic to filter out the all but the green emission.

 

               - Dave 

 

https://www.amazon.c...363308990&psc=1

 

 

 

[BGRE]

My large area Mercury green line source used a ring style BL bulb mounted in a large die cast box with a 6" diameter yellow optical glass window/filter from Surplus Shed. The interior of the diecast box is painted matte white and the ring fluorescent doesn't directly illuminate the following white acrylic diffuser. This arrangement avoids hot spots due to proximity to the lamp and ensures that most of the light produced exits through the filter.

 

I also have a small Neodymium glass filter that can be used in front of the eye/camera.

Neodymium glass has a series of narrow passbands one of which includes the Mercury green line.

Additional colored optical glass filters are used to block the unwanted transmission bands.

The filter components are bonded together with Canada Balsam as UV cured optical cement cannot be used.   

Edited by BGRE, 04 December 2024 - 04:48 PM.

[ngc4910]

I managed to source a CFL bulb and a green camera filter locally, and got see the first "fringes" today.

 

As a first test, I simply laid the prism flat on top of the optical flat and let it sit there till fringes formed

 

 

Then, I inverted the setup, and placed the flat on top of the hypotenuse. Unfortunately, I did not manage to get fringes as wide as previously (two pictures, one straight through the camera, the other holding the green filter in front of the camera lens)

 

 

 

What potentially appears as dust in the pictures are tiny tiny air bubbles within the optical flat. Now, I've no clue if the curvature in the fringes is due to the prism, or due to the flat (which, rated at 1/20th wave when new, has a chipped edge which may impact the figure).

 

 

[Arjan]

Can you test on top of the clamshell? I would expect the fringes to be stronger curved if the flat has been affected.
Also see if the curves under the prism are appr the same shape on different locations. If so, it's pobably the prism that's unflat.

Edited by Arjan, 09 December 2024 - 02:44 PM.

[jelloptic]

Good start, Mr. NGC.... so you've generated fringes and taken some respectable photos of the set up.  You've "floated" one optical surface on another that are both sufficiently clean/dust-free to allow a thin air wedge to develop and show "tilt" fringes.  If both surfaces were perfectly flat, these fringes would be straight, and the number of fringes is in proportion to the wedge or tilt between the surfaces.

To answer your question, the fringe curvature is due to the combination of both surfaces; the trick is to decouple these... is one at fault, or the other, or is the error shared between them in some proportion?  But for now, let's assume that your master flat (1/20th wave) is free of curvature (ie, perfectly flat).

 

Your next assignment is to determine the sign of the curvature... is it concave, or convex?   This is readily done by light pressing upon the optical flat to see which way the fringes move.  In your last pic, if you pressed the flat at ten o'clock (near the fracture) ... and the fringes fanned out towards the left and became fewer in number, then you are reducing the wedge (tilt)... then the curvature would indicate that the prism has a convex curve relative to the flat.  I suspect this to be the case for another reason... notice that the right-side fringe is very black, and as you go to the left, they get less black and more colored.  The black fringe indicates the thinnest line (apex) of the air wedge.

 

So play around with your set up to see how the fringes react/move. Of course, keep surfaces clean, float one along the other, don't forcefully push one on the other, and don't press too hard on the black fringe, as you may go into optical contact and get them stuck together.

 

...just passing experiences on, hope this is helpful.   kbl 

[Oberon]

Terrific thread! 

[BGRE]

If the camera/eye is too close to the airgap, fringe curvature can occur even if the surfaces bounding the airgap are flat.

The OPD is 2*t*cos(theta) so if the angle of incidence varies too much across the airgap of thickness t then the fringes will be curved rather than straight.

[davidc135]

There's a distinctive kink in the fringes and so, as Arjan says, moving the prism around will show which surface is responsible. Probably on the prism.

 

David

[italic]

I'm glad you started showing fringes, as I built my own flat box last year and have some flats laying around to show it. I've put all these images with short descriptions on imgur if you like, but I've included them here for everyone.

 

My box is modeled after a helium emission box at the local workshop, but less sophisticated (ie no beam collimator fresnel lens). I use the green mercury emission line from a standard 26W CFL tube. In the first image you can see the green filter acrylic, which is sitting on a sheet of diffuser acrylic. The lamp fixture is a standard clip lamp from the hardware store with the clamp removed. It simply sits on top of the box and I keep it from sliding around with the acrylic pucks.

 

 

Below the diffusion sheet I have a beamsplitter made of acrylic and half-mirror film. The sheet is supported across three of four edges to prevent warping under weight, but it's not the best build. I painted the box matte black to reduce stray light and increase contrast. In the second photo, you can see a reflection of the floor in the splitter.

 

 

Third image is with the light on. You can see the diffuser above and some light leaking around the splitter.

 

 

Fourth photo is where you should be viewing or imaging your test from. The splitter is there to give you a vertical view into the test glass to reduce parallax, which BGRE has described above. Viewing from some distance gives you more accurate fringes across the whole surface instead of just in the center.

 

 

This image is taken from inside the box directly viewing the glass, showing clearly curved fringes and reflections from the environment. This is how you should not be measuring fringes.

 

 

Lastly, the same configuration of glass (untouched from previous image), but taken from directly above through the beamsplitter. This shows nearly perfectly straight fringes, which I expect from two 1/10-wave flats together. You can see I didn't get perfectly centered over the glass, but it's close enough to show the difference. If you want, you could put this fringe image into DFTFringe and analyze how flat your flats are.

 

 

Hopefully this helps you going forward. Making the box was pretty simple, if more money than I thought it would be. It's still cheaper than finding an excitation tube and variac, and the lamp is readily replaceable with a new lamp should it burn out or break. It's not perfect, but it works pretty darn well for what I needed it for (testing a secondary mirror).

[Oregon-raybender]

One suggestion. Shot mono or B&W. The sensor of the smart phone or camera

has more green pixels, which can be enhanced easy to bring out the fringes

more. 5-7 fringes are enough to tell how good the surface is. I made a sting

(thick thread, black) harp to aid in looking at the straightness of fringes (think a archery bow)

Lay the string on the test flat and read the fringe bow-ness(?).

 

Starry Nights

 

I took your image to B&W and sharpen.

 

 

Attached Thumbnails

• 

Edited by Oregon-raybender, 09 December 2024 - 11:24 PM.

[italic]

One suggestion. Shot mono or B&W. The sensor of the smart phone or camera

has more green pixels, which can be enhanced easy to bring out the fringes

more. 5-7 fringes are enough to tell how good the surface is. I made a sting

(thick thread, black) harp to aid in looking at the straightness of fringes (think a archery bow)

Lay the string on the test flat and read the fringe bow-ness(?).

 

Starry Nights

 

I took your image to B&W and sharpen.

A string is a great idea. I'll have to make one and put a hook on the box to store it. If I ultimately decide to try to reflatten this secondary, that'll be very handy while working.

[italic]

One suggestion. Shot mono or B&W. The sensor of the smart phone or camera

has more green pixels, which can be enhanced easy to bring out the fringes

more. 5-7 fringes are enough to tell how good the surface is. I made a sting

(thick thread, black) harp to aid in looking at the straightness of fringes (think a archery bow)

Lay the string on the test flat and read the fringe bow-ness(?).

 

Starry Nights

 

I took your image to B&W and sharpen.

In about 15 minutes I had one of these last night, just big enough for a 6" flat. I sure do love having a 3D printer.

 

[duck]

all you need is to bend a coat hanger.

[italic]

all you need is to bend a coat hanger.

If I used a coat hanger every time I could use one, I wouldn't have any left

[ngc4910]

Thanks a lot for the comments so far. "Italic": I really like your DIY beam splitter setup, and I appreciate how sensitive to camera positioning the curvature of the fringes appears.

 

Today, I took pictures of the fringes for two different star diagonals, and moved each of them across the flat to see how the patterns change depending on where on the flat the diagonal is placed.

 

See below for the first diagonal (a).

 

 

Here, the fringes had a rather funky shape (b). The general shape did not change much when moving the diagonal across the flat; that said, when the air gap was wider and more fringes could be seen, the fringes appeared somewhat straighter ( c ), as expected. When moving the diagonal prism close to the position with the defect in the flat, the patterns changed again, I hence decided only to use the half of the flat opposite to the visible defect; within that half, the fringes did not appear to change considerably with position of the prism. (They do, of course, evolve as the air gap decreases over time).

 

 

Then, the second diagonal (d), which is the same diagonal prism I used in my previous post ( https://www.cloudyni...lat/?p=13842308 ).

 

 

Here, the fringes are much straighter compared to diagonal (a), see pictures (e) and (g). When I wait a few minutes, the number of fringes decreases considerably as the air gap decreases, and the fringes appear more curved (f). As before, as long as I stay away from the immediate vicinity of the defect in the flat, the fringes hardly change with the position of the prism.

 

I took the images standing with the camera roughly 1.5 meters above the optical flat;  to avoid a direct reflection of my CFL bulb, I had to take the images from a small angle with respect to vertical, but I can confirm that the appearance of the fringes did not evolve visibly even when positioning the camera at shallower angles.

 

I did not yet dare to press on the prism to determine the curvature. I may try to find a "disposable" prism first to learn with before risking scratches to the prism that I actually use.

 

 

Now, the following is only anecdotal but perhaps worth noting: it seems to me that the second prism (d) has the flatter surface compared to (a) -- and this is consistent with the impression that I generally preferred the second diagonal (d) over the first when when observing planets.

 

 

 

 

 

 

 

[Oregon-raybender]

B has a few fringes of astigmatism, that is why the shape is odd.

 

F is OK, about a 1/4 wave (1/2 a fringe of power) or less, it has ast to, but

may not be noticeable.

 

Starry Nights

[ngc4910]

F is OK, about a 1/4 wave (1/2 a fringe of power) or less, it has ast to, but

may not be noticeable.

Thanks for the fast response -- unfortunately I seem to be missing some very basic concept here; could you kindly elaborate how you reached the conclusion that the surface accuracy is 1/4 wave (and not 1/2 wave?)

 

I added a straight black line for reference, and highlighted two fringes of identical color in pink. Over the length of the diagonal (from point (i) to point (ii) ), the black curve intersects with one other fringe; hence point (ii) is lambda/2 higher (or lower) than point (i) , as each fringe corresponds to a height difference of lambda/2.  But why is the accuracy then lambda/4 , and not simply lambda/2 ?