Flat Optics: Considerations When Buying
Sep. 30, 2024
Flat Optics: Considerations When Buying
Michael Naselaris, Sydor Optics Inc.
If you are looking for more details, kindly visit our website.
Flat optics generally perform three main functions: They transmit light (windows), reflect light (mirrors), and fold light (prisms). While most optical manufacturers make spherical and flat optics, a smaller percentage focus specifically on making only flat optics.
Knowing the intended use of the optical component, shared by about half of all customers, helps the manufacturer to understand which specifications are the most critical to the optic&#;s performance, and to discuss specifications that might have been overlooked or overspecified, as they all could be cost and/or delivery drivers.
Common flat optical components used in the UV, visible, and IR spectrums include:
&#; Debris shields &#; Reference surfaces &#; Encoder disks &#; Reticles &#; Filters &#; Substrates &#; Gratings &#; Wafers &#; Lightpipes &#; Wave plates &#; Mirrors &#; Wedges &#; Optical flats &#; Windows
Optical materials
The first and foremost item to consider is the optical material. Important factors include homogeneity, stress birefringence, and bubbles; all of these affect product quality, performance, and pricing. Homogeneity plays a major role with transmissive optics. As homogeneity decreases, so does the optician&#;s ability to achieve the desired transmitted wavefront specification. Stress birefringence, on the other hand, affects the mechanical stability of optics requiring surface flatness. Bubbles could affect cosmetics if they break the surface during grinding and polishing stages.
Other relevant factors that can impact processing, yield, and pricing include chemical, mechanical, and thermal properties, along with the form of supply. Optical materials can vary in hardness, making manufacturability difficult and processing cycles possibly lengthy. The ideal optical materials for flat optics manufacturing are BK7, Borofloat, and fused silica. Other optical materials require careful handling and special processing techniques, as they can easily stain or may be sensitive to environmental changes, such as temperature and humidity.
Keep in mind that often, equivalent material types can be used interchangeably. Some engineers will document a specific material (e.g., Schott&#;s N-BK7), whereas others may state a preferred material and add the clause, or equivalent, to their specifications. Having this option may shorten lead times and even decrease pricing to some extent.
&#; B270 &#; Silicon &#; BK7 &#; Sapphire &#; Borofloat® &#; ULE® &#; Crystalline material, such as
CaF2, MgF2, BaF2 &#; Zerodur® &#; Filter glass &#; Zinc selenide &#; Fused quartz and fused silica &#; ZSinc sulfide &#; Germanium
Reflected wavefront vs. transmitted wavefront, or both
One issue requiring clarification in approximately one in four inquiries involves reflected wavefront and transmitted wavefront. Prints received are often vague enough to make one question the intent of the reflected wavefront, which is the accuracy of the surface with respect to a reference plane. Transmitted wavefront is the permissible wavefront deformation involving the surface flatness of both surfaces, the parallelism of the optic, and the homogeneity of the optical material. For the most part, flat optics require reflected wavefront or transmitted wavefront, with the primary exceptions being plate beamsplitters and prisms. Prints oftentimes state the word flatness, yet the function may be that of a window, thereby requiring transmitted wavefront as the ideal specification for optical performance. Mirrors, on the other hand, require reflected wavefront as the key indicator of performance in this respect. Head-up displays, for example, are plate beamsplitters and require both good optical transmission and optimal reflected wavefront for performance.
Filter glass on a double-sided machine.
The terms used for specifying reflected wavefront and transmitted wavefront are optical in nature &#; waves and fringes (half-wave) and lately more requests are in terms of nanometers &#; but on rare occasions, surface flatness may be specified as a mechanical callout in microns (0.001 mm). It is important to distinguish the difference between two commonly used specifications: peak to valley and rms. Peak to valley is the maximum measurement and the worst-case scenario, taking into account the difference between the surface&#;s lowest and highest points. It is by far the most widespread flatness specification used today. A more accurate measurement of surface flatness is rms, as it takes into account the entire optic and calculates deviation from the ideal form. Traditionally, optical flats have measured surface flatness in fringes; today, however, laser interferometers at 632.8 nm measure most optical components.
The clear aperture, also known as the usable aperture, is important. Normally optics are specified with an 85% clear aperture. For optics requiring larger clear apertures, attention must be taken during the production process to extend the performance area closer to the part&#;s edge, making it more difficult and costly to fabricate.
A large double-sided machine.
Parallel or wedged
Components such as filters, plate beamsplitters, wafers, and windows are required to be of very high parallelism, whereas prisms and wedges are intentionally wedged. The method of grinding and polishing plays an important role in the manufacturer&#;s ability to achieve the parallelism specs. For parts requiring exceptional parallelism (<1 arc second) and transmitted wavefront (<1 wave), double-sided grinding and polishing is the best method to use. Parallelism can be easily measured using an interferometer.
Tropel wafer interferometer.
Wedges and prisms require angled surfaces at demanding tolerances and are usually processed via a much slower process using pitch polishers. Pricing increases as angle tolerances become tighter. Wedge is specified in degrees, minutes, and seconds, and occasionally it will be stated as a thickness measurement at the parts&#; thin edge, thick edge, or center. Wedge-angle tolerance of several arc seconds falls into the higher level of precision, whereas tolerances of minutes or degrees fall into the medium and looser levels of precision. Typically, an autocollimator, goniometer, or a coordinate measurement machine is used for wedge measurements.
Side view of a pitch polisher.
Dimensions and tolerances
Size, in conjunction with other specifications, will dictate the best processing method, along with the size of equipment to use. Although flat optics can be any shape, round optics seem to achieve the desired specifications more quickly and uniformly. Overly tightened size tolerances can be the result of a precision fit or simply an oversight; both have an adverse effect on pricing. Bevel specifications are at times overly tightened, also resulting in increased pricing.
Surface quality
Surface quality is influenced by cosmetics, also known as scratch-dig or surface imperfections, as well as surface roughness, both with documented and universally accepted standards. In the U.S., MIL-PRF-B is popular with an increasing use of ISO-7 or its American-based counterpart ANSI/OEOSC OP1.002.
Top view of a pitch polisher.
Scratch-dig is represented with two numbers (e.g., 20-10) that generally fall into predetermined sets, such as 20-10, 40-20, 60-40, etc. The first number is arbitrary and denotes the scratch appearance, best matched to a calibrated standard. The second number refers to the dig size, which is designated in 0.01-mm increments. Scratch-dig values of 80-50 and above refer to commercial quality, 60-40 refers to general optics quality, and surface qualities of 20-10 and 10-5 are utilized more for laser optics and high-end optics applications. Lower numbers mean a higher level of precision and increased pricing. Keep in mind that, as the area of a part increases the difficulty, achieving a higher level of precision for scratch-dig increases difficulty at an even greater rate.
metrology that can objectively determine cosmetic thresholds for each of the existing cosmetic standards has become more prevalent and is taking the pre-existing subjectivity out of cosmetic evaluation.
Surface roughness refers to the overall texture of an optical surface and can influence the production process or the need for different or additional polishing steps to achieve lower surface roughness requirements, both having cost implications. Surface roughness generally falls into five categories: superpolishing (&#;1 angstrom rms), high-precision laser grade (1 to 5 angstroms rms), standard optics (5 to 15 angstroms rms), commercial optics (15+ angstroms rms) and those with no specification. The detail to remember is that lower roughness equals higher price. Generally, surface roughness is measured with noncontact optical profilometers. One universal and often overlooked problem with roughness specifications is the omission of a measurement&#;s length.
Quantity
For the most part, the smaller the quantity, the higher the processing costs per piece and vice versa. Quantities too low may involve lot charges, as a group of components may need to be processed to properly fill and balance the machine to achieve the desired specifications. The goal is to maximize each production run to amortize processing costs over the largest quantity possible. Although the same optical component can be made using different processing methods, the dominant one is usually indicated by the quantities and specifications.
A small double-sided machine.
The most commonly used processes for flat optics involve double-sided polishing and single-sided polishing on pitch polishers. Double-sided grinding and polishing, a batch-type procedure, can process both faces of the optic simultaneously for parallel optics. Economical batch sizes are determined by the size of the optic and the machine size. Pitch polishing, however, is a more time-consuming process generally utilized for requirements specifying fractional wave surface flatness and/or improved surface roughness. Double-sided polishing is deterministic, involving hours, while pitch polishing may involve days for the same quantity of parts. If transmitted wavefront and/or total thickness variation are your primary specifications, double-sided polishing is best, whereas polishing on pitch polishers is ideal if reflected wavefront is of primary importance.
Trends
Over the past several years, we have observed a few trends regarding precision flat optics. More and more customers are making the assumption of quality, in turn making delivery more important than pricing for the most part. The assumption of quality needs to be substantiated with more questions to ensure that the proper metrology and levels of verification are used. Increasingly, the demand is for thinner and thinner optics, along with tighter surface flatness and higher levels of cleanliness.
My recommendation would be to call your optics vendor. Share more information up front in the quoting process and be open to suggestions, as your vendor may point out key cost, delivery, and quality drivers. This will yield the most effective and accurate quotation, as compared to just sending an .
Using a Bath Interferometer to test Optical flats
Bruce Griffiths
#
Its possible to use a Bath interferometer to test an optical flat in conjunction with an axuliary "collimator" lens as depicted below:
where the test surface is located at the front focal point of the collimator lens.
Residual errors are measured by testing a known perfect flat such as a liquid flat.
The Bath is modified slightly by using a pentamirror setup to replace the mirror in the right angle Bath to ensure that both the test beam and return test beam foci lie in the same plane.
A somewhat more compact arrangement is possible if a Petzval style collimator lens is used.
Barry Jensen
#
This sounds interesting. I am interested in making a 16 inch flat and have acquired a 3 inch thick blank for that purpose, but alas have not worked it because I have no way to test it. I wonder if modifying the Bath is beyond my abilities/resources. I have enough difficulties getting good igrams as it is. I am curious to to hear if anyone tackles this.
Barry
Bruce Griffiths
#
Barry
This method still requires a suitable collimator objective wih a diaeter slightly larger than the flat being tested that is corrected for coma and preferably astigmatism.
Its probably better suited when one wishes to check coated and uncoated diagonals and immersion as in the Rayleigh water test isn't an option. It will work well for testing using a refractor objective.should one have a suitable one.
In your case using a reflecting collimator is likely the cheapest option.
A reflecting collimator has been used at UoA for testing much larger flats than yours in a Fizeau interferometer setup using a large diameter liquid flat.They used a non immersion test using a liquid flat in a container with a glass bottom and the flat was suspended over the liquid flat.
Bruce
toggle quoted message
Show quoted textOn 12 June at 09:06 Barry Jensen <barryjensen@...> wrote:
This sounds interesting. I am interested in making a 16 inch flat and have acquired a 3 inch thick blank for that purpose, but alas have not worked it because I have no way to test it. I wonder if modifying the Bath is beyond my abilities/resources. I have enough difficulties getting good igrams as it is. I am curious to to hear if anyone tackles this.
Barry
Bruce Griffiths
#
Barry
Either an F/5 spherical mirror with an aperture slightly greater than 16" or a 17" diameter F/8 float glass singlet lens with spherical surfaces should be satisfactory for use with the Rayleigh water test (for a 16" diameter flat) with up to 40 or so tilt fringes. The error due to collimator aberrations will be around 1nm or less as long as the surfaces are smooth with errors of no more than a few waves from spherical.
NB the water film thickness is 1mm or less for the Rayleigh water test.
There will be some distortion in imaging the test surface due to collimator SA, however this is easily measured and corrected for.
Bruce
toggle quoted message
Show quoted textOn 12 June at 10:26 Bruce Griffiths <bruce.griffiths@...> wrote:
Barry
This method still requires a suitable collimator objective wih a diaeter slightly larger than the flat being tested that is corrected for coma and preferably astigmatism.
Its probably better suited when one wishes to check coated and uncoated diagonals and immersion as in the Rayleigh water test isn't an option. It will work well for testing using a refractor objective.should one have a suitable one.
In your case using a reflecting collimator is likely the cheapest option.
A reflecting collimator has been used at UoA for testing much larger flats than yours in a Fizeau interferometer setup using a large diameter liquid flat.They used a non immersion test using a liquid flat in a container with a glass bottom and the flat was suspended over the liquid flat.
Bruce
On 12 June at 09:06 Barry Jensen <barryjensen@...> wrote:
This sounds interesting. I am interested in making a 16 inch flat and have acquired a 3 inch thick blank for that purpose, but alas have not worked it because I have no way to test it. I wonder if modifying the Bath is beyond my abilities/resources. I have enough difficulties getting good igrams as it is. I am curious to to hear if anyone tackles this.
Barry
astroelectronic
#
that means you can't use this test to distinguish between a flat and a long-radius convex or concave sphere.
Michael
Bruce Griffiths
#
Michael
If one starts with a liquid flat and then places the test surface in the same position as the liquid flat without adjusting focus or ignoring Z3 then the difference in curvature (if any)should be detectable.
Bruce
toggle quoted message
Show quoted textOn 12 June at 23:33 astroelectronic <astroelectronic@...> wrote:
that means you can't use this test to distinguish between a flat and a long-radius convex or concave sphere.
Michael
astroelectronic
#
I think it's almost impossible to bring it to the same position, because the permissible tolerance is very small. May be 10µm or so.
Michael
Bruce Griffiths
#
1 micron repeatability should be achievable with suitable kinematic design at least for solid flats.
The attached interferometer also has the same weakness.
If the flat is intended for use as the flat in a DPAC test then large radius convex or concave surfaces are acceptable.
Goto RuiQi to know more.
Bruce
toggle quoted message
Show quoted textOn 13 June at 00:15 astroelectronic <astroelectronic@...> wrote:
I think it's almost impossible to bring it to the same position, because the permissible tolerance is very small. May be 10µm or so.
Michael
Bruce Griffiths
#
The shift in position of the Bath interferometer diverger focus with respect to the collimator objective between measuring a liquid flat (or other flat with known error) and measuring the test flat should be less than
8*eps*(F/D)^2
Where
F is the focal length of the collimator
D is the diameter of the test flat
eps is the maximum desired sagitta of the best fit sphere
Example 1
F = 1m
D = 0.2m
eps = 1nm
Max focal shift < 200nm
F = 2m
D =0.1m
eps = 10nm
Max focal shift < 32 micron
The difficulty of achieving this is mitigated somewhat by the fact that it only has to be maintained over the time required to measure the reference and test flats.
For the first example plastic parts are unlikely to be stable enough.
However metal parts should easily meet this constraint.
The TCE of the lens and the TC of its refractive index also combine to shift the collimator focus when the temperature changes.
BK7 has a lower thermal focus shift than fused silica. An acrylic lens has a significantly higher thermal focus shift than either fused silica or BK7.
Bruce
#
If the flat is intended for use as the flat in a DPAC test then large radius convex or concave surfaces are acceptable.Why can you use this layout to test flats with a long focus spherical mirror? The flat is at 45 degrees, Newotnian style.
On Tue, Jun 12, at 05:47 am, Bruce Griffiths wrote:Why can you use this layout to test flats with a long focus spherical mirror? The flat is at 45 degrees, Newotnian style.
Bruce Griffiths
#
Mladen
The image of the entire surface being tested test has to be in focus at the detector.
This is difficult to impossible to achieve with Ritchey-Common style test setups.
If the entire surface isn't in focus then one has to use raytracing and diffraction calculations to help disentangle the resultant interferogram and extract the surface error of the test surface.
Bruce
toggle quoted message
Show quoted textOn 14 June at 13:06 "Mladen, Florida, USA" <mkvranjican@...> wrote:Mladen, Florida, USA
On Tue, Jun 12, at 05:47 am, Bruce Griffiths wrote:If the flat is intended for use as the flat in a DPAC test then large radius convex or concave surfaces are acceptable.Why can you use this layout to test flats with a long focus spherical mirror? The flat is at 45 degrees, Newotnian style.
#
Bruce,
The nature of the setup (just as in the Newtontian telecope) is that both, the 45 degree flat and the primary mirror, look head-on and are in focus. Why wouldn't the test mirror be in focus?
Bruce Griffiths
#
Mladen
The distance from the camera lens to the surface of the test flat varies with the position on the test flat. consequently the image of the surface of the test flat doesn't lie in a plane perpendicular to the axis of the test system. Thus one has to tilt the image sensor to meet the Schiemplug condition which isn't practical with that illumination setup.
Its also impossible (with conventional optics) that both the "primary" and "diagonal" images formed by the camera lens are simultaneously in focus at the detector as the diagonal and the primary do not share the same conjugate.
Most publications on the Ritchey-Common test only discuss measuring the RROC of the flat via measurement of the astigmatism.
The double reflection from the test flat is problematic when the flat errors are large as the point at which the forward and return rays intersect the surface of the flat do not coincide ie retrace errors are present.
Bruce
toggle quoted message
Show quoted textOn 14 June at 13:31 "Mladen, Florida, USA" <mkvranjican@...> wrote:
Bruce,
The nature of the setup (just as in the Newtontian telecope) is that both, the 45 degree flat and the primary mirror, look head-on and are in focus. Why wouldn't the test mirror be in focus?
Bruce Griffiths
#
That should have been Scheimpflug.
Bruce
toggle quoted message
Show quoted textOn 14 June at 14:38 Bruce Griffiths <bruce.griffiths@...> wrote:
Mladen
The distance from the camera lens to the surface of the test flat varies with the position on the test flat. consequently the image of the surface of the test flat doesn't lie in a plane perpendicular to the axis of the test system. Thus one has to tilt the image sensor to meet the Schiemplug condition which isn't practical with that illumination setup.
Its also impossible (with conventional optics) that both the "primary" and "diagonal" images formed by the camera lens are simultaneously in focus at the detector as the diagonal and the primary do not share the same conjugate.
Most publications on the Ritchey-Common test only discuss measuring the RROC of the flat via measurement of the astigmatism.
The double reflection from the test flat is problematic when the flat errors are large as the point at which the forward and return rays intersect the surface of the flat do not coincide ie retrace errors are present.
Bruce
On 14 June at 13:31 "Mladen, Florida, USA" <mkvranjican@...> wrote:
Bruce,
The nature of the setup (just as in the Newtontian telecope) is that both, the 45 degree flat and the primary mirror, look head-on and are in focus. Why wouldn't the test mirror be in focus?
#
Bruce, I will take your word for it, although it's not what I remember. Now, I am not sure what to think of the raytrace results which basically treat the setup as an on-axis test. The traditional Ritchie-Common test is usually performed at a more acute angle to accentuate astigmatism, and the test flat appears oval, not circular as in the example I gave.
Here's an example of a test for a Newtonian diagonal mirror. The elliptical flat appears perfectly round and in focus at 46 degrees relative to the optcal axis. The igrams came out very dim and noisy. The large red field is the reference beam, and the two bright lights are the Bath's parasitic reflections.
I wanted to repeat the test but never got around to it. Of course, the reference wavefront is obtained by testing the ref mirror alone, and then with a flat. The two wavefronts should then be subtracted in DFTFringe.
Mladen
Bruce Griffiths
#
Mladen
One has to be careful with raytraces as the wavefront is usually referred to the exit pupil of the system and thus depend on which surface is the STOP.
Unless your camera can tilt either the lens and/or the detector then its not possible to have the entire periphery or the surface in focus.
Lookup Scheimpflug principle.
How did you get a circular outline for the diagonal mirror?
Was the axis of the camera aligned with the test system axis?
Bruce
toggle quoted message
Show quoted textOn 14 June at 16:23 "Mladen, Florida, USA" <mkvra its not physically possible njican@...> wrote:
Bruce, I will take your word for it, although it's not what I remember. Now, I am not sure what to think of the raytrace results which basically treat the setup as an on-axis test. The traditional Ritchie-Common test is usually performed at a more acute angle to accentuate astigmatism, and the test flat appears oval, not circular as in the example I gave.
Here's an example of a test for a Newtonian diagonal mirror. The elliptical flat appears perfectly round and in focus at 46 degrees relative to the optcal axis. The Bath igrams showed a lof of noise. The large red field is the reference beam, and the two bright lights are the Bath's parasitc relfections.
I first obtained a wavefront for the mirror at its center of curvature and then with the flat. The two wavefornts were subtracted using DFTFringe. The fringes appear rather straight and uniform, and the mirror edge looks well defined all the way around.
Mladen
For more Optical Flat Mirrorinformation, please contact us. We will provide professional answers.
27
0
0
Comments
All Comments (0)