You are on page 1of 8

J Opt

DOI 10.1007/s12596-016-0357-y

RESEARCH ARTICLE

A simple setup for the measurement of conic surfaces using


a plane-parallel plate as a null corrector
Youssef Al-Khateb1 • Mhd. Fawaz Mousselly1 • Mamoun Naim1

Received: 14 March 2016 / Accepted: 25 July 2016


 The Optical Society of India 2016

Abstract Optical traditional interferometers are usually surface; and then, the reflected rays will be precisely
used to evaluate the form of spherical and plane surfaces; coincident with the incident rays, making this method very
those systems could also assess the quality of aspheric sensitive to any sub-micron deviation over the whole of the
surfaces which are quasi-spherical. But if a general tested surface [2].
aspheric surface is to be evaluated, this procedure requires However, general aspheric surfaces cannot be directly
the purchase of complicated and very expensive devices, measured using standard interferometers, and a suit-
such as aspheric interferometer, null-lenses, and computer able null corrector has to be introduced between the
generated hologram for each aspheric surface. This article interferometer and the aspheric surface [3]. The null cor-
presents a solid method allowing the measurement of a rector function is to convert a spherical or plane wave-front
wide range of conical aspheric surfaces. This method is into a form matching the aspheric surface under test; and as
based on inserting plane-parallel plate (PPP) within the a result, the interferogram will display straight, parallel,
path of a laser beam coming out of a traditional Fizeau and equally spaced fringes if the aspheric surface under test
interferometer. This article also presents a mathematical has the profile requested in the lens drawing. A variety of
justification that relates the conical surface parameters to designs of null correctors could be seen in optical articles;
the thickness of the inserted PPP. The article also presents those correctors has been designed and manufactured to
the supporting computer simulations and some practical measure specific aspheric surfaces, for example Fig. 1
results of applying this method and its range of use. shows Shafer null corrector consisting of 3 lenses [4];
while Fig. 2 shows an Offner null corrector comprising two
Keywords Conical aspheric surface  Wave front  Fringe  lenses [5], and in this figure, the first lens is called the field
Interferogram lens and its function is to image the aspheric surface onto
the second lens which is called the relay lens, and the role
of the relay lens is to produce spherical aberration equaling
Introduction the difference between the aspheric surface and the best-fit
spherical surface [6–8].
The accurate metrology of optical surfaces is usually The design and handling of a null corrector is usually
achieved by means of optical interferometers [1]. In such difficult because it is made of several lenses which have to
systems, the laser wave-front has to be spherical with be perfectly manufactured, and have also to be accurately
radius of curvature matching that of the surface under test, assembled. Even if the null corrector has been produced
so the laser rays fall perpendicularly upon the tested within the acceptable tolerances, the accuracy of the
measurement requires the corrector to be precisely placed
relative to the aspheric surface and the interferometer; and
& Youssef Al-Khateb in this context it is appropriate to recall the problem of the
khatyos76@gmail.com
Hubble Space Telescope where an incorrectly placed null
1
Higher Institute for Applied Science and Technology, lens resulted in the primary mirror being ground to an
Damascus, Syrian Arab Republic

123
J Opt

Fig. 1 Shafer null corrector [4]

Fig. 2 Offner null corrector [8]


Fig. 3 Effect of PPP on a converging light beam: each circular zone
of the beam converges to a different point
inaccurate surface figure, and thus producing strong
spherical aberration [9].
This paper presents a simple null corrector which con-
sists only of parallel faces plate PPP made of optical glass.
A mathematical relation has been proved to show that one
can make a null optical test using this null corrector PPP to
measure a conical convex surface with negative conic
constant or a conical concave surface with positive conic
constant.

Definition of conic surfaces

Conical surfaces are rotationally symmetric, and are often


expressed by their projection in the plane OYX; and this
type of surfaces is known in the optical designing drawing Fig. 4 Deviation of one ray after passing the PPP
made for manufacturing by the next equation [10]:
Y 2 ¼ 2R0 x  ð1 þ kÞx2 ð1Þ emerging ray without refraction; d: coordinate difference
on y axis between the incoming ray, and the emerging ray
where R0 is vertex curvature and k is the conical constant. with refraction; n: refractive index of PPP.
Finally, D = D - d, which can be also calculated as
follows:
Analytical justification
D ¼ ðd þ DÞ  d ¼ t tan h  t tan h0 ¼ tðtan h  tan h0 Þ
When a light beam having a specific wave-front shape falls ð2Þ
on a surface of the same shape, then the light beam will be using the trigonometric relation
reflected exactly on itself, producing a ‘‘null test’’ in the
interferometer setup. Therefore, this analytical presentation sin h0 sin h0
tan h0 ¼ ¼ p ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

begins by identifying the modifications undergone by the cos h0 1  sin2 h0
incident spherical wave-front when passing through the
with Snell’s law sin h ¼ n sin h0 we find:
PPP.
Figure 3 shows the effect of PPP on a converging light sin h =n sin h
tan h0 ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð3Þ
beam, whereas Fig. 4 shows the deviation of one ray after 2  n  sin2 h
2
1  sin h n2
traversing the PPP.
Let the following letterings: h: incidence angle; h0 : And by denoting m = tan h, we have sin h ¼ pffiffiffiffiffiffiffiffiffiffiffiffi
tan h
¼
1þtan2 h
refraction angle; t: thickness of PPP; D: coordinate dif- pffiffiffiffiffiffiffiffi
m ffi
: by substitution in (3):
1þm2
ference on y axis between the incoming ray, and the

123
J Opt

pffiffi
0m 1 þ m2 m y ¼ M ðx  R 0 Þ þ D
tan h ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2
¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi :
n2  m =ð1 þ m2 Þ n :ðm þ 1Þ  m2
2 2
(ray slope before and after passing the PPP has the same
0
value M) multiplying the last equation by the derivative y
Combined with (2) gives: we get:
!
0 m y 0 y ¼ y 0 M ð x  R0 Þ þ y 0 D
D ¼ tðtan h  tan h Þ ¼ t m  pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
n ð m þ 1Þ  m 2
2 2
and because the ray is perpendicular to the wave, we have
or: M.y0 = - 1, which leads to:
0 1 y0 y ¼ ðR0  xÞ þ y0 D
B 1 C
D ¼ tm@1  qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiA integrating the last equation gives the equation of the sur-
2 1
n 1 þ m2 n2 n
face perpendicular to the beam e0 :
 Z
2 2
 2
 0
considering that m2 n n1
2  1, which is valid for small y ¼ 2R0 x  x þ C þ 2 y Ddx :
angles of incidence (this approximation will be discussed
in next section), and using Taylor’s expansion we can Where C is a constant, the previous equation consists of
    
write:
2
D ffi tm 1  1n 1  12 m2 n n1 ¼ 1  1n tm two terms: the first is the equation of a circle, and the
2

2
second term represents the deviation off the sphericity, and
þ n2n1 3
3 tm the previous relation may be rewritten as: we can write it as follows:
Z
D ffi am þ bm3 ð4Þ J ð xÞ ¼ 2 ðy0 DÞdx
n1 2 
where: a ¼ n t; b ¼ n2n1 3 t.
Then, the light wave-front converges to its center of substituting the relation (4) in the last equation gives:
curvature C located on the X axis at distance R0, and the Z Z
 
equation of this ray may be written as: J ð xÞ ¼ 2 a þ bm2 dx ¼ 2ax þ 2b m2 dx:

ys ¼ M ð x  R0 Þ ð5Þ And since the ray slopes before and after the PPP are
where M = -m is the ray slope. Then, the surface who has equal, and making use of use of (5) and (6).we get:
Z Z Z
y0s as a derivative (except at the vertex where the slope is y2s 2R0 x  x2
undefined) and is perpendicular to the incident light beam, m2 dx ¼ 2
dx ¼ dx
ð R0  x Þ ðR0  xÞ2
will verify the following relation: Z Z
R20 ðR0  xÞ2
¼ dx  dx
M y0s ¼ 1 ðR0  xÞ2 ðR0  xÞ2
substituting the previous relation in (5), leads to: R20 R0
¼ x¼ x
R0  x 1  Rx0
y0s ys ¼ R0  x
noting that 1  Rx0 which is effectively valid for the current
by integration we get the equation of the surface perpen-
dicular to the incident light beam: case where the angles are small (this approximation will be
discussed within the next section), and using Taylor’s
y2s ¼ 2R0 x  x2 þ b expansion, we can write:
Z 
the last surface passes through the point O, so the constant 2 x x2 x2
b = 0, then: m dx ¼ R0 1 þ þ 2  x ¼ R0 þ
R0 R0 R0
y2s ¼ 2R0 x  x2 ð6Þ by substitution in the equation of J:

which represents a circle with the center of curvature at C, x2 b
and its radius equals R0. J ð xÞ ¼ 2ax þ 2b R0 þ ¼ 2bR0 þ 2ax þ 2 x2 :
R0 R0
When the convergent beam e passes through the PPP,
each ray undergoes a specific shift D depending on its Consequently, the equation of the surface perpendicular
intersection height along the y axis; and the new beam to the light beam after passing the PPP, becomes:

is, labeled e’, whose equation may be written as b
y2 ¼ 2bR0 þ 2ðR0 þ aÞx  2 þ 1 x2 þ C:
following: R0

123
J Opt

Since this surface passes through the point O, so has an objective of numerical aperture f# = 1.5 (i.e.
C ¼ 2bR0 , and the surface equation gets the form: mmax = 1/3) and a PPP made of glass BK7 (n = 1.515 at
 k = 0.6328) so the maximal relative error is about 10-3.
b The second approximation was:
y2 ¼ 2ðR0 þ aÞx  2 þ 1 x2
R0
1 x x2
Comparing the last equation with the Eq. (1), it becomes x ¼ 1þ þ 2
1  R0 R0 R0
apparent that this surface (which is the output wave-front
after passing the PPP) is of conical form and its conical the maximal value of x is can be calculated as following:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 2ffi
constant is k ¼ 2 Rb0 , and the center of the wave-front
  D
curvature has shifted by a ¼ t n1 compared to the inci- xmax ¼ R0  R20  :
n 2
dent wave-front.
substituting the value of b from the relation (4) we get: This approximation was also compatible with the same
2 equipments as in the previous paragraph so the maximal
n 1
k¼ t: relative error is about 2 9 10-4.
n 3 R0
The preceding analytical study demonstrates that an
incident spherical wave-front become a conical wave-
Computer simulation
front after passing through the PPP, so we expect the
The main objective of this simulation is to confirm the
possibility of measuring any aspheric surface of conical
validity of mathematical relations and the strength of using
form whose radius at vertex is R0 and its conical constant
the PPP as a null corrector. This simulation was based on
is k using a parallel faces glass plate which has the fol-
using the well-known optical design program ZEMAX, and
lowing thickness:
data fitting had been done using the mathematical program
n3 ‘‘Matlab’’.
t¼ kR0 : ð7Þ
n2  1 Figure 5 shows the incident rays coming from the inter-
This relation shows that the PPP thickness will have a ferometer (assumed to be to left), the rays reflected off the
positive value (i.e. a real and material thickness) if the aspheric surface, and the resulted interferogram; whereas
product R0.k is negative, i.e. the surface is concave with a Fig. 6 shows the effect of the PPP on the rays bundle and the
positive conical constant or the surface is convex with a resulted interferogram. The straight, parallel, and equally
negative conical constant. Actually these two types of spaced fringes with P–V error (peak to valley wave-front
conical surfaces are the important and commonly used error [2]) less than k/10 shows that the PPP is definitely a
aspheric surfaces in optical design. the last relation can be suitable null optic for this aspheric surface.
found in a lot of optical design textbooks during derivation Many conical surfaces having different vertex curva-
of spherical aberration and image displacement caused by a tures Ro and conical constants K, respecting the inequality
plane-parallel plate in converging/diverging light [11, 12], Ro* K \ 0, had been simulated in Zemax using the PPP as
but our derivation to relation (7) shows that a spheric wave- a null corrector. The thickness of the used PPP was the only
front would modify to conic wave-front after passing variable in the optimization process designed to obtain a
through PPP permitting the null test of this kind of null corrector. The following tables shows the final results:
surfaces. • for K = -0.1 and n = 1.515, Table 1 shows the wave-
front peak to valley error, and the required PPP
thickness as a function of the vertex curvature Ro of
Accuracy of the analytical study the tested conical surface.
• for R0 = 50 and n = 1.515, Table 2 shows the wave-
In the preceding analytical study we made two approxi- front peak to valley error and the required PPP
mations to obtain the final relation of t. The first approxi- thickness as a function of the conical constant of the
mation was: tested surface.
1 1 n2  1 • for K = -0.5 and R0 = 50, Table 3 shows the wave-
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 1  m2
2 2 n2 front peak to valley error and the effect of refractive
1 þ m2 n n1 2
index n on the PPP thickness.
The validity of this approximation is compatible with The PPP thickness resulted from the optimization pro-
the measuring equipments since the used interferometer cess using Zemax matches the results predicted by the

123
J Opt

Fig. 5 Testing a conical surface


without a PPP

Fig. 6 Testing a conical surface


using a PPP as null corrector

analytical relation previously derived in this paper. The Table 1 PPP thickness t as a function of the vertex curvature Ro
simulation results show that the P–V wave-front error is
R0 (mm) t (mm) P–V (k/2) R0 (mm) t (mm) P–V (k/2)
small compared to the general manufacturing tolerances of
aspheric surfaces [13]; therefore, the suggested PPP can be 30 7.843 0.0218 80 21.390 0.0002
classified as a good null corrector specialized in testing 40 10.576 0.0057 90 24.086 0.0001
common conical surfaces. 50 13.291 0.002 100 26.779 0.0001
60 15.997 0.008 110 29.421 0.0001
70 18.697 0.0004 150 39.726 0
Experimental results
• Curvature at vertex R0 = 57.15 mm.
In order to endorse the previous results and the feasibility
• Conical constant K = -0.074.
of using PPP as a null corrector, this technique was applied
• Clear aperture D = 45 mm.
in testing a Germanium lens having an aspheric surface
with parameters as shown in Fig. 7:

123
J Opt

Table 2 PPP thickness t as a function of the conical constant K it is impossible to derive useful information about the
K t (mm) P–V (k/2) K t (mm) P–V (k/2)
profile of the surface.
Then, the PPP was inserted perpendicularly to the
-0.05 6.636 0.0011 -0.4 53.635 0.0008 optical axis of the tested surface as shown in the left photo
-0.1 13.291 0.0020 -0.5 67.243 0.0019 in Fig. 10; the distance between the interferometer and the
-0.15 19.966 0.0025 -0.6 80.931 0.0060 tested surface had been readjusted to obtain fringes as
-0.2 26.66 0.0025 -0.7 94.699 0.0110 straight as possible. It is worth mentioning that if the PPP is
-0.25 33.374 0.0027 -0.8 108.551 0.0175 not perpendicular to the optical axis, then some aberration
-0.3 40.108 0.0024 -0.9 122.479 0.025 (similar to coma or astigmatism) would appear in the
-0.35 46.862 0.0018 -1 136.503 0.0343 interferogram, as can be seen in the middle photo in
Fig. 10. Therefore, the PPP must be adjusted to obtain the
required perpendicularity, and the fringes would resemble
the right photo in Fig. 10.
Table 3 Effect of refractive index n on the PPP thickness t
The practical measurement confirms that the PPP with
n t (mm) P–V (k/2) n t (mm) P–V (k/2) the previous specifications compensated for the spherical
1.3 79.601 0.002 1.7 64.987 0.002 aberration generated by the conical surface under test; and
1.4 71.458 0.008 1.8 65.089 0.008 thus, it becomes possible to measure such aspheric surfaces
1.5 67.501 0.0004 1.9 65.699 0.0004
using traditional interferometers.
1.6 65.641 0.0002 2 66.667 0.0002

Discussion

Comparing between the PPP as a null corrector and an


ordinary null corrector used in metrology clearly shows the
usefulness and the limitation of this method: a null system
usually consists of more than one optical element—and in
some cases it may contain diffractive elements—so, the
following notes have to be taken in consideration when
designing, manufacturing, and assembling any conven-
tional null corrector:
• Optical elements must be fabricated and assembled in
extremely high accuracy, because they are made to
measure aspheric surfaces.
• The placement of the null corrector must be defined and
Fig. 7 Germanium lens having a conical surface maintained accurately, refer to Hubble Space Telescope
A suitable PPP for testing this conical surface had been primary mirror problem [9].
made of glass BK7 having the following specifications (see • The optical axis of the interferometer and the null
Fig. 8): corrector must be accurately coincident. Usually this
step is achieved using expensive auxiliary elements,
• Refractive index n = 1.515 at k = 0.6328 lm. and needs a lot of experience and time.
• Thickness, obtained using Zemax, t = 11.35 mm.
• Accuracy of faces parallelism & 700 . In contrast, one may note that using the PPP as a null
• Optical surface irregularity DN ¼ k=4. corrector has the following advantages:

Firstly, the conical surface was directly measured, • PPP is easier to fabricate since it is a plate of glass with
without any null corrector, by a common interferometer as only two parallel surfaces.
shown in the left photo of Fig. 9, and the middle photo • PPP has not specific optical axis, so it can be shifted up
displays the interferogram resulted from a null test at the and down, right and left without effecting the accuracy
vertex; as for the right photo, it shows an interferogram of the measurement.
obtained when the null test is not at the vertex. It is evident • PPP has no optical power so its position between the
that the large number of fringes hinders the interferometer interferometer and the aspheric surface has no effect on
software from even resolving most of the fringes, and thus the measurement.

123
J Opt

Fig. 8 Real PPP for testing a


conical surface

Fig. 9 Testing a conical surface without a PPP

Fig. 10 Testing a conical surface using a PPP as null corrector

• PPP with changeable thickness may be made-up of two Referring to the relation (7) and to the previous com-
prisms having the same refractive index and comple- parison one may clearly conclude that PPP is the optimum
mentary angles—with a special oil between them to option to measure a conical surfaces if we have the fol-
avoid any internal reflection—see Fig. 11, this PPP lowing conditions:
could be used to measure a range of conical surfaces (or
1. The product R0*k is negative, i.e. the conical surface is
reverse engineering some unknown surfaces, but for a
convex and the conical constant is negative or the
system made of two wedged glass plates, the alignment
conical surface is concave and the conical constant is
error would be seriously higher).

123
J Opt

Fig. 11 Plane-parallel plate


with variable thickness

positive. It must be said here that the majority of References


optical design that use aspheric surface do use this kind
of surfaces in order to minimize the aberrations. 1. G. Schulz, Imaging performance of aspherics in comparison with
spherical surfaces. Appl. Opt. 12, 5118–5124 (1987)
2. The values of most practical R0*k leads to realistic
2. D. Malacara, Optical shop testing, Wiley Series in Pure and
values of the PPP thickness. This is especially true for Applied Physics, ISBN 0-471-52232-5, (1992)
the cases of aspheric lens used in portable optical 3. K. Oka, S. Sparrold, Asphere design for dummies, SPIE 8487,
systems, such as IR cameras. 84870B (2012)
4. J.H. Burge, Advanced techniques for measuring primary mirrors
for astronomical telescopes, Ph.D. The University of Arizona,
(1993)
5. A. Offner, A null corrector for paraboloidal mirrors. Appl. Opt. 2,
Conclusions 153–155 (1963)
6. C. E. Devoe, Limitations on aspheric surface testing with simple
null correctors, Master Thesis, Optical Sciences Center, Univer-
In this paper, a new robust method for interferometric sity of Arizona Tucson (1989)
measurement of aspheric surfaces of conical form was 7. J.M. Sasian, Optimum configuration of the Offner null corrector.
proposed. The analytical study and computer simulation Testing An F#/1 Paraboloid, in surface Characterization and
results were also presented to show the validity of using Testing, Proc. SPIE 1164, 8–17 (1989)
8. J.M. Geary, Introduction to Lens Design with Practicle Zemax
PPP as a null corrector to measure specific kinds of Examples, (Willmann-Bell). ISBN: 0-943396-75-1 (2002)
conical surfaces. This article also presented some prac- 9. M. Bottema, Reflective correctors for the Hubble Space Tele-
tical results confirming the possibility of using a PPP as scope axial instruments. Appl. Opt. 32(10), 1768–1774 (1993)
a null corrector for a specific aspheric lens. Finally, a 10. B. Braunecker, R. Hentschel, H. Tiziani Advanced optics using
aspherical elements. SPIE Bellingham, Washington 98227-0010
comparison between PPP and typical optical null lenses USA (2008)
showed the superiority of this method when measuring a 11. D. Malacara, Z. Malacara, Handbook of optical design, by Marcel
specified range of conical surfaces, This easy setup for Dekker, Inc. ISBN: 0-8247-4613-9 (2004)
the measurement of conic surface and the obtained 12. J. Wyant, C. Creath, Basic wavefront aberration theory for optical
metrology,Academic press, Inc. ISBN:0-12-408611-x (1992)
results may be useful for traditional optical work-shops 13. J. David Briers, Optical testing: a review and tutorial for optical
in developed countries. Engineers. Opt. Lasers Eng. 32, 111–138 (1999)

123

You might also like