Вы находитесь на странице: 1из 8

Manufacturable mobile phone optics: higher order aspheres are not

always better

James P. McGuire, Jr.

Optical Research Associates, 3280 E. Foothill Blvd., Suite 300, Pasadena CA 91107


ABSTRACT

We provide an overview of the design drivers for mobile camera phone cameras, with particular attention paid to
aspheric order. The designer is often tempted to use high order aspheres (e.g. 18
th
order). However, better as-built
performance can often be obtained at lower orders. 2010 Optical Society of America, Inc.

Keywords: Optical design, aspheric lens design, Forbes aspheric surfaces, mobile phone cameras, cell phone cameras,
plastic optics

1. INTRODUCTION

Mobile phone cameras are ubiquitous. They are in 80% of the nearly 1 billion handsets sold annually. Camera
resolution steadily increased to the point that megapixel sensors are common today. This increase in sensor resolution
and size drove complexity into the optical designs. Two, three, and even four elements are commonly used. Besides,
cost, which constrains the element count and heavily favors plastic optics, there are several other design drivers. Table 1
lists a representative design specification.

Aspheric surfaces are the key to obtaining the performance required for a modern mobile phone camera. Aspheric
surfaces can be represented in many ways; in this paper we will use the standard power series expansion as well as the
newer Forbes aspheres
1
. The standard aspheric representation is

z =
c h
2
1+(1-|1+k])c
2
h
2
+ Ab
4
+Bb
6
+Cb
8
+b
10
+Eb
12
+Fb
14
+0b
16
+Eb
18
+[b
20

z =
c h
2
1+(1-|1+k])c
2
h
2
(1)

where z is the sag parallel to the z-axis, c is the curvature, h is the radial distance, k is the conic constant, and
A,B,C,D,E,F,G,H,J are the aspheric deformation coefficients. The Forbes strong asphere expansion is

+u
4
_ o
m

m
con
(u)
M
m=0

m
con
(2)

where z, c, and h are as above, u=h/h
max
is the normalized, sag parallel to the z-axis, and

are the orthogonal basis
functions. Youngworth and Betensky discuss the application of Forbes aspheres to CRT lenses, Schmidt cameras, and
CD pick-up lenses
2
. A number of other aspheric representations have been used in optical design including splines
3
,
Chebyshev polynomials
4
, transcendental functions
5
, Zernike polynomials
6
, implicit surface definitions
7
, and sub/super
conics
8
.

Typically, mobile phone camera lenses are composed largely (or entirely) of plastic elements. Plastic optics allow the
designer to introduce aspheric surfaces with essentially no cost penalty in a large production run: molded spheres and
aspheres have comparable costs.

It is important to realize that it is essential to consider the as-built lens rather than then nominal lens
9, 10
, when
determining the order of the asphere to use. One method for determining the order is to create an all-spherical design
that meets the package constraints and analyze the ray aberration curve. The order of the aberration in this curve is a
good indicator of the order of asphere that is needed. Another method is to design with more orders than are necessary
and then to remove the extraneous order. This latter procedure is more effective with Forbes aspheric descriptions than
conventional aspheric descriptions.

Table 1. Typical mobile phone camera specifications
Parameter Value
Wavelengths (nm) 656, 588, 486 with 1:2:1 weights
Focal length (mm) 4
F-number 2.8
Sensor (mm) 4.6
Field (deg) 30
Relative
illumination (%)
> 50
Distortion (%) < 1
Telecentricity < 30 (needs to match sensor)
Total track (mm) < 4

1.1 Constraints

In the design of mobile phone lenses, the overall length is a driving factor. Thinner is better. This often results in thin,
tightly spaced elements. During the design process, the edge and center thickness of the elements must be constrained to
manufacturable values.

The next most important factor is the telecentricity. CMOS sensors contain micro lens arrays, which limit the chief ray
angle at the image plane. If the sensor only accepts light that is near normal incidence, the lens must be telecentric with
the stop must be located one focal length away. This extra air space is not tenable in a mobile phone lens.

Fortunately, for several years now, sensor manufacturers have been offsetting the position of the micro lenses from the
center of the active region of the pixel. The optical designer needs to match the chief ray to the sensors Chief Ray
Angle (CRA) requirements (or, if you are placing a large enough order, you might persuade the sensor manufacture to
design a sensor to your specifications). A quick Internet survey at the time this paper was written showed CRAs from 0
(non-shifted micro lenses) to 30 deg.

The next most important constraint is typically distortion. Mobile phone cameras are usually specified to have a
maximum distortion of around 1%. This can result in unusual-looking (for glass) rear elements, which bend the ray
bundles over to simultaneously meet the distortion and CRA constraints.

The final factor in mobile phone camera design is the pixel size. With a smaller pixel, the same field can be covered by a
smaller focal length and in less space. Unfortunately, in order to take advantage of these smaller pixels, more resolution
is required at higher spatial frequencies.

1.2 Degrees of freedom

Most mobile phone cameras utilize injection molded plastic optics, which has two important consequences. First, there
is a limited material choice, which means a severely restricted glass map. Second, since most of the expense of a
plastic optic is determined by the length of time the part occupies the injection molding machine (not materials), aspheric
lenses are only marginally more than spherical lenses (an aspheric molds cost more but this is amortized over hundreds
of thousands of lenses). While low order aspheres are very helpful, higher order aspheres can actually lead to worse as-
built performance as illustrated below. Aspheric order should be used in moderation.

Typically, the relative uniformity can fall-off to around 50% at the corner of the field of view, without noticeably
impacting a customers perception of the image quality. It is important to make use of this degree of freedom.



2. DESIGN PROCESS

In this section, we discuss two effective methods for the design of mobile phone cameras. The first is to start with an all
spherical system and slowly increase the aspheric order until the maximum as-built performance is obtained. The second
method is to start with an excessively high aspheric order and then to reduce the order, until the maximum as-built
performance is obtained.

2.1 Increasing aspheric order

You need to start somewhere. Typically, a designer will start from one of his old designs or from the patent literature.
The choice of the starting point can heavily influence the final design, especially if the designer relies heavily on local
optimization. In order to avoid a bad starting point and to gain insight into the design drivers, we start from
monochromatic, spherical, plane parallel plates and progressively added complexity until we reached the final
polychromatic aspheric solution


(a) (b) (c)
mono_plates_start.len Scale: 26.00 ORA 19-May-10
0.96 MM
mono_plates.len Scale: 28.00 ORA 19-May-10
0.89 MM
best_mono_plates.len Scale: 28.00 ORA 19-May-10
0.89 MM

(d) (e)
best_poly_plates.len Scale: 28.00 ORA 19-May-10
0.89 MM
best_a_term_IODC.len Scale: 27.00 ORA 31-May-10
0.93 MM

Figure 1. Progression of design from parallel plates (a) to local spherical design (b) to global spherical design (c) to best
polychromatic spherical design (d) to 4
th
order aspheric minimum (e)

Figure 1 shows the progression of the design form. We started out with the stop out in front of plane parallel plates. The
stop position was dictated by the need to minimize the length and maximize the telecentricity; the majority of patent
lenses share this stop position for the same reason. The merit function included constraints to desensitize the lens to tilt
and decenter errors
11, 12
and Lagrange constraints on the CRA. However, distortion was not controlled (it was held with
the aspheric terms later in the design process). The figure on the top left shows the result after the local optimization and
the bottom left shows the best result after replacement of the fictitious glasses with plastics using the Glass Expert
feature in CODE V on the most manufacturable of the Global Synthesis lens
13
. We restricted the glasses to the
most common plastics (PMMA, polycarbonate, SAN, polystyrene, Zeonex, and OKP4). Glass Expert picked Zeonex
E48R, PMMA, SAN, and PMMA (from left to right).


(a) (b)
ORA 31-May-10
best_poly_plates.len
RAY ABERRATIONS ( MILLIMETERS )
ORA 31-May-10
best_a_term_IODC.len
RAY ABERRATIONS ( MILLIMETERS )
Figure 2. Rim rays for best all spherical design (a) and for best 4
th
order aspheric design (b). Both plots are on the same scale.


(a) (b)
Figure 3. Astigmatism/distortion for the best all spherical design (a) and the best 4
th
order aspheric design (b). The scale for the all (a)
is 2.5x larger than for (b).

Figures 2 and 3 show the rim rays and the distortion curves for the spherical and 4
th
order aspheric designs. The residual
aberrations and distortion of the spherical design on the left are approximately 4
th
order, which is why only 4
th
order
terms were used in the aspheric design (Fig. 1e). The best 4
th
order aspheric design was generated by running Global
Synthesis to generate local minima, with the best design selected based on the lowest as-built wavefront errors (as
discussed in the Section 3). Both the standard aspheric representation and the Forbes aspheric representation gave
similar results, but the standard aspheric representation was faster computationally.

2.2 Decreasing aspheric order

In this section, we started from a spherical design as in the prior section. However, in this case the plastics were changed
to E48R, E48R, OKP4HT, and E48R to take advantage of the higher indices. Next we assigned 18
th
order Forbes
con

aspheres and optimized all the radii, thicknesses and aspheric coefficients with Global Synthesis using weighted
constraints for distortion and Lagrange constraints for telecentricity.
655.0000 NM
587.5600 NM
486.1300 NM
-0.025
0.025
-0.025
0.025
0.00 RELATIVE
FIELD HEIGHT
( 0.000 )
O
-0.025
0.025
-0.025
0.025
0.31 RELATIVE
FIELD HEIGHT
( 10.00 )
O
-0.025
0.025
-0.025
0.025
0.63 RELATIVE
FIELD HEIGHT
( 20.00 )
O
-0.025
0.025
-0.025
0.025
0.81 RELATIVE
FIELD HEIGHT
( 25.00 )
O
-0.025
0.025
-0.025
0.025
TANGENTIAL 1.00 RELATIVE SAGITTAL
FIELD HEIGHT
( 30.00 )
O
655.0000 NM
587.5600 NM
486.1300 NM
-0.025
0.025
-0.025
0.025
0.00 RELATIVE
FIELD HEIGHT
( 0.000 )
O
-0.025
0.025
-0.025
0.025
0.46 RELATIVE
FIELD HEIGHT
( 15.00 )
O
-0.025
0.025
-0.025
0.025
0.63 RELATIVE
FIELD HEIGHT
( 20.00 )
O
-0.025
0.025
-0.025
0.025
0.81 RELATIVE
FIELD HEIGHT
( 25.00 )
O
-0.025
0.025
-0.025
0.025
TANGENTIAL 1.00 RELATIVE SAGITTAL
FIELD HEIGHT
( 30.00 )
O
ASTIGMATIC
FIELD CURVES
ANGLE(deg)
S T
30.00
23.41
16.10
8.21
-0.50 -0.25 0.0 0.25 0.50
FOCUS (MILLIMETERS)
DISTORTION
ANGLE(deg)
30.00
23.41
16.10
8.21
-5.0 -2.5 0.0 2.5 5.0
% DISTORTION
best_poly_plates.len ORA 19-May-10
ASTIGMATIC
FIELD CURVES
ANGLE(deg)
S T
30.00
23.41
16.10
8.21
-0.2 -0.1 0.0 0.1 0.2
FOCUS (MILLIMETERS)
DISTORTION
ANGLE(deg)
30.00
23.41
16.10
8.21
-2 -1 0 1 2
% DISTORTION
best_a_term_IODC.len ORA 19-May-10

Because the Forbes aspheric coefficients are orthogonal, the coefficients on a single surface will play well with each
other. For example, if the design needs an adjustment of the 6
th
order spherical aberration, only one term needs to
change, whereas in the standard representation, the piston, radius, r
4
, and r
6
terms all need to vary. This limitation of the
standard aspheric representation becomes more burdensome, as the aspheric order increases. In our mobile phone
camera example, we initially optimized with 18
th
order coefficients on each surface and Global Synthesis finds better
solutions with the Forbes aspheres than the standard aspheres (although it does take longer).

We also note that, while the Forbes aspheres play well with other terms on the same surface, they do not necessarily
play well with Forbes aspheric terms on adjacent surfaces. Some of the elements are thin and closely spaced; these
closely spaced Forbes coefficients can (and do) fight with each other.

Figure 4 shows the progression of the design form best polychromatic spherical minima to best design before aspheric
order reduction to best reduced order aspheric lens. The molds are typically made by diamond turning, so the change in
the sign of curvature in the last element in Fig. 1e and 4c does not present a manufacturing issue. However, if the
surfaces were being conventionally ground and polished, we would want to constrain the surface to have no inflection
points.



aperture. We then reoptimized the lenses and repeated the order reduction, until the as-built lens performance no longer
(a) (b) (c)
sphere.len Scale: 26.00 ORA 31-May-10
0.96 MM
Forbes18_IODC209.len Scale: 27.00 ORA 31-May-10
0.93 MM
best_forbes18_reduce10.len Scale: 27.00 ORA 31-May-10
0.93 MM
Figure 4. Progression of design from best polychromatic spherical minimum (a) to best design before aspheric order reduction (b) to
best reduced order aspheric lens (c).

Figures 5 and 6 show the rim rays and the distortion curve for the best 18
th
and reduced order aspheric designs. The rim
rays, astigmatic curves, and distortion for the 18
th
order asphere all have many ripples, which are symptomatic of the
excessive aspheric order used in design. We used too high of an aspheric order, which complicated the merit function
space unnecessarily. The result was that even the designs with the lowest merit functions are ugly with noticeable
inflection points in the field and distortion curves. There are too many wiggles. Some of these effects might have been
mitigated by increasing the sampling in the aperture and the field.

The wiggles are ugly for several reasons. First, the designs may give significantly worse performance if the fields or
pupil sampling are shifted slightly. Secondly, the performance typically degrades rapidly with small manufacturing and
alignment errors, as the errors introduced at one surface are not cancelled by aberrations at a second surface, as they are
in the nominal design.

For the best lenses, we removed all the terms with negligible contributions (< 10 microns of sag at the edge of the
aperture), which was easy since the
con
surface components are orthogonal and represent sags at the edge of the
improved. The best reduced order aspheric design was selected based on the lowest as-built wavefront errors (as
discussed in the next section). The best reduced aspheric order design is shown in the Figure 4c. In this case, all the
aspheric terms reduced to 4
th
order (this was not the case with all the Global Synthesis lenses, just the one that became
the best after order reduction). We see the aberrations in the nominal design are reduced and the wiggles have been
removed. The Forbes aspheric representation gave better results than the standard aspheric representation at 18
th
order
and was easier to reduce the order. It was also computationally slower.


(a) (b)
im rays spheric order reduction (b). Both plots are on the
(a) (b)
der reduction (a) and the best re The
ameras are often specified by the MTF. Figure 7 shows the nominal MTF for the best design before order reduction (a)
3. AS-BUILT PERFORMANCE
The performance of the optical system must be sufficient to eet the required performance specifications in production.
OR A 1 7 -F e b -1 0 F/2.8, 60 deg full FOV, 4.60 mm RAYAB ERRATIONS( M ILL IMETERS ) 65 5 . 0 00 0 N M 58 7 . 5 60 0 N M 48 6 . 1 30 0 N M -0 .0 2 5 0 .0 2 5 -0 . 02 5 0 . 02 5 0 .0 0 R E LA TI VE FI EL D HE I G H T ( 0. 00 0 )O -0 .0 2 5 0 .0 2 5 -0 . 02 5 0 . 02 5 0 .4 6 RE LA TI VE FI EL D HE I GH T ( 15 .0 0 ) O -0 .0 2 5 0 .0 2 5 -0 . 02 5 0 . 02 5 0 .6 3 RE LA TI VE FI EL D HE I GH T ( 20 .0 0 ) O -0 .0 2 5 0 .0 2 5 -0 . 02 5 0 . 02 5 0 .8 1 RE LA TI VE FI EL D HE I GH T ( 25 .0 0 ) O -0 .0 2 5 0 .0 2 5 -0 . 02 5 0 . 02 5 TANGENTI AL 1 .0 0 RE LA TI VE SAGITT AL FI EL D HE I GH T ( 30 .0 0 )O
ORA 31-May-10
Forbes18_IODC209.len
RAY ABERRATIONS ( MILLIMETERS )
655.0000 NM
587.5600 NM
ORA 31-May-10
best_forbes18_reduce
10.len
RAY ABERRATIONS ( MILLIMETERS )
655.0000 NM
587.5600 NM
Figure 5. R for best design before aspheric order reduction (a) and after a
same scale.


Figure 6. Astigmatism/distortion for the best design before or duced order aspheric design (b).
scale for a) is 2.5x larger than for (b).



C
and after order reduction (b). The order reduction has significantly improved the performance of this Global Synthesis
generated local minima, as might be expected from the rim ray plots.


m
The performance of the nominal design (before tolerances) is interesting, but more important is the performance that can
be achieved with a cost-effective set of manufacturing and alignment tolerances.

486.1300 NM
-0.025
0.025
-0.025
0.025
0.00 RELATIVE
FIELD HEIGHT
( 0.000 )
O
-0.025
0.025
-0.025
0.025
0.31 RELATIVE
FIELD HEIGHT
( 10.00 )
O
-0.025
0.025
-0.025
0.025
0.63 RELATIVE
FIELD HEIGHT
( 20.00 )
O
-0.025
0.025
-0.025
0.025
0.81 RELATIVE
FIELD HEIGHT
( 25.00 )
O
-0.025
0.025
-0.025
0.025
TANGENTIAL 1.00 RELATIVE SAGITTAL
FIELD HEIGHT
( 30.00 )
O
486.1300 NM
-0.025
0.025
-0.025
0.025
0.00 RELATIVE
FIELD HEIGHT
( 0.000 )
O
-0.025
0.025
-0.025
0.025
0.31 RELATIVE
FIELD HEIGHT
( 10.00 )
O
-0.025
0.025
-0.025
0.025
0.63 RELATIVE
FIELD HEIGHT
( 20.00 )
O
-0.025
0.025
-0.025
0.025
0.81 RELATIVE
FIELD HEIGHT
( 25.00 )
O
-0.025
0.025
-0.025
0.025
TANGENTIAL 1.00 RELATIVE SAGITTAL
FIELD HEIGHT
( 30.00 )
O
ASTIGMATIC
FIELD CURVES
ANGLE(deg)
S T
30.00
23.41
16.10
8.21
-0.50 -0.25 0.0 0.25 0.50
FOCUS (MILLIMETERS)
DISTORTION
ANGLE(deg)
30.00
23.41
16.10
8.21
-5.0 -2.5 0.0 2.5 5.0
% DISTORTION
Forbes18_IODC209.len ORA 31-May-10
ASTIGMATIC
FIELD CURVES
ANGLE(deg)
S T
30.00
23.41
16.10
8.21
-0.2 -0.1 0.0 0.1 0.2
FOCUS (MILLIMETERS)
DISTORTION
ANGLE(deg)
30.00
23.41
16.10
8.21
-2 -1 0 1 2
% DISTORTION
best_forbes18_reduce10.len ORA 31-May-10

(a) (b)
igure 7. Nominal (before m ing and alignment tolerances) MTF for the for the best des efore order reduction (a) an e
order to predict the MTF of the as-built system, we assigned the state-of-the-art injection molding tolerances in

Table 2. Plastic tolerances (from the Triformix website).
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
M
O
D
U
L
A
T
I
O
N
15 30 45 60 75 90 105 120
SPATIAL FREQUENCY (CYCLES/MM)
Forbes18_IODC209.len
DIFFRACTION MTF
ORA 04-Jun-10
DIFFRACTION LIMIT
AXIS
T
R
0.5 FIELD ( ) 15.00
O
T
R
0.6 FIELD ( ) 20.00
O
T
R
0.8 FIELD ( ) 25.00
O
T
R
1.0 FIELD ( ) 30.00
O
WAVELENGTH WEIGHT
655.0 NM 1
587.6 NM 2
486.1 NM 1
DEFOCUSING 0.00000
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
M
O
D
U
L
A
T
I
O
N
20 40 60 80 100 120
SPATIAL FREQUENCY (CYCLES/MM)
Forbes18_IODC209_red
uce10.len
DIFFRACTION MTF
ORA 31-May-10
DIFFRACTION LIMIT
AXIS
T
R
0.5 FIELD ( ) 15.00
O
T
R
0.6 FIELD ( ) 20.00
O
T
R
0.8 FIELD ( ) 25.00
O
T
R
1.0 FIELD ( ) 30.00
O
WAVELENGTH WEIGHT
655.0 NM 1
587.6 NM 2
486.1 NM 1
DEFOCUSING 0.00000
F anufactur ign b d th
best reduced aspheric order design (b).

In
Table 2 and assumed that a focus compensator would be used at final assembly to maximize performance. For initial
ranking, we used the RMS wavefront error, as it is faster and effectively ranks even large errors (MTF can be zero for
large errors, which makes it less useful for ranking). The CODE V TOR option quickly evaluates the RMS wavefront
error of each lens. With the above tolerances, the best 4
th
order power series aspheric design increased from 0.382 RMS
to 0.453 RMS. On the other hand, the reduced order aspheric lens had a worst nominal field: 0.337 RMS, which
increased to 0.412 rms RMS tolerances with the thickness tolerances causing the largest errors followed by the radius
tolerances. For comparison, a lens designed using 18
th
order aspheres to the same specifications, without regard to
manufacturability, had a nominal error of 0.198 RMS that degraded to 0.625 RMS with these tolerances. Also, the 18
th

order Global Synthesis lens shown in Figure 4b had had a nominal error of 0.902 RMS that degraded to 1.322 RMS with
the Table 2 tolerances. Higher order aspheres may lead to a lower design error, but they do not always translate to better
as-built performance.
Parameter Value
Radius (%) 0.5
Thickness (um) 15
Index 5 x 10
-3

Irregularity (waves) 0.25
Decenter (um) +/-2

igure 8 shows the predicted as-built MTF for the best design before and after order reduction. 97.7% of the as-built
e note that it is not mathematically meaningful to tolerance the coefficients of a power series asphere
14
. The Forbes
F
systems are predicted to have higher MTF than that shown in the curves. The reduced order design has usable radial and
tangential MTF past 120 lp/mm for all but the corner of the field. While the 18
th
order design has higher MTF on-axis
and higher radial MTF at 15 deg off-axis at 120 lp/mm, it has no useable MTF at 120 lp/mm over the rest of the field at
120 lp/mm.

W
polynomial coefficients (or other orthogonal polynomial representations) are however more useful for tolerance analysis.
Even then, it is recommended that the designer place a specification on the RMS difference from the design surface, as
there will be asymmetric errors and errors at higher order than specified in the surface.


(a) (b)
ORA 04-Jun-10
As-Built MTF: Forbes18_IODC209

Frequency (lines/mm)
A
s
-
B
u
i
l
t

M
T
F
0.00 20.00 40.00 60.00 80.00 100.00 120.00
0.0
0.2
0.4
0.6
0.8
1.0
TAN 0 deg
RAD 0 deg
TAN 15 deg
RAD 15 deg
TAN 20 deg
RAD 20 deg
TAN 25 deg
RAD 25 deg
TAN 30 deg
RAD 30 deg ORA 31-May-10
As-Built MTF: Forbes18_IODC209_r
educe10
Frequency (lines/mm)
A
s
-
B
u
i
l
t

M
T
F
0.00 20.00 40.00 60.00 80.00 100.00 120.00
0.0
0.2
0.4
0.6
0.8
1.0
TAN 0 deg
RAD 0 deg
TAN 15 deg
RAD 15 deg
TAN 20 deg
RAD 20 deg
TAN 25 deg
RAD 25 deg
TAN 30 deg
RAD 30 deg
Figure 8. Predicted as-built MTF for the for the best design before order reduction (a) and the best reduced aspheric order design (b).

4. CONCLUSIONS

We have demonstrated that, in some cases, better as-built system performance can be obtained using lower order
aspheric surfaces than higher order aspheric surfaces. Starting with a spherical solution and incrementally increasing the
aspheric order to match the aberration content is one strategy. Another strategy is to start with an excessively high
aspheric surface and then reduce the design order. Order reduction is straightforward, when the surfaces are described
by orthogonal polynomials (e.g. the Forbes polynomial surface representation).

REFERENCES

[1] Forbes, G. W., Shape specification for axially symmetric optical surfaces, Opt. Ex. 15 (8) 5218-5226 (2007).
[2] Youngworth, R. N. and Betensky, E. I., Lens design with Forbes aspheres, Proc. SPIE Vol. 7100 71000W-
71000W-14, (2008).
[3] Rigler, A. K. and Vogl, T. P., Spline functions: an alternative representation of aspheric surfaces, Appl. Opt. 10,
1648-1651 (1971).
[4] Braat, J., Chebychev polynomial expansion of aspherical surfaces, JOSA A 73, 1890 (paper TuI5), (1983).
[5] Rodgers, J. M., Nonstandard representations of aspheric surfaces in a telescope design, Appl. Opt. 23, 520-522
(1984).
[6] Kross, J. and Schuhmann, R., Zur Korrektur optischer Systeme mit asphrischen Flchen, Optik 70, 76-85 (1985).
[7] Lerner, S. A. and Sasian, J. M., Optical design with parametrically defined aspheric surfaces, Appl. Opt. 39(28),
5205-5213 (2000).
[8] Greynolds A.W., Superconic and subconic surface descriptions in optical design, Proc. SPIE Vol. 4832 (2002).
[9] Rogers, J. R., Using Global Synthesis to find tolerance-insensitive design forms, Proc. SPIE Vol. 6342, (2006).
[10] McGuire, Jr., J.P., Designing easily manufactured lenses using a global method, Proc. SPIE Vol. 6342, (2006).
[11] Hopkins, H.H. and Tiziani, J. J., A theoretical and experimental study of lens centring errors and their influence on
optical image quality, Brit. J. Appl. Phys., 17, 33 (1966).
[12] Grey, D.S., Tolerance sensitivity and optimization, Appl. Optics, 9, 523 (1970).
[13] CODE V

, Optical Research Associates, 3280 E. Foothill Blvd, Pasadena, CA 91107.


[14] Rogers, J. R., Slope Tolerances, ODF08 Taipei, (2008).

Вам также может понравиться