Optical Design Using Stop-Shift Theory

Dave Shafer
• The use of first and 3rd order stop shift theory can lead to
new types of designs and a better understanding of existing
designs.

• No computations are necessary to benefit from stop-shift

theory – it just involves a few basic principles and some
temporary changes in aperture stop position.

• Experiments can be carried out in your head. Computer

calculations only happen after you are done with the
conceptual work.
 Copernicus
Ptolemy system

The view of Copernicus, that the sun is the center of the solar system, is widely
considered to be the correct view and the very complicated system of Ptolemy, with
epicycles and with the earth the center of the solar system, is considered wrong. But
neither is right or wrong, if they correctly predict the apparent motions of the planets.
One system is much simpler and easier to understand. Stop shift, especially temporary
shift, helps understanding in optical design through simplicity – just like Copernicus.
 Let’s start out with 1st order stop-shift theory, which
relates lateral and axial color.

1) If a system has axial color then lateral color is linear

with stop position. That means that there must be a stop
position that makes primary (1st-order) lateral color be zero.

2) If a system is corrected for primary axial color, then

primary lateral color is independent of stop position.

3) A thin lens with the stop in contact has no lateral color.

4) A thin lens at a focus has no axial or lateral color.

 Field lens

Design with broad spectral range

3 silica elements and a spherical mirror

gives a deep UV high NA objective.
 What is the aberration theory behind
this very simple design?

Answer – it involves stop-shift theory

 Both lenses are same glass type

Axial color is linear with lens power, quadratic with beam

diameter, so color here cancels between the lenses

Schupmann design with virtual focus

 Field lens

Offner improvement – a field lens at the intermediate focus

The field lens images the other two lenses onto each other
 Field lens
Low-order
theory of design

1) Put stop on first lens, then choose power of field lens to image it
onto the lens/mirror element. Stop is then effectively at both places.
2) Then neither of those elements has lateral color. Power of
lens/mirror element corrects axial color.
3) Field lens imaging and only one glass type corrects for secondary
axial color too (Offner theory).
4) Then can put stop anywhere.
• A key point – the aperture stop was only
temporarily located at a place where the theory is
simple to understand and the aberration correction
method becomes obvious.

• Then later the stop is moved to where it needs to be

– like in order to have a telecentric system.

• Once the aberrations are well-corrected they do not

change (at the lower-order levels) when the stop is
moved.
 • Lateral color depends
Telecentric design
on aperture stop
position, since axial
color is not corrected.

All same glass type • Move the stop around

and find out what
position makes lateral
color be zero.
Lateral color
for front stop • Then correct axial color
position at that location. Let’s
try using a diffractive
surface.
11
 Aperture stop position that
corrects lateral color

We move the stop position back and forth until we get lateral color = zero

12
 Aperture stop position that
corrects lateral color

If we correct axial color here, with a diffractive surface, then both

axial and lateral color will be corrected. Then we can move the stop
back to where we want it, and both color types will still be corrected.
13
• This same design method indicates where to add
lenses for color correction

• It minimizes the number of extra lenses needed for

color correction

• But it may indicate adding color correcting lenses

where we don’t want them, because of space
constraints

• Then we rely on conventional color correcting

techniques
14
 Aperture stop
Telecentric
image

Long working
distance design

Diffraction-limited monochromatic f/1.0 design with 5.0 mm field diameter

15
 Aperture stop
position for no
lateral color

Axial color not corrected

Aperture stop position for no lateral color may not be in a

desirable, place - as in this long working distance design. We don’t
want to put axial color correcting lenses there, in the long working
16
distance space.
 Cemented triplet Cemented doublets

In these cases you have to use two separated groups of color correcting
lenses, instead of just one, for axial and lateral color correction.
17
Correcting secondary lateral color
 Stop position for
best performance

SF2
SK16

Design is corrected for primary axial and lateral color, has

secondary axial and secondary lateral color.
• Suppose primary axial and lateral color are corrected.

• If a design has secondary axial color then secondary

lateral color is linear with stop position.

• So there must be a stop position then that corrects for secondary

lateral color.

• If you fix secondary axial color at that stop position, then both
secondary axial and lateral color will be corrected.

• Then you can put the stop anywhere with no effect.

Stop position for best
monochromatic performance

Stop position for no

secondary lateral color
Semiconducter wafer metrology inspection design
• KLA-Tencor in 2005 wanted a “perfect” .80 NA design for
.488u - .720u

• Requires correction of primary, secondary, and tertiary axial

color to get .999 polychromatic Strehl over that spectral range.

• Needs correction of primary lateral color, and secondary lateral

color is a very big problem – doesn’t hurt image quality but
gives wafer measurement error.

• Olympus, Tropel, and ORA all worked on this and could not get
any better than 10 to 20X more secondary lateral color than was
acceptable (needed = 1.0 nanometer over a 80u field diameter).

• I tried a design and also was about 10X too much lateral color
• The solution – use stop-shift theory

• I corrected primary axial and lateral color and partially

corrected secondary axial color.

• I found out where the stop position is then where

secondary lateral color is zero

• I corrected the remaining secondary axial color at that stop

position

• Then I moved the stop back to its telecentric position

 Telecentric stop position

Stop position for no secondary lateral color, when secondary axial

color is partly uncorrected.
Here I add a very low power “dense flint” lens (SF6 glass) with
anomalous dispersion and fixed secondary and tertiary axial color.
Result is a design with <1.0 nanometer of lateral color, but with
telecentric stop in very different stop position from this lens location.
 Aperture stop for telecentric design

Semiconducter wafer metrology inspection design

.80 NA microscope objective, 80u field, .999 polychromatic

Strehl from .488u-.720u, lateral color <1.0 nanometer.
 A 1.0X catadioptric relay system
developed using stop shift theory
 Bad coma
Small obscuration

Spherical mirrors, same radius, corrected for 3rd order spherical aberration
• If a design has spherical aberration then coma is
linear with stop position and astigmatism is
quadratic with stop position

• If spherical aberration is corrected then coma is

constant with stop position and astigmatism is
linear with stop position. Then, for non-zero
coma, there is always a stop position that corrects
for astigmatism.

• If both spherical aberration and coma are

corrected then astigmatism is a constant
Pupil position for no astigmatism

Two symmetrical systems make coma cancel, give a 1.0X magnification aplanat

Each half has a stop position which eliminates

astigmatism, since each half has coma. But
pupil can’t be in both places at the same time.
Astigmatism-correcting pupil positions are imaged onto each other by
positive power field lens.

System is then corrected for spherical

aberration, coma, and astigmatism, but there is Petzval
from field lens.
Thick meniscus field lens pair has positive power but no Petzval or axial or lateral color

Result is corrected for all 5 Seidel aberrations, plus axial and

lateral color. This shows how a simple building block of two
spherical mirrors was turned into something quite useful.
Plus, how stop shift theory is useful for thinking of a new design.
Aft-Schmidt Design
• If spherical aberration is uncorrected then coma is linear
with stop position and astigmatism is quadratic with stop
position.

• So then, for non-zero spherical aberration, there is always

a stop position that corrects for coma and either 2 or none
that correct for astigmatism.

• In some cases (like the Schmidt telescope) the stop

position which corrects coma also corrects astigmatism.
 Aperture stop at center of
curvature of M1

Much
spherical
aberration
Field mirror

Pupil at center of curvature of

M3, due to field mirror power

Three spherical mirrors with decentered pupil

Field mirror images M1 center of curvature onto M3 center of curvature
 Aspheric plate

Not there Exit pupil

Because of field mirror power the aspheric acts like it is in both the
aperture stop and the exit pupil, at the centers of curvature of M1 and M3
Design is good for rectangular strip fields
 Aspheric plate

Smaller aspheric but more higher-order aberrations

Aspheric acts like it is at the centers of curvature

of both M1 and M3, due to power of field mirror
 Aspheric mirror
and aperture stop

All-reflective - 3 spheres and one asphere

In all of these designs the image is curved

 After the system is given good correction, with the
Schmidt aspheric, the aperture stop can be moved if
that is wanted, maybe to minimize the size of M1.
Higher-order aberrations will be affected and the
best stop position is at the centers of curvatures of
M1 and M3
 2 X afocal pupil relay

For afocal case, Petzval is zero

Aspheric plate at either pupil or a concentric Bouwers lens

in either place does the spherical aberration correction
 Afocal version of system

Can be a building block in other designs

Best for rectangular fields, with long direction in X field direction.

Infrared Target Simulator Design
 External pupil of simulator matches internal pupil of missile head

A system from 1984 – customer wanted an infrared target simulator

to test missile heat seeking heads. Requires a distant external pupil.
Goals – all-reflective, inexpensive, 8 X 8 degree square
field, f/4.5, 200 mm aperture, unobscured, .05 to .10 millirad spot on a
flat image
 Two oblate spheroid mirrors

Part of the solution – two aspheric mirrors with same radius. Corrected
for spherical aberration, coma, astigmatism and Petzval. One of
Schwarzschild’s designs from the 1890’s
 Field is all set to one side of axis. Stop could be on either mirror.
Here it is on the larger mirror to minimize its size due to field size.
Now how do we get an external pupil?
Reed patent – images one pupil to another. Offner
independently invented this system but with finite
conjugates, imaging an object to an image, not pupils – which is
done here.
 Reed 1X afocal
pupil relay

Center of curvature of monocentric

Reed system is imaged by convex
Schwarzschild mirror onto concave
Schwarzschild mirror
Also a pupil
Schmidt aspheric needed to
correct Reed system could be
placed either at first pupil or
at second one.
By putting Schmidt aspheric onto this pupil an
oblate spheroid becomes a sphere!!!
Only one asphere and
that is a centered
one, not an off-axis one

Fold flat is made a very

long radius sphere.
 New idea for design – get almost constant astigmatism over field
and then correct with weak sphere on tilted fold flat mirror
Only one asphere and
that is a centered
one, not an off-axis one

Fold flat is made a very

long radius sphere.
This gives a 3X improvement
in performance.
Two-Axis Asphere Design
 Hard to
baffle
image

Schmidt aspheric is sum of what corrects the spherical aberration of

the primary mirror + what corrects for the secondary mirror
 Easy to
Two-Axis Aspheric Design baffle

Separate part of aspheric for primary mirror from that for secondary
mirror, and place on opposite sides of aspheric plate. Then tilt secondary
mirror and decenter its aspheric to follow secondary’s center of curvature.
 Instead of two rotationally symmetric aspherics on opposite
sides of the Schmidt plate, with decentered axis, combine
aspherics into a single non-rotationally symmetric aspheric.
Early warning missile defense system

Work I did in 1972, 40 years ago.

 If a missile comes over
the rim of the earth it will
be seen here by a satellite
against a black sky, but it
will be very close to an
extremely bright
earth, which gives an
unwanted signal that vastly
exceeds the missile’s heat
signal. But that is the easy
case. Much worse is when
the satellite is on the night
side and the missile is seen
against a sun-lit earth’s
limb.
 With the sun behind the horizon, the earth’s limb is
1.0 e+10 times brighter than the missile signal.
 Rim of aperture stop is source of diffracted light

Light
Two confocal
from
parabolic mirrors
earth
give well-corrected
limb
imagery
(Mersenne design)

Second aperture stop is smaller than image of first

Lyot stop stop, blocks out-of-field diffracted light from earth limb.
principle
 Aperture stop M1
Image from M3 is
not accessible

M2

Lyot stop
M3
Put Schmidt aspheric for M3 onto Add M3, a spherical mirror with
M2, then M2 parabola becomes a M2 at center of curvature
hyperbola
 Parabola + Schmidt aspheric Accessible image with
= hyperbola conventional
aspheres, but a long
system

parabola sphere
Image of M1 by M2, at
center of curvature of M3

Alternate design, with Schmidt aspheric

added to M1 instead of M2
 parabola

Parabola + decentered Schmidt

aspheric = 2-axis aspheric

2-axis aspheric sphere

Well-corrected image in Image is curved

an accessible location because of Petzval
High NA laser beam expander
 Surface at focus of
first surface Aplanatic surface

Surface radius chosen to correct

spherical aberration of first surface

(There are two different values that do this, on either side of the
perpendicular incidence condition. One speeds up the divergence, and
we choose that, while the other one slows down the beam divergence.)
Put stop at center of
curvature of first surface

1st surface has no coma or

astigmatism. 2nd surface is
Choose curvature of surface at at an image, 3rd surface is
the focus to make the chief ray aplanatic, so no coma or
go through the center of astigmatism, 4th surface has
curvature of the 4th surface no coma or astigmatism
because of where pupil is.
Spherical aberration cancels
between 1st and 4th surface
Stop can be placed anywhere, once
aberrations are corrected. Then
computer optimize the design
So system is insensitive to tilt of
entering collimated beam
Cascaded Conic Mirrors
• A conic mirror with the aperture stop at
either of its focii has no astigmatism of any
order.

• This can be proven mathematically with

the Coddington equations.

• Some interesting designs are possible

using this fact.
 Corrected for
Eye pupil astigmatism
ellipse
and Petzval

OSLO can’t
No common axis draw this part
of surface
of mirrors

hyperbola hyperbola

Collimated pupil
Part of a fundus camera to look at the eye’s retina
 Corrected for
astigmatism
ellipse
pupil and Petzval
Hand
drawn
No common axis part
of mirrors

hyperbola hyperbola
pupil

Each conic mirror shares one of its focii with the next mirror
2.2X afocal pupil relay
• Spherical aberration and coma are uncorrected in this
design but the pupil size is very small so they don’t
matter very much

• But still this means that the aperture stop and pupils
cannot be moved from the mirror focii without
hurting the zero astigmatism situation of the system
 Conclusion

• Stop shift theory gives insight into the aberration theory

of a design and also suggests new design possibilities

• Temporary stop shift is a powerful design tool and does

not usually require changing the actual final position of
the stop, which may be set by the telecentric condition
or other constraints