See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/326310164

Optimal baffle design for flat illumination with an integrating sphere

Conference Paper · July 2018

DOI: 10.1117/12.2311359

CITATIONS READS
0 240

1 author:

J. Haupt
Brookhaven National Laboratory
14 PUBLICATIONS   1,166 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Large Synoptic Survey Telescope View project

LSST Project View project

All content following this page was uploaded by J. Haupt on 11 February 2019.

The user has requested enhancement of the downloaded file.

 PROCEEDINGS OF SPIE
SPIEDigitalLibrary.org/conference-proceedings-of-spie

Optimal baffle design for flat

illumination with an integrating
sphere

J. Haupt

J. Haupt, "Optimal baffle design for flat illumination with an integrating

sphere," Proc. SPIE 10706, Advances in Optical and Mechanical
Technologies for Telescopes and Instrumentation III, 1070647 (10 July 2018);
doi: 10.1117/12.2311359

Event: SPIE Astronomical Telescopes + Instrumentation, 2018, Austin, Texas,

United States

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 7/12/2018 Terms of Use: https://www.spiedigitallibrary.org/terms-of-use

 Optimal baffle design for flat illumination with an integrating sphere
J. Haupta,
a
Brookhaven National Laboratory, Upton, NY 11973, United States

ABSRACT
An integrating sphere is often used to approximate a flat illumination field at some distant surface, but in the design of a
tunnel which bridges the integrating sphere and the surface it illuminates, many opportunities exist to contaminate the
geometric purity that would exist in the tunnel’s absence. It is shown that when constraining the tunnel/baffle geometry
to a cylindrical tube with flat baffles and circular apertures, a potentially optimized first-order design exists and provides
a suitable basis for more nuanced elaborations, such as the addition of secondary baffles or non-flat surfaces. A
parametric design formula is derived and provided for real-world use, non-sequential stray light analyses with
comparisons to alternate baffle designs is provided, and two reference designs as used in the CCD test facilities for the
Large Synoptic Survey Telescope are shown.

1. INTRODUCTION
Previous treatments of optical baffle design have been scarce, and always sufficiently broad as to give the reader a
starting point in approaching stray light suppression for a general optical system. In particular, a 1980 paper by E. R.
Freniere [1] and a 1977 paper by R. P. Breault [2] introduce key concepts that are built upon in order to present a
solution for a specific case, rather than a universal method for the general case.

1.1 Concept
Considering the integrating sphere aperture to be a Lambertian surface, the field uniformity at some distance is
approximated by a law. To maintain this relationship, the ideal setup would place the integrating sphere and the
surface it illuminates (hereafter, the detector) inside a perfect black body. In practice one must chose some enclosure,
using baffles to reduce reflections from the enclosure’s walls while not simultaneously introducing new reflections.

Starting with the apparent, any baffle system is made more effective by reducing the absorptivity of the surfaces, and any
such highly absorptive (black) coating is not ideal, especially at grazing incidences. Further recognizing that the
attenuation of light impingent on a system of baffles with reflectivity goes as , consideration of surface
preparation is discarded in favor of devising a geometry that maximizes reflections of unwanted rays. Having designed
said system, the question of optimizing the surface preparation is left to “best effort” and can be considered separately.

1.2 Design Principle

It is clear that one wants only the rays that emanate directly from the integrating sphere (without reflection) to impinge
on the detector, though this is a practical impossibility for any optical tunnel of finite extent. The presence of some kind
of baffling would seem to be indicated, except that increasing baffle quantity is not a panacea for suppressing stray light,
as baffle edges “glint” however thin they may be. Indeed, in an ideally-baffled system, the structured pattern resulting
from such grazing-incidence glint reflections may represent the primary contamination of the illumination field at the
surface under test.

With this in mind, the goal is set to optimize a design in which only a single aperture is exposed to glints directly from
the source. This harkens to the concept of a “critical object” [2], whereby any surface directly visible from the vantage of
the surface under test is a potential scattering surface. Extending that concept, critical objects that are also directly visible
from the vantage of the source can be considered particularly critical, which for clarity we call definite scatter surfaces.

For simplicity, the only tunnel form that will be considered is a tube with flat endcaps and the only baffles that will be
considered are flat planes with circular apertures, parallel with the tube end caps. The detector will be considered to be
coincident with the end cap of the optical tunnel, although in the case of an imaging detector inside a chamber, the end
cap becomes either the boundary of the detector’s chamber or an explicit field stop. For a system of given detector size,

Advances in Optical and Mechanical Technologies for Telescopes and Instrumentation III, edited by
Ramón Navarro, Roland Geyl, Proc. of SPIE Vol. 10706, 1070647 · © 2018 SPIE
CCC code: 0277-786X/18/$18 · doi: 10.1117/12.2311359

Proc. of SPIE Vol. 10706 1070647-1

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 7/12/2018
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
 source apertuure, and detectoor-source distaance, the param
meters that fullyy describe the tunnel
t are the tube
t diameter, the
baffle quantitty, the longitud
dinal baffle positions, and thee baffle aperturre diameters.

Figures 1-6 showcase

s somee baffle configuurations with noteworthy
n scaattered light chaaracteristics, illlustrating the presence
p
or lack of sinngle-bounce specular and difffuse reflectionss as a proxy forr overall effecttiveness. Definnite scatter surffaces are
colored red.

Starting withh Figure 1, it is seen that the entire

e interior of
o a cylindricall tunnel withouut baffles is a definite
d scatter surface.

SINGLE-BOUNCE SPECULAR REFLETCTIONS

-----'
SOURCE DEI

..... ... .........

.......
SINGLE-BOLTh'ICE DIFFUSE REFLIECTIONS

Figure1: A tunnnel without bafffles. The entire interior is a defiinite scatter surfa
face (red) and is prone
p to speculaar and diffuse refflections.
As tunnel diam
meter is decreaseed, specular refleections approachh grazing incidennce, making the limit of a small diameter tunnell is a light
pipe.

Adding a muultitude of baffl fles to the tunneel depicted in Figure

F 1 eliminnates single-bounce specular and diffuse
reflections, and
a indeed morre than that, buut when every baffle b aperture has the same (or
( nearly the same)
s diameterr, each
aperture is a definite scatterr surface, and will
w glint. We see s from this case,
c depicted in Figure 2, thaat baffles with
apertures thaat are in the fielld of view of both
b b added gratuitously.
the detectoor and the sourcce should not be

BM,FLE EDGE SINGLE -IBOUNCE SPECULAIR REFLECTIONS

SOURCE

Figure 2: A tuunnel with many

y baffles is pronee to specular “glint” reflections, as each baffle apperture is a definnite scatter surfaace (red).

The goal is thherefore taken to eliminate siingle-bounce sppecular and difffuse reflectionns while simulttaneously reduucing the
number of baaffles which arre definite scattter surfaces. Seeen in Figure 3, the addition of
o a single bafffle to the case depicted
d
in Figure 1 eliminates much h of the definitte scatter surfacce therein.

Proc. of SPIE Vol. 10706 1070647-2

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 7/12/2018
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
 SOURCE

Figure 3: A tunnel with a sin

/ SINGLE - BOUNCE SPECULAR REFLE(CTIONS

..........
...

SINGLE -BOUNC:E DIFFUSE REFLEC;TIONS

ngle baffle elimiinates much of thhe surface from which

w
..... ,
DEI

single-boounce reflectionss would normallyy occur.

And as showwn in Figure 4, by

b reducing thhe baffle apertuure sufficiently, all single-bouunce specular reflections
r can be
eliminated.

SOURCE
/ ...........
...........
..
NO SINGLE -BIOUNCE SPECULAR REFLECTIONS

SINGLE -BOON(CE DIFFUSE REFLEC:LIONS

Fiigure 4: A tunnel with a single narrow

n baffle.

Moving on, adding

a a secon
nd baffle as in Figure
F 5 can reeduce all singlee-bounce diffusse reflections, eliminating
e all definite

/
scatter surfacces on the tunn
nel’s inner walll.

NO SINGLE -BOUN(CE SPECULAR REFLECITONS

SOURCE

NO SINGLE -BOUN ICE DIFFUSE REFLEICTIONS

F
Figure 5: A tunnnel with two narrrow baffles.

What has nott yet been men ntioned is the diiameter of the optical tunnel itself, which has
h been held coonstant in Figuures 1-5.
c be seen thaat in general, as the diameter of the optical tunnel
However, it can t is decreased the quanttity of baffles needed
n
to eliminate single-bounce
s specular and diffuse
d reflectioons increases, and
a it is for diddactic reasons that
t the figuress so far
presented reaach this conditiion with just tw
wo baffles. It iss now natural to
t see that for a tunnel of suffficiently large

Proc. of SPIE Vol. 10706 1070647-3

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 7/12/2018
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
 diameter, a single baffle off a certain size and
a position will
w eliminate alll single-bouncce specular andd diffuse reflecttions.
This scenarioo is depicted byy Figure 6.

NO SINGLE-BOU!NCE SPECULAR REFwLECTIONS

SOURCE DE

NO SINGLE-BOUNCE DIFFUSE REF LECTIONS

Figure 6:: A tunnel with a single narrow baffle

b and sufficciently wide to elliminate definitee scatter surfacess on interior surfface.

It is now noteed that regionss exist which arre hidden from

m the view of thhe detector or of
o the source. That
T is, regionss exist
where baffless can be placed d without havinng definite scattter surfaces, allowing
a for thee possibility off additional scaattered
light suppresssion beyond thhe 1-bounce reggime.

2. SYSTEM PRESCR
RIPTION
The general case
c for this id
deal system is represented
r in Figure
F 7, where the hatched regions
r indicate “safe zones” for
baffles in adddition to the sin
ngle/primary baffle.
b It is usefful to call bafflles placed in thhis region seconndary baffles, having
h
no risk of expposing the deteector to single--bounce glints (no definite scatter surfaces).. Also defined here are labelss for the
system param meters. In desig gning the systeem we take as input
i the source radius , thee detector radiuus , and the tunnel
t
length , annd as outputs we w obtain the minimum
m tunnell radius , thee baffle aperturre radius , annd the source-baffle
spacing .
HEu

Zr
I
S-B

j
FXTF

C.

TEcToR
FIELD
V of
OFyW/, . - gOi1RCEp1Et
DETECTOR

NT OF SOURCE R.]LUMINAT[ON PAT

ri

rig

Vre
.motz A ZZ //ii..._
SIPACE FOR SECONIJARY BAFFLING F'REE FROM APER?

Figure 7:
7 General form of an optimizedd circular tunnel with
w a single priimary baffle andd any number off secondary bafflles.

A set of paraametric equatio

ons that definess the baffle system depicted inn Figure 7 is now derived.

We start by obtaining
o the baffle
b radius from the diffeerence of the deetector and souurce radii, − , the source-baffle
distance , the tunnel len
ngth ,, and thhe source radiuss :

Proc. of SPIE Vol. 10706 1070647-4

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 7/12/2018
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
 = + (1)

We also find using two additional sets of variables, = ( , , , , )= ( , , , , ):

= − (2)
And,

=− + (3)

We combine (1) and (2) and solve for , then combine (1) and (3) to solve for :

− +
+ = −

=− (4)

− +
+ =− +

( )
= (5)

(4) and (5) are combined, reduced to quadratic form, and factored:

2 ( − )
− =
−2 − 2 − +

−3 − = −

− ( + )−3 =0

1 /
= − − ±( + 14 + )
2

/
Which gives strictly in terms of and . Anecdotally it is seen that = − − −( + 14 + ) is
always the correct form, so for clarity we use:

/
= + +( + 14 + ) (6)

The other output parameters we’re looking for are = ( , , , , and/or ), and = ( , , ). We get
separately from (2) and from (3). From (2):

( )
= (7)
And from (3):

( )
= (8)

Proc. of SPIE Vol. 10706 1070647-5

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 7/12/2018
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
 comes from combining (7) and (8):

( + ) ( − )
=
+ +

−2 − = ( + )

= (9)

So, equations (6), (7), and (9), or equivalently, (6), (8), and (9) fully define the system. The complete formula for
designing the optimized optical tunnel is summarized below:

: :
: :
: ℎ : −

/
Minimum radius of tunnel: = + +( + 14 + ) (6)

( ) ( )
Distance of baffle from source: = or = (7) or (8)

Radius of baffle: = (9)

As it is impossible to ignore the strange coefficient in (6), it is acknowledged that further reduction may be possible.

3. NON-SEQUENTIAL ANALYSIS
This section presents a stray light analysis case study for some of the configurations described in 1.2. The purpose is to
extend the analysis presented there to two-bounce specular reflections as well as to provide a check of the assertions
regarding single bounce reflections.

All tunnels were of the same length, width, source size, and detector size. Baffles were modeled to be .5mm thick and
the 3D renderings (Figures 8, 10, 12, and 14) show only the direct (blue), single-bounce (green), and two-bounce (red)
specular rays of 500 launched from a circular Lambertian source, which is always on the left side. The detector plots
(Figures 9, 11, 13, and 15) were each the result of launching 1-million rays, and for visibility Gaussian smoothing was
turned on and set to 5 x 5. The detector plots are divided to show only the direct hits on the left plots, only the 1-bounce
hits on the center plots, and only 2-bounce hits on the right plots, both as an image and as a horizontal section. Intensity
and size is scaled the same for every plot.

Note that all of the “direct hits” plots (left plots on Figures 9, 11, 13, and 15) are identical. This is expected as the Zemax
non-sequential ray trace always launches the same ray pattern, and the distance from source to detector is invariant
throughout. The 1-bounce and 2-bounce plots (center and right plots on Figures 9, 11, 13, and 15) represent unwanted
stray light (contamination).

Baffle-edge “glint” reflections are not well represented in any of the plots, as the quantity of rays able to be traced with
the available computer was insufficient for a statistically significant number of hits on baffle edges. Therefore, in
interpreting the detector plots of Figures 9, 11, 13, and 15, we want to see low numbers of hits on the 1-bounce and 2-
bounce (center and right) plots, and as few primary baffles as possible, as we expect primary baffles to be contributing 1-
bounce specular reflections that are not represented in the detector plots.

Proc. of SPIE Vol. 10706 1070647-6

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 7/12/2018
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
 Figures 8 andd 9 present the case of a tunnnel without bafffles, which corrresponds to that presented inn Figure 1 abovve. Note
that for a refllective tunnel contaminated
c light overwhelm ms the direct illlumination, as we see by com
mparing the cennter and
right plots to the left plot in
n Figure 9. Alsoo note the 1-boounce and 2-boounce ray pathss in Figure 8, where
w it’s imm
mediately
obvious that a single primaary baffle locateed it the tunnell’s approximatee mid-point wiill do nothing to
t suppress 2-bbounce
specular refleections.

`\.
Figure 8: Non-sequential raaytrace for case depicted
d in Figurre 1 (tunnel withh no baffles). Raays emitted direcctly from the souurce are
colored blue before striking a surface, green after 1 bounce, red
r after 2 bouncces, and mustardd after 3.
MUTP4OGO

MUTP4OGO
I C l=
aoMlnrits vil linen value ate value
i>
Si=

c3i
li

hi

:!
.;

,f irv
A s Nn
11° 7:
Direct Hits Hits afterr 1 Specular Boounce Hits after 2 Specular Bounnces

Figure 9: Dettector plots resullting from non-sequential ray traace shown in Figgure 8. Left: Direect hits only. Cennter: 1-bounce hits
h only.
Right: 2-bounce hits only. Note the high illlumination due to 1- and 2- bouunce reflections asa compared to the t direct illumnnation.

Figures 10 annd 11 present thet case correspponding to Figgure 2 above, where

w a tunnel is naively popuulated with a multitude
m
of similar priimary baffles. Suppression off 1- and 2- bouunce reflectionss is good, but thhere are as many definite scaatter
surfaces (i.e. glinting bafflee edges) as therre are primary baffles (7 in thhis example).

Proc. of SPIE Vol. 10706 1070647-7

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 7/12/2018
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
 I
Figure 10: Non-sequential
N raaytrace for case depicted
d in Figuure 2 (tunnel withh many primary baffles). Rays emitted directly from
f the
souurce are colored
d blue before strikking a surface, green
g after 1 bouunce, red after 2 bounces, and muustard after 3.

Loa
.w.. .mwwo .

C_
:card/note volut in. ..,..
6
4

^
4 4

1.
,
4

Ii
hi
4

Hill
4
4
4
4

X coordinate value

-
JO I

eT7kfle

Direct Hits Hits afterr 1 Specular Boounce Hits after 2 Specular Bounnces

Figure 11: Detector

D plots reesulting from nonn-sequential ray trace shown in Figure
F 10. Left: Direct hits only. Center: 1-bounnce hits
only. Right: 2-bounce
2 hits onnly. Stray light iss well suppresedd, but reflectionss from definite sccatter sources (primary baffle eddges) are
noot represented.

In Figures 122 and 13 we see the case of a single primaryy baffle of optimmal placementt and size correesponding to Figure 6
above. Singlee-bounce reflecctions are welll suppressed wiith only a singlle definite scattter surface, buut .0no effect is had on
2-bounce refflections.

Proc. of SPIE Vol. 10706 1070647-8

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 7/12/2018
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
 Figure 12: Noon-sequential ray
ytrace for case deepicted in Figuree 6 (tunnel with single well-placced primary baffl fle). Rays emittedd directly
from thhe source are collored blue beforee striking a surfaace, green after 1 bounce, red aftter 2 bounces, annd mustard after 3.
I

...... :it ...lu.

I

-? -a
I 2-
2,-

6'`'s

2-
!:

hi

/JP/NA
X ceartilnate value

Direct Hits Hits afterr 1 Specular Boounce Hits after 2 Specular Bounnces

Figure 13: Detector

D plots reesulting from nonn-sequential ray trace shown in Figure
F 12. Left: Direct hits only. Center: 1-bounnce hits
only. Right: 2-bounce hits onlly. A single prim
mary baffle supprresses 1-bounce reflections well but has little im
mpact on 2-bouncce scatter.

Figures 14 annd 15 show thee optimal case, in which a sinngle well placedd and sized priimary baffle is supplementedd by
secondary baaffles, which arre in the shadoww of the primaary baffle relatiive to the sourcce, or equivalenntly, in the shaadow of
the primary baffle
b relative to
t the detector,, or perhaps booth. Single and 2-bounce (andd in fact higherr) scatter is welll
a the numberr of primary sccatter surfaces is kept singulaar.
suppressed, and

Proc. of SPIE Vol. 10706 1070647-9

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 7/12/2018
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
 `i
Figure 14: Non-sequential ray ytrace for optimaally designed tunnnel as describedd by Figure 7 andd equations (6)-((9). Rays emittedd directly
from thhe source are collored blue beforee striking a surfaace, green after 1 bounce, red aftter 2 bounces, annd mustard after 3.

mß
tlinab value
aatrupwat

Ineehsrant IretIlAnta
i
P
i
i i:

X COONIIIt VIIIII

I
if

I ti
ii!

Direct Hits Hits afterr 1 Specular Boounce Hits after 2 Specular Bounnces

Figure 15: Detector

D plots reesulting from nonn-sequential ray trace shown in Figure
F 14. Left: Direct hits only. Center: 1-bounnce hits
only. Right: 2-bounce
2 hits onnly. The single primary
p baffle suuppresses 1-bounnce reflections well,
w and secondaary baffles whichh do not
have definite sccatter surfaces arre added to supppress 2-bounce reeflections.

Perhaps the most

m illustrativ ve comparison is found when examining Figgures 8/9 againnst Figures 12/13, and again in i
considering Figures
F 10/11 against
a 14/15. In the former case
c we see both the deficienncy of the bafflee-free tunnel inn
Figures 8 andd 9 and a samp pling of the limmitations of a siingle baffle in Figures
F 12 andd 13. In the latter case we first see the
effectivenesss of a multitudee of baffles in Figures
F 10 andd 11, but are addvised against this
t method duue to the dangerr of
baffle-edge glints.
g We thenn see in Figuress 14 and 15 thaat there exists a configurationn using a singlee primary bafflee (in
addition to seecondary bafflees) that is in faact more effectiive than the arbbitrarily multittudinous primaary baffles of Figures
F
10/11.

The configurration shown inn Figures 14 annd 15 show a reepresentative finalf design usiing the optimizzed configuratiion of
Figure 7 and equations (6)--(9) as a startinng point. The fiirst-order desiggn characteristiics (primary baaffle number, siize,
placement, annd tunnel diam
meter) are preseent, and seconddary baffles weere added in thee hatched regioons of Figure 7. 7 Other
refinements are
a possible annd expected. Foor example, a non-sequential
n ray trace may be used to infoorm more nuannced
placement off secondary bafffles, or non-flaat shaping for any and all of the baffles. Othher characterisstics are then leeft to
best-effort im
mprovement, foor example the black/absorptiive coating choosen for interioor surfaces, andd the thickness of the

Proc. of SPIE Vol. 10706 1070647-10

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 7/12/2018
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
 exposed edge of the primary baffle, which should be as thin (sharp) as practicable. In the next section, two as-built
systems are taken as representative examples of this kind of building-up of the basic optimized configuration.

4. REFERENCE DESIGNS
Two designs as used in the CCD test facilities for the Large Synoptic Survey Telescope (at BNL and SLAC) are shown
here. These systems conform to the requirements of tunnel diameter, baffle size, and baffle placement dictated by the
optimized design described in Figure 7 and equations (6)-(9), and give an idea of the range of variation possible even
given the constraints of Figure 7 as a starting point.

An obvious limitation of real-world designs is that sensors requiring a self-contained enclosure (e.g. a vacuum chamber)
increase the minimum quantity of definite scatter surfaces from 1 (the primary baffle itself) to 2 (an “end cap” in
addition to the primary baffle). The end cap’s aperture is in effect a field stop, and as with the primary baffle it is
advantageous that its aperture be as thin as practicable. For the sake of equations (6)-(9), the aperture of the output end
cap (referred to in the below designs as the field stop) is considered to be the detector, which must itself be sized so that
its penumbra does not impinge on the actual sensor surface. The existence of the field stop is evident in both designs
below.

4.1 Reference Design A

The LSST single-CCD electro-optical test station consists of a cylindrical tunnel with a 3D printed conical primary
baffle having knife-edge apertures, three 3D printed secondary baffles with plain edges, a 3D printed field stop (end cap)
with knife-edge apertures, black-flocked inner walls, and a dark slide. The layout of the baffles is evident in the
rendering given in Figure 16, where the knife-edge features on the primary baffle and field stop can also be seen. Note
the source is on the left, as indicated by the blue integrating sphere.

In this system, the tunnel length is 735mm, which was set by space constraints on the test stand’s optical table. The
detector radius is taken as 33.23mm, which is the half-diagonal of the 47mm x 47mm square field stop, as sized for a
42mm square LSST CCD some distance downstream. The source radius is 31.75mm.

From equation (6) we have ≥ 97.46, from (9) we have = 32.49, and from (7) we have = 365.41. The actual
tunnel radius is 131.75mm as determined by the availability of ready-made tubes, the actual primary baffle radius is
~33mm, having margin for decenter and other errors, and the primary baffle’s aperture was placed within 1mm of
365.41mm from the source.

Non-sequential ray tracing shows that a cone-shaped primary baffle often has a beneficial quality as compared with the
flat alternative. In the case of this reference design, it was found that a cone angle of 15° had the effect of reflecting rays
coming directly from the source back into the source, or else onto the source bulkhead at an angle that dramatically
increased the number of bounces needed before contact with the detector. This improvement was attained without
expense to the system’s other stray light suppression qualities and was incorporated as an enhancement to the optimized
design otherwise described by (6) – (9).

The choice of having placed secondary baffles only on the source-side of the primary baffle, however, was not informed
by analysis and nothing is implied here about the preference of putting them on one side, the other, or both.

Proc. of SPIE Vol. 10706 1070647-11

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 7/12/2018
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
 FIELD STOP/END CAP

INTEGRATING
SPHERE PRIMARY BAFFLE
SECONDARY BAFFLES

Figure 16: Optical tunnel used for single-CCD electro-optical testing in LSST sensor test and integration facility.

In addition to the possibility of a conical primary baffle, the freedom of geometry permitted by 3D printing motivated the
addition of a knife-edge aperture formed by an array of conventional razors on both the primary baffle and the field stop.
A photograph of the primary baffle is shown in Figure 17. Eleven razor blades were overlapped into a fixed-aperture iris
and mounted to the mating geometry designed into the conical baffle. The number of razors was set by the diameter of
the opening given by equation (9) and by the access requirements of mounting each razor with a single screw. Individual
razors were spray painted flat black before assembly, and the completed assembly was given a final coat of flat black
spray paint on both sides.

KNIFE EDGES
(RAZORS)

Figure 17: Primary baffle for LSST single-CCD eletro-optical test tunnel. Conical form with aperture defined by knife edges.

The field stop was manufactured in a similar way, and the interior of the tunnel was coated with black flocking material
with all baffles mounted and positioned by friction-fit against said flocking.

Proc. of SPIE Vol. 10706 1070647-12

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 7/12/2018
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
 4.2 Referen
nce Design B

The electro-ooptical test stattion for the LSSST 9-CCD mosaic subassembbly is compriseed of a more coonventional boox
enclosure with flat sheet meetal baffles andd apertures cutt from black aluuminum foil. TheT 3D printedd field stop (endd cap)
provides the interface to thee sensor testingg cryostat, but otherwise empploys the same style of black foil aperture as is
employed forr the baffles. The
T interior is powder-coated
p black and is acccessed via a full-size
fu side-oppening hatch with
w air-
spring assistss. A light-tight seal lines the hatch
h and conssists of an elasttomeric bulb seeal covered witth black rubberrized
fabric from Thorlabs.
T The layout
l is shownn in Figure 18,, with the integgrating sphere source
s on the leeft.

Here, the tunnnel length is

i 953mm, as set
s by optical taable space. Thee detector radiuus is taken as
a 96.17mm (thhe half-
diagonal of thhe 136mm x 136mm square field
f stop), whhich was sized for
f the 127mm m square 9-CCDD mosaic somee
distance dowwnstream. The source
s radius is 50.00mm.

From equatioon (6) we have ≥ 213.68, from (9) we have h = 71.23, and from (77) we have = 438.14. Thhe closest
distance of anny of the box enclosure’s
e waalls to the optical axis is 355mmm, and the furrthest is 870mm m. The actual primary
p
baffle radius is 72mm, haviing margin for decenter and other
o errors, annd the primary baffle’s apertuure was placed within
2mm of 338.14mm from th he source.

Due to the coost of 3D printiing parts of thee scale needed for this systemm, knife-edge baffle
b and field stop aperturess were
not considereed, nor was a conical
c primaryy baffle. Insteadd, the large recctangular sheett metal baffle cards
c were giveen
circular cutouuts larger than that specified by the baffled design. Black--coated aluminnum foil was taaped to these baaffles,
and the aperttures of the apppropriate sizes were cut into the
t foil, yieldinng edges muchh thinner than thhe sheet metal,,
although not razor-sharp ass in Reference Design
D A.

The box itsellf is constructed of T-slotted aluminum struuctural extrusioons and sheet metal
m walls. The non-cylindriccal form
remains conssistent with thee optimized dessign from abovve in that speecifies the minnimum tunnel raadius and doess not
preclude the possibility of a non-circular cross
c section.

SECOND
DARY BAFFLES
-
i

F
FIELD STOP/EN
ND CAP

IN
NTEGRATING
G
PR
RIMARY BAFF
FLE
SPHERE

F
Figure 18: Opticaal tunnel used foor 9-CCD electroo-optical testing in LSST sensor test and integrattion facility.

Proc. of SPIE Vol. 10706 1070647-13

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 7/12/2018
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
 SECONDARY BAFFLES

PRIMARY BAFFLE

Figure 19: Close-up photograph of region extending from integrating sphere to primary baffle. Enviornment is illuminated from the
direction of the detector to show all the baffle edges.

5. CONCLUSION

Stray light suppression is a broad topic with a design landscape that changes readily from one optical system to another.
This treatment has attempted to distill the topic for a single, very specific case; that of projecting a flat illumination field
onto some surface by use of an integrating sphere.

By imposing the modest starting constraints of considering all optical tunnels to be some permutation of flat baffles in a
circular tunnel, by considering the effect of exposed baffle edges which “glint” as a primary driver, and by analyzing
baffle geometries on the basis of small numbers of specular bounces, a potentially globally-optimized design has been
described to first order. That is, a particular optimal geometry is presented as a firm foundation for the critical
dimensions of the system, and for further elaboration using modern non-sequential optical design tools. The concern of
baffle edge glints is formalzied via the concept of a definite scatter surface, and baffles are categorized as one of two
types (primary or secondary) depending on whether its aperture is a definite scatter surface. A set of parametric
equations has been derived to calculate the design given the dimensions of the source and the detector, and two real-
world systems build using this design perscription have been presented.

REFERENCES

[1] E.R. F eniere, “First-order design of optical baffles”, Proc. SPIE A81-36878 16-74, 19-28 (1980)

[2]
2] R. P reault, “Problems and techniques in stray radiation suppression”, Proc. SPIE 0107 (1977)

Proc. of SPIE Vol. 10706 1070647-14

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 7/12/2018
Terms of Use:View
https://www.spiedigitallibrary.org/terms-of-use
publication stats