Design Considerations for Precision Solar Simulation

D. L. Fain

Four basic types of optical system are described which are suitable for precision solar simulation where

the collimation angle is precisely held to 32 min of arc. Maximum beam diameters possible with each

type of system are given, and the optical requirements for all components are discussed in detail. Various

light sources for solar simulation are considered in terms of maximum beam intensity and beam aper-

ture. An approach to beam homogenization is discussed as a possible future development.

1. Introduction

Engineering efforts to simulate the sun in the labora-tory have been mainly concerned with duplicating theradiant intensity and spectral distribution of solarenergy. These efforts were oriented to materials test-ing and thermal transfer studies for spacecraft, re-quiring only a reasonable facsimile of the solar spectrumand sufficient incident power on the workpiece toequal space solar intensity. Collimation angle con-siderations were important only to avoid excessive errorsdue to shadowing, and beam angles of 2° to 5° were notuncommon.

Solar attitude sensing for space vehicle attitudecontrol is being used in the orbiting solar observatoryprograms and in certain space probe experiments. It islikely that the interest in determining the solar vectorfrom space vehicles will be increasing with time asmore sophisticated satellites and probes are developed.In order to predict accurately the performance of solarattitude control systems, it is necessary to provide asimulated solar source having high optical quality andduplicating the sun in angular subtense and parallaxproperties. While it is desirable to have the precisionsolar simulator equal the intensity and spectral dis-tribution of the sun, these considerations are secondaryto reproducing its optical properties.

The technical discussion in this article will be limitedto precision solar simulators which produce an imageof the simulated sun in the focal plane of a lens placedin the simulator beam and focused at infinity. Theimage formed must subtend accurately an angle of 32min of arc with reference to the focal length of the lens.For purposes of definition, the usable diameter of thebeam at a given distance from the solar simulator will beconsidered to be the beam limits which will produce a

The author is with the Infrared and Optics Branch, ResearchDivision, Melpar, Inc., Falls Church, Virginia.

Received 4 August 1964.

near-perfect solar image in the focal plane of a lensfocused at infinity.

This article discusses optical designs, light sources,and special techniques as they aff ect angular precision,beam uniformity, beam diameter, and stability of pre-cision solar simulators in general. Specific recommen-dations are given for the development of special-purposecandidate systems.

11. Technical Discussion

A. Optical Designs

There are four basic types of optical systems whichmay be used for special-purpose precision solar simula-tion. For purposes of simplicity, these systems will beanalyzed in terms of refracting optical designs, althoughreflecting systems and combinations of reflecting ele-ments and lenses would probably be used in practicalsystems.

1. Type I Simulator

Figure 1 is an optical schematic of a type I solarsimulator. This consists primarily of a relay lens whichreimages an emitting source in the plane of a pre-cision circular aperture. An objective lens projects theimage of the aperture to infinity, producing a beamwhose collimation angle is determined by the focallength of the objective lens and the diameter of theaperture.

Some relationships of interest which define the opera-tion of a type I sun are given as:

Da = Of,

O(Dof, + Dfo + Off2)DL = D,

Since

f2 Ofom = - = If, D(

Eq. (2) can be rewritten

(1)

(2)

(3)

December 1964 / Vol. 3, No. 12 / APPLIED OPTICS 1389

OBJECTIVE LENS

RELAYSOURCE ELS APERTURE

Fig. 1. Type I solar simulator

DL = -1- + Ofo + I fo (4)

Also

D, Djfo O f foDO = - - (5)

or,, ft D,

where Da is the diameter of the aperture, 0 is the solarangle equals 32 min of arc equals 9.3 X 10-s rad, fo isthe focal length of objective lens, D is the diameter ofobjective lens, DL is the diameter of relay lens, fi is thefocal distance from source to relay lens, f2 is the focaldistance from relay lens to aperture, D is the diameterof useful portion of source, m is the magnification ofrelay lens, and F1 is the optical F number of relay lens= fI/DL.

The useful maximum diameter of the beam is equalto D at the objective lens. This is reduced by R asthe distance R from the objective lens is increased.If the workpiece must be located rather far from thesystem, or if translation of the work in the beam is re-quired, it would be highly desirable to provide a largediameter beam. The limit on useful beam diameter isimposed by the source diameter, the minimum F num-ber of the relay lens, and the back focal length of therelay lens required to provide adequate clearance tothe source.

The quality of the relay lens must be such thataberrations in the image do not destroy the uniformityof aperture illumination. This creates the need forcamera-quality, low F number relay lenses. A speciallens could be designed for this application, but it ismore practical to use a pair of highly corrected cameralenses placed front-to-front and each focused at in-finity for this function. The major drawback to thisapproach is the rather large light losses which arecaused by the sizable number of optical elements usuallyfound in lenses of this type. Typically, the trans-mission through a lens pair of this nature is approxi-mately 60%.

The objective lens is used essentially on axis andneed only be corrected for spherical and chromaticaberrations. If a small telescope focused at infinityis traversed across the objective lens, the sun imageshould remain fixed with relation to the telescope axis.Any movement of the sun image indicates the presenceof spherical aberration. In practice, a precision align-ment autocollimator-telescope is aligned within afraction of see of arc on a fixed, reflecting optical flatmounted to obscure half of the objective lens area.

The alignment autocollimator is positioned so thathalf of its entrance aperture views the mirror andthe other half looks into the simulator objective lens.The alignment instrument is first placed along the op-tical axis of the simulator, and autocollimation is per-formed on the mirror. The position of the sun imageis noted in the telescope, and the measurement is re-peated at another position across the objective lensface. In this manner a quantitative measure of theeffect of spherical aberration can be obtained. It isapparent that very high quality objective lenses arerequired if the simulated solar vector is to remain fixedacross the usable beam diameter.

Chromatic aberration of the objective lens willcause color rings to appear around the solar imagewhen viewed through a telescope. This may seriouslydefocus the solar image and will be undesirable in anyprecision instrument. Telescope quality objectivelenses are color-corrected and have little spherical aber-ration. However, these lenses are not usually correctedfor near infrared wavelengths, and in many applicationsit may be desirable to use a precision, off-axis parabolicreflector instead of a refracting optical element.

The uniformity of beam intensity across Do may be ofinterest in certain applications. Any shadowed ormasked regions in the plane of the relay lens will createcorresponding shadows or "holes" in the beam. If it isdesired to use reflecting optics to perform the relay func-tion, only off-axis types which cause no vignetting orbeam blockage should be considered. If it can beassumed that reasonable care is exercised in selecting therelay optics, the output beam intensity will be free fromlocal intensity variations but will have a gross intensitygradient such that the edge is approximately cos4 a lessintense than the center, where a is the half-anglebetween the source center and the first relay lens edges.Depending on the geometry of the lens and the departureof the source from Lambert's law, the exact intensityfalloff will vary slightly. As a typical example, the useof F1 = 1.0 would result in an edge intensity of approx-imately 64% of the beam center intensity.

At the expense of a further reduction in over-all beamintensity a filter having a suitable density gradient canbe placed at the relay lens or over the objective lens tocorrect this defect. A simple procedure for fabricatingthe filter from photographic film is to expose the filmto the beam directly over the objective lens. Whensuitably developed, a reasonably accurate correctionfilter results. Unfortunately, in order to obtain theemulsion linearity needed for accurate correction, theover-all density must be quite large, resulting in ex-cessive beam losses. Other techniques using photo-graphic materials have been suggested as well as dyediffusion methods and filters that incorporate narrow,circular opaque zones having variable widths. In anyevent, a correction filter having 100% efficiency wouldreduce the beam intensity at the center by 64% (usingF1 = 1.0).

For illustrative purposes a sample calculation of themaximum value of Do which can be obtained using asource having D = 2.54 mm and F = 1 is

1390 APPLIED OPTICS / Vol. 3, No. 12 / December 1964

presented. For mechanical clearance and thermalprotection of the relay lens, assume fi = 5.08 cm.Note that the back focal distance from a complex,multielement lens is considerably less than its focallength. From Eq. (5) we obtain

0.254 0.254(50.8) (9.3 X 10-3)2 50.8(5.08)Dn = -- _ _ _ _____

9.3 X 10-3 5.08 0.254

assuming fo = 50.8 cm and 0 = 32 min of arc = 9.3 X10-3 rad. Thus

Do = 27.3 - 2.54 - 8.8 X 10-4 = 24.8 cm.

Note that the maximum usable beam diameter willbe reduced to half this value at a distance of 1335 cmfrom the objective lens. If a larger beam diameter isrequired, it will be necessary to provide a larger sourcearea; or, alternately, a number of collimators can bemounted alongside each other, taking into considerationthe dead areas between the individual units.

The central beam spectral intensity out of the simula-tor Io? is independent of the choice of optical componentsexcept for transmission losses.

O're W8X10 = 4 (6)

4

where eTV,, is the total blackbody spectral radiant emit-tance of the source modified by the spectral emissivityof the surface over the wavelength interval of interest;and T is the product of all optical losses (, 2, T3 . ..

produced by the lenses and correction filters.A sample calculation is given, based on a tungsten

filament source at 3000'K, e = 0.4, and a wavelengthinterval of 0.4 jt to 1.15 l (the spectral response regionfor a silicon solar cell). is assumed to be 0.6 for multi-element lenses.

rX2 = 1.15 aIo = 6.33 X 10-6 WxdX = 1.05 X 10-3 W/cm2.

J1 = 0.4

Solar intensity outside the earth's atmosphere and atthe earth's orbital radius from the sun is approximately0.14 W/cm2. The fraction of this power in the wave-length interval of 0.4 A to 1.15 ,u is 9.28 X 10-2 W/cm2 .It can be seen that the solar simulator described aboveproduces 1.13 X 10-2 solar constant in this wavelengthinterval. For a silicon solar cell an approximate gainof 2.0 can be assumed to account for the better matchingof spectral response with a 3000'K source than for solarenergy. Thus, for a silicon photovoltaic detector a3000'K tungsten source solar simulator produces theequivalent of approximately 2.3 X 10-2 solar constant.If a filter is inserted to increase beam uniformity, anadditional loss factor of 0.6 must be included. Unlessthe source temperature is increased, there is little possi-bility that a substantially more intense simulator couldbe developed. Optical losses could be reduced andthe surface emissivity of the tungsten might be in-creased by a suitable coating of black metallic oxide,although little success is expected with this technique.In any case the increased output will not exceed fourtimes the value shown unless the source is operated at a

higher temperature. The limit for tungsten is ap-proximately 3200'K, and at this temperature shortfilament life should be expected.

The solar image when viewed through a telescopewill contain the brightness irregularities of the portionof the source surface which is reimaged on the simulatoraperture. For this reason a source must be selectedwhich has a uniform emitting surface when used in atype I solar simulator.

2. Type II SimulatorAnother type of solar simulator optical system which

is useful for certain purposes is shown schematically inFig. 2. Notice that the optical design is similar tothat used in a slide projector. In this configuration thesource is imaged by the relay lens onto the objectivelens with sufficient magnification entirely to fill the ob-jective lens with the source image. The objectivelens is focused to project an image of the relay lens at adistance R, forming a large, evenly illuminated circleof light. When observed at R within the illuminatedarea, the objective lens appears as a uniform circle oflight subtending an angle defined by R and Do. Forsolar simulation the ratio Do/R is maintained at 9.3 X10-3.

The primary advantage of a type II simulator is thelarge work area available. Unfortunately, this type ofsystem suffers because the sun does not appear at in-finity and because the vector to the sun varies withinthe illuminated circle at R.

The equations which describe the geometry of a typeII simulator are given below, using the same notationpreviously defined.

(7)

(8)

and

f0 = f2,

R = DB0

D = DDL = D,fi0 F1°

(9)

where Di is the diameter of the illuminated area atrange R.

The intensity of the beam at R is identical to that ofa type I simulator and is given in Eq. (6).

A parallax error of S/2R rad will be present, whereS is the distance to the optical axis of the simulatorat R. At the beam edge this has a maximum value ofDi/2R rad.

The illuminator intensity across the image at R is amaximum on the optical axis and falls to cos4 a at the

IMAGEPLANE

SOURCE LENSCTIVI RELAY LENLENS--- --

R - -

Fig. 2. Type II solar simulator.

December 1964 / Vol. 3, No. 12 / APPLIED OPTICS 1391

edge, as defined previously. This can be corrected byfilters similar to those described for a type I simulator.In addition, the intensity of the apparent sun will falloff at the edge of the solar disk by an amount approxi-mated by cos4B, where B is the half-angle subtended bythe objective lens at the center of the relay lens.

The relay lens should be of sufficiently high qualityto illuminate uniformly the objective lens, since thegranularity of the source image formed in the objectivelens plane will be present in the image of the sun whenviewed by a telescope placed at R. Objective lensaberrations may cause a nonuniformly illuminatedarea at R, a factor which may not be important in manyapplications.

A type II simulator has often been used with a high-wattage, coiled filament tungsten lamp having a largeeffective emitting area. A spherical mirror positionedbehind the lamp serves partially to fill in the inter-stices between the filament coils. Low-cost projectorcondenser lenses comprise the relay lens assembly,and an inexpensive objective lens completes the unit.The primary use of a system of this sort is to provide atarget for coarse and medium-fine solar attitude controlsensors positioned over a relatively wide area on a spacesatellite during testing. When desired, the beam canbe introduced through suitable optical ports into vac-uum environmental chambers. However, this systemis unsuitable where high precision is demanded of thesimulator.

3. Type III SimulatorAn interesting modification of a type II simulator

gives rise to an optical system which is sufficientlyunique to warrant the separate classification as a typeIII simulator. The major difference is that the ob-jective lens projects the image of the relay lens at in-finity and 0 is defined by the angular subtense of the re-lay lens to the center of the objective lens. Figure 3provides a representation of this type of system.

Employing the same notation as before, the geometryof the optical system can be described by the followingequations:

fo = f (10)

DL = Ofo, (11)

D = fDL - FO (12)

The useful beam diameter is reduced by OR, where Ris the distance from the objective lens to the workpiece.

Fig. 3. Type III solar simulator.

As in any solar simulator the intensity of the outputbeam is given by Eq. (6), which is independent of theoptical design.

The beam intensity at the edge of the objective lensis reduced by cos4B, where B is the half-angle sub-tended by the objective lens from the center of therelay lens. When viewed through a telescope, theedge of the simulated solar disk will appear to have anintensity approximately cos4 a less than at the imagecenter, where a is the half-angle subtended by the relaylens from the center of the source. This latter defectcan be corrected by a suitable filter placed at the relaylens similar to those discussed previously.

In most respects the type III simulator is opera-tionally similar to a type I simulator. An importantdifference is that the granularity of the source is notreproduced in the image formed by an optical systemplaced in the outgoing beam of the simulator. If thesource consists of filament coils, the solar image is notadversely affected. However, the image of the fila-ment on the objective lens will cause nonuniformityof illumination over the beam diameter. In manyinstances this may not be objectionable, and large-area, nonuniform sources may be incorporated withlarge-diameter objective lenses to cover a large targetarea or to enable the system to use high brightness butnonuniform sources such as carbon arc lamps or xenon-mercury discharge lamps.

Considerations of objective lens quality are similarto those for a type I simulator, and spherical and chro-matic aberrations will appear as serious system defects.The requirements for relay lens quality are similarto those discussed for a type II simulator, and thisoptical element may be selected on the basis of the beamuniformity desired over the useful beam diameter.

4. Type IV SimulatorA major drawback to the development of a large-

aperture type III simulator results from the large-diameter relay lens needed to provide adequate clear-ance between the relay lens and the source. The largerelay lens necessitates a long focal length objective lens,as shown in Eq. (11). This may result in a bulky sys-tem having long, rigid mounting structures. For ex-ample, assume that for adequate clearance fi must be5.08 cm, and a DO of 12.7 cm is required. If a usablesource diameter of 0.25 cm is available, DL must be2.36 cm. In this case fo would have to be 254 cm tosimulate the solar angle. The type IV simulator in-corporates an additional relay lens assembly which re-images the source at an aperture. The aperture pro-vides a virtual source for the optical system of a stand-ard type III simulator, as shown in Fig. 4. This per-mits the use of a short focal length, low numbersecond relay lens and a short focal length objectivelens in conjunction with a large back focal length firstrelay lens having sufficient clearance from the sourceenvelope to avoid overheating of the lens.

The equations describing the optical geometry of atype IV simulator are somewhat more complex than

1392 APPLIED OPTICS / Vol. 3, No. 12 / December 1964

OBJECTIVE LENS

Fig. 4. Type IV solar simulator.

for the simpler type III simulator, due to the con-siderations concerning the first relay lens which takeinto account the larger size needed to avoid vignetting.In this respect the analysis is similar to that which wasnecessary to derive the equations for a type I simulator.

(13)

(14)

f4 = f,

DL2 = Ofj _

D,(fo - DoF2)and

By rearranging Eq. (15),

Do =- DLID~fo (16)

Ofifo + DMLF2D OFJfO + F2D8

where, in addition to the previous notation, DL1 is thediameter of the first relay lens, DL2 is the diameter of thesecond relay lens, fi is the focal distance of the firstrelay lens to the source, f4 is the focal distance of thesecond relay lens to the objective lens, F1 is the opticalF number of the rear half of the first relay lens (F1 =

fi/DLl), and F2 is the optical F number of the front halfof the first relay lens (F2 = f 2 /DLl).

In calculating the output beam intensity it must beremembered that only a portion of the first relay lensarea is employed to form the image of the source in theaperture plane. Different portions of the source utilizedifferent but overlapping areas of the lens. The in-stantaneous effective diameter of the first relay lens DLIe

is given as

DLIe = ODf, (17)D.

Straightforward calculation of the output beam in-tensity now yields the familiar Eq. (6).

The extent of limb-darkening evident with a type IVsimulator will depend on the choice of source size andoptical elements. It will not be as severe as in the caseof the type III simulator, due to the larger ratio be-tween virtual image size and the size of the secondrelay lens. The intensity gradient over the usefulbeam diameter will likewise be less. Residual gra-dients may be determined in practice and compensatingfilters incorporated.

The optical quality of the objective lens determines,to the same extent as in a type I simulator, the systemperformance. The relay lenses need only uniformlyilluminate the objective lens. The major drawback to

this type of simulator is the low optical F number de-

manded of the first relay lens. However, the qualityof the first relay lens can be rather poor without se-riously degrading performance, allowing the use of pro-jector-type condenser lenses possibly utilizing asphericsurfaces. In addition, a minor amount of beam block-age can be tolerated in the plane of the first relay lens dueto the overlapping of the ray bundles from differentportions of the source surface. This may permit theuse of sources such as a carbon arc lamp with a frontsurface ellipsoidal condenser mirror without concernfor the blockage caused by the carbon rods and theirsupports.

If high precision and extreme beam uniformity arerequired in a specific application, it will be necessaryto perform a detailed ray trace of the proposed systembefore the extent of beam "holing" and solar imageintensity gradient can be accurately predicted.

5. Image Homogenization

With the exception of a tungsten ribbon filamentlamp (this will be discussed in greater detail later),sources for use in solar simulation do not exhibit in-tensity uniformity over their useful areas. Some sourcetypes may even have emitting areas which wanderabout in a more or less random fashion. In any simu-lator type where the source granularity persists throughthe optical system and can be seen through a telescopeviewing the objective lens it would be very desirable tohave a device that would homogenize the image. Anideal device of this sort would be essentially lossless,would not alter the angular relationships between therays, and would create a random displacement of rayorigins in the image plane of the relay lens. Thisjumbling of ray origins must remain within the con-fines of the original image dimensions or losses willoccur.

Simple experiments were performed by introducinglight rays into a hollow glass tube internally coatedwith a high reflectance gold film. The length of thetube was approximately 8 times the diameter. When adiffuse source was used which filled the entrance of thetube, visual inspection of the output, by looking intothe tube and by allowing the output to fall on a whitesurface indicated an apparently diffuse source whichfilled the tube output aperture. A point source wasimaged by an external lens system at the tube entranceand the output showed several bright bands which wereattributed to the geometrical properties of an internallyreflecting cylinder. Noncylindrical shapes, partiallydiffuse surfaces and bending the tube are possible meansof improving the situation.

OBJECTIVE LENS

WAPERTURE ERTUERUESOURCE RELAY

Fig. 5. Type I solar simulator with image homogenizer.

December 1964 / Vol. 3, No. 12 / APPLIED OPTICS 1393

An alteration of the ray angles is produced if theinternal surface of the tube is not optically smooth,or if the diameter of the tube varies along its length.For example, if the tube has a slight flare toward theoutput end, a general reduction in the ray angles will beproduced and a taper to the tube produces the oppositeeffect. As an alternative, the reflective tube can bebent or coiled through a long radius arc to alter theangular relationships between entering rays. Thepossibility of utilizing this technique to scramble anglesas well as ray positions has not been studied. If theresults are positive, beam "holing" can be eliminatedeven though considerable optical blockage is present.

B. Emitting Sources

The choice of emitting sources for solar simulation islimited today to incandescent tungsten lamps, metallicarc lamps, carbon arcs, and high-pressure xenon orxenon-mercury arc lamps. The various sources whichshow promise for use in practical systems are dis-cussed next.

1. Tungsten Filament LampsThere are two general types of tungsten filament

lamps which have application in solar simulation:coiled filament and ribbon filament lamps. A coiledtungsten wire, often coiled further upon itself to form acoiled-coil, is the most common electrical light sourceused today. In high-power lamps consuming 500 Wto more than 1000 W, the temperature of the filamentmay be operated as high as 32000 K for short periods oftime.

Type II solar simulators have been built using thisform of source with a spherical, front-surface mirrorplaced behind the lamp. The filament is placed at theradius of curvature of the mirror and slightly off-axis.When adjusted properly the mirror forms an image ofthe coils between the actual coils, tending to fill thevoids between the wires.

The coils cause an increase in the apparent emissivityof the tungsten (due to multiple reflection), but thevoids between the turns negate this effect. In a high-wattage lamp specifically designed for projector use,having a flat array of coils and a rear spherical mirror,the effective total emissivity may approach that of thesolid material.

When operated at high filament temperature, con-siderable evaporation of tungsten onto the glass en-velope limits the useful life of the lamp to a few hundredhours or less. New techniques incorporating halogenswithin the envelope retard this effect to some extentand permit sustained operation at 31000K in certainlamps. At this time no flat-type coil arrays utilizingthis technique are available which would be suitablefor solar simulation.

The major problem with the coiled filament lamps isthe granularity of the simulated solar image thatwould be produced by many types of optical systems.In some applications this is offset by the relativelylarge emitting area which is available in some of thelarger lamps.

If it is necessary to have a uniformly emitting source,a likely candidate is found in the ribbon filament lamp.There is available a line of ribbon filament lamps whichhave useful source dimensions ranging from 1 mm to 4mm in width and many times that in length. Whenproperly selected and used, these lamps provide ahighly uniform surface which is constant to within 20 Kover the desired area at a temperature of 3000'K.Life of the lamps operated at this temperature variesfrom 200 h for the smallest filaments to more than1000 h for the largest filaments. The emissivity of thelamps varies slightly with wavelength, but it can beconsidered to be 0.4 for calculation purposes.

A number of type I simulators have been constructedwhich use an 18-A vertical ribbon filament lamp withconsiderable success. These simulators exhibited anoutput beam intensity which ranged from 0.010 to0.016 solar constant (for a silicon detector), dependingon the optical losses.

2. Metallic Arc LampsA tungsten arc lamp using mercury as a starter has

been available as a "sun lamp" for some time. Meas-urements performed on this lamp type show the effec-tive arc temperature to be approximately 3100'K, notappreciably above that of the incandescent metalitself. Therefore, these lamps serve no useful purposefor solar simulation at the present time.

The author has tested a series of zirconium arc lampswhich operate at approximately 3400'K. The smallsize bulbs of this type have an emitting area of theorder of 0.01 mm in diameter, but high-power zirconiumarcs are available having a 0.305-cm diameter source.The bulbs are filled with an inert gas and a definitestreaming effect can be seen due to convection aboutthe emitting arc. Some motion of the arc is also dis-cernible and an intensity variation is present which hasboth long- and short-term components. Source lifeis limited to the 100-h range.

It seems highly improbable that a zirconium arcsource could be successfully employed in a solarsimulator unless some improvement was made in thelamps or an optical system was used which reduced theseeffects to a tolerable level. The variations in total out-put intensity were the most serious of the defectsfound, since these are not amenable to solution by im-age homogenization or optical design techniques.

3. Xenon or Xenon-Alercury Arc LampsHigh-pressure Xe and XeHg lamps in the 5000-W

size produced small, very intense areas having a bright-ness considerably greater than the sun but falling offrapidly a short distance from the maximum brightnesspoint. Based on the manufacturer's data for themicrobrightness of a 5-kW xenon arc, a 1.0-mm X1.0-mm area directly adjacent to the cathode providesthe highest brightness within the arc, and this isapproximately 2.7 solar constant in terms of candela/cm2 . The integrated brightness over a 5-mm X 5-mmarea is approximately one-half solar brightness and

1394 APPLIED OPTICS / Vol. 3, No. 12 / December 1964

Table I. Typical Brightness Contour Measurements

Distance fromcenter (mm) Horizontal (cd/mm 2) Vertical (cd/mm 2 )

8 150 680

6 550 10004 1050 1250

2 1150 1300

0 1150 1250

-2 975 1100

-4 725 800

-6 450 425

-8 100 150

contains almost all of the effective emitting area.These light sources produce an emission spectrum whichconsists of a relatively weak continuum superimposedupon line spectra, broadened by the high plasma tem-perature and the pressure within the lamp. Unlesssubstantial spectral filtering is incorporated in theoptical system, a solar simulator using a Xe or XeHglamp will have a spectral output which departs fromsolar radiation to a sufficient extent to require difficultextrapolations in interpreting the response of varioussolar-sensing detectors. For example, a silicon photo-voltaic diffused junction "solar cell" will exhibit sub-stantial spectral response variations depending on thedepth of the diffused layer. Unless the simulator hasa spectral output matching that of the sun, it would benecessary to plot the spectral responsivity of the sensorand normalize the response to that of the particularsimulator spectrum if accurate extrapolation to spacesun response is desired.

Spectral filters may be incorporated to modify thesource spectrum, but there have been difficulties inobtaining stable filters which are unaffected by theintense ultraviolet output of the lamp. Furthermore,recent studies* have shown limited stable life and se-rious spectral intensity variations over the life of typicalhigh-wattage Xe and XeHg lamps.

4. Carbon Arc Lamps

A number of companies manufacture carbon arclamps which are suitable for solar simulation use.Considerable development of these sources has beenperformed for theater projection use and solar simula-tion, and a wide selection of arc lamp models is available.The useful emitting area of these sources is sufficientlylarge to permit the use of simple relay optics withoutsacrificing beam quality or imposing a restriction on theoutput beam diameter to less than the desired size.

The maximum brightness of a 400-A, 16-mm positivecarbon arc is 1300 cd/mm 2 , which is approximately0.81 of the solar brightness. Typical brightness con-tour measurements performed over the arc diameterhave produced the results shown in Table I. Similarmeasurements on a 13.6-mm positive carbon operating

* N. Z. Searle, P. Giesecke, R. Kinmonth, and R. C. Hirt,

Appl. Opt. 3, 923 (1964).

at 185 A show a peak intensity of 682 cd/mm 2 over thecentral 3.6 mm, dropping to 480 cd/mm 2 at 5.5 mm fromthe center.

The 400-A arc unit can provide an 11-mm-diam areahaving an approximate integrated brightness of 800 cd/mm2 , or one-half that of the sun. The 185-A arc wouldprovide approximately 500 cd/mm 2 over the same area,which is 0.31 that of the sun.

The spectral distribution of a carbon arc closelymatches that of space sunlight. The similarity of thetwo spectral curves will permit the use of a widevariety of solar sensors without extrapolations or com-pensation for spectral match, using any sun-calibratedsensor to measure the total power density in the outputbeam.

With an emitting area as large as 1.27 cm in diameter,it is practical to build a reflecting optical system whichwould produce a 50.8-cm diam (or greater) beam.Careful choice of optical components is necessary, how-ever, to avoid darkening and damage of mirror sur-faces and lenses from the high heat, fumes, and ultra-violet radiation produced by these arc sources.

The major drawbacks to the use of carbon arc sourcesfor precision solar simulation are caused by the arcitself, which exhibits streaming, positional variation, andsome instability, and by the practical problems ofcarbon rod replacement, noise, fumes, and high powerconsumption. Nevertheless, the carbon arc lampprovides the most likely candidate for a precision solarsimulator having large aperture and relatively highintensity. The type IV simulator should be used oran image homogenizer incorporated to eliminate theeffects of arc wander, streaming, and instability.

Ill. Conclusions

In selecting a solar simulator for a particular appli-cation, it is essential that the primary desired attributesof the system be carefully delineated. On the basisof the requirements, the appropriate optical design andsource can be selected. Extreme accuracy and sta-bility, such as might be required for the checking ofprecision solar sensors, would probably require the useof a tungsten filament source, limiting the beam in-tensity and maximum beam diameter as explainedearlier.

Where high intensity is desired, combined with alarge beam aperture, a simulator containing a carbonarc lamp is the obvious choice. While there is amplereason to believe a precision solar simulator can bebuilt that would have an aperture of more than 50 cm,there is little hope that such a system can be devisedproducing much more than 0.4 solar constant unlessmore intense large-area sources can be found.

One solar constant or more is possible with the largeXe or XeHg lamps if their spectral properties can betolerated. The latter system aperture would belimited to 7.5 cm to 15 cm, due to the relatively smallemitting area, and image homogenation (or possibly atype IV system) is indicated.

December 1964 / Vol. 3, No. 12 / APPLIED OPTICS 1395

Patents ctttnued from page 1863,130,309 21 Apr. 1964 (Cl. 250-83.3)Infrared visual image converter with lens mirror coated withinfrared absorbing material.J. H. SNYDER. Assigned to Minneapolis-Honeywell RegulatorCo. Filed 3 Oct. 1961.

This invention relates to an ir image system which converts ir radiationinto an image visible to the eye. This is accomplished by a schlierenlikesystem, as shown in the figure, whereby the thermal radiation causes distor-tions which are viewed by an observer with optical methods. D.J.L.

3,133,198 12 May 1964 (Cl. 250-199)Traveling wave light modulator.I. P. KAMINOW, R. KOMPFNER, AND W. H. LOUISELL. Assignedto Bell Telephone Laboratories, Inc. Filed 12 Jan. 1962.

This traveling wave light modulator consists of two coupled waveguides.Coupling is accomplished through a slot between the guides. This causesthe coupling coefficient to increase in an exponential manner in the directionof light travel. Input power is divided between the two waveguides;in one waveguide is an electrooptical material such as DP; and the purposeof the second waveguide is to replenish the power which would be normallylost due to attenuation in the lossy electrooptical material. In order toassure synchronism between light beam and modulating wave, each guideis fitted with dielectric loading members. Several configurations for thissystem are described. M.L.

ENVERGY /19SOURCE

Vapor cell light amplifier.J. W. HORTON. Assigned to International Business MachinesCorp. Filed 22 July 1959.

This device consists of a cell containing- (for example) sodium vapor.Provision is made to allow two beams of light to pass through it. The cellis in a magnetic field causing Zeeman splitting of the hyperfine states. Onebeam (the pumping beam) causes the gas to become transparent to thesodium D line due to transitions between the ground state and the P/2excited state. In the presence of a second beam (the depumping beam)the gas is restored to its normal condition of being opaque to the pumpingradiation. Tus the device may be used as an amplifier or as a logic element.

M.L.

3,134,297 26 May 1964 (Cl. 88-24)Optical information display system having metachromatic means.C. 0. CARLSON, D. B. CONGLETON, AND D. A. GRAFTON. As-signed to The National Cash Register Co. Filed 27 Dec. 1960.

An ordinary photographic slide is irradiated with 4 50-g radiation andimaged on a metachromatic plate of variable transmittance. An image ofthe plate is then projected onto a display screen. The transmittance at anypoint on the metachromatic plate may be semipermanently (but reversibly)reduced to 1% when irradiated with about 0.3 J/cm2 of 380-A, to 400-Aradiation. Thus, information may be written on the plate by a flying-spotscanner having an uv light source. The transmittance may be increased byconduction heating or ir irradiation. This process erases information fromthe metachromatic plate. J.A.F.

3,134,837 26 May 1964 (Cl. 88-1)Optical maser.P. P. KISLIUK AND D. A. KLEINMAN. Assigned to Bell TelephoneLaboratories, Inc. Filed 16 Oct. 1961.

A technique for laser mode suppression is claimed, including specificallythe use of three or more plane-parallel reflectors for a Fabry-Perot type laserresonator. S.F.J.3,137,757 16 June 1964 (Cl. 88-14)Apparatus for the analysis of substances by absorption of radia-tion.A. E. MARTIN AND J. SHIELDS. Assigned to Sir Howard GrubbParsons & Co. Ltd. Filed 4 Feb. 1959 (in Great Britain 7 Feb.1958).

This invention relates to apparatus for the analysis of substances by opticalabsorption. This is done, in the present invention, by a simple beam-switching, rotating shutter which forms two beams. A monochromatorprovides the illumination desired, and the beams are alternately passed toa detector, which senses any alternations in the two signals. D.J.L.3,137,759 16 June 1964 (Cl. 88-14)Spectal-flame burner apparatus and spectral-flame burnerstherefor.J. ISREELI. Assigned to Technicon Instruments Corp. Filed 1Sept. 1959.

Two flames, one containing a known substance and the other the sample,are compared for the quantitive analysis of the sample. The referenceflame is filtered to allow its strong spectral lines to pass into one photodetec-tor. A second filter passes the strong spectral lines of a substance that prob-ably could be found in the sample into a second photodetector. The ratioof the outputs of the two photodetectors then is recorded. Use is made ofmagnesium oxide coatings to aid in the passage and the uniformity of thelight produced by the flames. The ratio of lithium to sodium in blood is theexample cited. W.W.

- p -(eZ-611eru, L MAM)'CIRC POARIZED

~ A~ L F O OI M V PO R I o I 4 Liy /V eD /UA,Ci t e ARGO f 5UP EA TJOOV r in |(M w 0 e J~rs e ~e a er a ! Z ( ' d l I VI MILLML~rr24'Rr ¾11 1i

iResv s R As'L --f- G . i e >4 //rrsw~. Al .a 7 'iTVAI iFVn HAZE -tm- `"t~-

2L A $ W - - LO GAte2c6 O MI 7

\/4.- continued

on page 1416

1396 APPLIED OPTICS / Vol. 3, No. 12 / December 1964