Sie sind auf Seite 1von 84

7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

EssentialRadioAstronomy
2RadiationFundamentals 4FreeFreeRadiation

Chapter3
RadioTelescopesand
Radiometers
3.1AntennaFundamentals
Anantennaisapassivedevicethatconvertselectromagneticradiationinspaceinto
electricalcurrentsinconductorsorviceversa,dependingonwhetheritisbeingusedfor
receivingorfortransmitting,respectively.Radiotelescopesarereceivingantennas,and
radartelescopesarealsotransmittingantennas.Itisofteneasiertocalculatetheproperties
oftransmittingantennasandtomeasurethepropertiesofreceivingantennas.Fortunately,
mostcharacteristicsofatransmittingantenna(e.g.,itsradiationpattern)areunchanged
whenthatantennaisusedforreceiving,soanyanalysisofatransmittingantennacanbe
appliedtoareceivingantennausedinradioastronomy,andanymeasurementofa
receivingantennacanbeappliedtothatantennawhenusedfortransmitting.

3.1.1RadiationfromaShortDipoleAntenna(HertzDipole)

Figure3.1:Thecoordinatesystemusedtodescribetheradiationfromashort(total
lengthl )dipoledrivenbyacurrentsourceoffrequency.

Thesimplestantennaisashort(totallengthlmuchsmallerthanonewavelength)
dipoleantenna,whichisshowninFigure3.1astwocollinearconductors(e.g.,wiresor
conductingrods).Whentheyaredrivenatthesmallgapbetweenthembyanoscillating
currentsource(atransmitter),thecurrentgoingintothebottomconductoris180degrees
outofphasewiththecurrentgoingintothetopconductor.Theradiationfromadipole
dependsonthetransmitterfrequency,soconsiderasinusoidaldrivingcurrentIwith
angularfrequency 2:

http://www.cv.nrao.edu/~sransom/web/Ch3.html 1/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

I = I 0cos(t),
(3.1)

whereI 0isthepeakcurrentgoingintoeachhalfofthedipole.Itiscomputationally
convenienttoreplacethetrigonometricfunctioncos(t)withitscomplexexponential
equivalent(AppendixB.3),therealpartof

e i t = cos(t) isin(t), (3.2)

sothedrivingcurrentcanberewrittenas

I = I 0e i t
(3.3)

withtheimplicitunderstandingthatonlytherealpartofIrepresentstheactualcurrent.
Thedrivingcurrentaccelerateschargesintheantennawires,soLarmorsformulacanbe
usedtocalculatetheradiationfromtheantennabyconvertingfromchargesand
accelerationstotimevaryingcurrents.

Theelectriccurrentinawireisdefinedastheflowrateofelectricchargealongthe
wire:

dq
I .
dt (3.4)

Forawireonthezaxis,

dq dq dz dq
I= = = v, (3.5)
dt dz dt dz

wherevistheinstantaneousflowvelocityofthecharges.

Manypeopleincorrectlybelievethatthevelocityvofindividualelectronsinawire
iscomparablewiththespeedoflightcbecauseelectricalsignalsdotraveldownwiresat
nearlythespeedoflight.However,awirefilledwithelectronsislikeagardenhose
alreadyfilledwithwater,anearlyincompressiblefluid.Whenthefaucetisturnedon,
waterflowsfromtheotherendofafullhosealmostimmediately,eventhoughindividual
watermoleculeshavemovedonlyashortdistancealongthehose.Theexampleabove
showsthatelectronsmovesoslowlyinawirethatLarmorsnonrelativisticequation
accuratelypredictstheradiationfromantennas.

http://www.cv.nrao.edu/~sransom/web/Ch3.html 2/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

Example.Estimatethespeedofelectronsflowingthroughacopperwireofcross
section = 1mm = 2 10 6m 2andcarryingacurrentof1ampere.
Thenumberdensityoffreeelectronsisaboutequaltothenumberdensityofcopper
atomsinthewire,n 10 29m 3.InMKSunits,thechargeofanelectronis

1 coul
e 4.80 10 10 statcoul 9
1.60 10 19 coul.
3 10 statcoul

Oneampereisdefinedasonecoulombpersecond,sothenumberofelectrons
flowingpastanypointalongthewireinonesecondis

I 1 coul s 1
N = 19
6.25 10 18 s 1.
|e| 1.60 10 coul

Theaverageelectronvelocityisonly

N 6.25 10 18 s 1
v 6 2 29 3
6 10 5 m s 1 c.
n 10 m 10 m

ThusthenonrelativisticLarmorequationmaysafelybeusedtocalculatethe
radiationfromawire.

Equation2.136fromthederivationofLarmorsformula

qvsin
E =
rc 2

canbeappliedtoyieldthedE contributedbyeachinfinitesimaldipolesegmentof
lengthdz.Ifthedipoleisshort(l ),alloftheseelectricfieldsareinphaseandadd
directlytogivethetotalE producedbythedipole:

+ l / 2 dq v sin
E = z = l / 2 z 2 .
dz rc (3.6)

Atdistancesr l,(1 / r)isnearlyconstantoverthewholeantennaandcanbetaken


outsidetheintegral.Forasinusoidaldrivingcurrent,v = ivand

isin + l / 2 dq isin + l / 2
E =
rc 2
l / 2 dz v z =
rc 2
l / 2 I z. (3.7)

Thatis,theradiatedelectricfieldstrengthE isproportionaltotheintegralofthe
currentdistributionalongtheantenna.Thecurrentatthecenteristhedrivingcurrent

i t
,andthecurrentmustdroptozeroattheendsoftheantenna,wherethe
http://www.cv.nrao.edu/~sransom/web/Ch3.html 3/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

I = I 0e i t,andthecurrentmustdroptozeroattheendsoftheantenna,wherethe
conductivitygoestozero.Thecurrentdistributionalongashortdipoleisthetailendofa
standingwavesinusoid,whichdeclinesalmostlinearlyfromthedrivingcurrentatthe
centertozeroattheends:

I(z) I 0e i t 1
|z|
[
(l / 2)
. ] (3.8)

Then

+l/2 I 0l
l / 2 I z e i t (3.9)
2

and

isin I 0l i t
E e . (3.10)
rc 2 2

Substituting = 2c / gives

i2csin I 0l i t isin I 0l e i t
E e = . (3.11)
rc 2 2 c r

ThetimeaveragedPoyntingflux(powerperunitarea)followsfromEquation2.139itis

c 2
S = E .
4 (3.12)

Thus

( )( )
c 1 I 0l 2 sin 2
S = , (3.13)
4 2 c r2

wherethefactor(1 / 2)reflectsthefactthatsin 2(t) = cos 2(t) = 1 / 2.(Thisisagood


relationtorememberandaneasyonetoderivebecausesin 2(t) + cos 2(t) = 1and
sin 2(t) = cos 2(t).)

Thepowerpatternofatransmittingantennaistheangulardistributionofits

http://www.cv.nrao.edu/~sransom/web/Ch3.html 4/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

radiatedpower,oftennormalizedtounityatthepeak.FromEquation3.13thenormalized
powerpatternofashortdipoleis

P sin 2. (3.14)

Theradiationfromashortdipolehasthesamepolarizationandthesamedoughnut
shapedpowerpatternasLarmorradiationfromanacceleratedchargebecauseallofthe
chargesintheshortdipolearebeingacceleratedalongonelinemuchshorterthanone
wavelength.Fromtheobserverspointofview,thepowerreceiveddependsonlyonthe
projected(perpendiculartothelineofsight)lengthlsinofthedipole.Theelectricfield
strengthreceivedisproportionaltotheapparentlengthofthedipole,andtheradiation
fromthedipoleislinearlypolarizedparalleltotheprojecteddipole.Thetimeaveraged
totalpoweremittedisobtainedbyintegratingthePoyntingfluxoverthesurfaceareaofa
sphereofanyradiusr lcenteredontheantenna:

( )( )
c 1 I 0l 2 2 sin 2
P = SA =
c = 0 = 0 r2
rsin d r d (3.15)
4 2

( )( )
c 1 I 0l 2
= 2 = 0 sin 3 d. (3.16)
4 2 c


Recallthat 0 sin 3 d = 4 / 3,sothetimeaveragedpowerradiatedbyashortdipoleis

P =
3c ( )
2 I 0l 2

, (3.17)

whereI 0cos(t)isthedrivingcurrentandl / isthetotallengthofthedipolein


wavelengths.

Mostpracticaldipolesarehalfwavedipoles(l / 2)becausehalfwavedipoles
areresonant,meaningthattheyprovideanearlyresistiveloadtothetransmitter.When
eachhalfofthedipoleis / 4long,thestandingwavecurrentishighestatthecenterand
naturallyfallsasI = I 0cos(2z / )tozeroattheendsoftheconductors.

Figure3.2:Agroundplaneverticalantennaisjusthalfofadipoleaboveaconducting
plane.Thelowerhalfofthedipoleisthereflectionoftheverticalinthemirror
providedbytheconductinggroundplane.Theimageverticalis180degreesoutof
http://www.cv.nrao.edu/~sransom/web/Ch3.html 5/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

phasewiththerealvertical.Abovethegroundplane,theradiationfromtheground
planeverticalisexactlythesameastheradiationfromthedipole.

ThegroundplaneverticalantennashowninFigure3.2isverysimilartothe
dipole.Agroundplaneverticalisonehalfofadipoleaboveaconductingplane,whichis
calledagroundplanebecausehistoricallytheconductingplaneforverticalantennas
wasthesurfaceoftheEarth.Thetransmitterisconnectedbetweenthebaseofthe
vertical,whichisinsulatedfromtheground,andthegroundplanenearthebase.Many
AMbroadcasttransmittingantennasaretall(at 1MHz, 300manda / 4
verticalantennaisabout75mhigh),insulatedtowersactingasquarterwaveverticals.
Theconductinggroundplaneisamirrorthatcreatesthelowerhalfofthedipoleasthe
mirrorimageoftheupperhalf.Electricfieldsproducedbytheverticalantennainduce
currentsintheconductingplanetomakethehorizontalcomponentoftheelectricfieldgo
tozeroontheconductor.Thevirtualelectricfieldsfromtheimageverticalhavethesame
amplitudebutare180degreesoutofphase,exactlyasinahalfwavedipole.
Consequentlytheradiationfieldfromagroundplaneverticalisidenticaltothatofa
dipoleinthehalfspaceabovethegroundplaneandzerobelowthegroundplane.

Figure3.3:Mosthighfrequencyfeedsarequarterwavegroundplaneverticalsinside
waveguidehorns.Theonlytrueantennainthisfigureisthe / 4groundplanevertical,
whichconvertselectromagneticwavesinthewaveguidetocurrentsinthecoaxialcable
extendingdownfromthewaveguide.

Accordingtothestrictdefinitionofanantennaasadeviceforconvertingbetween
electromagneticwavesinspaceandcurrentsinconductors,theonlyantennasinmost
radiotelescopesarehalfwavedipolesandtheirrelatives,quarterwavegroundplane
verticals.Thelargeparabolicreflectorofaradiotelescopeservesonlytofocusplane
wavesontothefeedantenna.(Thetermfeedcomesfromradarantennasusedfor
transmittingthefeedantennafeedstransmitterpowertothemainreflector.Receiving
antennasusedinradioastronomyworktheotherwayaround,andthefeedactually
collectsradiationfromthereflector.)

Actualhalfwavedipoles,backedbysmallreflectorsabout / 4behindthemtofocus
thedipolepatterninthedirectionofthemaindish,arenormallyusedasfeedsatlow
frequencies( < 1GHz)orlongwavelengths( > 0.3m)becauseoftheirrelatively
smallsize.However,theradiationpatternsofhalfwavedipolesbackedbysmall
reflectorsarenotwellmatchedtomostparabolicdishes,sotheirperformanceislessthan
optimum.

Forshorterwavelengths,almostallradiotelescopefeedsarequarterwaveground
planeverticalsinsidewaveguidehorns.Radiationenteringtherelativelylarge(size > )
rectangularorcircularapertureofthetaperedhornisconcentratedintoarectangularor
circularwaveguidewithparallelconductingwalls.Inthecaseoftherectangular
http://www.cv.nrao.edu/~sransom/web/Ch3.html 6/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

waveguidewhosecrosssectionisshowninFigure3.3,thesidewallsareseparatedby
slightlyover / 2sothatverticalelectricfieldscantraveldownthewaveguidewithlow
loss.Thetopandbottomwallsareseparatedbysomewhatlessthan / 2soonlythemode
withverticalelectricfieldscanpropagate(Section3.4).The / 4verticalantennainserted
throughasmallholeinthebottomwallcollectsmostofthisverticallypolarizedradiation
andconvertsitintoanelectriccurrentthattravelsdownthecoaxialcabletothereceiver.
Thebackshortwallabout1/4oftheguidewavelength w(Equation3.144)behindthe
dipoleensuresthatthedipoleseesonlyradiationcomingfromthedirectionofthehorn
opening.

Bothdipolesandquarterwaveverticalsarelinearlypolarizedfeeds.Thevoltage
responseofalinearlypolarizedfeedtoalinearlypolarizedsourceisproportionaltocos,
whereistheanglebetweenthefeedandthesourceelectricfield,andthepower
responseisproportionaltocos 2 = [cos(2) + 1] / 2.Consequentlythedegreeof
polarizationandthepolarizationpositionangleofapartiallylinearlypolarizedradio
sourcecanbemeasuredbyrotatingthelinearlypolarizedfeedofaradiotelescopewhile
trackingthesource.Thedegreeofpolarizationpofapartiallylinearlypolarizedsource
definedbyEquation2.58is

Ip Ip
p = , (3.18)
I Ip + Iu

whereI pisthepolarizedfluxdensityandI = I p + I uisthetotalfluxdensityofthe


source.ThepowerresponseofthefeedR() I pcos 2 + I u / 2willbeR I p + I u / 2
whenthefeedandsourcepolarizationsareparallel( = 0)andR = I u / 2whentheyare
perpendicular( = / 2).IntermsoftheobservablesR andR ,thedegreeof
polarizationis

Ip Ip + Iu / 2 Iu / 2 R R
p= = = . (3.19)
Ip + Iu Ip + Iu / 2 + Iu / 2 R + R

Figure3.4showshowtherelativepoweroutputR()ofalinearlypolarizedfeedvariesas
itisrotatedthrough = radiansrelativetothepolarizationpositionangleofsources
withfractionalpolarizationsp = 1.0,0.1,and0.0.

http://www.cv.nrao.edu/~sransom/web/Ch3.html 7/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

Figure3.4:Therelativepoweroutputfromalinearlypolarizedantennaasafunctionof
thepolarizationpositionangledifferencebetweenthesourceandtheantennafor
sourceswithfractionalpolarizationsp = 1.0(dashedcurve),p = 0.1(solidcurve),and
p = 0.0(dottedline).Abscissa:Positionangledifference(rad).Ordinate:Relative
poweroutputR.

TomeasureallfourStokesparametersofanarbitrarilypolarizedsource,itis
necessarytocombinethevoltageoutputsoftwoorthogonallypolarizedfeeds.For
example,twoorthogonalquarterwaveverticalscanbeinsertedintoasquarewaveguide
toreceiveboththehorizontallyandtheverticallypolarizedcomponentssimultaneously.
Iftheiroutputvoltagesareaddedinphase(phasedifference = x y = 0in
Figure2.15),thefeedcombinationwillrespondtoradiationlinearlypolarizedinposition
angle / 4 = 45 .Ifaphasedifference = / 2isinsertedeithermechanically(by
movingonefeed / 4behindtheother)orelectrically(byinsertinga / 4longercable
betweenonefeedandthepointwherethetwooutputsareadded),then = / 2andthe
feedcombinationwillrespondtocircularpolarization.

3.1.2RadiationResistance
Thepowerflowingthroughacircuitis

P = VI, (3.20)

whereVisthevoltage(definedasenergyperunitcharge)andIisthecurrent(definedas
thechargeflowingthroughthecircuitperunittime),soPhasdimensionsofenergyper
unittime.ThephysicistGeorgeSimonOhmobservedthatthecurrentflowingthrough
most(butnotall)materialsisproportionaltotheappliedvoltage,somostobjectshavea
welldefinedresistanceRdefinedbyOhmslaw,

V
R . (3.21)
I

WhenOhmslawholds,

http://www.cv.nrao.edu/~sransom/web/Ch3.html 2 8/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

V2
P= I 2R = .
R (3.22)

Theaveragepowerinaresistivecircuitwithtimevaryingcurrentsis

P = I 2R. (3.23)

2
IntheparticularcaseofsinusoidalcurrentsI = I 0cos(t)andI 2 = I 0 / 2,so

2
I 0R
P = . (3.24)
2

Thusthe(frequencydependent)radiationresistanceofanantennaisdefinedby

2P
R 2
.
I0 (3.25)

Forashortdipole,thepoweremittedisgivenbyEquation3.17andtheradiation
resistanceis

R=
3c ()
2 2 l 2

. (3.26)

Example.Ahalfwave(lengthl = / 2)dipoleisaresonantantenna.Resonant
antennasareusedinmostrealapplicationsbecausetheimpedanceofaresonant
antennaisresistivenonresonantantennashavelargecapacitiveorinductive
reactancesaswell.Mostoftheantennacurrentinahalfwavedipoleisastanding
wavewithacurrentdistributionI = I 0e i tcos(2z / )thathasamaximum
I = I 0e i tatthez = 0feedpointanddeclinescosinusoidallytozeroatthe
endpointsz = / 4.Withthe(nolongersoaccurate)assumptionthattheradiation
fromallpartsofthedipoleemitinphase,Equation3.7stillholds:

isin + l / 2
E =
rc 2
l / 2 I z.
Thecurrentdistributionofthehalfwavedipoleis

I = I 0e i tcos ( ) 2z

so
http://www.cv.nrao.edu/~sransom/web/Ch3.html 9/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy
so

( )
+l/2 +/4 2z I 0
l / 2 I z I 0e i w t / 4 cos
z = e i t,

whichisafactorof2 / (l)largerthanthecomparableintegralforashortdipole
(Equation3.8):

+l/2 I 0l
l / 2 I z e i w t.
2

TheaveragepowerPradiatedbyagivenI 0isproportionaltothesquareofthis
factor(seeEquations3.10through3.17),or

( ) 2 2
l
.

ThustheradiationresistanceRofahalfwavedipoleistheradiationresistanceofa
shortdipole(Equation3.26)multipliedbythefactorsquared:

R= [ ( ) ]( )
2 2 l 2
3c
2 2
l
=
8
3c
8 8
= 10 1
10 10 s cm 1.
3 3 10 cm s 9
EngineersandrealtestinstrumentsusetheMKSohm(symbol)astheunitof
resistance.Theconversionfactoris1=(10 11 / 9) s cm 1,so

R= ( 8
9
10 10 s cm 1 / )( 1
9
10 11 s cm 1 ohm 1 = 80 . )
ThisisprettyclosetotheR 73 resultfromanexactcalculationthatdoesntuse
thel approximation.

Agroundplaneverticalofheightl / 2emitsexactlylikeadipoleoflengthlabovethe
groundplaneandnothingbelowthegroundplane.Thusthetotalpoweremittedbythe
verticalishalfthepoweremittedbythedipole,andtheradiationresistanceofthevertical
ishalftheradiationresistanceofthedipole.

Theradiationresistance offreespace(sometimescalledtheimpedanceZ 0of


freespace)canbeobtainedfromtherelations

c V2
| S| = E 2 and P = . (3.27)
4 R

TheelectricfieldEisjustthevoltageperunitlengthV / landthefluxisthepowerper
unitareal 2,so

http://www.cv.nrao.edu/~sransom/web/Ch3.html 2 2 10/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

cV 2 V2
| S| = =
4l 2 R 0l 2 (3.28)

and

4 4
R0 = = = 4.19 10 10 s cm 1.
c 3 10 10
cm s 1 (3.29)

ConvertingfromCGStoMKSunitsyieldstheradiationresistanceofspaceinohms:

4
R0 = = 120 377 .
3 10 10 cm s 1 1 / 9 10 11 s cm 1 1 (3.30)

Thetaperedopeningofawaveguidehornfeed(Figure3.3)actsasanimpedance
transformertomatchtheimpedanceofthewaveguidetotheimpedanceoffreespaceto
minimizestandingwavesandcouplepowerefficientlybetweenthewaveguideandspace,
justasthebellofatromboneisanacoustictransformermatchingsoundvibrationsofair
inthetrombonetotheoutsideenvironment.

Ablackholeisaperfectabsorberofradiation,soitsresistancemustalsobe120
tomatchthatoffreespace.Ablackholespinninginanexternalmagneticfieldcan
generateelectricalpowerwithavoltage/currentratioof120 ,andthisprocessmaybe
importantinpoweringquasarjets[12].

3.1.3ThePowerGainofaTransmittingAntenna

ThepowergainG(, )ofatransmittingantennaisdefinedasthepowertransmittedper
unitsolidangleindirection(, )relativetoanisotropicantenna,whichhasthesame
gaininalldirections.Frequently,thevalueofGisexpressedlogarithmicallyinunitsof
decibels(dB):

G(dB) 10log 10(G).


(3.31)

Foranylosslessantenna,energyconservationrequiresthatthegainaveragedover
alldirectionsbe

G = 1. (3.32)

Consequently,alllosslessantennasobey

sphereG d = 4. (3.33)

http://www.cv.nrao.edu/~sransom/web/Ch3.html 11/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

Differentlosslessantennasmayradiatewithdifferentdirectionalpatterns,buttheydonot
alterthetotalamountofpowerradiated.Consequently,thegainofalosslessantenna
dependsonlyontheangulardistributionofradiationfromthatantenna.Ingeneral,an
antennahavingpeakgainG 0mustbeammostofitspowerintoasolidanglesuchthat
4 / G 0.Thismotivatesthedefinitionofthebeamsolidangle A:

4
A .
G0 (3.34)

Thusthehigherthegain,thesmallerthebeamsolidangle.

Theantennaefficiencyisdefinedastheratioofinputpowertoradiatedpower.If
ohmiclossesreduce,thenthegainGinEquations3.31through3.34shouldbereplaced
bythedirectivitydefinedbyD G / .

http://www.cv.nrao.edu/~sransom/web/Ch3.html 12/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

Example.Whatisthepowergainofalosslessshortdipole?Itissufficienttorecall
onlytheangulardependenceoftheshortdipolepowerpattern(Equation3.14)

P sin 2,

whereistheanglefromthedipoleaxis.Thus

G sin 2 = G 0sin 2.

ThemaximumgainG 0isdeterminedbyenergyconservation:

2
sphereG = = 0 = 0G 0sin 2 d sin d = 4,


2G 0 0sin 3 d = 4.


Recallthat 0 sin 3 d = 4 / 3so

4 3 3
G0 = =
2 4 2

and

3sin 2
G(, ) = .
2

ExpressedindB,themaximumgainG 0ofashortdipoleis

G 0 = 10log 10(3 / 2) 1.76 dB.

NotethatG(, )isnearlyindependentoftheantennalengthsolongasl
becausethepowerpatternofashortdipoleisnearlyindependentofl.Varying
l affectsonlytheradiationresistance.

3.1.4TheEffectiveAreaofaReceivingAntenna
Thereceivingcounterpartoftransmittingpowergainistheeffectiveareaoreffective
collectingareaofareceivingantenna.ImagineanidealantennaofgeometricareaAthat
couldcollectalloftheradiationfallingonitfromadistantpointsourceandconvertitto
electricalpoweraraingaugeforcollectingphotons.Thetotalspectralpowerincident
ontheantennaistheproductofitsgeometricareaandtheincidentspectralpowerperunit
area,orfluxdensityS .However,anysingleantennacanrespondtoonlyone
polarization,soitsoutputP canequalalloftheinputspectralpower(P = AS )froma
fullypolarizedsourcewhosepolarizationmatchesthatoftheantenna,butonlyhalfofthe
incidentpower(P = AS / 2)fromanunpolarizedsourceandnothingatallfroman
orthogonallypolarizedsource.Theoutputofarealantennaisalwayssmallerthanthis
andmostradiosourcesarenearlyunpolarized,soradioastronomersfinditusefulto
http://www.cv.nrao.edu/~sransom/web/Ch3.html 13/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

definetheeffectivecollectingareaA eofanantennawhoseoutputspectralpowerisP in
responsetoanunpolarizedpointsourceoftotalfluxdensityS by

2P
Ae .
S (3.35)

Theaveragecollectingarea

4 A e 1
4 4 e
A e = A (3.36)
4

ofanylosslessantennacanbecalculatedviaanotherthermodynamicthoughtexperiment.

Figure3.5:AcavityinthermodynamicequilibriumattemperatureTcontaininga
resistorRiscoupledtoanantenna,alsoattemperatureT,throughafilterblocking
electromagneticradiationbutpassingcurrentshavingfrequenciesintherangeto
+ d.

Imagineanantennainsideacavityinfullthermodynamicequilibriumattemperature
Tconnectedthroughatransmissionlinetoamatchedresistor(whoseresistanceequals
theradiationresistanceoftheantenna)inasecondcavityatthesametemperature
(Figure3.5).Afilterbetweenthecavitiespassesonlycurrentsinanarrowrangeof
frequenciesbetweenand + d.Becausethisentiresystemisinthermodynamic
equilibrium,nonetpowercanflowthroughthewiresconnectingtheantennaandthe
resistor.Otherwise,onecavitywouldheatupandtheotherwouldcooldown,inviolation
ofthesecondlawofthermodynamics.ThetotalspectralpowerP fromalldirections
collectedinonepolarizationishalfthetotalspectralpowerintheunpolarizedblackbody
radiation,so

1
2 4 e
P = A (, )B
(3.37)

mustequaltheNyquistspectralpowerP producedbytheresistor.InsertingtheNyquist
formulafromEquation2.119andPlanckslawfromEquation2.85,

http://www.cv.nrao.edu/~sransom/web/Ch3.html 14/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

[ ] [ ]
h h
kT 2kT kT
P = kT and B = (3.38)
h 2 h
exp ( )
kT
1 exp( )
kT
1

leadsto

2kT
kT = A e(, ) ,
2 2 4 (3.39)

andfinally,

4 A e(, ) = 4A e = 2. (3.40)

WithoutusingMaxwellsequationswehaveobtainedtheremarkableresult

2
A e = (3.41)
4

whichimpliesthatalllosslessantennas,fromtinydipolestothe100mdiameterGreen
BankTelescope(GBT),havethesameaveragecollectingarea.A eisproportionalto 2
becausespacehastwomoredimensionsthanatransmissionlinehas.

Thecollectingareaofanisotropicreceivingantennaisproportionalto 2,somost
satellitebroadcastservices,GPS(GlobalPositioningSystem)orsatelliteFMradiofor
example,operateatrelativelylongwavelengths(10to20cm).Likewise,practicalradio
telescopesconstructedfromarraysofdipolesarereasonablysensitiveonlyatlong
wavelengths.

ByanalogywithEquation3.34,thebeamsolidangleofalosslessreceivingantenna
isdefinedas

A e(, )
A 4 ,
A0 (3.42)

whereA 0isthemaximumeffectivecollectingarea,so

A 0 A = 2.
(3.43)

http://www.cv.nrao.edu/~sransom/web/Ch3.html 15/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

ThemuchlargerpeakcollectingareaoftheGBTimpliesithasamuchsmallerbeam
solidangle A.

3.1.5ReciprocityTheorems
Manyantennapropertiesarethesameforbothtransmittingandreceiving.Itisoften
easiertocalculatethegainofatransmittingantennathanthecollectingareaofa
receivingantenna,anditisofteneasiertomeasurethereceivingpowerpatternofalarge
radiotelescopethantomeasureitstransmittingpowerpattern.Thusthis
receiving/transmittingreciprocitygreatlysimplifiesantennacalculationsand
measurements.ReciprocitycanbeunderstoodviaMaxwellsequationsorby
thermodynamicarguments.

BurkeandGrahamSmith[20]statetheelectromagneticcaseforreciprocityclearly:
Anantennacanbetreatedeitherasareceivingdevice,gatheringtheincomingradiation
fieldandconductingelectricalsignalstotheoutputterminals,orasatransmittingsystem,
launchingelectromagneticwavesoutward.Thesetwocasesareequivalentbecauseof
timereversibility:thesolutionsofMaxwellsequationsarevalidwhentimeisreversed.

Thestrongreciprocitytheoremstates,

IfavoltageisappliedtotheterminalsofanantennaAandthecurrentis
measuredattheterminalsofanotherantennaB,thenanequalcurrent(in
bothamplitudeandphase)willappearattheterminalsofAifthesame
voltageisappliedtoB.(Figure3.6)

ItcanbeformallyderivedfromMaxwellsequations(seeapartialderivationinWilson
etal.[116,AppendixD])orbynetworkanalysis(seeKrausetal.[63,Antennas,
p.252]).

Figure3.6:ThestrongreciprocitytheoremimpliesthatthetransmittervoltagesV Aand
1 1
V BarerelatedtothereceivercurrentsI AandI BbyI B V A = I A V Bforanypairof
antennasAandB.

Mostradioastronomicalapplicationsdonotdependonthedetailedphase
relationshipsofvoltagesandcurrents,soitissufficienttouseaweakreciprocity
theoremthatrelatestheangulardependencesofthetransmittingpowerpatternandthe
receivingcollectingareaofanyantenna:Thepowerpatternofanantennaisthesamefor
transmittingandreceivingthatis,

http://www.cv.nrao.edu/~sransom/web/Ch3.html 16/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

G(, ) A e(, ).
(3.44)

Theweakreciprocitytheoremcanbeprovenbyanothersimplethermodynamic
thoughtexperiment:Anantennaisconnectedtoamatchedloadinsideacavityinitiallyin
equilibriumattemperatureT.Theantennasimultaneouslyreceivespowerfromthecavity
wallsandtransmitspowergeneratedbytheresistor.Thetotalpowertransmittedinall
directionsmustequalthetotalpowerreceivedfromalldirectionsbecausenonetpower
canbetransferredbetweentheantennaandtheresistorotherwisetheresistorwouldnot
remainattemperatureT.Moreover,inanydirection,thepowerreceivedandtransmitted
bytheantennamustbethesame,elsethecavitywallindirectionswherethetransmitted
powerwasgreaterthanthereceivedpowerwouldriseintemperatureandthecavitywall
indirectionsoflowertransmitted/receivedpowerratiowouldcool,leadingtoaviolation
ofthesecondlawofthermodynamics.

TheconstantofproportionalityrelatingGandA ecanbederivedfrom
Equations3.32and3.41:

2
A e = and G = 1. (3.45)
4

Thusenergyconservationandtheweakreciprocitytheoremimply

2G(, )
A e(, ) = (3.46)
4

foranyantenna.Thisextremelyusefulequationshowshowtocomputethereceiving
powerpatternfromthetransmittingpowerpatternandviceversa.

Example.Usethetransmittingpowerpatternofashortdipole(Equation3.14)to
calculatetheeffectivecollectingareaofashortdipoleusedasareceivingantenna:
2G(, ) 2 3sin 2
A e(, ) = = ,
4 4 2
3 2sin 2
Ae = .
8
Theeffectivecollectingareaofashortreceivingdipoledoesnotdependonthe
lengthlofthedipoleitselfbecausethetransmittingpowerpatternofashort(l )
dipoleisindependentofl.

3.1.6AntennaTemperature
Aconvenientpracticalunitforthepoweroutputperunitfrequencyfromareceiving
antennaistheantennatemperatureT A.Antennatemperaturehasnothingtodowiththe
physicaltemperatureoftheantennaasmeasuredbyathermometeritisonlythe
temperatureofamatchedresistorwhosethermallygeneratedpowerperunitfrequencyin
http://www.cv.nrao.edu/~sransom/web/Ch3.html 17/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

thelowfrequencyNyquistapproximation(Equation2.117)equalsthatproducedbythe
antenna:

P
TA . (3.47)
k

Itiswidelyusedforthefollowingreasons:

1.1Kofantennatemperatureisaconvenientlysmallpowerperunitbandwidth.
T A = 1Kcorrespondsto
P = kT A = 1.38 10 23 J K 1 1 K = 1.38 10 23 W Hz 1.
2.Itcanbecalibratedbyadirectcomparisonwithhotandcoldloads(anotherword
formatchedresistors)connectedtothereceiverinput.
3.TheunitsofreceivernoisearealsoK,socomparingthesignalinKwiththe
receivernoiseinKmakesiteasytocomparethesignalandnoisepowers.

CombiningEquations3.35and3.47showsthatanunpolarizedpointsourceofflux
densitySincreasestheantennatemperatureby

P A eS
TA = = , (3.48)
k 2k

whereA eistheeffectivecollectingarea.Itisoftenconvenienttoexpressthepointsource
sensitivityofaradiotelescopeinunitsofkelvinsperjanskyratherthaninunitsof
effectivecollectingarea(m 2).Theeffectivecollectingareacorrespondingtoasensitivity
of1 K Jy 1is

2kT A 2 1.38065 10 23 J K 1 1 K
Ae = = 26 2 1
= 2761 m 2. (3.49)
S 10 Wm Hz

InanarbitraryradiationfieldI (, ),Equation3.37becomes

1
2 4 e
P = A (, ) I (, ) . (3.50)

ReplacingP byantennatemperatureT AusingEquation3.47andinserting


I (, ) = 2kT b(, ) / 2(Equation2.33)gives

1 2k
2 4 e
kT A = A (, ) T b(, ) ,
2 (3.51)

http://www.cv.nrao.edu/~sransom/web/Ch3.html 18/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

1 (3.52)
T A = 2 4 A e(, ) T b(, ) .

InthelimitofaveryextendedsourcehavingnearlyconstantT bacrosstheentire
beam,

Tb
TA =
2
4 A e(, ) (3.53)

so

T A = T b.
(3.54)

Inwords,theantennatemperatureproducedbyasmoothsourcemuchlargerthanthe
antennabeamequalsthesourcebrightnesstemperature.

Ifalosslessantennaispointedatacompactsourcecoveringasolidangle smuch
smallerthanthebeamandhavinguniformbrightnesstemperatureT b,then

A 0T b s
TA = , (3.55)
2

whereA 0istheonaxiseffectivecollectingarea.SubstitutingA 0 A = 2(Equation3.43)


givestheresult:

TA s
= .
Tb A (3.56)

Statedinwords,theantennatemperatureequalsthesourcebrightnesstemperature
multipliedbythefractionofthebeamsolidanglefilledbythesource.AT b = 10 4K
sourcecovering1%ofthebeamsolidanglewilladd100Ktotheantennatemperature.
Theratio( s / A)iscalledthebeamfillingfactor.

Themainbeamofanantennaisdefinedastheregioncontainingtheprincipal
responseouttothefirstzeroresponsesoutsidethisregionarecalledsidelobesor,very
farfromthemainbeam,strayradiation.Themainbeamsolidangle MBisdefinedas

1
G 0 MB
MB G(, ) d.
(3.57)

http://www.cv.nrao.edu/~sransom/web/Ch3.html 19/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

Thefractionofthetotalbeamsolidanglelyinginsidethemainbeamiscalledthemain
beamefficiencyor,loosely,thebeamefficiency:

MB
B . (3.58)
A

3.2ReflectorAntennas

3.2.1ParaboloidalReflectors
Antennasusefulforradioastronomyatshortwavelengthsmusthavecollectingareas
muchlargerthanthe 2 / (4)collectingareaofanisotropicantennaandmuchhigher
angularresolutionthanashortdipoleprovides.Becausearraysofdipolesareimpractical
atwavelengths < 1morso,mostradiotelescopesuselargereflectorstocollectand
focuspowerontotheirsmallfeedantennas,suchaswaveguidehornsordipolesbacked
bysmallreflectors,thatareconnectedtoreceivers.Themostcommonreflectorshapeisa
paraboloidofrevolutionbecauseitcanfocustheplanewavefromadistantpointsource
ontoasinglefocalpoint.

Tofocusplanewavesontoasinglepoint,thereflectormustkeepallpartsofanon
axisplanewavefrontinphaseatitsfocalpoint.Thusthetotalpathlengthstothefocus
mustallbethesame,andthisrequirementissufficienttodeterminetheshapeofthe
desiredreflectingsurface.Clearlythesurfacemustberotationallysymmetricaboutits
axis.Inanyplanecontainingtheaxis,thesurfacelookslikethecurveinFigure3.7.

Figure3.7:Aplanecontainingtheaxisofaparaboloidalreflectorwithfocallengthf.
Planewavefrontsfromadistantpointsourceareshownasdottedlinesperpendicular
tothezaxis.Fromawavefrontatheighthabovethevertex(thepointr = 0,z = 0)of
theparaboloid,theraypath(dashedline)lengthsatallradialoffsetsrdowntothe
reflectoranduptotheprimefocusatz = fmustbeequal.

Therequirementofconstantpathlengthcanbewrittenbyequatingtheonaxispath
length(f + h)fromanyheighthtothereflectorandthenbacktotheprimefocusat
heightfwiththeoffaxispathlength:

http://www.cv.nrao.edu/~sransom/web/Ch3.html 20/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

(3.59)
(f + h) = r 2 + (f z) 2 + (h z).

Thisyieldsthereflectorheightzasafunctionofradiusr:

r 2 + (f z) 2 = f + z,

r 2 + f 2 + z 2 2fz = f 2 + z 2 + 2fz

theresultis

r2
z= . (3.60)
4f

Thisistheequationofaparaboloidwithfocallengthf.

TheratioofthefocallengthftothediameterDofthereflectoriscalledthe / ratio
orfocalratio.Notethatthegain,collectingarea,andbeamwidthofareflectorantenna
dependonlyweaklyandindirectlyonf / D,viatheeffectoff / Donilluminationtaper.In
principle,f / Disafreeparameterforthetelescopedesigner,butinpracticeitis
constrained.Ifthereflectorf / Distoohigh,thesupportstructureneededtoholdthefeed
orsubreflectoratthefocusofalargeradiotelescopebecomesverylongandunwieldy.
Consequentlylargeradiotelescopesusuallyhavef / D 0.4,anorderofmagnitudelower
thanthetypicalfocalratioofanopticaltelescope.Thedrawbackofalowf / Disasmall
fieldofview.Thefocalellipsoidisthevolumearoundtheexactfocalpointthatremains
inreasonablygoodfocus,andthefocalcircleisdefinedbytheintersectionofthefocal
ellipseandthetransverseplaneatz = f.Ruze[95]showedthattheangularradiusofthe
focalcircleisproportionalto(f / D) 2,andonlyasmallnumber(aboutseven)ofdiscrete
feedscanfitinsidethefocalcircleofanf / D 0.4paraboloid.Largearraysoffeedsor
imagingcamerasrequirelargerf / Dratios,obtainedeitherbyusingashallower
paraboloidorbyusingmagnifyingsubreflectorstoincreasetheeffectivefocallength.

Theprimarymirrorsofmostradiotelescopesarecircularparaboloidsorsections
thereofforthefollowingreasons:

1.TheeffectivecollectingareaA eofareflectorantennacanapproachitsprojected
geometricareaA = D 2 / 4.
2.Theyareelectricallysimple(comparedwithaphasedarrayofdipoles,for
example).
3.Asinglereflectorcanworkoverawiderangeoffrequencies.Changing
frequenciesonlyrequireschangingthefeedantennaandreceiverlocatedatthe
focalpoint,notbuildingawholenewradiotelescope.

3.2.2TheFarFieldDistance
Howfarawaymustapointsourcebeforthereceivedwavestosatisfytheassumption
http://www.cv.nrao.edu/~sransom/web/Ch3.html 21/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

thattheyarenearlyplanaracrossthereflector?Theanswerdependsonboththe
wavelengthandthereflectordiameterD.Figure3.8showsthesphericalwaveemitted
byapointsourceafinitedistanceRfromaflataperture,animaginarycircularholethat
coversthereflector.Itcouldbelocatedattheplanez = hshowninFigure3.7,for
example.

Figure3.8:Thesphericalwavefront(dashedcircle)emittedbyapointsourceat
distanceRdeviatesfromaplanebyattheedgeofanapertureofdiameterD.

Themaximumdeparturefromaplanewaveoccursattheedgeoftheaperture.The
farfielddistanceR ffissomewhatarbitrarilydefinedbyrequiringthat < / 16.Atthe
apertureedge,thePythagoreantheoremgives

R 2 = (R ) 2 + () D 2
2
. (3.61)

Thus

D2
R= + . (3.62)
2 8

Inthelimit D,wehave / 2 D 2 / (8)and

D2
R . (3.63)
8

Giventhe = / 16criterion,thefarfielddistanceis

2D 2
R ff . (3.64)

http://www.cv.nrao.edu/~sransom/web/Ch3.html 22/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

IfR < R ff,thepathlengtherrorswillintroducesignificantphaseerrorsinthewaves


comingfromtheoffaxisportionsofthereflector,reducingtheeffectivecollectingarea
anddegradingtheantennapattern.

Example.WhatisthefarfielddistanceoftheGreenBankTelescope(D = 100m)
observingat = 1cm?
Equation3.64gives

2(100 m) 2 2 10 4 m 2
R ff = = = 2 10 6 m = 2000 km.
1 cm 0.01 m

SuchalargefarfielddistancemakesgroundbasedmeasurementsoftheGBT
antennapatternimpractical.TomeasuresmallerrorsintheGBTreflectorsurface
usingradioholography,itisnecessarytoobserveageostationarysatellitehavingan
orbitalaltitudeR 36,000 2000km.Similarly,theeasiestwaytodetermine
thetransmittingpowerpatternforalargeradarantennasuchastheD = 305m
Areciboreflectoristoscanacrossacelestialpointsourceinthefarfieldandusethe
reciprocitytheoremtoequatethetransmittingandreceivingpatterns.

3.2.3PatternsofApertureAntennas
Inoptics,thetermaperturereferstotheopeningthroughwhichallrayspass.For
example,theapertureofaparaboloidalreflectorantennawouldbetheplanecircle,
normaltotheraysfromadistantpointsource,thatjustcoverstheparaboloid(Figure3.9).
Thephaseoftheplanewavefromadistantpointsourcewouldbeconstantacrossthe
apertureplanewhentheapertureisperpendiculartothelineofsight.

Figure3.9:TheapertureplaneassociatedwithaparaboloidaldishofdiameterD.

Anotherexampleofanapertureisthemouthofawaveguidehornantenna
(Figure3.10).

http://www.cv.nrao.edu/~sransom/web/Ch3.html 23/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

Figure3.10:DocEwenlookingintotherectangularapertureofthehornantenna
usedtodiscoverthe=21cmlineofneutralhydrogen.Imagecredit:
NRAO/AUI/NSF.

Howcanthebeampattern,orpowergainasafunctionofdirection,ofanaperture
antennabecalculated?Forsimplicity,firstconsideraonedimensionalapertureofwidth
D(Figure3.11)andcalculatetheelectricfieldpatternatadistant(R R ff)point.

Figure3.11:Coordinatesystemforaonedimensionallinearaperturespanning
D / 2 < x < + D / 2.Forasourcedistantfromthecenteroftheaperture(R D),the
fractionalchangeinthedistancerbetweenthesourceandanyapertureelementata
distance|x|fromtheaperturecenterissmall,sothevariabler(x)inthedenominatorof
Equation3.66canbereplacedbytheconstantR.However,thevariationofr(x)across
theaperturecanbemuchlargerthanthewavelength,sotheoscillatingnumeratorof
http://www.cv.nrao.edu/~sransom/web/Ch3.html 24/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

Equation3.66cannotbereplacedbyaconstant.

Whenusedinatransmittingantenna,thefeedcanilluminatetheapertureantenna
withasinewaveoffixedfrequency = / (2)andelectricfieldstrengthg(x)thatvaries
acrosstheaperture.Theilluminationinducescurrentsinthereflector.Thecurrentswill
varywithbothpositionandtime:

I g(x)exp( it). (3.65)

Theconstantofproportionalitydoesntmatteryetitcanbecalculatedlaterfromenergy
conservation.Huygenssprincipleassertsthattheaperturecanbetreatedasacollection
ofsmallelementswhichactindividuallyassmallantennas.Huygenssprincipleactually
appliestowavesofanytype,soundwavesforexample.Theelectricfieldproducedby
thewholeapertureatlargedistancesisjustthevectorsumoftheelementalelectricfields
fromthesesmallantennas.Thefieldfromeachelementextendingfromxtox + dxis

exp( i2r(x) / )
df g(x) dx,
r(x) (3.66)

wherer(x)isthedistancebetweenthesourceandtheapertureelementatpositionx
(Figure3.11).Inthefarfield(Equation3.64),theFraunhoferapproximation

r R + xsin
(3.67)

isvalid.Thisequationisusuallywrittenintheform

r R + xl, (3.68)

where

l sin. (3.69)

Forthesmallangles 1radrelevanttolargeD apertures,l = sin .

Atlargedistances,thequantity

1 1
(3.70)
r R

isnearlyconstantacrosstheapertureandcanbeabsorbedbytheconstantof
proportionalityinEquation3.66.Althoughexp( i2R / )isaconstantbecauseRisfixed,
thevariablepartxlofr = R + xlinthenumeratorofEquation3.66cannotbeignoredat
anydistance:
http://www.cv.nrao.edu/~sransom/web/Ch3.html 25/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

df g(x)exp( i2xl / )dx.


(3.71)

When 0thephasexl / sin / varieslinearlyacrosstheaperture,anddifferentparts


oftheapertureaddconstructivelyordestructivelytothetotalelectricfieldf(l).Defining

x
u
(3.72)

toexpresspositionalongtheapertureinunitsofwavelengthyields

f(l) = apertureg(u)e i 2 l udu.


(3.73)

Inwords,thisveryimportantequationsaysthatinthefarfield,theelectricfieldpattern
f(l)ofanapertureantennaistheFouriertransform(AppendixA.1)oftheelectricfield
distributiong(u)illuminatingthataperture.

3.2.4TheElectricFieldandPowerPatternsofaUniformly
IlluminatedAperture
Whatistheelectricfieldpatternofauniformlyilluminatedonedimensionalapertureof
widthDatwavelength?Uniformilluminationmeansthatthestrengthofthe
illuminationisconstantovertheaperture:

D D
g(u) = constant, <u<+ .
2 2

Thisquestionisbestansweredintwosteps:firstfindthefarfieldpatternofaunit
aperture(D = )andthenusethesimilaritytheorem(EquationA.11)forFourier
transformstoscalethefirstresulttoanapertureofanysize.

Theunitrectanglefunctionisdefinedas

(u) 1, 1 / 2 < u < + 1 / 2,


(3.74)

and(u) = 0otherwise.Thefunctionsymbol(anuppercasepi)iseasytoremember
becauseitlookslikethefunctiongraphshowninthetoppanelofFigure3.12.

http://www.cv.nrao.edu/~sransom/web/Ch3.html 26/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

Figure3.12:Thesymbolisshapedliketheunitrectanglefunctionitrepresents.The
functionsinc(l) sin(l) / (l)istheFouriertransformoftheunitrectanglefunction
andistheelectricfieldpatternofauniformlyilluminatedunitaperture.Thepower
patternofauniformlyilluminatedunitapertureisshowninthebottompanel.Forlarge
(D )apertures,thezerosatl = 1, 2, appearattheangles
/ D, 2 / D, .

Inserting(u)intoEquation3.73givesthefieldpatternf(l)oftheuniformly
illuminatedunitaperture:


f(l) = (u)e i 2 l uu. (3.75)

Thus

+1/2 i 2 l u e i 2 l u + 1 / 2 e i l ei l
f(l) = 1/2 e du =
i2l
| 1/2 =
i2l
. (3.76)

Next,differencethemathematicalidentities(AppendixB.3)

http://www.cv.nrao.edu/~sransom/web/Ch3.html il 27/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

e i l = cos(l) + isin(l),
e i l = cos(l) isin(l)

toderive

e i l e i l = 2isin(l).

InsertingthisresultintoEquation3.76gives

2isin(l) sin(l)
f(l) = = sinc(l). (3.77)
2il (l)

TheusefulsincfunctiondefinedinEquation3.77isplottedinthemiddlepanelof
Figure3.12.

Thepowerpatternp(l)isthesquareofthefieldpatternf(l).Thepowerpattern
p(l) = sinc 2(l)ofauniformlyilluminatedunitapertureisgraphedinthebottompanelof
Figure3.12.Thecentralpeakofthepowerpatternbetweenthefirstzerosatl = 1is
calledthemainbeam.Thesmallerpeaksarecalledsidelobes.Theyareseparatedby
zerosornullsinthepowerpatternatl = 1, 2, .

NextapplythepowerfulsimilaritytheoremforFouriertransforms:iff(l)isthe
Fouriertransformofg(u),then

1
|a|
f ()
l
a

istheFouriertransformofg(au),wherea 0isadimensionlessscalingfactor.
Accordingtothesimilaritytheorem,makingafunctiongwider(0 < a < 1)ornarrower(
a > 1)makesitsFouriertransformfnarrowerandtaller,orwiderandshorter,
respectively,alwaysconservingtheareaunderthetransform.Consequentlythe
beamwidthofanapertureantennaisinverselyproportionaltotheaperturesizein
wavelengthsandtheonaxisfieldstrengthisdirectlyproportionaltotheaperturesizein
wavelengths.

ThescalefactorforauniformlyilluminatedonedimensionalapertureofwidthD
operatingatwavelengthisa = / D,sotheelectricfieldpatternbecomes

f(l) = ()D sin(lD / )


(lD / )

D

sinc ( )
lD

.

Iftheapertureislarge(D / 1),therelevantanglesaresosmall( 1radian)that


l = sin and

http://www.cv.nrao.edu/~sransom/web/Ch3.html 28/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

f() =
D

sinc ( )
D

.
(3.78)

Thepowerpatternisproportionaltothesquareoftheelectricfieldpattern,so

P(l) ()D 2

sinc 2 ( )
lD

.

If 1radian,then

P() = ()D 2

sinc 2 ( ) D

. (3.79)

Radioastronomersusetheanglebetweenthehalfpowerpointstospecifythe
angularwidthofthemainbeam,callingitthehalfpowerbeamwidth(HPBW)orthe
fullwidthbetweenhalfmaximumpoints(FWHM).Thenarrowbeamwidth
HPBW 1 radofalarge(D )onedimensionaluniformlyilluminatedaperture
satisfies

( )
1 HPBWD
P( HPBW / 2) = = sinc 2 , (3.80)
2 2

HPBWD
0.443 , (3.81)
2


HPBW 0.89 . (3.82)
D

Thesimilaritytheoremimpliesthegeneralscalingrelation


HPBW .
D (3.83)

Theconstantofproportionalityvariesslightlywiththeilluminationtaper.Evenanideal
apertureantennaoffinitesizehasafiniteresolvingpowerthatislimitedbydiffraction,
thespreadingofrayspassingthroughafiniteaperture,andEquation3.82specifiesthe
diffractionlimitedresolutionofauniformlyilluminatedapertureantenna.

Theweakreciprocitytheorem(Section3.1.5)saysthattheprecedinganalysisofthe
http://www.cv.nrao.edu/~sransom/web/Ch3.html 29/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

transmittingpowerpatternofanapertureantennaalsoyieldsitsreceivingpowerpattern,
orthevariationofA ewithorientation.Inreceivingterms,theanalogofthepowerpattern
iscalledthepointsourceresponse.Forauniformlyilluminatedaperture,scanninga
radiotelescopebeaminangleacrossapointsourcewillcausetheantennatemperature
tovaryassinc 2(),andthewidthofthehalfpowerresponsewillequalthetransmitting
HPBW.ThereceivingHPBWissometimescalledtheresolvingpowerofatelescope
becausetwoequalpointsourcesseparatedbytheHPBWarejustresolvedbythe
Rayleighcriterionthatthetotalresponsehasaslightminimummidwaybetweenthe
pointsources.

3.2.5TheElectricFieldandPowerPatternswithTaperedIllumination
Practicalfeedssuchassmallwaveguidehornsorhalfwavedipolesbackedbysmall
subreflectorscannotilluminatealargeapertureuniformly.Abetterapproximationtotheir
illuminationisthecosinetaperedfieldpattern(cosinesquaredtaperedpowerpattern)


g(u) = cos(u), 1 / 2 < u < + 1 / 2,
2 (3.84)

andg(u) = 0otherwise(Figure3.13).The( / 2)normalizationfactorinEquation3.84


ensuresthat

+1/2
1 / 2 g(u)u = 1. (3.85)

Thecorrespondingfieldpatternofaonedimensionalunitapertureisgivenby

+1/2
f(l) = 1 / 2 cos(u)e i 2 l uu. (3.86)
2

ThisFouriertransformcanbeevaluatedasfollows:

+1/2 i u
+ e i u)e i 2 l uu
4 1/2
f(l) = (e (3.87)

=
4 [
ei ( 1 2 l ) u + 1 / 2 e i ( 1 + 2 l ) u + 1 / 2
|
i(1 2l) 1 / 2
+ |
i(1 + 2l) 1 / 2 ] (3.88)

=
4 [
ei ( 1 / 2 l ) e i ( 1 / 2 l )
i2(1 / 2 l)
+
ei ( 1 / 2 + l ) e i ( 1 / 2 + l )
i2(1 / 2 + l) ] (3.89)

=
4 [
2isin[(1 / 2 l)]
i2(1 / 2 l)
+
2isin[(1 / 2 + l)]
i2(1 / 2 + l) ] (3.90)

=

[ cos(l)
4 (1 / 2 l)
+
cos(l)
(1 / 2 + l) ] =

4
cos(l) (
2 / 4 2l 2 ) (3.91)

http://www.cv.nrao.edu/~sransom/web/Ch3.html 30/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

Figure3.13:Thecosinetaperedfieldilluminationg(u)(top)yieldsthefieldpatternf(l)
andpowerpatternP(l)onaunitaperture.ThelowsidelobesofP(l)areclearlyvisible
onlyonaplotofP(dB)(bottom).

toyieldthefieldpattern

cos(l)
f(l) =
1 4l 2 (3.92)

ofaonedimensionalunitaperturewithcosinetaperedilluminationgivenby
Equation3.84.Boththefieldpatternandthepowerpattern

http://www.cv.nrao.edu/~sransom/web/Ch3.html 31/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

P(l) = [f(l)] 2 = [ cos(l) 2


1 4l 2 ] (3.93)

areshowninFigure3.13.ThesidelobesaresoweakthataplotofP(dB) = 10log 10(P)is


neededtoshowthemclearly(bottompanelofFigure3.13).

Taperingincreasesthehalfpowerbeamwidth.IfD ,thenormalizedpower
patternis

P() = [ cos(D / )
1 4(D / ) 2 ] 2
, (3.94)

and

( ) [ ]
HPBW 1 cos[ HPBWD / (2)] 2
P = = (3.95)
2 2 1 4[ HPBWD / (2)] 2

canbesolvednumericallytoyield


HPBW 1.2 . (3.96)
D

Thisbeamwidthistypicalofmostradiotelescopes.

Theperfectlysharpcutoffofilluminationattheedgeoftheapertureshowninthe
toppanelofFigure3.13cannotbeachievedinpractice.Anyilluminationextending
beyondthereflectoriscalledspillover.Inthecaseofareceivingantenna,aprimefocus
feedlookingdownatanaperturealsoseesspilloverradiationfromthesurrounding
ground.Mostsoilsaregoodabsorbers,whichemitblackbodyradiationattheambient
temperatureT 300 K,andgroundradiationcanaddsignificantlytothesystemnoise
temperatureofaradiotelescope.Thepurposeofthe15mhighannulargroundscreen
surroundingtheAreciboreflector(Figure8.2)istointerceptmostofthespillover
radiationandredirectittothecoldskyinorderminimizethesystemtemperature.

3.3TwoDimensionalApertureAntennas

3.3.1TheFieldPatternofaTwoDimensionalAperture
Themethodusedtoshowthatthefieldpatternofaonedimensionalapertureistheone
dimensionalFouriertransformoftheaperturefieldillumination(Equation3.73)can
easilybegeneralizedtothemorerealisticcaseofatwodimensionalaperture:

http://www.cv.nrao.edu/~sransom/web/Ch3.html 32/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

f(l, m) g(u, v)e i 2 ( l u + m v ) du dv, (3.97)

wheremistheyaxisanalogoflonthexaxis,and

y
v . (3.98)

Inwords,Equation3.97statesthattheelectricfieldpatternofatwodimensional
apertureisthetwodimensionalFouriertransformoftheaperturefieldillumination.

3.3.2TheUniformlyIlluminatedRectangularAperture

Figure3.14:AtwodimensionalrectangularaperturewithsidelengthsD xandD y.
Dividinglengthsintheapertureplanebythewavelengthyieldsthenormalized
coordinatesu x / andv y / .Thedirectionfromtheorigintoanydistantpointcan
bespecifiedbyl sin xandm sin y,where xistheanglefromthe(y, z)planeand
yistheanglefromthe(x, z)plane.

Thetwodimensionalcounterpartofauniformlyilluminatedonedimensional
apertureisauniformlyilluminatedrectangularaperturewithsidelengthsD xandD y.
Dividinglengthsintheapertureplanebythewavelengthyieldsthenormalized
coordinatesu x / andv y / .Thedirectionfromtheoriginofthe(u, v)planetoany
distantpointcanbespecifiedbyl sin xandm sin y,where xistheanglefromthe
(y, z)planeand yistheanglefromthe(x, z)plane(Figure3.14).Iftheillumination
g(x, y)isconstantovertheaperture,theintegralsoveruandvintheFouriertransformare
separableand

lD x mD y
f(l, m) sinc
( ) ( )

sinc

. (3.99)

Squaringtheelectricfieldpatterngivestherelative(normalizedtounityatthepeak)
powerpattern

http://www.cv.nrao.edu/~sransom/web/Ch3.html 33/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

( ) ( )
lD x mD y
P n(l, m) = sinc 2 sinc 2 . (3.100)

TheabsolutepowergainGinanydirectioncanbecalculatedfromtherelativepower
patternbyinvokingenergyconservation:

+1 +1
G = 4 = G 0 1 1 P n(l, m)l m, (3.101)

[ ] [ ]
+1
sin(lD x / ) 2 +1
sin(mD y / ) 2
4 = G0 1 lD x /
l 1 mD y /
m. (3.102)

Definingthetemporaryvariableaas

lD x D x
a , so da = dl, (3.103)

gives,forD x ,

[ ] [
sin(lD x / ) 2
]
2
+1 sin a
1 lD x /
dl 2
da
a D x
=
Dx (3.104)

becausethevalueofthedefiniteintegralinsquarebracketsis.[Toprovethis,simply
applyRayleighstheorem(EquationA.7)totheFouriertransformpairsinc(l)
(Equation3.77)and(u)(Equation3.74).]Then

2
4 = G 0 .
D xD y (3.105)

Thusthepeakpowergainis

4D xD y
G0 = , (3.106)
2

andthepowerpatternofauniformlyilluminatedrectangularaperturewithsidelengths
D xandD yis

http://www.cv.nrao.edu/~sransom/web/Ch3.html 34/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

( ) ( )
4D xD y lD x mD y
G= sinc 2 sinc 2 , (3.107)
2

and

( ) ( )
4D xD y xD x yD y
G sinc 2 sinc 2 (3.108)
2

when xand yaremuchsmallerthan1radian.

Ingeneral,thepeakpowergainofanapertureantennaisproportionaltothe
geometricareaA geom(A geom = D xD yinthiscase)oftheaperture.Theconstantof
proportionalityis4 / 2forauniformlyilluminatedapertureandsomewhatlessforany
otherilluminationpattern.

UsingEquation3.46

2G
Ae = , (3.109)
4

wefindthattheonaxiseffectivecollectingareais

2G 0 4 2D xD y
A0 = = = D xD y = A geom. (3.110)
4 4 2

Thepeakeffectiveareaofanidealuniformlyilluminatedapertureequalsitsgeometric
area,independentofwavelength.Withanyotherilluminationtaper,theeffectiveareais
smallerthanbutproportionaltothegeometricarea.Itisusefultodefinetheaperture
efficiency Aastheratiooftheeffectiveareatogeometricarea:

A0
A .
A geom (3.111)

Thus A = 1foranidealuniformlyilluminatedapertureand A < 1otherwise.The


apertureefficienciesofmostradiotelescopesare A 70%,althoughphasedarrayfeeds
controltheilluminationwellenoughtoletASKAP(Figure8.6)reach A 80%.

Large(D )rectangularwaveguidehornsarenearlyuniformlyilluminated
unblockedapertures,sotheiractualgainsandeffectivecollectingareascanbecalculated
accurately.Thismakesthemusefulformeasuringtheabsolutefluxdensitiesofstrong
sourcessuchasCasAandCygAanddefiningthepracticalfluxdensityscalesusedby
http://www.cv.nrao.edu/~sransom/web/Ch3.html 35/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

radioastronomers[6].

Mostaperturesassociatedwithreflectorsandlensesarecircular.Thepowerpattern
ofauniformlyilluminatedcircularapertureisknownastheAirypattern.6

3.3.3GaussianBeamSolidAngleandBeamwidth

Figure3.15:ThebeamsofmostradiotelescopesarenearlyGaussian,andtheir
beamwidthsareusuallyspecifiedbytheangle HPBWbetweenthehalfpowerpoints.
Abscissa:offsetfromthebeamcenterinunitsoftheHPBW.Ordinate:Effective
apertureA enormalizedbythepeakeffectiveapertureA 0.

Foranyrealisticilluminationtaper,thebeamsolidangle(Equation3.42)

A e(, )
A 4
A0

ofaradiotelescopeisaboutequaltothesquareofthehalfpowerbeamwidth HPBW.In
fact,thebeamsofmostradiotelescopesarenearlyGaussianandcanbewrittenas

Ae
= exp( x 2), (3.112)
A0

whereistheanglefromthebeamcenterandxisascalingfactorsuchthatA e / A 0 = 1 / 2
when = HPBW / 2(Figure3.15):

[( )]
1 HPBW 2
= exp x . (3.113)
2 2

http://www.cv.nrao.edu/~sransom/web/Ch3.html 36/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

Thus

4ln2
x= ,
2 (3.114)
HPBW

[ ( )]
Ae 2
= exp 4ln2 , (3.115)
A0 HPBW

and

2
[ (
A = = 0 = 0 exp 4ln2

HPBW
2
. )] (3.116)

Integratingoverandsubstitutingthedummyvariabley = 4ln2( / HPBW) 2yields

2
HPBW
A = 2 ( 8ln2 )
y = 0 exp( y) y, (3.117)

sothebeamsolidangleofaGaussianbeamis

A = ( )


4ln2 HPBW
2
1.133 HPBW.
2
(3.118)

3.3.4ReflectorAccuracyRequirements
Realradiotelescopesdonthaveperfectlysmoothparaboloidalreflectors.Small
deviationsfromthebestfitparaboloidmaybecausedbypermanentmanufacturing
errors,changinggravitationaldeformationsasthereflectoristilted,thermaldistortions
resultingfromsolarheating,andbendingbystrongwinds.Therewillbesomeshortest
wavelength minbelowwhichthesesurfaceerrorsdegradethereflectorperformanceso
severelythatthetelescopebecomesunusable.Thereflectorsurfaceefficiency sis
definedasthepowergainoftheactualreflectordividedbythepowergainofaperfect
paraboloidalreflectorwiththesamesizeandillumination.Thefollowingcalculationof
how svarieswiththerms(rootmeansquare)surfaceerrorinwavelengths( / )isbased
ontheclassicmethodofRuze[96].

http://www.cv.nrao.edu/~sransom/web/Ch3.html 37/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

Figure3.16:Deviationsoftheactualreflectorsurface(thickcurve)fromthebestfit
paraboloid(thincurve)degradeshortwavelengthperformance(top).Vectorsumsof
theelectricfieldsEproducedbyNelementsofperfectandimperfectaperturesare
shownbottomleft.Bumpsintheimperfectapertureproducephaseshifts 4 /
whichlowerthevectorsumofelectricfieldsfromNEtoNEcos(bottomright).

Wheretheactualreflectorsurfacedeviatesfromthebestfitparaboloidbyadistance
(Figure3.16),thepathlengthofthereflectedwavewillbeinerrorbyalmost2and
thephaseerror(radians)ofthereflectedwavewillbe

2 4
(2) = .
(3.119)

Anoversimplifiedexamplewouldbeabumpysurface,halfcoveredwithsmallbumpsof
height andhalfcoveredwithsmalldipsofthesamedepth.Thenthe
contributionofeachareaelementtothefar(electric)field(Figure3.16)isreducedbya
factorcos.Inthelimit 1rad,cos 1 2 / 2 + and

E() 2
1 + , (3.120)
E(0) 2

sotherelativepowergainis

G()
G(0)
[ ]
E() 2
E(0)
1 2 1 ( )
4 2

. (3.121)

Thisroughestimateshowsthatthesurfaceerrorsmustbeanorderofmagnitudesmaller
thantheshortestusablewavelength,asevererequirementindeed.

Amorerealisticcalculationmakesuseofthefactthemosterrorshaveroughly

http://www.cv.nrao.edu/~sransom/web/Ch3.html 38/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

Gaussianamplitudedistributions.SupposethatthesurfaceerrorshaveaGaussian
probabilitydistributionP()withrms:

P() =
2
1
exp ( ) 2
2 2
.
(3.122)

Thentherelativefieldstrengthisobtainedastheweightedsumoverallpossible:

E / E(0)

cos ( )
4


2
1
exp ( )
2
2 2
.
(3.123)

Substitutinge i z = cosz + isinzturnsthisintegralintoamorefamiliarone,theFourier


transformofaGaussian:

E / E(0)

exp ( )i
4


2
1
exp ( )
2
2 2
.
(3.124)

Notethattheisinzpartdropsoutimmediatelybecauseitisantisymmetricinanotherwise
symmetricintegral.Tomakethislookevenmorefamiliar,lets 2 / ,x ,and
1
a (2) .Then


E / E(0) exp( i2sx)exp (ax) 2 x. ( ) (3.125)

RecallthattheFouriertransformoff(x) = exp( x 2)isF(s) = exp( s 2)(AppendixB.4)


andapplythesimilaritytheorem(EquationA.11)toget

E / E(0) =
|a|2
1
[ ( )]
exp
s 2
a (3.126)

= exp[ 2 2 2s 2] (3.127)

= exp ( 8 2 2
2 )
. (3.128)

PowerisproportionaltoE 2sothereflectorsurfaceefficiencyissimply

s = exp [ ( )] 4 2

. (3.129)

http://www.cv.nrao.edu/~sransom/web/Ch3.html 39/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

Equation3.129isoftencalledtheRuzeequationitisplottedinFigure3.17.

Figure3.17:Thesurfaceefficiency sdeclinesrapidlyasthermssurfaceerrorin
wavelengths / exceeds1 / 16 0.06.

Thesurfaceefficiency siscloselyrelatedtotheStrehlratioSusedbyoptical
astronomerstospecifythepeakintensitylosscausedbyopticalaberrationsor
atmosphericturbulence.TheStrehlratioisnormallyexpressedintermsoftherms
wavefronterrorinwavelengths,whichisabouttwicethermssurfaceerrorin
wavelengths / ,soEquation3.129implies

S = exp[ (2) 2]. (3.130)

Atraditionalruleofthumbfortheshortestwavelength minatwhicharadio
telescopeworksreasonablywellis

min
(3.131)
16

becausethesurfaceefficiencyat = minisonly

[( )]
s exp
2
4
0.54 (3.132)

andfallsexponentiallyatshorterwavelengths.Forexample,the100mdiameterGBTis
intendedtooperateatfrequenciesashighas 100GHz,or min 3mm.Tomeetthis
specification,thermsdeviationfromaperfectparaboloidmustnotexceed
3 mm / 16 200 m,thethicknessoftwosheetsofpaper.Thepowergainofaperfect
paraboloidalreflectorisproportionalto 2.IfthereflectorsurfacehasaGaussianerror
distributionwithrms,thenitsgainincreasesas 2atlowfrequencies,reachesa
maximumat

http://www.cv.nrao.edu/~sransom/web/Ch3.html 40/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

= 4, (3.133)

anddecreasesquicklyathigherfrequencies.

3.3.5PointingAccuracyRequirements
Realradiotelescopesdonthaveperfectlyaccuratepointing.Smallerrorsintrackinga
targetsourcereducethegaininthesourcedirectionandcontributetotheuncertaintyin
fluxdensitymeasurementsofcompactsources.Trackingerrorsarejustasimportantas
surfaceerrorsinlimitingtheshortwavelengthperformanceoflargeradiotelescopes.

ThepowerpatternsofmostradiotelescopesarenearlyGaussiannearthepeak.In
termsofthebeamwidthbetweenhalfpowerpoints HPBW,therelativegainatapoint
offsetbyanglefromthebeamaxisis

G
G0 [ (
= exp 4ln2

HPBW )]
2
. (3.134)

Iftheonedimensionaltrackingerrorineachcoordinate(e.g.,azimuthorelevationangle)
hasaGaussiandistributionwithrms 1,thetrackingerrorintwodimensionshasa
Rayleighdistribution

P() =

2
1
( )
exp
2
2
2 1
.
(3.135)

Themeansquaredtrackingerroris

2
2 = 0 2P() = 2 1. (3.136)

Thermsvalueofthetwodimensionaltrackingerroris 2 = 2 1 / 2 1,sosmalltracking
errorsreducetheaverageonsourcegainbythefactor

2 1

[ ( )]
2
G / G 0 = 1 + 4ln2 . (3.137)
HPBW

Moreimportantly,thefluctuatingonsourcegaincausedbytrackingerrorscontributesa
fractionaluncertainty7

http://www.cv.nrao.edu/~sransom/web/Ch3.html 41/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

S z (3.138)
= 1/2
,
S (1 + 2z)

where

( )
2 2
z 4ln2 , (3.139)
HPBW

toameasurementofsourcefluxdensityS.Thusanrmstrackingerrorof0.2 HPBWwill
contributea10%rmsfluxdensityuncertainty.For5%accuracy, 2 / HPBW 0.14(or
1 / HPBW 0.10ineachcoordinate)isneeded.

Forexample,wecancalculatethelargesttrackingerrorinarcseccompatiblewith
makingfluxdensitymeasurementswith5%rmserrorsusingtheGBT100mtelescopeat
= 33GHz.FromEquation3.138, S / S = 0.05when 2 / HPBW 0.14.Thehalfpower
beamwidthoftheGBTat = 33GHz( 9.1 mm)is

1.2D 1.2 0.0091 m


HPBW = 1.09 10 4 rad 23 arcsec. (3.140)
100 m

Thusthetotaltrackingerrormustbesmallerthan 2 = 0.14 23 arcsec = 3.2 arcsec,or


1 2 1 / 2 2 2.2 arcsec 10 5 radinazimuthandinelevationangle.

Thethermalexpansioncoefficientofsteelisabout10 5C 1,sochangingthe
temperaturedifferentialacrossthesteelGBTsupportstructurebyonly1centigrade
degreecouldproducea10 5 rad 2 arcsecpointingshift.Forthisreason,high
frequencyobserversmustmonitorpointingcalibrationsourcesandcorrecttheGBT
pointingeveryhourorso,particularlyjustaftersunriseonsunnydays.Windgustsalso
degradepointingaccuracy,buttheyfluctuateonmuchshortertimescales.

3.4Waveguides

http://www.cv.nrao.edu/~sransom/web/Ch3.html 42/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

Figure3.18:Theupperdrawingshowsthecrosssectionofarectangularwaveguide
havinginteriorwidthainthexdirectionandheightb a / 2intheydirection.The

componentofanyelectricfieldEparalleltoaconductingwallmustgotozeroatthe
wall,justaswithradiationinacavity(Figure2.16).Thecurvesindicatethreeallowed
xdistributionsofelectricfieldstrength(calledmodes).Thelowerdrawingisaplan
viewofthewaveguide.ItisanalogoustoFigure2.18showingwavesinaconducting
cavity.Radiationofwavelengthmusttravelthroughthewaveguideinthedirection
indicatedbythelargearrowtosatisfytheboundaryconditionn x = 2a / x = (2a / )cos
,wheren x = 1, 2, 3, (Equation2.66).OnlytheTE10modeshown,withn x = 1(
a = x / 2)andnovariationofE = E ywithheight(n y = 0),isnormallyused.

Waveguidesarelowlossshieldedpipesusedtotransportelectromagneticwaves
betweenantennasandreceiversorbetweensectionsofareceiver.Thesimplest
waveguideisahollowrectangulartubewithconductingwalls(Figure3.18,top)
separatedbydistanceainthehorizontal(x)directionandb a / 2inthevertical(y)
direction.Attheconductingwalls,theparallelcomponentofanyelectricfieldinsidethe
waveguidemustbezero.ThreepermitteddistributionsoftheelectricfieldstrengthE
alongthehorizontalaxisareshownascurvesinthetoppanelofFigure3.18,whichis
similartoFigure2.16forstandingwavesinacavity.However,onlythelongest
wavelengthdominantmodewithn x = 1isnormallyused,andhigherordermodeswith
n x = 2, 3, aredeliberatelysuppressedbecausetheytraveldownthewaveguidewith
differentgroupvelocities.

ThebottompanelofFigure3.18presentstheplanviewofthedominantradiation
modetravelingthroughthewaveguidewithawavenormalinthedirectionofthelarge
arrowandawavenode(|E| = 0)indicatedbythedashedline.Itisanalogousto
Figure2.18showingwavesinaconductingcavity.Radiationofwavelengthtraveling
throughthewaveguideinthedirectionindicatedbythearrowmustsatisfytheboundary

http://www.cv.nrao.edu/~sransom/web/Ch3.html 43/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

conditionn x = 2a / x = (2a / )cos,wheren x = 1, 2, 3, (Equation2.66).Whenn x = 1,


x / 2 = a = ( / 2)cos.

Themaximumwavelength(cos = 1)thatcanpropagate( 0)inthewaveguideis


thecutoffwavelength

c = 2a,
(3.141)

andthecorrespondingminimumfrequency

c = c / c
(3.142)

iscalledthecutofffrequency.Waveguidesareextremelyeffectivehighpassfilters.

Thegroupvelocityofpropagationdownthewaveguideis

c 2 1 / 2
v g = csin = c(1 cos 2)
1/2
[ ( )]
=c 1

, (3.143)

whichvariesquiterapidlywithfrequencyasapproaches c(sinapproaches0).The
waveguidephasevelocityv p = c 2 / v g c,sotheguidewavelength

c 2 1 / 2
w =
c
[ ( )]
1
(3.144)

issomewhatgreaterthanthefreespacewavelength.

Tominimizedispersion(thevariationofv gwithfrequency),waveguidesarerarely
usedatfrequenciesbelow 1.25 c.Higherordermodeswithn x = 2, 3, have
frequencies > 2 c, 3 c, andpropagatewithdifferentgroupvelocities.Tosuppress
them,waveguidesarenotusedforfrequencies > 2 c,thecutofffrequencyofthen x = 2
mode.Practicalwaveguidesareusuallylimitedtofrequencies < 1.9 c.Therequirement
b a / 2ensuresthat / 2 > bforall < 2 cson y = 0,soonlythisTE10(Transverse
Electricfieldwithn x = 1, n y = 0)modecanpropagate.TheTE10modeelectricfieldis
verticallypolarizedanditsstrengthisindependentofy.

Thecombinationoftheseupperandlowerfrequencylimitsrestrictmostwaveguide
applicationstooctavebandwidths,andwaveguidesofdifferentsizescoverdifferent
octaves.Manyofthewaveguidebandnamesinusetodayoriginatedasdeliberately
confusingcodenamesforWorldWarIIradarbands.Theyandtheirfrequencyrangesare
listedinAppendixF.5.Forexample,thestandardXbandwaveguidehasinterior
http://www.cv.nrao.edu/~sransom/web/Ch3.html 44/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

dimensionsa = 0.9 inches 2.286 cm,b = 0.4 inches 1.016 cm.Itscutoffwavelength


is c = 2a 4.572 cmanditscutofffrequencyis c = c / c 6.557 GHz.Itsnominal
frequencyrangeextendsfrom1.25 c 8.2 GHzto1.9 c 12.4 GHz.Unfortunately,
thewaveguidebandnamesaresodeeplyembeddedinradioastronomyjargonthatradio
observerscannotavoidthemanymorethanopticalastronomerscanavoidmagnitudes.

Eachfeedandreceiveronaradiotelescopecoversonlyonewaveguideband,so
severalfeedsandreceiversareneededtospanthemuchwiderusefulfrequencyrangeof
thetelescopeitself.AttheVLA,thefrequencyrangefrom1to50GHziscoveredby
eightsetsoffeedsandreceiversineightwaveguidebands:L(12GHz),S(24GHz),
C(48GHz),X(812GHz),Ku(1218GHz),K(1826.5GHz),Ka(26.540GHz),
andQ(4050GHz).

3.5RadioTelescopes
Theradiobandistoowide(fivedecadesinwavelength)tobecoveredeffectivelybya
singletelescopedesign.Thesurfacebrightnessesandangularsizesofradiosourcesspan
anevenwiderrange,soacombinationofsingletelescopesandaperturesynthesis
interferometersareneededtodetectandimagethem.Itisnotpracticaltobuildasingle
radiotelescopethatisevenclosetooptimumforallofradioastronomy.

Theidealradiotelescopeshouldhavealargecollectingareatodetectfaintsources.
TheeffectivecollectingareaA e(, )ofanyantennaaveragedoveralldirections(, )is
(Equation3.41)

2
A e = , (3.145)
4

solargepeakcollectingareasimplyextremelydirectiveantennasatshortwavelengths.
Onlyatlongwavelengths( > 1m)isitfeasibletoconstructsensitiveantennasfrom
reasonablenumbersofsmall,nearlyisotropicelementssuchasdipoles.Janskys 15
mwireantenna(Figure1.7)isanarrayofphaseddipoles.Itproducesawidefanbeam
nearthehorizonbuthasalargecollectingareabecause 2issolarge.Directiveaperture
antennasareneededforadequatesensitivityathigherfrequencies.

Thesimplestapertureantennaisawaveguidehorn.Radiationincidentonthe
openingisguidedbyataperedwaveguide.Atthenarrowendofthetaperedhornisa
waveguidewithparallelwalls,andinsidethiswaveguideisaquarterwavegroundplane
verticalantennathatconvertstheelectromagneticwaveintoanelectricalcurrentthatis
senttothereceiverviaacable.

Hornantennaspickupverylittlegroundradiationbecause,unlikemostparaboloidal
dishes,theiraperturesarenotpartiallyblockedbyexternalfeedsandfeedsupport
structures,whichscattergroundradiationintothereceiver.Thisfreedomfromground
pickupallowedPenziasandWilson[80]toshowthatthezenithantennatemperatureof
theBellLabshorn(Figure3.19)was3.5Khigherat 4GHzthanexpectedthefirst
detectionofthecosmicmicrowavebackgroundradiation.

http://www.cv.nrao.edu/~sransom/web/Ch3.html 45/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

Figure3.19:ThehornantennaatBellLabs,Holmdel,NJusedbyPenziasandWilson
todiscoverthe3Kcosmicmicrowavebackgroundradiationin1965.Reprintedwith
permissionofAlcatelLucentUSAInc.

Theapertureofawaveguidehornisnotblockedbyanyfeedsupportstructure,soit
isalsoeasiertocalculatethegainofahornantennafromfirstprinciplesthantocalculate
thegainofapartiallyblockedreflectingantenna.Thussmallhornantennashavebeen
usedbyradioastronomerstomeasuretheabsolutefluxdensitiesofverystrongsources
suchasCasA.Radioastronomersobservingwithlargedishestypicallydonotmeasure
theabsolutefluxdensitiesofsources,onlytheirrelativefluxdensitiesbycomparison
withsecondarycalibrationsourceswhosefluxdensitiesrelativetothatofCasAare
knowninadvance.Thepainstakingprocessofmeasuringtheabsolutefluxdensitiesof
CasAandcomparingthemwiththefluxdensitiesofweakerpointsourcessuitablefor
calibratingobservationsmadewithlargeradiotelescopeswasdescribedindetailby
Baarsetal.[6].

http://www.cv.nrao.edu/~sransom/web/Ch3.html 46/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

Figure3.20:The140foot(43m)telescopeinGreenBank,WVisthelargesttelescope
withanequatorialmount.Imagecredit:NRAO/AUI/NSF.

Mostradiotelescopesusecircularparaboloidalreflectorstoobtainlargecollecting
areasandhighangularresolutionoverawidefrequencyrange.Becausethefeedisonthe
reflectoraxis,thefeedandlegssupportingitpartiallyblockthepathofradiationfalling
ontothereflector.Thisapertureblockagehasanumberofundesirableconsequences:

1.Theeffectivecollectingareaisreducedbecausesomeoftheincomingradiationis
blocked.
2.Thebeampatternisdegradedbyincreasedsidelobelevels.
3.Radiationfromthegroundthatisscatteredoffthefeedanditssupportstructure
increasesthesystemnoise.
4.RadiationfromtheSunandartificialsourcesofradiofrequencyinterference(RFI)
farfromthemainbeamwillbemixedwiththedesiredsignal.

Radiotelescopesaresolargethatparaboloidswithhighf / Dratiosareimpractical
typicallyf / D 0.4.Thusradiodishesarerelativelydeep,asshowninFigure3.20.
Anotherconsequenceofalowf / Dratioisatinyfieldofviewattheprimefocus.The
instantaneousimagingcapabilityofalargesingledishisseverelylimitedbythesmall
numberoffeedsthatcanfitintothetinyfocalcircle.

Nearlyallradiotelescopeshavealtazmountsconsistingofahorizontalazimuth
trackonwhichthetelescopeturnsinazimuth(theanglemeasuredclockwisefromnorth
inthehorizontalplane)andahorizontalelevationaxleaboutwhichthetelescopetipsin
altitudeorelevationangle(twonamesfortheangleabovethehorizon).The140foot
telescopeinGreenBankisuniqueamonglargeradiotelescopesinhavinganequatorial
mount(Figure3.20).Theadvantageofaequatorialmountistrackingsimplicitythe
declinationaxisisfixedandthehourangleaxisturnsataconstantratewhiletrackinga
distantcelestialsource.(Thehourangleistheanglepastthemeridian,measuredin
hours.Themeridianisthegreatcirclepassingthroughthenorthpole,southpole,and
http://www.cv.nrao.edu/~sransom/web/Ch3.html 47/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

zenith.)Incontrast,boththealtitudeandtheazimuthofacelestialsourcechange
nonlinearlywithtime.Whenthe140foottelescopewasbeingdesigned,theabilityof
computerstoperformtherealtimecalculationsneededforanaltaztelescopetotracka
sourceaccuratelywasindoubt.Thedisadvantageofaequatorialmountismechanical
theslopedhourangleyokeandpolaraxlewithitshugetailbearingareverydifficultto
buildandsupport.

Figure3.21:Crosssectionofaradiotelescoperotationallysymmetricaroundthezaxis
andhavingaCassegrainsubreflector.Parallelraysfromadistantradiosourceare
reflectedbyacircularparaboloidwhoseprimefocusisatthepointmarkedf 1.The
convexCassegrainsubreflectorisacircularhyperboloidlocatedbelowtheprime
focus.Itreflectstheseraystothefeedlocatedatthesecondaryfocusf 2justabovethe
vertexoftheparaboloid.Theangle2 1subtendedbythemainreflectorviewedfrom
theprimefocusismuchlargerthantheangle2 2subtendedbythesubreflectorviewed
fromthesecondaryfocus,soCassegrainfeedshavetobemuchlargerthanprimary
feeds.

Figure3.20clearlyshowstheCassegrainopticalsystemofthe140foottelescope.
Radiationreflectedfromthemaindishisreflectedasecondtimefromtheconvex
Cassegrainsubreflectorlocatedjustbelowthefocalpointdowntofeedhornsand
receiversnearthevertexoftheparaboloid.Asubreflectorsystemhassomeadvantages
overaprimefocussystem:

1.Themagnifyingsubreflectorcanmultiplytheeffectivef / Dratiovaluesof
f / D 2aretypical.Thisgreatlyincreasesthesizeofthefocalellipsoid.Multiple
feedscanbelocatedwithinthefocalellipsoidtoproducemultiplesimultaneous
beamsforfasterimaging.
2.Thesubreflectorismanywavelengthsindiametersoitcanbeusedtotailorthe
illuminationtapertooptimizethetradeoffbetweenhighapertureefficiencyand
lowsidelobes.
3.Receiverscanbelocatednearthevertex,notthefocalpoint,wheretheyareeasier
toaccess.
4.Feedspilloverradiationisdirectedtowardthecoldskyinsteadofthewarm
ground,loweringoverallsystemtemperatures.
5.Thesubreflectorcanbenutated(rockedbackandforth)rapidlytoswitchthe
beambetweentwoadjacentpositionsonthesky.Suchdifferentialobservationsin
timeandspacecanbeusedtoremovereceiverbaselinedriftintimeandlargescale
spatialfluctuationsofatmosphericnoise.
http://www.cv.nrao.edu/~sransom/web/Ch3.html 48/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

6.Thesubreflectorcanbetiltedtoselectoneofseveralfeedsatthesecondaryfocus,
sothattheobservingfrequencybandcanbechangedrapidly.

Asubreflectorsystemhassomedisadvantages:

1.Relativelylargefeedsarerequiredtoproducethenarrowbeamsneededto
illuminatethesubreflector,whichtypicallysubtendsonlyasmallangleasviewed
fromthevertex.
2.Standingwavesintheleakycavityformedbythereflectorandsubreflectorcause
sinusoidalrippleswithfrequencyperiod c / (2f)intheobservedspectraof
strongcontinuumradiosources.Theseripplescanbeminimizedbyalternately
defocusingthesubreflectorradiallyby / 8andaveragingthedatafromboth
subreflectorpositions.
3.ACassegrainsubreflectorblockstheprimefocusposition,soprimefocusfeeds
cannotbeusedwhentheCassegrainsubreflectorisinposition.

ThegeometryofasymmetricalradiotelescopewithaCassegrainsubreflectoris
showninFigure3.21.Theparaboloidalshapeoftheprimaryreflectorwasdeterminedby
therequirementthatallincomingraysparalleltothezaxistravelthesamedistanceto
reachtheprimefocusatf 1.Likewise,thesecondaryreflectorshapeisdeterminedbythe
requirementthattheseraystravelthesamedistancetoreachthesecondaryfocusatf 2.
Forasubreflectorlocatedbelowtheprimefocus,therequiredshapeisahyperboloid
whosemajoraxiscoincideswiththemajoraxisoftheparaboloid.Theequation

z2 r2
2 =1
a2 b (3.146)

witha > bdefinessuchahyperboloid.Fromanypointonthehyperboloid,thedifference


betweenthedistancetof 2andthedistancetof 1is2a.Thedistancebetweenthefociis
1/2
2(a 2 + b 2) .Thetwofreeparametersaandbcanbeadjustedtosetboththediameter
ofthesubreflectorasneededtointerceptraysfromtheedgeoftheprimaryandtheheight
ofthesecondaryfocusonthezaxis.Themagnificationprovidedbythesubreflectoris

tan( 1 / 2)
M= , (3.147)
tan( 2 / 2)

where 1isthehalfanglesubtendedbytheprimaryviewedfromf 1and 2isthehalf


anglesubtendedbythesecondaryviewedfromf 2.Asmallsubreflectorislight,easyto
tilt,andreducesstandingwaves,butitsubtendsasmallangle2 2atf 2soafeedhorn
severalwavelengthsindiameterisrequiredtoilluminateitproperly.

TheParkes210foot(sincerenamedto64m)telescope(Figure3.22)inAustralia
wasbuiltaboutthesametimeasthe140foottelescope,butitsaltazmountandcentrally
concentratedreflectorbackupstructurepointedthewaytothedesignofmodernradio
telescopes.

http://www.cv.nrao.edu/~sransom/web/Ch3.html 49/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

Figure3.22:TheParkes64mtelescope.PhotoShaunAmy.

Elevationdependentgravitationaldeformationsdegradetheshortwavelength
performanceoftiltingreflectors.Thedeformationscanbecontrolledbydesigningthe
backupstructuresothatthedeformedsurfaceremainsparaboloidal.Thedeformations
causethefocalpointtoshiftslightlyinelevation,butthisshiftcanbeaccommodatedby
movingthefeedslightlytotrackthefocus.Thefirstlargehomologoustelescope
deliberatelydesignedtodeformthiswayisthe100mtelescope(Figure3.23)oftheMax
PlanckInstitutfrRadioastronomie(MPIfR)nearEffelsberg,Germany.Despiteitshuge
size,itspassivesurfaceremainsaccurateenoughtoworkatwavelengthsasshortas = 7
mmoverarangeofelevations.

http://www.cv.nrao.edu/~sransom/web/Ch3.html 50/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

Figure3.23:The100mtelescopenearEffelsberg,Germany.Thefirstdeliberately
homologoustelescope,itworksto 7mm.NotethelargeGregoriansubreflector
abovetheprimefocus.PhotobyMatthiasKadler.

The100mtelescopehasaconcaveGregoriansubreflectorabovetheprimefocus.
ThegeometryofasymmetricGregoriansystemisshowninFigure3.24.Aswiththe
Cassegrainsubreflector,theGregorianreflectorshapeisdeterminedbytherequirement
thatallparallelaxialraystravelthesamedistancetoreachthesecondaryfocusatf 2.For
asubreflectorlocatedabovetheprimefocus,therequiredshapeisanellipsoidwhose
majoraxiscoincideswiththemajoraxisoftheparaboloid.Theequation

z2 r2
+ =1 (3.148)
a2 b2

witha > bdefinessuchanellipsoid.Fromanypointontheellipsoid,thesumofthe


1/2
distancetof 2andthedistancetof 1is2a.Thedistancebetweenthefociis2(a 2 b 2) .

Figure3.24:Crosssectionofaradiotelescoperotationallysymmetricaroundthezaxis
andhavingaGregoriansubreflector.Parallelraysfromadistantradiosourceare
reflectedbythecircularparaboloidwhoseprimefocusisatthepointmarkedf 1.The
Gregoriansubreflectorisacircularellipsoidlocatedabovetheprimefocus.Itreflects
theseraystothefeedlocatedatthesecondaryfocusf 2justabovethevertexofthe
paraboloid.

TheAreciboradiotelescope(Figures8.2and3.25)wasoriginallydesignedasa
radarfacilitytostudytheionosphereviaThomsonscatteringof430MHz( = 70cm)
radiowavesbyfreeelectrons.Thermalmotionsoftrulyfreeelectronswouldgreatly
Dopplerbroadenthebandwidthoftheradarechoandlowerthereceivedsignaltonoise
ratio,soaverylargeantennawasbuiltforsensitivity.However,ionosphericelectronsare
coupledtothemuchheavierionsonscaleslargerthantheionosphericDebyelength,
whichisonlyafewmm.Thisismuchsmallerthanthe70cmwavelength,sotheactual
bandwidthisdeterminedbythermalmotionsofthemuchheavierionsandislowerby
twoordersofmagnitude.Thusafarsmallerdishwouldhavesufficed!Astronomershave
benefitedfromthisoversightanduseAreciboshugecollectingareaatfrequenciesupto
about10GHzforSolarSystemradar(planets,moons,asteroids),pulsarstudies,HI21
cmlineobservationsofgalaxies,andotherobservationsthatneedhighsensitivity.

http://www.cv.nrao.edu/~sransom/web/Ch3.html 51/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

Figure3.25:TheArecibofeedsupportplatformcansteerthebeamanywhereupto20
degreesfromthezenitheventhoughthesphericalreflectorisfixed.Thecurved
azimutharmrotatesabouttheverticalunderacircularringatthebaseofthefixed
triangularstructure.Thecarriagehouseundertheleftsideoftheazimutharmcarriesa
waveguidelinefeedthatcorrectsforsphericalaberration.Thedomeunderthecarriage
houseontherightsidecontainstheGregoriansecondarymirrorandtertiarycorrecting
mirror,illuminatedbywaveguidehornfeeds.Thecarriagehousescanmovealong
tracksatthebottomoftheazimutharmtochangethezenithangleofthebeam.

Thesphericalreflectorcanbeverylargebecauseitisdoesnotmove.Asphereis
symmetricaboutanyaxispassingthroughitscenter,sotheArecibobeamcanbesteered
bymovingthefeedinsteadofthereflector.ThecurvedfeedsupportarmvisibleinFigure
3.25is300feetlongandrotatesinazimuthbelowthefixedtriangularstructure.The
feedsaremountedundertwocarriagehousesthatmovealongtracksonthebottomofthe
feedarmandpermittrackingatzenithanglesupto20degrees.Thefeedillumination
spillsovertheedgeofthefixedreflectorathighzenithangles,soalargegroundscreen
surroundsthesphericalreflectortoreflectthespilloverontothecoldskyandkeepit
awayfromthewarmandnoisyground.

Asphericalreflectorfocusesadistantpointsourceontoaradiallinesegment,soa
radiallinefeed(seeFigure3.25)upto96feetlongisneededtoilluminatetheentire
apertureefficientlyfromtheprimefocus.Thelinefeedisaslottedwaveguidetaperedto
controlthegroupvelocity(Equation3.143)andphaseupradiationarrivingfromallover
thereflector.However,longslottedwaveguidelinefeedsareinherentlynarrowband,and
ohmiclossesinthelongslottedwaveguideincreasethesystemtemperaturesignificantly
atshortwavelengths.ThegolfballunderthefeedarmatArecibo(Figure3.25)houses
anenormousGregoriansubreflectorandatertiaryreflectorthatallowlownoise
widebandpointfeedstoilluminateanellipseabout200mby225minsizeonthemain
reflector.

http://www.cv.nrao.edu/~sransom/web/Ch3.html 52/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

Figure3.26:VerticalcrosssectionshowingthesymmetryplaneoftheGBT.Theactual
dishshownbythecontinuouscurveisanasymmetricsectionofthesymmetricparent
paraboloid(dottedcurve)whosediameteris208m.TheinneredgeoftheGBT
reflectoris4mtotherightofthezaxisofsymmetrysothefociandfeedsupport
structuretotheleftofthezaxisneverblocktheincomingradiation.Theprimaryfocal
lengthisf 1=60m,andthedistancefromf 1tothesecondaryfocusf 2is11m.The
secondaryfocusisoffsetby1.068mfromthesymmetryaxistominimize
instrumentalpolarization.ThediameteroftheGregoriansubreflectoris8m.The
secondaryfocusisfarabovethevertexoftheparentparaboloid,buttheoffaxisfeed
supportarmoftheGBTisstrongenoughtosupportalargefeed/receivercabin
(Figure3.27)atthisheight.

The100mRobertC.ByrdGreenBankTelescope(GBT)(Figure8.1)isthe
successortothecollapsed300foottelescopeinGreenBank,anditincorporatesanumber
ofnewdesignfeaturestooptimizeitssensitivityandshortwavelengthperformance.

Theactualreflectorisa110m 100moffaxissectionofanimaginarysymmetric
paraboloid208mindiameter.Projectedontoaplanenormaltothebeam,itisa100m
diametercircle.Becausetheprojectededgeoftheactualreflectoris4mawayfromthe
axisofthe208mparaboloid,thefocalpointdoesnotblocktheaperture.TheGBT
enjoysthesameclearaperturebenefitsofwaveguidehornsaverycleanbeamandlow
spillovernoisebutismuchlargerthananypracticalhornantenna.Thecleanbeamis
especiallyvaluableforsuppressingradiofrequencyinterference(RFI)andstrayradiation
fromveryextendedsources,suchasHIemissionfromtheGalaxy.

http://www.cv.nrao.edu/~sransom/web/Ch3.html 53/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

Figure3.27:TheconcaveGregoriansubreflectorjustabovetheprimefocusofthe
GBTimagessourcesontoconicalhornfeedsextendingthroughthetopofthe
rectangularreceivercabin.Theprimefocusfeedarmisshownstowedoutoftheway
ofthesubreflector.Noneoftheseoffsetstructuresblockradiationreflectedfromthe
mainaperture.Imagecredit:NRAO/AUI/NSF.

TheverticalcrosssectionoftheGBTplottedinFigure3.26showshowtheoffset
Gregoriansubreflectordoesnotblockanyradiationfallingontotheprimaryreflector.
TheGregoriansubreflectorisabovetheprimefocusatf 1,soprimefocusoperationis
possiblebyraisingaswingingboomcarryingtheprimefocusfeedsintopositionbelow
thesubreflector,althoughthistemporarilyblockstheGregoriansubreflector.Thehuge
feedsupportarmisover60mlong,thefocallengthofthe208mparaboloid.Thefeed
supportarmhasamuchlargercrosssectionthanthefeedsupportstructuresof
symmetricaltelescopes,whichmustbekeptasthinaspossibletominimizeblockage.
ThisGBTarmisverystrongandcansupportheavysubreflectors,feeds,equipment
rooms,andanelevator.Atthetopofthearmandabovetheprimefocusistheconcave
Gregoriansubreflector.Thissubreflectorilluminatesfeedsemergingthroughtheroofofa
largereceivercabinattachedtothefeedarmashortdistancebelow(Figure3.27).
Becausethesefeedsarerelativelyclosetothesubreflector,evenamoderatelysmall
subreflectorsubtendsalargeangleasviewedfromthefeeds,whichcanthenbe
moderatelysmallthemselves.Mostofthereceiversandfeedsneededtocoverthe
frequencyrange1 < (GHz) < 100canfitintothereceivercabinsimultaneouslyandare
availableforuseonshortnotice.

Themainreflectorissupportedbyabackupstructurethatdeformshomologouslyto
ensuregoodefficiencyatwavelengthsasshortas = 2cm.Theactivereflectingsurface
consistsofapproximatelytwothousandpanels,eachabout2monaside.Thecornersof
individualpanelsaremountedoncomputercontrolledactuatorsthatcanmovethepanels
upordownasneededtocontinuouslycorrecttheoverallshapeofthesurface.
Photogrammetrywasusedtomeasurethesurfaceattheriggingelevation(theelevationat
http://www.cv.nrao.edu/~sransom/web/Ch3.html 54/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

whichthesurfacewasoriginallyset).Thegravitationaldeformationsatotherelevation
anglespredictedbythefiniteelementcomputermodeloftheGBTarecontinuously
removedbytheactuatorsasthetelescopemoves.Asaresult,thermssurfaceerrorisonly
0.2 mmandtheGBThasahighsurfaceefficiencyatwavelengthsasshortas 3
mm.

The30mIRAM(InstitutdeRadioastronomieMillimtrique)telescope
(Figure3.28)isthelargesttelescopeoperatingat3,2,1,and0.8mm.Itsrmssurface
errorisonly55 m,anditspointingaccuracyisabout1arcsec.

Figure3.28:The30mIRAMtelescopeonPicoVeletainSpain.Imagecredit:IRAM.

3.6Radiometers
Naturalradioemissionfromthecosmicmicrowavebackground,discreteastronomical
sources,theEarthsatmosphere,andthegroundisrandombroadbandnoisethatisnearly
indistinguishablefromthenoisegeneratedbyawarmresistor(Section2.5)orbyreceiver
electronics.Aradioreceiverusedtomeasuretheaveragepowerofthenoisecomingfrom
aradiotelescopeinawelldefinedfrequencyrangeiscalledaradiometer.Thenoise
voltagehasaGaussianamplitudedistributionwithzeromean,anditfluctuatesonthe
veryshorttimescales(nanoseconds)comparablewiththeinverseoftheradiometer
bandwidth.Asquarelawdetectorintheradiometersquarestheinputnoisevoltage
toproduceanoutputvoltageproportionaltotheinputnoisepower.Noisepowerisalways
greaterthanzero,andthenoisefrommostastronomicalsourcesisstationary,mean
powerissteadywhenaveragedovermuchlongertimescales(secondstohours).The
NyquistShannonsamplingtheorem(AppendixA.3)statesthatanyfunctionhaving
finitebandwidthanddurationcanberepresentedby2 independentsamples
spacedintimeby(2 ) 1.ByaveragingalargenumberN = (2 )ofindependent
noisesamples,anidealradiometercandeterminetheaveragenoisepowerwitha
fractionaluncertaintyassmallas(N / 2) 1 / 2 = ( ) 1 / 2 1anddetectfaintsources
thatincreasetheantennatemperaturebyonlyatinyfractionofthetotalnoisepower.The
idealradiometerequationexpressesthisresultintermsoftheradiometerbandwidthand
theaveragingtime.Gainvariationsinpracticalradiometers,fluctuationsinatmospheric
emission,andconfusionbyunresolvedradiosourcesmaysignificantlydegradetheactual
sensitivitycomparedwiththatpredictedbytheidealradiometerequation.

3.6.1BandLimitedNoise
http://www.cv.nrao.edu/~sransom/web/Ch3.html 55/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

Figure3.29:TheoutputvoltageVofaradiotelescopevariesrapidlyonshort
timescales,asindicatedbytheupperplotshowing100independentsamplesofband
limitednoisedrawnfromaGaussianprobabilitydistributionP(V / V rms)(lowerplot)
havingzeromeanandfixedrmsV rms.SeeAppendixB.5foramathematical
descriptionoftheGaussiandistribution.

Thevoltageattheoutputofaradiotelescopeisthesumofnoisevoltagesfrommany
independentrandomcontributions.Thecentrallimittheorem[15]statesthatthe
amplitudedistributionofsuchnoiseisnearlyGaussian.Figure3.29(lowerpanel)shows
thehistogramofabout20,000independentvoltagesamplesrandomlydrawnfroma
GaussianparentdistributionhavingrmsV rmsandmeanV = 0.Figure3.29(upper
panel)showsN = 100successivesamplesdrawnfromtheGaussiannoisedistribution.
Thissequenceofvoltagesisrepresentativeofbandlimitednoiseinthefrequencyrange
from0toduringatimeintervalsuchthat( ) = N / 2 = 50,e.g.,noisewithall
frequenciesupto = 1MHzsampledevery(2) 1 = 0.5 sfor = 50 s.Thisiswhat
thebandlimitednoiseoutputvoltageofaradiotelescopelookslike.

Itisconvenienttodescribenoisepowerinunitsoftemperature.Thenoisepowerper
unitbandwidthgeneratedbyaresistoroftemperatureTisP = kTinthelowfrequency
limit,sowecandefinethenoisetemperatureofanynoiselikesourceintermsofits
powerperunitbandwidthP :

P
TN , (3.149)
k

wherek 1.38 10 23jouleK 1isBoltzmannsconstant.

Thetemperatureequivalenttothetotalnoisepowerfromallsourcesreferencedto
theinputofaradiometerconnectedtotheoutputofaradiotelescopeiscalledthesystem
noisetemperatureT s.Itisthesumofmanycontributorstotheantennatemperatureplus

http://www.cv.nrao.edu/~sransom/web/Ch3.html 56/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

theradiometernoisetemperatureT r:

T s = T cmb + T rsb + T source + [1 exp( A)]T atm + T spill + T r + .


(3.150)

TherearesevenantennatemperaturecontributionslistedexplicitlyinEquation3.150:

1.T cmb 2.73Kisfromthenearlyisotropiccosmicmicrowavebackground.


2.T rsbistheaverageskybrightnesstemperaturecontributedbyallbackground
radiosources.Extragalacticsourcesadd[31]

( ) ( )
T rsb 2.7
(3.151)
0.1 K 1.4 GHz

inalldirections,andtheGalacticplaneisabrightdiffusesourceatlow(
< 0.5 GHz)frequencies[43].
3.T sourceisfromtheastronomicalsourcebeingobserved,writtenwithato
emphasizethatitisusuallymuchsmallerthanthetotalsystemnoise:
T source T s.Forexample,inthe RF 4.85GHzskysurveymadewiththe
300foottelescope,thesystemnoisewasT s 60K,butthefaintestdetected
sourcesaddedonlyT source 0.01K.
4.[1 exp( A)]T atmisthebrightnessofatmosphericemissioninthetelescopebeam
(Section2.2.3).
5.T spillaccountsforspilloverradiationthatthefeedpicksupindirectionsbeyond
theedgeofthereflector,primarilyfromtheground.
6.T ristheradiometernoisetemperatureattributabletonoisegeneratedbythe
radiometeritself,referencedtotheradiometerinput.Allradiometersgenerate
noise,andanyradiometercanberepresentedbyanequivalentcircuitconsistingof
anoiselessradiometerwhoseinputisconnectedtoaresistoroftemperatureT r.
Radiometernoiseisusuallyminimizedbycoolingtheradiometertocryogenic
temperatures.However,radiometersarenotjustmatchedresistors,soT rmaybe
eitherlowerorhigherthanthephysicaltemperatureoftheradiometeritself.
7.representsanyothernoisesourcesthatmightbeimportant.Anexampleis
emissionresultingfromohmiclossesinthelongslottedwaveguidefeedatArecibo
(Figure3.25).

3.6.2Radiometers
Thepurposeofthesimplesttotalpowerradiometeristomeasurethetimeaveraged
poweroftheinputnoiseinsomewelldefinedradiofrequency(RF)range


RF to RF + , (3.152)
2 2

whereisthereceiverbandwidth.Forexample,thereceiversusedonthe300foot

http://www.cv.nrao.edu/~sransom/web/Ch3.html 57/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

telescopetomakethe 6cmcontinuumsurveyofthenorthernskyhadacenterradio
frequency RF 4.85 10 9Hzandabandwidth 6 10 8Hz.

Thesimplestradiometer(Figure3.30)consistsoffourstagesinseries:(1)alowloss
bandpassfilterthatpassesinputnoiseonlyinthedesiredfrequencyrange(2)asquare
lawdetectorwhoseoutputvoltageV oisproportionaltothesquareofitsinputvoltage
thatis,V oisproportionaltoitsinputpower(3)asignalaveragerorintegratorthat
smoothstherapidlyfluctuatingdetectoroutputand(4)avoltmeterorotherdeviceto
measureandrecordthesmoothedvoltage.

Figure3.30:Thesimplestradiometerfiltersthebroadbandnoisecomingfromthe
telescope,multipliesthefilteredvoltagebyitself(squarelawdetection),smoothsthe
detectedvoltage,andmeasuresthesmoothedvoltage.Thefunctionofthedetectoristo
convertthenoisevoltage,whichhaszeromean,tonoisepower,whichisproportional
tothesquareofvoltage.

Afterpassingthroughaninputfilterofwidth < RF,thenoisevoltageisno


longercompletelyrandomitlooksmorelikeasinewaveoffrequency RFwhose
amplitudeenvelope(dashedcurveinFigure3.31)variesrandomlyontimescales
1
t () 1 > RF.Thepositiveandnegativeenvelopesaresimilarsolongas RF
.

Figure3.31:ThevoltageoutputV(t)ofthefilterwithcenterfrequency RFand
bandwidth < RFisasinusoidwithfrequency RFwhoseenvelope(dashedcurves)
fluctuatesontimescales() 1 > ( RF) 1.

ThefilteredoutputissenttoasquarelawdetectorwhoseoutputvoltageV ois
proportionaltoitsinputpower.Foranarrowband(quasisinusoidal)inputvoltage
V i cos(2 RFt)atfrequency RF,thedetectoroutputvoltagewouldbe
V o cos 2(2 RFt).Thiscanberewrittenas[1 + cos(4 RFt)] / 2,afunctionwhosemean

http://www.cv.nrao.edu/~sransom/web/Ch3.html 58/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

valueisproportionaltotheaveragepoweroftheinputsignal.InadditiontotheDC(zero
frequency)componentthereisanoscillatingcomponentattwicetheinputfrequency RF.
Thedetectoroutputspectrumforafinitebandwidthandatypicalwaveformisshown
inFigure3.32.

Figure3.32:TheoutputvoltageV oofasquarelawdetector(Figure3.33)is
proportionaltothesquareoftheinputvoltage.Itisalwayspositive,soitsmean(DC,
orzerofrequencycomponent)ispositiveandproportionaltotheinputpower.Thehigh
frequency( 2 RF)fluctuationsaddnoinformationaboutthesourceandarefiltered
outinthenextstage.

Theoscillationsundertheenvelopeapproachzeroeveryt (2 RF) 1.Thusthe


oscillatingcomponentofthedetectoroutputiscenterednearthefrequency2 RF.The
detectoroutputalsohasfrequencycomponentsnearzero(DC)becausethemeanoutput
voltageisgreaterthanzero.

Figure3.33:TheupperplotshowstheoutputvoltageV oofasquarelawdetector
whoseinputistheGaussiannoiseshownbytheupperplotinFigure3.29.Theoutput
voltagehistogram(lowerplot)ispeakedsharplynearzeroandhasalongpositivetail.
ThemeandetectedvoltageV oequalsthemeansquaredinputvoltage,andthermsof
thedetectedvoltagedistributionis2 1 / 2V o.Forafullderivationofthedetectoroutput
distributionanditsrms,seeAppendixB.6.

http://www.cv.nrao.edu/~sransom/web/Ch3.html 59/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

Boththerapidlyvaryingcomponentatfrequenciesnear2 RFanditsenvelopevary
ontimescalesthatarenormallymuchshorterthanthetimescalesonwhichtheaverage
signalpowerTvaries.Theunwantedrapidvariationscanbesuppressedbytakingthe
arithmeticmeanofthedetectedenvelopeoversometimescale () 1byintegrating
oraveragingthedetectoroutput.Thisintegrationmightbedoneelectronicallyby
smoothingwithanRC(resistancepluscapacitance)filterornumericallybysamplingand
digitizingthedetectoroutputvoltageandthencomputingitsrunningmean.

Integrationgreatlyreducesthereceiveroutputfluctuations.Inthetimeinterval
thereareN = (2 )independentsamplesofthetotalnoisepowerT s,eachofwhichhas
anrmserror T 2 1 / 2T s.ThermserrorintheaverageofN 1independentsamplesis
reducedbythefactorN,sothermsreceiveroutputfluctuation T(seeAppendixB.6for
aformalderivationofthisresult)isonly

2 1 / 2T s
T = . (3.153)
N1 / 2

Intermsofbandwidthandintegrationtime,

Ts
T
(3.154)

aftersmoothing.Thecentrallimittheoremofstatisticsimpliesthatheavilysmoothed(
1)outputvoltagesalsohaveanearlyGaussianamplitudedistribution.This
importantequationiscalledtheidealradiometerequationforatotalpowerreceiver.
TheweakestdetectablesignalsTonlyhavetobeseveral(typicallyfive)timesthe
outputrms Tgivenbytheradiometerequation,notseveraltimesthetotalsystemnoise
T s.Theproduct( )maybequitelargeinpractice(10 8isnotunusual),sosignalsas
faintasT 5 10 4T swouldbedetectable.Figures3.34and3.35illustratetheeffects
ofsmoothingthedetectoroutputbytakingrunningmeansoflengthsN = 50andN = 200
samples.

http://www.cv.nrao.edu/~sransom/web/Ch3.html 60/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

Figure3.34:Thesmoothedoutputvoltagefromtheintegratorvariesontimescale
withsmallamplitude Tgivenbytheidealradiometerequation.Thetoppartofthis
figureshowsthedetectedvoltagesmoothedbyanN = 50samplerunningmean,and
thebottompartshowstheamplitudedistributionofthesmoothedvoltage.This
amplitudedistributionhasmeanV oandrms(2 / N) 1 / 2V o = V o / 5.AsNgrows,
thesmoothedamplitudedistributionapproachesaGaussian.Thesamplingtheorem
(AppendixA.3)statesthatN = (2 )so( ) = 25forthisexample.

Figure3.35:WhenthesamedetectoroutputissmoothedoverN = 200samplesinstead
ofN = 50samples(Figure3.34),themeanremainsthesamebutthermsfallsbya
factorof4 1 / 2 = 2toV o / 10.Inthisexample( ) = 100.

3.6.3SomeCaveats
Theidealradiometerequationsuggeststhatthesensitivityofaradioobservation

improvesas 1/2
forever.Inpractice,systematicerrorssetafloortothenoiselevelthat
http://www.cv.nrao.edu/~sransom/web/Ch3.html 61/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

improvesas 1 / 2forever.Inpractice,systematicerrorssetafloortothenoiselevelthat
canbereached.Receivergainchanges,erraticfluctuationsinatmosphericemission,or
confusionbytheunresolvedbackgroundofcontinuumradiosourcesusuallylimitthe
sensitivityofsingledishcontinuumobservations.

ReceiverGainandAtmosphericFluctuations

Radiometerscontainaseriesofamplifiersthatmultiplytheweakinputpowers
P in = kT s 10 14 Wtomilliwattlevels.Theoutputvoltageofatotalpowerreceiver
isdirectlyproportionaltotheoverallpowergainGofthereceiver.IfGisntperfectly
constant,thechangeinoutputvoltagecausedbyagainfluctuationGinapractical
radiometerproducesafalsesignalwhoseapparenttemperature

G = Ts ( )
G
G (3.155)

isindistinguishablefromacomparablechange Tcausedbynoiseinanidealradiometer.
Receivergainfluctuationsandnoisefluctuationsareindependentrandomprocesses,so
theirvariances(thevarianceisthesquareoftherms)add,andthetotalreceiveroutput
fluctuationbecomes
2 2 2
T = noise + G
(3.156)
2
= Ts [
1

+
G 2
G
. ( )] (3.157)

Thepracticaltotalpowerradiometerequationisthus

T Ts [

+
1
( )]G 2 1 / 2
G
. (3.158)

Clearly,radiometergainfluctuationswilldegradethesensitivityofanobservationunless

( )
G
G


1
.
(3.159)

Forexample,the5GHzreceiverusedtomaketheskysurveywiththe300foottelescope
had 6 10 8Hzand 0.1s,sothefractionalgainfluctuationsontimescalesupto
afewseconds(thetimetoscanonebaselinelength)hadtosatisfy

G 1
= 1.3 10 4.
G 10 8 (3.160)
6 Hz 0.1 s

http://www.cv.nrao.edu/~sransom/web/Ch3.html 62/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

Thisisdifficulttoachieveinpractice.Gainfluctuationstypicallyhave1 / fpower
spectra,wherefisthepostdetectionfrequency,sotheyarelargeronlongertimescalesand
increasingeventuallyresultsinahigheroutputnoiselevel.Thegainstabilityofa
receiverisoftenspecifiedbythe1 / fkneefrequencyf k,thepostdetectionfrequencyat
which noise = G.Integrationslongerthan 1 / (2f k)willlikelyincreasethereceiver
outputfluctuations.Dependingonthestabilityandbandwidthoftheradiometer,
1 Hz < f k < 1 kHz.

Figure3.36:Blockdiagramofabeamswitchingdifferentialradiometer.Thetotal
powerreceiverisswitchedbetweentwofeeds,onepointingatthesourceandone
displacedbyafewbeamwidthstoavoidthesourcebutmeasureemissionfromnearly
thesamesampleofatmosphere.Theoutputofthetotalpowerreceiverismultipliedby
+ 1whenthereceiverisconnectedtotheonsourcefeedandby 1whenitisconnected
tothereferencefeed.Fluctuationsinatmosphericemissionandinreceivergainare
effectivelysuppressedforfrequenciesbelowtheswitchingrate,whichistypicallyin
therange10to1000Hz.

Fluctuationsinatmosphericemissionalsoaddtothenoiseintheoutputofasimple
totalpowerreceiver.Watervaporisthemainculpritbecauseitisnotwellmixedinthe
atmosphere,andnoisefromwatervaporfluctuationscanbeasignificantproblemat
frequenciesof 5GHzandup.

Onewaytominimizetheeffectsoffluctuationsinbothreceivergainand
atmosphericemissionistomakeadifferentialmeasurementbycomparingsignalsfrom
twoadjacentfeeds.Themethodofswitchingrapidlybetweenbeamsorloadsiscalled
DickeswitchingafterRobertDicke,itsinventor.Figure3.36showstheblockdiagramof
abeamswitchingDickeradiometer.IfthesystemtemperaturesareT 1andT 2inthetwo
positionsoftheswitch,thenthereceiveroutputisproportionaltoT 1 T 2 T 1andthe
effectofgainfluctuationsisonly

G G
G (T 1 T 2) T1 . (3.161)
G G

Likewise,theatmosphericemissionintwonearlyoverlappingbeamsthroughthe
troposphereisnearlythesame,somostofthetroposphericfluctuationscancelout.The
maindrawbackwithDickeswitchingisthatthereceiveroutputfluctuations,relativeto
thesourcesignalinasinglebeam,aredoubledbecausethesourcesignalisbeing
receivedonlyhalfthetimewhilethenoisepowerispresentallthetime.Theideal
radiometerequationforaDickeswitchingreceiveris
http://www.cv.nrao.edu/~sransom/web/Ch3.html 63/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

2T s
T = .
(3.162)

Confusion

Figure3.37:Aprofileplotcovering45deg 2ofskyimagedbythe300foottelescopeat
1.4GHzwith = 12arcminresolution[25].Theubiquitousfluctuationswithrms
20 mJy beam 1arecausedbythesuperpositionofnumerousfaintsources,not
receivernoise.

Figure3.38:Thecontourimage[
http://www.cv.nrao.edu/~sransom/web/Ch3.html ]isa4deg 2subsetoftheareashownin 64/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

Figure3.38:Thecontourimage[25]isa4deg 2subsetoftheareashownin
Figure3.37.Thecontoursstartat45 mJy beam 1 2 candarespacedbyfactorsof
2 1 / 2,sosourceswithfewerthanfourcontoursarebelowthe5 cconfusionlimit.The
grayscaleplotisa1.4GHzVLAimagemadewith = 45arcsecresolution.Someof
thefaintsourcesseenbythe300foottelescopeareblendsoftwoormorefainter
sourcesresolvedbytheVLA.

Singledishradiotelescopeshavelargecollectingareasbutrelativelybroadbeamsat
longwavelengths.Nearlyalldiscretecontinuumsourcesareextragalacticandextremely
distant,sotheyaredistributedrandomlyandisotropicallyonthesky.Theskybrightness
fluctuationscausedbynumerousfaintsourcesineverytelescopebeamarecalled
confusion,andconfusionusuallylimitsthesensitivityofsingledishcontinuum
observationsatfrequenciesbelow 10 GHz.Figure3.37isaprofileplotofconfusion
fluctuationsinalowresolutionimage.Figure3.38showscontoursfromaportionofthat
lowresolutionimagesuperimposedonanoverlappinghighresolutiongrayscaleimage.

AlthoughtheamplitudedistributionofconfusionisdistinctlynonGaussian,the
rmsconfusion ccalculatedbyignoringthelongpositivetailisawidelyquoted.Atcm
wavelengths,thermsconfusioninaGaussiantelescopebeamwithFHWMis

{ ( ) ( )
0.7
2
0.2 ( > 0.17 arcmin),

( )
c GHz arcmin
(3.163)
mJy beam 1
2.2 ( ) ( )

GHz
0.7

arcmin
10 / 3 ( < 0.17 arcmin).

Individualsourcesfainterthantheconfusionlimit 5 ccannotbedetectedreliably,no
matterhowlowthereceivernoise.Mostcontinuumobservationsoffaintsourcesat
frequenciesbelow 10 GHzaremadewithinterferometersinsteadofsingledishes
becauseinterferometerscansynthesizemuchsmallerbeamwidthsandhencehave
significantlylowerconfusionlimits.

Confusionbysteadycontinuumsourceshasamuchsmallereffectonobservations
ofspectrallinesorrapidlyvaryingsourcessuchaspulsars.

3.6.4SuperheterodyneReceivers
Fewactualradiometersareassimpleasthosedescribedabove.Nearlyallpractical
radiometersaresuperheterodynereceivers(Figure3.39),inwhichtheRFamplifieris
followedbyamixerthatmultipliestheRFsignalbyasinewaveoffrequency LO
generatedbyalocaloscillator(LO).Theproductoftwosinewavescontainsthesumand
differencefrequencycomponents,

2sin(2 LOt)sin(2 RFt) = cos[2( LO RF)t] cos[2( LO + RF)t],


(3.164)

sothemixeractsasafrequencyshifter.Forexample,if LO = 12 GHzand RF = 9 GHz

http://www.cv.nrao.edu/~sransom/web/Ch3.html 65/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

,themixeroutputfrequency,calledtheintermediatefrequency(IF),willbe
LO RF = 3 GHz.

Theadvantagesofsuperheterodynereceiversinclude

1.shiftingthesignalstolowerfrequencies IF < RFwheretheyareeasierto


amplify,transmitoverlongdistances,filter,anddigitize
2.tunabilityoverawiderangeof RF
3.tuningbyadjustingonlythelocaloscillatorfrequencysothat
4.theIFamplifierandbackenddevicessuchasmultichannelfilterbanksordigital
spectrometerscanalloperateoverfixedfrequencyranges.

Figure3.39:Blockdiagramofasimplesuperheterodynereceiver.Onlythelocal
oscillatoristunedtochangetheobservingfrequencyrange.

3.6.5Spectrometers

Figure3.40:AnanalogfilterbanksplitsthebroadbandoutputofanIFamplifierintoN
contiguousfrequencychannelsofwidtheach.Ineffect,eachchannelisa
narrowbandIFamplifierwhoseoutputvoltageisdetected(multipliedbyitself),
smoothed,andrecorded.

Thesimplestsuperheterodyneradiometermeasuresthetotalpowerinitsnormally
broadIFpassbandofwidth.Aspectrometerisabackendthatdividesthatpassband
intoNadjacentnarrowfrequencyrangesofwidth / Nandsimultaneously
measuresthepowerinallNchannelstoquicklylocateandresolvespectralfeaturessuch
asatomicandmolecularlines(Section7.1).

http://www.cv.nrao.edu/~sransom/web/Ch3.html 66/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

Themoststraightforwardspectrometerisafilterbankofnarrowbandanalogfilters
connectedinparallelandwithcenterfrequenciesuniformlyspacedby(Figure3.40).
EachchannelactsasaseparateIFandhasitsowndetector.However,thechannelgains,
bandpasses,anddetectorresponsesofananalogfilterbankmustbeverycloselymatched
andstabletoyieldsmoothspectralbaselines,soanalogfilterbankswithmorethan
N 10 2channelsaredifficulttobuildandtune.Analogfilterbanksarealsoinflexible
becausetheirchannelbandwidthsandnumbersNcannotbechangedeasily.Flexible
spectrometerswithN 10 3orevenN 10 4frequencychannelsrequiredigitalsignal
processing(DSP)techniques.

Formanyyears,mostdigitalspectrometerswereautocorrelationspectrometers
usingtheWienerKhinchintheorem(EquationA.18)tocomputepowerspectrafrom
digitallysampledtimeseries(seeAppendixA.3)ofthebandlimitedIFoutput.A
sampledcopyofaportionoftheinputradiosignalisdelayedbyaseriesofprogressively
longertimedelays,thedelayedsignalsaremultipliedwiththeoriginalsignal,andtheir
productsareintegrated.Thisseriesofoperationsisanautocorrelation(AppendixA.7).
Ifthedigitalsamplescontainonlyoneortwobits(twoorthreelevels)each,
autocorrelationcanbeperformedinhardwareandofteninasinglechip,withrelatively
simpledigitallogic.Theseautocorrelationfunctions(ACFs)canbeintegratedtobuildup
signaltonoiseandthenfinallyconvertedintoapowerspectrumviaadiscreteFourier
transform(usuallyanFFTseeAppendixA.2)oftheACFviatheWienerKhinchin
theorem.Autocorrelationspectrometersallowtheintegrationofverydeep(i.e.,long
duration)spectrausingrelativelysimpledigitalhardwareandwithoutcomputingmany
costlyFFTsdirectlyonincomingNyquistsampleddataonlyoneFFTiscomputedat
theveryendoftheintegration.Similartechniques,butusingcrosscorrelationofthe
signalsfromdifferentantennas,areoftenusedtocalculatespectrafromradio
interferometers.

WiththecontinuingimprovementsinthespeedsandcapabilitiesofDSPsystems,
spectraareincreasinglybeingcomputeddirectlyviaFFTsofaNyquistsampledband.
TheFourieramplitudesaresquaredtomakepowerspectra,andthepowerspectraare
accumulatedfordeepspectralintegrations.SuchsystemsareknownasFourier
transformspectrometers,andtheFFTscanbecomputedinavarietyofways.Many
recentspectrometersuseFieldProgrammableGateArrays(FPGAs)tocomputethe
FFTs,integrate,andcomputepolarizationproducts,allonasinglechip.Otherhybrid
designsuseFPGAstodividethebandintocoarsechannelsandpassthoseeffectively
NyquistsampledsubbandsofftoCPUs,otherFPGAs,orGraphicalProcessingUnits
(GPUs)forfurtherprocessing,suchascoherentdedispersionandfoldingofpulsardatain
apulsarbackend,ormuchfinerfrequencyresolutionandperhapsevenactive
interferenceremovalforhighsensitivityspectroscopyapplications.ThenewVErsatile
GBTAstronomicalSpectrometer(VEGAS)isahybridFouriertransformspectrometer.
Thecapabilitiesofsuchsystems,especiallygiventhefidelityprovidedbysamplingwith
eightormorebitsprecision,ismakingthemthenewstandardbackendtechnologyfor
radioastronomy.

3.6.6MeasuringRadiometerNoise
Theradiometeritselfusuallycontributessignificantlytothetotalsystemnoise
temperatureT sys.Anyradiometercanbemodeledbyanequivalentcircuitconsistingof
anidealnoiselessradiometerplusaninputmatchedloadresistorattemperatureT r,where
T riscalledtheradiometerinputnoisetemperature.

http://www.cv.nrao.edu/~sransom/web/Ch3.html 67/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

ThesimplestwaytomeasureT ristoconnectamatchedhotloadresistorwhose
physicaltemperatureisT htotheradiometerinputandrecordthedetectoroutputvoltage
V h,andthenreplaceitwithacoldloadwhosephysicaltemperatureisT candrecordthe
outputvoltageV c.OftenthehotloadisjustaresistoratroomtemperatureT h 290 K
andthecoldloadisaresistorimmersedinliquidnitrogenatitsboilingtemperature
T c 77 K.

Foreachmeasurement,thesquarelawdetectoroutputvoltageisproportionaltothe
totalinputnoisepowergeneratedbytheactualloadplustheimaginaryresistorwhose
temperatureisT r.InthelowfrequencyNyquistapproximationP = kT,so

V h = P G = k(T h + T r) G,
(3.165)
V c = P G = k(T c + T r) G,
(3.166)
whereisthebandwidthandGistheoverallgainoftheradiometer.BothGand
canceloutintheYfactordefinedby

Vh Th + Tr
Y = , (3.167)
Vc Tc + Tr

sotheydonothavetobemeasured.Equation3.167canbesolvedfortheradiometer
noisetemperature

T h YT c
Tr = . (3.168)
Y1

ThistechniqueformeasuringT riscalledtheYfactormethod.

CommunicationsengineersoftenspecifytheradiometernoisefactorF ndefinedby

Tr + T0
Fn , (3.169)
T0

wherethestandardtemperaturedefinedasT 0 290 Kisclosetoroomtemperature.The


numeratorinEquation3.169isproportionaltothedetectedoutputvoltageofthe
radiometerconnectedtoanambienttemperatureloadandthedenominatoristheoutput
ofanoiselessradiometerconnectedtoanambienttemperatureload.IntermsofF n,the
radiometernoisetemperatureis

T r = (F n 1)T 0.
(3.170)

http://www.cv.nrao.edu/~sransom/web/Ch3.html 68/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

TherelatedradiometernoisefigureNFusedbymanycommercialmanufacturersof
amplifiersandradiometersisjustthenoisefactorF nexpressedindB:

NF 10log 10(F n).


(3.171)

3.7Interferometers
Everypracticalsingledishradiotelescope(Section3.5)hasrelativelylowangular
resolutionandpointingaccuracy,smallfieldofview,andlimitedsensitivity.Thelargest
fullysteerabledishhasdiameterD 100manditsangularresolutionisdiffraction
limitedto / Dradians,soimpossiblylargediameterswouldbeneededtoachieve
subarcsecondresolutionatradiowavelengths.Pointingandsourcetrackingaccuracyis
alsoaproblemforalargesingledish.Thetelescopebeamshouldbeabletofollowa
radiosourceontheskywithin / 10forreasonablyaccuratephotometryorimaging.
Theaccuracywithwhichtheactualbeamdirectionduringanobservationcanbe
recoveredbylaterdataanalysisdeterminestheaccuracywithwhichtheskypositionofa
radiosourcecanbemeasured.Gravitationalsagging,telescopedeformationscausedby
differentialsolarheating,andtorquescausedbywindgustscombinetolimitthe
mechanicaltrackingandpointingaccuraciesofthebestradiotelescopesto 1arcsec.
Mostopticaltelescopescanmakehighresolutionimagescoveringlargeareasofsky
rapidlybecausetheirlargefieldsofview FoV 2covermillionsorbillionsofpixels.
Incontrast,mostsingledishradiotelescopeshaveonlyoneorseveralbeams.The
geometricareaofasingledishisjustD 2 / 4,whilethegeometricareaND 2 / 4ofan
interferometerwithNdishescanbearbitrarilylarge.Thecontinuumsensitivityofasingle
dishisstronglylimitedbyconfusionatfrequenciesbelowabout10GHz.

AperturesynthesisinterferometerscomprisingN 2moderatelysmalldisheshave
mitigatedtheseandmanyotherpracticalproblemsassociatedwithsingledishes,suchas
vulnerabilitytofluctuationsinatmosphericemissionandreceivergain,radiofrequency
interference,andpointingshiftscausedbyatmosphericrefraction.Forexample,the
WesterborkSynthesisRadioTelescope(Figure8.3)consistsofN = 14,D = 25m
telescopesoneastwestbaselinesuptob 3 kminlength.Itstotalcollectingareaisthat
ofasingledishwithdiameterD tot N 1 / 2D 92m.Ithasthehighangularresolutionof
adiffractionlimitedtelescope3kmindiameter.Ithasthelargeinstantaneousfieldof
viewofa25mtelescope,soitcanimage (b / D) 2 10 6pixelsatoncewithonlyone
receiveroneachtelescope.Itcanmeasurepositionsofradiosourceswithsubarcsecond
accuracydespitethemuchlargersourcetrackingerrorsoftheindividualtelescopes.

Historically,thetotalbandwidthsandnumbersofsimultaneousfrequencychannels
ofaperturesynthesisinterferometerswithmanydisheswerelowerthanthoseofsingle
dishes.Recentadvancesincorrelatorelectronicsandcomputinghavelargelyovercome
thesepracticallimitations,soneworupdatedinterferometerssuchasALMA(Figure8.5)
andtheJVLA(Figure8.4)areplayinganincreasinglydominantroleinobservational
radioastronomy.Theprimaryusesofsingledishestodayare

(a)observingpulsars,whicharetimevariablesotheyareeasytoseparatefrom
confusionbytimeindependentcontinuumsources
(b)spectroscopicobservationsofextendedlowbrightnesssources,againlargely
immunetoconfusion
http://www.cv.nrao.edu/~sransom/web/Ch3.html 69/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

(c)complementinginterferometersbyprovidingzerospacingdataonvery
extendedsourcesorbyservingaselementsofverylongbaselinearrays.

3.7.1TheTwoElementQuasiMonochromaticInterferometer
Thesimplestradiointerferometerisapairofradiotelescopeswhosevoltageoutputsare
correlated(multipliedandaveraged),andeventhemostelaborateinterferometerswith
N 2antennas,oftencalledelements,canbetreatedasN(N 1) / 2independenttwo
elementinterferometers.

Figure3.41:Thisblockdiagramshowsthecomponentsofatwoelementquasi
monochromaticmultiplyinginterferometerobservinginaverynarrowradio
frequencyrangecenteredon = / (2).sistheunitvectorinthedirectionofadistant

pointsourceand bisthebaselinevectorpointingfromantenna1toantenna2.The
outputvoltageV 1ofantenna1isthesameastheoutputvoltageV 2ofantenna2,butit

isretardedbythegeometricdelay g = b s / crepresentingtheadditionallighttravel
delaytoantenna1foraplanewavefrontfromasourceatanglefromthebaseline
vector.Thesevoltagesareamplified,multiplied( ),andtimeaveraged( )bythe
correlatortoyieldanoutputresponsewhoseamplitudeRisproportionaltotheflux
densityofthepointsourceandwhosephase( g)dependsonthedelayandthe
frequency.Thequasisinusoidaloutputfringeshownoccursifthesourcedirectionin
theinterferometerframeischangingataconstantrated / dt.ThebroadGaussian
envelopeofthefringeshowstheprimarybeamattenuationasthesourcepasses
http://www.cv.nrao.edu/~sransom/web/Ch3.html 70/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

throughthebeamofthedishes.


Figure3.41showstwoidenticaldishesseparatedbythebaselinevector boflength
bthatpointsfromantenna1toantenna2.Bothdishespointinthesamedirection

specifiedbytheunitvectors,andistheanglebetween bands.Planewavesfroma

distantpointsourceinthisdirectionmusttravelanextradistance b s = bcostoreach
antenna1,sotheoutputofantenna1isthesameasthatofantenna2,butitlagsintime
bythegeometricdelay


b s
g = . (3.172)
c

Forsimplicity,wefirstconsideraquasimonochromaticinterferometer,onethat
respondsonlytoradiationinaverynarrowband 2 / gcenteredonfrequency
= / (2).Thentheoutputvoltagesofantennas1and2attimetcanbewrittenas

V 1 = Vcos[(t g)] and V 2 = Vcos(t).


(3.173)

Theseoutputvoltagesareamplifiedversionsoftheantennainputvoltagestheyhavenot
passedthroughsquarelawdetectors.Instead,acorrelatormultipliesthesetwovoltages
toyieldtheproduct

V 1V 2 = V 2cos[(t g)]cos(t) = ( )V2


2
[cos(2t g) + cos( g)] (3.174)

thatfollowsdirectlyfromthetrigonometricidentity
cosxcosy = [cos(x + y) + cos(x y)] / 2.Thecorrelatoralsotakesatimeaveragelong
enough(t (2) 1)toremovethehighfrequencytermcos(2t g)fromthe
correlatorresponse(outputvoltage)Randkeeponlytheslowlyvaryingterm

R = V 1V 2 = ( )
V2
2
cos( g). (3.175)

ThevoltagesV 1andV 2areproportionaltotheelectricfieldproducedbythesource


multipliedbythevoltagegainsofthetwoantennasandreceivers.Thusthecorrelator
outputamplitudeV 2 / 2isproportionaltothefluxdensitySofthepointsourcemultiplied
by(A 1A 2) 1 / 2,whereA 1andA 2aretheeffectivecollectingareasofthetwoantennas.

NoticethatthetimeaveragedresponseRofamultiplyinginterferometeriszero.
ThereisnoDCoutput,sofluctuationsinreceivergaindonotactonthewholesystem
http://www.cv.nrao.edu/~sransom/web/Ch3.html 71/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

temperatureT sasforatotalpowerobservationwithasingledish(Equation3.155).
Uncorrelatednoisepowerfromveryextendedradiosourcessuchasthecosmic
microwavebackgroundandtheatmosphereoverthetelescopes,alsoaveragestozeroin

thecorrelatorresponse.Shortinterferencepulseswithdurationt | b| / carealso
suppressedbecauseeachpulsedoesnotreachbothtelescopessimultaneously.Likewise,a
multiplyingradiointerferometerdiffersfromaclassicaladdinginterferometer,suchas
theopticalMichelsoninterferometer,thataddstheuncorrelatednoisepower
contributions.

ThecorrelatoroutputvoltageR = (V 2 / 2)cos( g)variessinusoidallyastheEarths


rotationchangesthesourcedirectionrelativetothebaselinevector.Thesesinusoidsare
calledfringes,andthefringephase


= g = bcos (3.176)
c

dependsonasfollows:

d
= bsin (3.177)
d c

= 2 ( )
bsin

. (3.178)

Thefringeperiod = 2correspondstoanangularshift = / (bsin).Thefringe


phaseisanexquisitelysensitivemeasureofsourcepositioniftheprojectedbaseline
bsinismanywavelengthslong.Notethatfringephaseandhencemeasuredsource
positionisnotaffectedbysmalltrackingerrorsoftheindividualtelescopes.Itdependson
time,andtimescanbemeasuredbyclockswithmuchhigheraccuracythanangles(ratios
oflengthsofmovingtelescopeparts)canbemeasuredbyrulers.Also,aninterferometer
whosebaselineishorizontalisnotaffectedbytheplaneparallelcomponentof
atmosphericrefraction,whichdelaysthesignalsreachingbothtelescopesequally.
Consequently,interferometerscandeterminethepositionsofcompactradiosourceswith
unmatchedaccuracy,asshowninFigure1.6.Absolutepositionswitherrorsassmallas
10 3arcsecanddifferentialpositionswitherrorsdownto 10 5arcsec
< 10 10radhavefrequentlybeenmeasured.

Iftheindividualantennascomprisinganinterferometerwereisotropic,the
interferometerpointsourceresponsewouldbeasinusoidspanningthesky.Suchan
interferometerissensitivetoonlyoneFouriercomponentoftheskybrightness
distribution:thecomponentwithangularperiod / (bsin).TheresponseRofatwo
elementinterferometerwithdirectiveantennasisthatsinusoidmultipliedbytheproduct
ofthevoltagepatternsoftheindividualantennas.Normallythetwoantennasare
identical,sothisproductisthepowerpatternoftheindividualantennasandiscalledthe
primarybeamoftheinterferometer.TheprimarybeamisusuallyaGaussianmuch
widerthanafringeperiod,asindicatedinFigure3.41.Theconvolutiontheorem
(EquationA.15)statesthattheFouriertransformoftheproductoftwofunctionsisthe
convolutionoftheirFouriertransforms,sotheinterferometerwithdirectiveantennas
respondstoafiniterangeofangularfrequenciescenteredon(bsin / ).Becausethe

http://www.cv.nrao.edu/~sransom/web/Ch3.html 72/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

antennadiametersDmustbesmallerthanthebaselineb(elsetheantennaswould
overlap),theangularfrequencyresponsecannotextendtozeroandtheinterferometer
cannotdetectanisotropicsourcethebulkofthe3Kcosmicmicrowavebackgroundfor
example.Themissingshortspacings(b < D)canbeprovidedbyasingledishtelescope
withdiameterD > b.ThustheD=100mGBTcanfillinthemissingbaselinesb < 25 m
thattheD=25mVLAdishescannotobtain.

Improvingtheinstantaneouspointsourceresponsepatternofaninterferometer
requiresmoreFouriercomponentsthatis,morebaselines.AninterferometerwithN
antennascontainsN(N 1) / 2pairsofantennas,eachofwhichisatwoelement
interferometer,sotheinstantaneoussynthesizedbeam(thepointsourceresponse
obtainedbyaveragingtheoutputsofallofthetwoelementinterferometers)rapidly
approachesaGaussianasNincreases.Theinstantaneouspointsourceresponsesofatwo
elementinterferometerwithprojectedbaselinelengthb,athreeelementinterferometer
withthreebaselines(projectedlengthsb / 3,2b / 3,andb),andafourelement
interferometerwithsixbaselines(projectedlengthsb / 6,2b / 6,3b / 6,4b / 6,5b / 6,andb)
areshowninFigure3.42.

Figure3.42:Theinstantaneouspointsourceresponsesofinterferometerswithoverall
projectedlengthbandtwo,three,orfourantennasdistributedasshownareindicated
bythethickcurves.Thesynthesizedmainbeamofthefourelementinterferometeris
nearlyGaussianwithangularresolution / b, butthesidelobesarestillsignificant
andthereisabroadnegativebowlcausedbythelackofspacingsshorterthanthe
diameterofanindividualantenna.Thusthesynthesizedbeamissometimescalledthe
dirtybeam.Theinstantaneousdirtybeamofthemultielementinterferometeristhe
arithmeticmeanoftheindividualresponsesofitscomponenttwoelement
interferometers.Theindividualresponsesofthethreetwoelementinterferometers
comprisingthethreeelementinterferometerandofthesixtwoelementinterferometers
comprisingthefourelementinterferometerareplottedasthincurves.
http://www.cv.nrao.edu/~sransom/web/Ch3.html 73/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

Mostradiosourcesarestationarythatis,theirbrightnessdistributionsdonot
changesignificantlyonthetimescalesofastronomicalobservations.Forstationary
sources,atwoelementinterferometerwithmovableantennascouldmakeN(N 1) / 2
observationstoduplicateoneobservationwithanNelementinterferometer.

3.7.2SlightlyExtendedSourcesandtheComplexCorrelator

TheresponseR c = (V 2 / 2)cos( g)ofthequasimonochromatictwoelement


interferometerwithacosinecorrelator(Figure3.41andEquation3.175)toaspatially
incoherentslightlyextended(muchsmallerthantheprimarybeamwidth)sourcewithsky
brightnessdistributionI (s)nearfrequency = / (2)isobtainedbytreatingthe
extendedsourceasthesumofindependentpointsources:


R c = I(s)cos(2 b s / c) = I(s)cos(2 b s / ). (3.179)

Noticethattheevencosinefunctioninthisresponseissensitiveonlytotheeven
(inversionsymmetric)partI Eofanarbitrarysourcebrightnessdistribution,whichcanbe
writtenasthesumofevenandodd(antisymmetric)parts:I = I E + I O.Todetecttheodd
partI Oweneedasinecorrelatorwhoseoutputisodd,R s = (V 2 / 2)sin( g).Thiscan
beimplementedbyasecondcorrelatorthatfollowsa / 2 rad = 90 phasedelayinserted
intotheoutputofoneantennabecausesin( g) = cos( g / 2).Then


R s = I(s)sin(2 b s / ). (3.180)

Thecombinationofcosineandsinecorrelatorsiscalledacomplexcorrelatorbecauseit
ismathematicallyconvenienttotreatthecosinesandsinesascomplexexponentialsusing
Eulersformula(AppendixB.3)

e i = cos + isin. (3.181)

Thecomplexvisibilityisdefinedby

R c iR s
(3.182)

whichcanbewrittenintheform

= Ae i , (3.183)

where

http://www.cv.nrao.edu/~sransom/web/Ch3.html 1/2 74/84


7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

2 2 1/2
A = (R c + R s )
(3.184)

isthevisibilityamplitudeand

= tan 1(R s / R c)
(3.185)

isthevisibilityphase.Theresponsetoanextendedsourcewithbrightnessdistribution
I(s)ofthetwoelementquasimonochromaticinterferometerwithacomplexcorrelatoris
thecomplexvisibility


= I(s)exp( i2 b s / ) d. (3.186)

3.7.3EffectsofFiniteBandwidthsandAveragingTimes
Equation3.186forquasimonochromaticinterferometersmaybegeneralizedto
interferometerswithfinitebandwidthsandintegrationtimes,whicharenecessaryforhigh
sensitivity.Inthesmallbutfinitefrequencyrangecenteredonfrequency c,Equation
3.186becomes

= [ + /2
cc / 2 I (s)exp( i2 b s / ) ]
(3.187)

[ ]
c + / 2
= I ()exp( i2 g)
c / 2 s
. (3.188)

Ifthesourcebrightnessandtheresponseoftheinterferometerarenearlyconstantover
,theintegraloverfrequencyisjusttheFouriertransformofarectanglefunction,so

I (s)sinc( g)exp( i2 c g).


(3.189)

Forafinitebandwidthanddelay g,thefringeamplitudeisattenuatedbythefactor
sinc( g).Thisattenuationcanbeeliminatedinanyonedirections 0calledthedelay
centerorthephasereferencepositionbyintroducingacompensatingdelay 0 gin
thesignalpathoftheleadingantenna,asshowninFigure3.43.AstheEarthturns, 0
mustbecontinuouslyadjustedtotrack gwithinatolerance| 0 g| () 1.Thisis
usuallydonewithdigitalelectronics.

http://www.cv.nrao.edu/~sransom/web/Ch3.html 75/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

Figure3.43:Thecompensatingdelay 0,shownhereasanextraloopofcablebetween
antenna2andthecorrelator,musttrackthegeometricdelay ginthedirections 0ofthe
delaycenteraccuratelyenoughtokeep| 0 g| () 1inordertominimize
attenuation.

Thegeometricdelayvarieswithdirection,sodelaycompensationcanbeexactin
onlyonedirection.Theangularradiusoftheusablefieldofviewisdeterminedbythe

variationof gwithoffsetfromthedirections 0.Becausec g = b s = bcos,
|c g| = bsin.Requiring

g 1
(3.190)

implies

(bsin) / c 1. (3.191)

Substituting = candusing s / (bsin)forthesynthesizedbeamwidth,wegetthe


requirement


.
s (3.192)

http://www.cv.nrao.edu/~sransom/web/Ch3.html 76/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

Atlargerangularoffsetsfromthephasereferenceposition,bandwidthsmearingwill
radiallybroadenthesynthesizedbeambyconvolvingitwitharectangleofangularwidth
/ .

Satisfactorywidefieldimagescanbemadewithalargertotalbandwidthonlyby
dividingthatbandwidthintoanumberofnarrowerfrequencychannelseachsatisfying
Equation3.192.Forexample,thesynthesizedbeamwidthoftheVLABconfiguration
(maximumbaselinelengthb 10 km)at = 20 cm( = 1.5 GHz)is
s [(0.2 m) / (10 4 m)] rad 4 arcsec.Toimageouttoanangularradius
= 15 arcmin = 900 arcsecequaltothehalfpowerradiusoftheVLAprimarybeam
requireschannelbandwidths

s 1.5 10 9 Hz 4 arcsec
= 7 MHz. (3.193)
900 arcsec

Likewise,thecorrelatoraveragingtimetmustbekeptshortenoughthattheEarths
rotationwillnotmovethesourcepositionintheframeoftheinterferometerbyasmuch
asthesynthesizedbeamwidth s / b.Forexample,ifthedelayissettotrackthenorth
celestialpole,asourceawayfromthenorthpolewillappeartomoveatanangular
rate2 / P,whereP 23 h56 m04 s 86164 sistheEarthssiderealrotationperiod.
Excessivecorrelatoraveragingtimeswillcausetimesmearingthattangentiallybroadens
thesynthesizedbeam.Tominimizetimesmearinginanimageofangularradius,we
require

2t t s
. (3.194)
P 1.37 10 4 s

Continuingwiththepreviousexample,toimageouttoanangularradius = 900 arcsec


when s = 4 arcsecrequiresaveragingtimestshortenoughthat

s 4 arcsec
t 1.37 10 4 s = 1.37 10 4 s 60 s. (3.195)
900 arcsec

3.7.4EarthRotationApertureSynthesis
TheEarthsrotationvariestheprojectedbaselinecoverageofaninterferometerwhose
elementsarefixedontheground.Inparticular,allbaselinesofaninterferometerwhose
baselinesareconfinedtoaneastwestlinewillremaininasingleplaneperpendicularto
theEarthsnorthsouthrotationaxisastheEarthturnsdaily.Confiningallbaselinesto
twodimensionshasthecomputationaladvantagethatthebrightnessdistributionofa
sourceissimplythetwodimensionalFouriertransformofthemeasuredvisibilities.

Figure3.44illustratesEarthrotationaperturesynthesisbyaneastwesttwo
elementinterferometeratlatitude+ 40 asviewedfromasourceatdeclination

.Let betheeastwestcomponentoftheprojectedbaselineinwavelengths
http://www.cv.nrao.edu/~sransom/web/Ch3.html 77/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

= + 30 .Letubetheeastwestcomponentoftheprojectedbaselineinwavelengths
andvbethenorthsouthcomponentoftheprojectedbaselineinwavelengths.

Figure3.44:Viewedfromadistantradiosource,atdeclination = + 30 forthis
drawing,theEarthrotatescounterclockwisewithaperiodofonesiderealdayaboutthe
northsouthaxisindicatedbythearrowemergingfromthenorthpole.Theantennasof
atwoelementeastwestinterferometeratlatitude+ 40 areshown,fromlefttoright,
astheywouldappearathourangles 6 h, 3 h,0 h,+ 3 h,and+ 6 h.Projectedontothe
planeofthepage,whichisnormaltothelineofsight,theinterferometerbaseline
rotatescontinuouslyfrompurelynorthsouthat 6 hthrougheastwestat0 handback
tonorthsouthat+ 6 h.Theprojectedantennaseparationalsochanges.Duringthis12
hourperiod,theprojectedbaselinetracesanellipseinthe(u, v)planeasshownbythe
dashedcurve,withpointsonthe(u, v)ellipsehighlightingtheinstantaneouscoverage
at 6 h, 3 h,0 h,+ 3 h,and+ 6 h.Thevaxisoftheellipseissmallerbyafactorsinthan
theuaxis.

Duringthe12hourperiodcenteredonsourcetransit,theinterferometertracesouta
completeellipseonthe(u, v)plane.Themaximumvalueofuequalstheactualantenna
separationinwavelengths,andthemaximumvalueofvissmallerbytheprojectionfactor
sin,whereisthesourcedeclination.Iftheinterferometerhasmorethantwoelements,
orifthespacingofthetwoelementsischangeddaily,the(u, v)coveragewillbecomea
numberofconcentricellipseshavingthesameshape.Thusthesynthesizedbeam
obtainedbyeastwestEarthrotationaperturesynthesiscanapproachanelliptical
Gaussian.Thesynthesizedbeamwidthis u 1radianseastwestand u 1cscradians
inthenorthsouthdirection.Thesynthesizedbeamiscircularforasourcenearthe
celestialpole,butthenorthsouthbeamwidthisverylargeforasourcenearthecelestial
equator.

3.7.5InterferometersinThreeDimensions
TheVLA(VeryLargeArray)showninFigure8.4isYshapedandisinstantaneouslya
nearlycoplanartwodimensionalarrayof2725mtelescopesonthehighPlainsofSan
AugustininNewMexico.Itbaselinesarenotconfinedtoaneastwestlinebutitisnearly
coplanar,sosnapshotobservationsmuchshorterthanasiderealdaycanbetreatedas
http://www.cv.nrao.edu/~sransom/web/Ch3.html 78/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

twodimensional.Onlongertimescales,EarthrotationcausestheVLAbaselinestofilla
threedimensionalvolume.Thenorthsouthbaselinesallowimagingwithanearly
circularsynthesizedbeamevennearthecelestialequator.Figure8.4showstheD
configurationspanningabout1km.Thetelescopescanbemovedalongrailroadtracksto
formtheC,B,andAconfigurationsspanning3.4,11,and36km,respectivelyfor
higherangularresolution.TheVLArecentlyunderwentamajorupgradetobecomethe
JVLA(theJstandsforJansky),withnewwidebandreceiverscompletelycovering
thefrequencyrange1to50GHzandafarmorepowerfulandversatilecorrelator.Itisup
toanorderofmagnitudemoresensitivethantheoriginalnarrowbandVLA.


The(u, v, w)coordinatesystemusedtodescribeanybaselinevector binthree
dimensionsisshowninFigure3.45.Thewaxisisinthereferencedirections 0usually
chosentocontainthetargetradiosource.Theuandvaxespointeastandnorthinthe

(u, v)planenormaltothewaxis.u,v,andwarethecomponentsof b / ,thebaseline
vectorinwavelengthunits.Anarbitraryunitvectorshascomponents(l, m, n)asdrawn,
1/2
wheren = cos = (1 l 2 m 2) .Thecomponents(l, m, n)arecalleddirectioncosines.

Figure3.45:The(u, v, w)coordinatesystemforinterferometers.Thewaxispointsin
thereferencedirections 0usuallycontainingthesourcetobeimaged.Projectedonto
theplanenormaltothewaxis,uistheeastwestbaselineinwavelengthsandvisthe
northsouthbaselineinwavelengths.l,m,andnareprojectionsoftheunitvectors
ontotheu,v,andwaxes,respectively.

Because

dl dm
d = ,
2 2 1/2 (3.196)
(1 l m )

thethreedimensionalgeneralizationofEquation3.186is

http://www.cv.nrao.edu/~sransom/web/Ch3.html 79/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

I (l, m) (3.197)
(u, v, w) = 1/2
exp[ i2(ul + vm + wn)]dl dm.
(1 l 2 m 2)

ThisisnotathreedimensionalFouriertransform.

However,ifw = 0,Equation3.197becomesatwodimensionalFouriertransform,
whichcanbeinvertedtogivethesourcebrightnessdistributionintermsofthemeasured
visibilities:

I (l, m)
2 2 1/2
= (u, v, 0)exp[ + i2(ul + vm)]du dv. (3.198)
(1 l m )

ThatisthecaseforanEarthrotationaperturesynthesisbyaneastwestinterferometerif
wechooses 0tocoincidewiththeEarthsrotationaxis,inwhichcase
1/2
(1 l 2 m 2) = cos = sin,whereisthedeclinationofthereferenceposition.

Foranyinterferometer,ifweconsideronlydirectionsclosetos 0,then
n = cos 1 2 / 2and

I (l, m)
(u, v, w) exp( i2w) exp[ i2(ul + vm w 2 / 2)]l m. (3.199)
2 2 1/2
(1 l m )

Thefactorexp( i2w 2 / 2)canbekeptclosetounitybykeepingw 2 1thatis,by


imagingonlyasmallfieldofviewwhoseradiusis w 1 / 2 ( / b) 1 / 2.Forexample,
0.01radiansissufficientlysmallforaninterferometerbaseline10 4wavelengths
long.Then

I (l, m)
exp(i2w) = 1/2
exp[ i2(ul + vm)]l m. (3.200)
(1 l 2 m 2)

Afieldwiderthan w 1 / 2canbeimagedwithtwodimensionalFouriertransforms
bybreakingitupintosmallerfacets,muchlikeaflyseye,andmergingthefacetsto
makethefinalimage.

3.7.6Sensitivity
Thepointsourcesensitivityofatwoelementinterferometercanbederivedfromthe
radiometerequationforatotalpowerreceiveronasingleantennabecauseasquarelaw
detectorisequivalenttoacorrelatormultiplyingtwoidenticalinputvoltagessuppliedby
oneantenna.Consideraninterferometerwithtwoidenticalelements,eachofwhichalso
hasasquarelawdetector,observingapointsource.Thecorrelatormultipliesthevoltages
fromthetwoantennas,whileeachsquarelawdetectormultipliesthevoltagefromone
http://www.cv.nrao.edu/~sransom/web/Ch3.html 80/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

antennabyitself,sothecorrelated/detectedoutputvoltagesoftheinterferometerandeach
singledishareequalinstrength.ThustheeffectivecollectingareaA eofthetwoelement
interferometerequalstheeffectivecollectingareaofeachelement.However,thenoise
voltagesfromthetwointerferometerelementsarealmostcompletelyuncorrelated(only
thepointsourcecontributescorrelatednoise),whilethenoisevoltagesgoingintothe
squarelawdetectorsarecompletelycorrelated(identical).Thecorrelatoroutputvoltage
distributionbeforesmoothingisshowninFigure3.46,andFigure3.47showsthe
correlatoroutputvoltagedistributionaftersmoothingoverN = 50samples.Inthelimit
wheretheantennatemperatureTcontributedbythepointsourceismuchsmallerthan
thesystemnoiseT s,thecorrelatoroutputnoiseis2 1 / 2lowerthanthesquarelawdetector
noisefromeachantenna.ForanunpolarizedpointsourceoffluxdensityS,then
kT = SA e / 2,soforasingleantenna,

2kT s
S =
A e( ) 1 / 2 (3.201)

andforatwoelementinterferometer,

2 1 / 2kT s
S = . (3.202)
A e( ) 1 / 2

Thepointsourcesensitivityofatwoelementinterferometeristherefore2 1 / 2timesbetter
thanthesensitivityofeachantenna,but2 1 / 2timesworsethanthatofasingledishwhose
areaisthatoftwoantennas.Thereasonthetwoelementinterferometerislesssensitive
thanasingledishhavingthesametotalcollectingareaisthattheinformationcontained
inthetwoindependentsquarelawdetectoroutputshasbeendiscarded.Togetherthey
have2 1 / 2timesthesensitivityofasingledish.Combinedwiththeindependentcorrelator
output,thetotalsensitivityis(2 + 2) 1 / 2 = twicethesensitivityofasingledish,or
exactlythesensitivityofasingledishwhoseareaequalsthetotalareaofthetwoelement
interferometer.

http://www.cv.nrao.edu/~sransom/web/Ch3.html 81/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

Figure3.46:Theunsmoothedoutputvoltageofacorrelatorwhoseinputsare
uncorrelatedGaussiannoisehasasymmetricdistributionwithzeromean,andtherms
fluctuationisafactor2 1 / 2timessmallerthanthatofasquarelawdetector
(Figure3.33).

Figure3.47:ThesmoothedoutputvoltageofacorrelatorapproachesaGaussianwith
zeromean,andthermsnoiseisreducedbythesquarerootofthenumberof
independentsamplesaveragedtogether.ThisfigureshowsnoisefromanN = 50
samplerunningmean.Thermsfluctuationisafactor2 1 / 2timessmallerthanthatofa
squarelawdetector(Figure3.34).

AninterferometerwithNdishescontainsN(N 1) / 2independenttwoelement
interferometers.Solongasthesignalfromeachdishcanbeamplifiedcoherentlybeforeit
issplituptobemultipliedbythesignalsfromtheN 1otherantennas,itspointsource
rmsnoiseis

http://www.cv.nrao.edu/~sransom/web/Ch3.html 82/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

2kT s
S = . (3.203)
A e[N(N 1) ] 1 / 2

InthelimitoflargeN,[N(N 1)] 1 / 2 Nandthepointsourcesensitivityofan


interferometerapproachesthatofasingleantennawhoseareaequalsthetotaleffective
areaNA eoftheNinterferometerantennas.Forexample,theVLAwithN = 27dishes
eachd = 25mindiameterhasthepointsourcesensitivityofasingledishwhose
diameterisD = [N(N 1)] 1 / 4d = [27(26)] 1 / 4 25 m = 129 m.Hadthesquarelaw
detectoroutputsbeenusedaswell,thepointsourcesensitivityoftheNelement
interferometerwouldbeexactlythesameasthesensitivityofasingledishhavingthe
sametotalcollectingarea.

Practicalinterferometersareslightlylesssensitivethanthisbecausetheircorrelators
usedigitalmultipliersthatsampleandquantizetheinputvoltage,notperfectanalog
multipliers.Forexample,adigitalmultiplierthatsamplesattwicetheNyquistratewith
threequantizationlevels( 1, 0, + 1)isonly0.89timesassensitiveasaperfectanalog
multiplier.ThechapterDigitalSignalProcessinginThompsonetal.[106]coversthis
andotherconsequencesofquantizationindetail.

Althoughthepointsourcesensitivityofaninterferometeriscomparablewiththe
pointsourcesensitivityofasingledishhavingthesametotalarea,bewarethatthe
brightnesssensitivityofaninterferometerismuchworsebecausethesynthesizedbeam
solidangleofaninterferometerismuchsmallerthanthebeamsolidangleofasingledish
ofthesametotaleffectivearea.Theangularresolutionofaninterferometerwith
maximumbaselinebis / bandtheangularresolutionofthesingledishwithdiameter
Dis / D,sothebeamsolidangleoftheinterferometerissmallerbyafactor
(D / b) 2.Thisisroughlytheareafillingfactoroftheinterferometer,definedasthe
ratiooftheareacoveredbyalloftheantennastotheareaspannedbytheinterferometer
array.Forexample,theVLAinitsb 11 kmBconfigurationhasafillingfactor
2
(129 m / 1.1 10 4 m) 1.2 10 4.Ahighresolutioninterferometercannotdetecta
sourceoflowsurfacebrightness,nomatterhowhighitstotalfluxdensity.

Theintensityaxisofanyastronomicalimagehasdimensionsofspectralbrightness
orspecificintensity(e.g.,unitsofJyperbeamsolidangleorMJysr 1orK),notflux
density(e.g.,Jy).Thepointsourcerms SinEquation3.203correspondstoimageflux
densityperbeamsolidangle,e.g.,Jybeam 1.Publishedradioimagesusuallyhave
intensityaxesinunitsofJybeam 1becausethefluxdensityofapointsourceequalsits
brightnessinthoseunitsandbecause Sisindependentofbeamsolidangle.However,a
properspectralbrightnessdependsonlyonthesource.Thespectralbrightnessspecified
inJybeam 1hasthedimensionsofspectralbrightness,butbewarethatthisisnota
properspectralbrightnessbecauseitdependsonthesynthesizedbeamsolidangleand
notjustontheradiosource.Infraredastronomersfrequentlyspecifyimageintensityin
MJysr 1,whichisaproperbrightness.ThebrightnesstemperatureTisaconvenient
properbrightnessforradioimages.Thermsbrightnesstemperaturesensitivity Tofan
imagemadewithbeamsolidangle Afollowsdirectlyfrom SandtheRayleighJeans
approximation:

http://www.cv.nrao.edu/~sransom/web/Ch3.html 83/84
7/25/2016 3 Radio Telescopes and Radiometers Essential Radio Astronomy

( )
S 2
T = . (3.204)
A 2k

MostinterferometerimagesarerestoredwithGaussianbeams.Thebeamsolidangle
(Equation3.34)ofaGaussianbeamwithHPBW HPBWis(Equation3.118)

2
HPBW
A = ,
4ln2

so

( )
2ln2 c 2 S
T = .
k 2 2
HPBW (3.205)

Forexample,allofthe1.4GHzNRAOVLASkySurvey(NVSS)imageshavermsnoise
S 0.45 mJy beam 1andwererestoredwithacircularGaussianbeamwhosehalf
powerbeamwidthis HPBW = 45 arcsec 2.18 10 4 rad.Consequently,NVSSrms
brightnesstemperaturenoiseis

2
T [ 2ln2 (3 10 8 m s 1)
1.38 10 23
JK 1
(1.4 10 Hz) 9 2 ] 0.45 10 29 W Hz 1
(2.18 10 4
rad)
2
0.14 K.

Thisisgoodenoughtodetect(5 T 0.7 K)normalspiralgalaxieswithmedian


T b 1 Kat1.4GHz.Bewarethatahighresolution(low A)imagewithagood
pointsourcesensitivity(low S)maystillhaveapoorbrightnesstemperaturesensitivity
(high T).

2RadiationFundamentalsBibliographyIndex 4FreeFreeRadiation
GeneratedonSatMar1218:31:372016byLATEXML

http://www.cv.nrao.edu/~sransom/web/Ch3.html 84/84

Das könnte Ihnen auch gefallen