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

net/publication/23825914

Optical antenna gain. II - Receiving antennas

Article · November 1974

Source: NTRS

CITATIONS READS

3 332

2 authors, including:

John J. Degnan
Semi-Retired Technical Consultant
243 PUBLICATIONS 4,334 CITATIONS

SEE PROFILE

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

Laser Altimetry View project

Optical Resonators View project

All content following this page was uploaded by John J. Degnan on 09 August 2018.

The user has requested enhancement of the downloaded file.

Optical Antenna Gain. 2: Receiving Antennas

John J. Degnan and Bernard J. Klein

Expressions are derived for the gain of a centrally obscured, circular optical antenna when used as the col-
lecting and focusing optics in a laser receiver which include losses due to (1) blockage of the incoming light
by the central obscuration, (2) the spillover of energy at the detector, and (3) the effect of local oscillator
distribution in the case of heterodyne or homodyne detection. Numerical results are presented for direct
detection and for three types of local oscillator distributions (uniform, Gaussian, and matched) in the case
of heterodyne or homodyne detection. The results are presented in several graphs that allow the rapid
evaluation of receiver gain for an arbitrary set of telescope and detector parameters. It is found that, for
uniform illumination by the LO, the optimum SNR is obtained when the detector radius is approximately
0.74 times the Airy disk radius. The use of an optimized Gaussian (spot size = 0.46 times the Airy disk ra-
dius) improves the receiver gain by less than 1 dB. These results are insensitive to the size of the central
obscuration.

1. Introduction and primary mirrors, respectively, and A is the area

In Part 1 of this paper, we presented several of the telescope primary (= 7ra2). The first term is,
graphs that allow the rapid evaluation of the near- of course, the antenna gain realized by an ideal unob-
and far-field gain of centrally obscured, circular opti- scured antenna of area A and has been plotted as a
cal antennas fed by a laser transmitter operating in function of aperture radius for several important
the TEMoo Gaussian mode. In this part, we consider wavelengths in Fig. 7 of Part 1. The second term in-
the gain of such antennas when used as the collecting dicates the dB loss due to blockage of the incoming
and focusing optics in a laser receiver. In evaluating light by the central obscuration and is plotted in Fig.
the receiver gain we consider the losses due to (1) 1. The third term represents the losses at the detec-
blockage of the incoming light by the central obscu- tor. In the case of direct detection, we will take this
ration, (2) the spillover of energy beyond the perime- to mean losses due to the spillover of signal energy
ter of a finite detector, and (3) the effects of the local beyond the detector perimeter. We shall see that,
oscillator distribution in the case of heterodyne and for heterodyne detection, the quantity wq depends on
homodyne detection. several quantities including the local oscillator field
distribution.
1. Theory When used as the collecting optics in a laser receiv-
er, the telescope performs the role of a large aperture
A. Definition of the Receiver Gain lens of focal length f. The detector is placed in the
In the present calculation, we assume that the sig- back focal plane of this lens. The amplitude of the
nal source is sufficiently far away so that plane waves electric field distribution in the plane of the detector
impinge on the receiver aperture. In keeping with is given by 2
the spirit of Part 1,1 we will write the receiver gain as
ES(Pf) = Epa [J,(pf ) _ Yj(Y f)]' (2)
47TA +1 o~
GR(dB) = 10 log VITA± 10 log(1 _ C2) + 10 logqn
(1) where p is the radial coordinate in the plane of the
detector, Eo is the peak amplitude of the uniform
where, as in Part 1, y = b/a, where b and a are the ap- plane wave impinging on the telescope entrance
erture radii of the secondary (or other obscuration) pupil, a is the primary radius, Fs is the antenna F
number, y is the obscuration ratio, and k = 2/X,
where Ais the wavelength.
B. Direction Detection

The authors are with NASA Goddard Space Flight Center, In a direct detection system, the over-all power
Greenbelt, Maryland 20771. into the detector is the important quantity. The
Received 5 November 1973. fraction detecte is simply given by

October 1974 / Vol. 13, No. 10 / APPLIED OPTICS 2397

 cillator field distributions, COT and WLO are the trans-
Is

nn mitter and local oscillator frequencies, and 0 is an ar-

CaJ
._ ~~~~I I bitrary lag in the longitudinal phase of the two
-4
0 beams. In general, the field distributions are com-
-J plex, that is, Es(pf) = s(pf) expULps(pf)Iand ELO(pf)
0
-4 = Lo(pf) expJpLo(pf)], where the exponential argu-
~0_ ments correspond to a possible variation of the beam
phase front across the detector. In examining Eq.
-1..0 (5), it will be assumed that the antenna design prop-
z erly takes into account the phase front matching re-
U)
0 quirements of a heterodyne receiver and that there
U: are no tilts or misalignments in the system. The
-J above assumption can be expressed mathematically
by the condition os(pf) fo(Pf) for pf < RD. If we
-2. _ also ignore the relatively minor contribution of the
m 0.0 0.1 0.2 0.3 0.4 0.5 signal power to the dc power component and compo-
0n
OBSCURATION RATIO,) nents associated with frequencies of order 2 oT (which
Fig. 1. Receiver loss due to the central obscu. ration in dB. do not lie within the frequency response of the
mixer), Eq. (5) becomes

2 7TR
P(t) = 2 ELO(Pf)PfdPf] + 2
T cos[(w)
I(pf)pf dpf
77
D =
27 I(Pf)pfdpf
- )LO)t + i]f R ELO(Pf)ES(pf)pfdpf, (6)

2 f RD[ ( ) P I dpf
where the signal amplitude function Es(pf) is given by
- Yj1(Y!2FLS)] PY ) (3) Eq. (2). The first term in Eq. (6) corresponds to the
time-averaged local oscillator power PLO actually in-
where RD is the radius of the detector (assumed to be cident on the detector while the second term is the
circular). By defining the variable u = kpf/2Fs, the instantaneous IF component PIF(t). The time-aver-
above equation simplifies to aged signal power entering the annular entrance ap-
erture is given by
kRD

2FDSY - J FS[ju- J& u) P= 2-E02T7a

2
(1 _ 2) (7)
(4)
where the fraction detected is a function of kRD/2Fs
and y. The fractional power detected is plotted in
dB in Fig. 2 against the parameter kRD/2Fs for sever-
al values of y. The points of zero slope in the various
curves correspond to radial nulls in the electric field
distribution. The curves clearly indicate that, as the
obscuration ratio -y increases, a smaller fraction of a
the total diffracted energy appears in the central U)
lobe.
z
0
C. Heterodyne and Homodyne Detection I
0
In a heterodyne detection system, the electric field
of the signal distribution is mixed with the local oscil- a
C:)
lator field. The instantaneous power into the detec-
tor is given by

P(t) = 2 r RDI(pf, t)pfdpf

3.0 4.0 5.0 6.0 10.0
= 2o [Re{Es(Pf) exp(jott)] kR0 /2F,

Fig. 2. Fractional power detected as a function of the antenna ob-

+ ReELo(pf) expjj(wLot - )}]2pfdpf (5) scuration ratio -y and the quantity kRD/2Fs, where k = 2r/X, RD is
the detector radius, and F is the F number of the antenna (direct
where Es(pl) and ELO(Pf) are the signal and local os- detection or matched heterodyne detection).

2398 APPLIED OPTICS / Vol. 13, No. 10 / October 1974

Using Eqs. (2) and (7), the mean square IF power can the shape of the local oscillator field distribution and
be expressed as not on its magnitude. The effect of the local oscilla-
tor power level on the mean square IF current is con-
tained in the factor PLO appearing in Eq. (11). This
XIF= 2 [2R pES(pf)pfdPf ]
(Pf) factor of PLO cancels with a similar factor in the ex-
pression for the LO-induced shot noise, which for a
PR y 2{ E LO(Pf) [J1( 2 F)
well-designed receiver, is the predominant noise term
in the denominator of Eq. (9).3 Thus, a dB change in
the factor 1OH converts directly into a dB change in
- YJ1(y 2f ) dPf}. (8) the system SNR. In succeeding paragraphs, we con-
sider some important special cases of Eq. (13).
The SNR at the detector is defined as
D. Uniform Local Oscillator Distribution

(S/NIPower= i 2/(Z iv ) (9) If we assume a uniform local oscillator distribution

n across the detector, Eq. (13) reduces to
where i 2 is the mean square current generated in the 2 )]{[J - JO(Ur)]/Ur[J
7)Hf(Y,r) = [4/(1 - 7 0 (yur)
detector by the IF signal, and in 2 is the mean square (14)
current generated by a particular noise process. The
quantity is2 is in turn given by where a =kRA/2FS, r = RD/RA, and RA is the Airy
spot size (defined as the distance from the center of
iS2 = (eTG/hv)
2
P, (10) the Airy pattern to the first null). From Eq. (2) we
see that a is the first nonzero root of the equation
where e is the electronic charge, r is the detector
quantum efficiency, G is the detector gain (if any), h J1 () - YJ1 (yu) = 0 (15)
is Planck's constant, and v is the photon frequency.
In most references dealing with heterodyne detec- and hence is a function of y. The quantity a relates
the Airy spot size to the wavelength and antenna F
tion,3 it is assumed that the signal and local oscillator number and plotted against y in Fig. 3. Equation
distributions uniformly illuminate the detector. (14) is plotted in Fig. 4 vs r for several values of y. It
This assumption allows the mean square signal cur- is clear from the latter figure that a minimum detec-
rent to be written in the form
tion loss is incurred when r 0.75. This might be
is = 2(eTG/hv) PLOPS, (11) considered a surprising result at first sight since,
from Eqs. (8) and (15), the power in the IF signal is
where Ps and PLO are the time averaged signal and clearly a maximum when r = 1. (The detector radius
local oscillator powers incident on the detector. is equal to the Airy spot radius.) System perfor-
Substituting Eq. (8) into Eq. (10) and comparing the mance, however, is determined by the SNR and not
resultant expression with Eq. (11) suggests the rela- the absolute magnitude of the IF signal. Therefore,
tionship while it is true that increasing the detector radius
from 0.75RA to L.ORAresults in a slightly larger IF
IlHPR
PS. = (12) signal, it also results in greatly increased local oscilla-
for the more general case where the two distributions tor-induced shot noise with the net result that the
are not uniform. The quantity r7Hthat is given by SNR actually decreases beyond r = 0.75. An alter-
native approach in viewing this result is to keep the
2
=1-
(
A
* {J RD [j(kP, -YJl (_Y,)] dpf} I~ I I 1

} 3.9
{J (LO(Pf)[J( 2FS) Y1(2Fs)
3.8
f RD LO (pf~pfdpf 3.7 0 0
3.6
(13) ZIL 3.5 _
11
3.4-
can be interpreted as a detection loss for a hetero- 3.3-
dyne receiver in complete analogy with IOD in the case 3.2_\
of direct detection. In deriving Eq. (13), we have 3.1_
used the explicit expression for PLO. With the signal A3.0I
power defined by Eq. (12), all the standard equations 0'.0 0.1 0.2 0.3 0.4 0.5
dealing with the SNR (power) in a heterodyne receiv- OBSCURATIONRATIO,Y
er, which were derived assuming uniform illumina- Fig. 3. The introduction of a central obscuration into the aper-
tion by the signal and local oscillator, become valid ture modifies the Airy pattern in the plane of the detector. This
for more general distributions. It should be noted curve relates the Airy spot radius RA to the antenna F number Fs
that the detection efficiency 77H is dependent only on and the obscuration ratio -y.

October 1974 / Vol. 13, No. 10 / APPLIED OPTICS 2399

 pected from the first outer ring of the signal Airy dis-
tribution. Thus, the detector efficiency decreases
only slowly in this range. A comparison of the opti-
mum uniform LO and optimum Gaussian situations
a (corresponding to the peaks in Figs. 4 and 5) is made
0
XZ in Fig. 6 where the minimum detection loss is plotted
vs the obscuration ratio y. The curves are labeled by
the type (uniform or Gaussian) of local oscillator dis-
z tribution function assumed. It is clear from the fig-
0
ure that the use of an optimized Gaussian distribu-
tion leads to smaller detection losses. This is to be
0
expected since the uniform distribution is a imiting
XI~ case where the Gaussian spot size co is allowed to be-
come large compared to the detector dimensions.

0.75 1.50
r=RD/RA
Fig. 4. Heterodyne (or homodyne) detection loss for a uniform
local oscillator distribution as a function of the obscuration ratio y
and the ratio of the detector radius to the Airy spot radius. a
U,
a
In

1
(n
C)

system noise constant by fixing the local oscillator 0

power PLO actually incident on the detector. As we
increase the detector diameter to capture the outer 0
z
I~
portions of the Airy disk (where the signal is much 0

reduced in intensity), then, to, hold PLO constant,

we must also reduce the amplitude of the local oscil-
lator wave, thereby lessening the contribution of the
more intense central portion of the Airy disk to the
over-all IF power. Thus, under the constraint of
constant LO power incident on the detector, the IF
signal begins to decrease beyond r = 0.75. r= RD/RA

E. Gaussian Local Oscillator Distribution Fig. 5. Heterodyne (or homodyne) detection loss for a Gaussian
local oscillator distribution. The Gaussian spot size is chosen to
If we now assume a Gaussian local oscillator distri- minimize the detection loss at each value of the obscuration ratio
bution, Eq. (13) becomes and the detector radius to Airy spot radius ratio r.

2
H()',rt) = - 2 U22 1
4-y - 14(-r 2 2)
exp 0 m 0.0
V l I I I
2
x exp( I t U2) [J () - YJ (U)] (16) 0 GAUSSIAN LO
-j
where again a = kRA/2FS and is defined by Eq. (15), z -1.0
0
r = RD/RA, and t = RAIW,where is the radius to the uJ
0
li/e2 intensity point in the Gaussian distribution. In Ia -2.01
Fig. 5, the detection efficiency is plotted against the
parameter r for several values of the obscuration UNFRLOX
ratio -y. For each combination of -yand r, a computer 0 -3.01
0
tested for the optimum Airy spot-to-Gaussian spot U-)
size ratio t. This optimum value was then used to I-
zUj

obtain the value of H plotted in the figure. Figure 5 U- -4.01

n
clearly indicates that optimum performance is ob- I I_

tained when the detector radius matches the Airy

-. 9.n L
spot radius (r 1). The optimum value for t in the -<.v

c 0.0 0.1 0.2 0.3

0.4 0.5
vicinity of r 1 varies only slightly as a function of y,
OBSCURATIONRATIO, Y
ranging from t = 2.2 for y = 0 to t = 2.1 for y = 0.5. Fig. 6. The minimum heterodyne (or homodyne) detection loss
The greatly reduced intensity of the Gaussian local attainable with uniform or Gaussian illumination by the LO as a
oscillator distribution in the region beyond r = 1 sup- function of the obscuration ratio -y. The curves correspond to the
presses the partial cancellation of the IF signal ex- peaks in Figs. 4 and 5.

2400 APPLIED OPTICS / Vol. 13, No. 10 / October 1974

F. Matched Local Oscillator Airy spot radius ratio of approximately 0.74.6 For
Gaussian illumination, the latter ratio is slightly less
We now consider a highly specialized case where
than 1.0 with an optimum Airy radius-to-Gaussian
the local oscillator distribution matches the signal spot radius (RA/w) of approximately 2.2. The num-
distribution except for a constant amplitude factor. bers just quoted vary only slightly as the obscuration
Such a situation might be realized physically by en- ratio y is varied within the 0-0.5 range. The value of
larging the local oscillator beam and then focusing it the q1Hfor these optimized cases is plotted against the
through a centrally obscured lens having a clear aper- obscuration ratio in Fig. 6. Slightly better perfor-
ture small compared to the enlarged beamwidth. mance can be obtained if the local oscillator field dis-
The F number of the focusing optics and the size of tribution matches the signal distribution. The ex-
the central obscuration relative to the clear aperture pression for the receiver gain is then identical to the
could then be chosen to match the corresponding expression obtained for direct detection.
quantities in the optical train of the signal beam. In the present article, we have considered the ef-
Equation (13) would then reduce to fects of obscurations, detector dimensions, and (for
heterodyne or homodyne systems) local oscillator dis-
(kRDS) '_2 duF
[J(u) yJj(yu)]2_, tributions on the receiver gain. Changes in the gain
21 s(I
f) 1 - y2 due to pointing errors or system misalignments have
(17) not been discussed, and it should be mentioned that
systems in which the local oscillator is focused onto
which is identical to the expression obtained for the the detector require more stringent alignment toler-
direct detection case [Eq. (4)] plotted in Fig. 2. The ances.4 Some general considerations regarding the
latter figure implies that the detection loss could be field of view of a heterodyne receiver have been dis-
reduced below that obtainable with a Gaussian local cussed by Siegman5 who concludes that the receiver
oscillator distribution by allowing the outer rings of antenna pattern depends not only on the over-all an-
the signal Airy distribution to fall on the detector. tenna dimensions, but also on the complex local oscil-
Because the field distributions are matched, the lator amplitude falling on the photodetector. We
outer rings contribute positively to the over-all IF hope to consider off-axis effects of this type in a fut-
power, whereas, in the situations considered pre- ure article.
viously, they lead to partial cancellation of the IF sig- In conclusion, the description of transmitting and
nal. receiving optical antennas in terms of the power gain
greatly simplifies the calculation of the SNR in a
Ill. Summary laser radar or communication system. The effect of
The gain of an actual receiver is degraded from the obscurations, local oscillator distributions, and detec-
ideal value of 47rA/X2 (see Fig. 7 in Part 1). The re- tor dimensions can be rapidly accounted for by utiliz-
duction in dB due to blockage of the incoming light ing the curves presented in Parts 1 and 2 of this
by the central obscuration is plotted in Fig. 1, while paper. The numerical values for the trannansmitter
Figs. 2 and 4-6 give the dB loss at the detector for di- and receiver gain obtained from the curves are sub-
rect detection and heterodyne detection, respective- stituted into the microwave range equation
ly. Ps = PTGTGRTTTRTA(X/4rR)', (18)
For a direct detection system, the total receiver
gain can be obtained by summing the separate con- where Ps is the effective signal power at the detector,
tributions from Fig. 7 in Part 1 of this paper and PT is the output power of the laser itself prior to en-
Figs. 1 and 2 in Part 2. Of course, the effect of lossy tering any external optics, GT is the transmitter gain
optical elements or nonunity detector gains and evaluated in Part 1 of this paper, GR is the receiver
quantum efficiencies must be separately accounted gain evaluated in Part 2, TT and TR account for re-
for.' flective and absorptive losses in the transmitter and
In heterodyne (or homodyne) detection systems, receiver optics, TA is the atmospheric transmission,
2 is the so-called space loss representing
the effective receiver gain depends on the electric and (X/47rR)
field distributions of the local oscillator and signal the effect of the target distance R on the signal
beams across the detector. An analytic expression strength.
for the detection loss as a function of a generalized References
local oscillator distribution was derived [Eq. (13)].
Numerical values were obtained for some important 1. B. J. Klein and J. J. Degnan, Appl. Opt. 13, 2134 (1974).
(1) uniform LO; (2) Gaussian LO; and 2. M. Born and E. Wolf, Principles of Optics (Pergamon, Oxford,
special cases:
1965), p. 416.
(3) a matched LO. For uniform illumination by the 3. See, for example, H. Melchior, M. B. Fisher, and F. R. Arams,
local oscillator, the total receiver gain is obtained by Proc. IEEE, 58, 1466 (1970).
summing the contributions from Fig. 7 in Part 1 4. B. J. Peyton, A. J. DiNardo, G. M. Kanischan, F. R. Arams, R.
2 ) and Figs. 1 and 4 in the present article. For
(47rA/X A. Lange, and E. W. Sard, IEEE J. Quantum Electron QE-8,
Gaussian or matched local oscillators, Fig. 4 is re- 225 (1972).
placed by Fig. 5 and Fig. 2, respectively. With uni- 5. A. E. Siegman, Proc. IEEE 54, 1350 (1966).
form illumination by the local oscillator, optimum 6. The optimum value of r ranges from 0.72 for y = 0 to 0.76 for y
performance is obtained with a detector radius-to- = 0.5.

October 1974 / Vol. 13, No. 10 / APPLIED OPTICS 2401

View publication stats