Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

THE EFFECT OF RANDOM AMPLITUDE AND PHASE ERRORS

ON PHASED ARRAYS PERFORMANCE

A. Sebak, L. Shafai and H. Moheb

Department of Electrical and Computer Engineering


University of Manitoba
Winnipeg, Manitoba, Canada R3T 2N2

A. Ittipiboon
Communication Research Centre
Ottawa, Ontario, Canada K2H 8S2

ABSTRACT
In this paper, the effect of amplitude and phase errors as well as amplifiers and
phase shifters failures on a phased array performance is investigated. It is found that
random phase errors and phase shifters failures are dominant factors, they influence the
peak sidelobe level, pointing angle accuracy, directive power and the half power
beamwidth.

I. INTRODUcnON
The peak sidelobe level, beam width and beam direction of a phased array
antenna are important parameters for MLS operation. Low sidelobes and narrow
beamwidth at different scan angles are normally achieved by an amplitude taper across
the array and proper phase shifts. Due to random errors as a result of manufacturing,
material, environmental and other tolerances, parasitic or grating lobes may become
significant Random errors are important for an antenna design and influence the con-
struction tolerances which determine the overall performance and the final cost.
Another kind of error is due to material characteristics, operating frequency and tem-
perature variations.
The effect of random errors in the performance of an array has been studied by
many investigators. The first work in this subject was carried out by Ruze [1] using a
statistical approach applied to a large number of identical antennas. Bao [2-3] has
extended Ruze' s work and obtained expressions to study the effect of phase errors on
the average power pattern taking into consideration the influence of illumination taper
and correlation interval on the radiation characteristic. Ruze' s work has also been
extended by Schanda [4] to treat both amplitude and phase errors with the restriction
that they are unrelated.
In this paper, a numerical simulation has been developed to investigate the effect
of random amplitude and phase errors on the array radiation pattern. In addition, we
are considering the effect of one or more radiating elements failures. Few numerical
results, to show the effect of these deficients on the radiation pattern of a phased array
antenna, are presented.
ANTEM'90 391

Authorized licensed use limited to: Cornell University Library. Downloaded on August 17,2020 at 01:54:59 UTC from IEEE Xplore. Restrictions apply.
n. NUMERICAL SIMULATION
For an array of N radiating elements, the far field E (9) is given by
E(9) =Ee(9) . f a;eU'III) e Uk XI sin 9) (1)
;=1
where
k = 27t1'A.
ai = amplitude of the excitation coefficient of the i th element
Xi =coordinate of the i the element
'lfi = phase of the excitation coefficient of the i th element
Ee (9) = element pattern in the 9 direction.
A computer program has been developed to simulate the effect of random errors
and elements failures on the performance of MLS antennas in terms of main beam
power loss, shape, pointing accuracy and width as well as the peak sidelobe level. A
linear array with 56 elements and -30 dB Taylor distribJltion is assumed. The interele-
ment spacing d is 0.599 'A. at 5.06 GHz and the scan angle is 90 , The corresponding
half-power beam width (HPBW) is about 2.030 • The effect of amplifiers failures on
the radiation pattern is studied for two cases:
1) complete failure, i.e. a failed element receives no power, and
2) 50% failure, i.e. a failed element receives one half of its assigned power.
Cases (1) and (2) correspond, respectively, to using a single module for each element
or using a hybrid combined module that contains two single modules. The effect of
phase elTors due to phase shifter failures is also carried out by considering both bit and
stick to state failures of each phase shifter. The number of radiating elements, number
of elements per subarray (if any), interelement spacing, operating frequency, design
sidelobe level, number of bits in a digital phase shifter and the scan angle are input
parameters for the errorfree case. On the other hand, the rms values for amplitude and
phase elTors, the number of failed amplifiers, number of failed phase shifters and the
type of failure as well as the steering algorithm are variable input data to the simula-
tion program.

A. Element Failure:

(1) Amplifier Failure:


Initially the number of failures is specified as an input parameter. For each failed
element, the program determines the type of failure, i.e. complete failure or 50%
failure, and then computes the corresponding radiation pattern. The loss in the power
normalization factor, maximum sidelobe level and the changes in the pointing direction
~90 and in the 3 dB power beamwidth IlHPBW for some cases are summarized in
Table 1. The change in the HPBW, from one case to another, is small and within the
the resolution of ± 0.002°. It is clear that the most sever results occur when the
failures are 100%. It is also found that the most sever results occur when central ele-
ments are failed. Note that neither phase random errors nor amplitude random elTors
are included in these computations.

392

Authorized licensed use limited to: Cornell University Library. Downloaded on August 17,2020 at 01:54:59 UTC from IEEE Xplore. Restrictions apply.
(2) Phase Shifter Failure:
Two types of failure are considered. bit failure and stick to state failure.

(a) Bit failure:


In the simulation program. initially the number of failed phase shifters. i.e. the
probability of failure, is specified and the location of the failed elements are randomly
determined assuming their locations are uniformly distributed. For each failed phase
shifter, the simulation program selects randomly which bit fails. i.e. lout 4 in a 4-bit
digital phase shifter or lout of 5 in a 5-bit digital phase shifter. One of two model is
then used to simulate such a failure. The first model. model I, corresponds to the
worst case (which may be unrealistic). The phase of the failed element in this case is
set to zero keeping the power output unaffected. In the second model, model II, the
phase of a failed element is set to the nearest phase that can be accomplished using the
other un failed bits.
The loss in the main beam power, peak side10be level and the changes in the
pointing direction and 3 dB beamwidth are summarized in Table 2.

(b) Stick to state failure:


A 4-bit digital phase shifter has 16 states (0. 22.5. 45 ..... 337.5 degree) which are
controlled by the steering algorithm to yield a stair-step phase gradient across the array
to correspond to a given scan angle. By a stick to state failure we mean that when the
main beam is steered from direction 8) to direction 82, a failed phase shifter does not
respond to the new command from the steering algorithm. Thus the phase of the sig-
nal stucks to a previous value. A proper simulation of such a type of failure requires a
complete definition of the steering algorithm associated with MLS antenna. The
development of such an algorithm is not the subject of this study. Instead, we
assumed that the coverage section (-40 0 ,40°) is covered by 11 beams (-40, -30, -24,
-16, -8. O. 8. 16. 24, 30. 40 degrees). In the simulation run. the probability of failure
is specified and accordingly the number of failed phase shifters is determined. Next,
for each failed phase shifter, the program determines randomly the beam position j,
lout of 11, to which a failed phase shifter is stuck to. The phase of each element.
except for the failed ones, is set to the required phase according to the steering algo-
rithm. On the other hand, the phase at a failed element is set to that corresponds to a
previous beam with index j. That beam corresponds to the previous phase of the
failed phase shifter.
The output of the simulation program is summarized in Table 3 for different
number of failed phase shifters. The quantization phase errors are not included in this
table.

B. Random Errors:
Random errors as a result of manufacturing, material and environmental toler-
ances. excitation errors or elements failures may deteriorate the radiation pattern of an
antenna. They will reduce the mean radiated power and deteriorate the shape. pointing
direction and width of the main beam. In this section. a simulation on the computer of
both amplitude and phase random errors will be given. Again, a 56-element linear
array with -30 dB Taylor distribution is assumed with d = 0.559/....
In the simulation program. the variances of the amplitude and phase errors are
input data. For each value(s). we calculate the corresponding radiation pattern.

393

Authorized licensed use limited to: Cornell University Library. Downloaded on August 17,2020 at 01:54:59 UTC from IEEE Xplore. Restrictions apply.
Initially the excitation distribution is detennined. Next two sets (56 each) of nonnal
random numbers are generated with zero means to be used to simulate the amplitude
and phase errors. Each of these two sets is then multiplied by the corresponding RMS
<J (i.e., <Ja or crill) to yield the amplitude and phase errors of each radiating element.
These random errors are then added to their respective excitation coefficients. The
radiation pattern for each case is then computed. The results are summarized in Table
4. It can be seen that sidelobe patterns deteriorate mush larger with the phase errors.
Also the peak sidelobe level increases with increasing the variance of the error. In all
these cases the peak sidelobe level does not exceed -20 dB.
The results at 300 scan angle are also given in Table 4.6 for phase only random
errors. Four-bit digital phase shifters are assumed. It can be seen that the phase quant-
ization error effect dominates for small variances of random phase errors. However
their effects increase with increasing the variance of the phase random error. The
effect of phase quantization errors can be minimized by using higher bit digital phase
shifters.

C. Mean Values:
Finally, the mean absolute values of the pointing error A9 0 and the variation of
the half-power beamwidth AHPBW corresponding to 5 different runs are given in Fig.
I, as functions of the rms phase error, for 90 = 0.. Each run corresponds to a seed
number for generating different random. nonnally distributed, phase errors with a
given rms value. The corresponding mean value of the directive power loss (not the
directive gain) is shown in Fig. 2.
It can be concluded from the results of this paper that a major effect of random
errors and element failures is to distort the sidelobe region of the radiation pattern and
have minor effect on the pointing accuracy.

REFERENCES
[1] J. Ruze, Antenna Tolerance Theory: A Review, Proc. IEEE, Vol. 54, pp. 633-640,
1966.
[2] V.T. Bao, Influence of correlation interval and illumination taper in antenna
tolerance theory, Proc. lEE, Vol. 116, pp.195-202. 1969.
[3] ..... , Nonconstant Correlation Interval in Antenna Tolerance Theory, IEEE Trans.
Ant. Propg., pp. 118-119, 1969.
[4] E. Schanda. The Effect of Random Amplitude and Phase Errors of Continuous
Apertures. IEEE Trans. Ant. Propg., pp. 471-473, 1967.

Acknowledgment
This work was supported by Transport Development Centre, Montreal and the
Communication Research Centre. Ottawa.

394

Authorized licensed use limited to: Cornell University Library. Downloaded on August 17,2020 at 01:54:59 UTC from IEEE Xplore. Restrictions apply.
Table I

No. or % Power Scan Norm. Factor SLL 6. HPBW


-
6.9~
Failures loss ane;le dB dB deE. deg.
None 0 0 31.2892 ·29.3 0 0
2 100 0 30.9469 ·23.39 0 0
4 100 0 30.691 ·23.39 0 0.022
6 100 0 30.536 ·22.8 0 0.07
2 50 0 31.1197 ·27.07 0 0
4 50 0 30.9953 -25.98 0 0.011
6 SO 0 30.896 ·24.52 0 0.023
8 SO 0 30.8508 ·24.45 0 0.06
16 SO 0 30.1113 -20.4 0 0.02
None 0 30 31.2338 -22.646 0 0
2 100 30 30.8899 ·22.096 ·0.001 ·0.002
4 100 30 30.6328 ·20.366 0.000 0.032
6 100 30 30.0845 -20.28 0.002 ·0.052
6 50 30 30.6781 -22.8 0.0 0.026

Table 2

No. or Type or Scan Norm. Factor SLL to8~ to HPUW--


Failures model angle dO dO deg. deg.
6 I 30 30.42 ·17.52 0.033 0.078
6 I 30 30.857 ·21.305 0.004 0.056
6 I 30 30.256 -15.61 0.018 ·0.104
6 I 30 30.1886 .16.95 ·0.011 -0.11
6 I 30 30.3886 ·16.95 -0.046 0.08
6 II 30 31.1496 -20.62 0.0 ·0.009
6 II 30 31.Z148 ·22.225 -0.006 0.004
6 II 30 31.2307 .23.263 ·0.006 0.004
6 II 30 31.2338 -22.646 0.0 0.0
6 II 30 31.2338 -19.875 0.0 0.0

Table 3.

No. or Seed Scan Norm. Factor SLL 60~ 6. IIPIlW


Failures index angle dB dB deg. deg.
6 1 30 31.0148 -24.505 -0.028 0.06
6 2 30 30.5299 -19.73 0.036 0.016
6 3 30 30.5153 ·16.61 -0.016 -0.028
6 4 30 30.351 -16.16 -0.053 ·0.014

395

Authorized licensed use limited to: Cornell University Library. Downloaded on August 17,2020 at 01:54:59 UTC from IEEE Xplore. Restrictions apply.
Table 4

0'., C11V Scan Power Loss SLL A90 A IIPBW


% deg. angle dO dO del!. deg.
0 5 0 0.0264 -27.11 -0.008 0.0()f··-
0 10 0 0.1057 -24.08 -0.016 0.013
0 15 0 0.238 -21.65 -0.024 0.019
0 20 0 0.4237 -19.68 -0.032 0.024
5 0 0 0.0056 -29.83 O. O• ..-
10 0 0 0.0113 -28.64 O. O.
15 0 0 0.017 -26.97 O. O.
20 0 0 0.0226 -24.84 O. O.
10 10 0 0.1242 -24.805 0.004 0.004
20 15 0 0.2693 -20.54 0.0105 D.OOS
0 0 30 0.0259 -24.73 0.002 -0.001
0 5 30 0.0506 -21.33 0.0 0.0
0 10 30 0.1586 -19.95 0.002 -0.001
0 15 30 0.3247 -18.526 0.002 -0.001
0 20 30 0.5496 -17.136 0.00 ·0.D02

5.00...---.------,-----r----r--"'7"'T- 0.10

-----~
4.00 -l------l---+---+------ut=--j.L---+o.o8 0.00

~J.OO ~_..---j~--4-____:#--+-----+0.06 • -0.20


~
o
~
In ~
~
~
aJ
~

'"
I '0
o 2.00 +--__l--_~~--+--+---+0.04 <I
a.
.,.-0.40
9 "-
a.
1.00 -I---.......,61L-----4---+--....L---t0.02 -0.150
-l!9
... t.HP8W

8 12
-0.80
o . 8 12 :6 20

a." deg a", deq

Fig. 1: Mean of the absolute values Fig. 2: Mean of the power loss vs
of the pointing errors vs rms phase errors.
rms phase errors.

396

Authorized licensed use limited to: Cornell University Library. Downloaded on August 17,2020 at 01:54:59 UTC from IEEE Xplore. Restrictions apply.

You might also like