- Email: [email protected]
Infrared Horizon Sensors for Attitude Determination
INFRARED HORIZON SENSORS FOR ATTITIJDE DETERMINATION by ABSTRACT Earth orbiting satellites usually require some means of determining their attitude with respect to the earth, either for control or monitoring purposes. Infrared sensors which detect the sharp thermal discontinuity at the earth-space horizon have been developed for this purpose and are being used in a number of American satellite systems. They may be classed under three general types, namely: conical scan sensors, horizon edge trackers, and radiometric balance sensors. The principles of operation of these general types, and examples of flight hardware are briefly described. The relative advantages and dis-advantages of the different types with respect to system application is discussed. Means for minimizing errors caused by cloud cover and horizon gradients are discussed and some data on flight performance given. I.
INTRODUCTION It is often necessary to establish the attitude of orbiting and interplanetary vehicles with respect to some external coordinate system. This is usually accomplished by sensing the position of some combination of the sun, stars, or planets. If the object is at a distance such that it appears as a point, the sensor is called a star tracker, while if the planet is close enough so that a definite disc is perceptible, it is usually called a horizon sensor. In the latter case, the direction toward the center of the disc is desired. This paper will consider the problem of determining the horizon position by means of the thermal self-emission of the planets. In order to sense the position of a planet by its thermal self-emission reliably, and with some degree of accuracy, three properties of the planet are desirable. First, it must be warm enough so that a detectable thermal discontinuity exists between the horizon and the space background. Second, for a
Robert W. Astheimer Barnes Engineering Company Stamford, Connecticut (U.S.A.) good accuracy, this discontinuity or gradient should be very sharp; and lastly, system problems are simplified if the radiance is reasonably uniform over the surface of the planet. The last condition can sometimes be attained fairly satisfactorily by selecting the spectral region of operation. A plot of the apparent temperature distribution along a diametral scan ofdE Earth, the Moon, Mars, and Venus is shown in Figure 1 (Ref. 1). A logarithmic radiance scale for the total thermal emission is also shown since it is the radiance rather than the temperature which produces the received signal. E?ch object is assumed at half phase with the terminator bisecting the disc in order to illustrate the difference in si~ nal level between the dark and sun illuminated sides. This figure is a great over-simplification but is u::teful for comparing the gross radiant characteristics of the four objects. It must be pointed out that these characteristics may be very different in restricted spectral regions, particularly in the absorption bands of atmospheric constituents; however, restricting the spectral region can only decrease the apparent radiance, although it may also reduce the dynamic range of signal that must be accepted. A minimum detectable radiance level is shown at 0.1 milliwatts/cm 2 -ster assuming an immersed thermistor detector with a 2-inch diameter collecting aperture, 20% optical efficiency, a 1 0 x 1 0 field of view, a 250 cps bandwidth, and a signal 10 times rms noise. These are realistic parameters for a system as will be shown. It is seen that the horizon of all planets is readily detectable with such a system. We conclude then that all these planets have enough thermal emission to make infrared horizon sensors of reasonable aperture practical. The accuracy with which the horizon can be determined will depend upon the steepness of the horizon gradient \mich
381
100.0 50.0
400' MOON
10.0
~ t;;
:::1
III Il'-
EARTH 7
I
1.0
~
05
cti I 50' ~
is
lr
0
I
~
MOONJ
...J
UJ I-
'CLOUD
I I
UJ
<..> Z et
200'~
r"'
I
~<..>
250'
-
VENUS
...J::lE
300'
,
5.0
...IN
350'
I 20'
z
f:!
I::E lr
g
DARK SIDE
:~
LIGHT SIDE
0.1
MINIMUM DETECTABLE RADIANCE
Figure I TEMPERATURE PROFILES OF EARTH, MOON, MARS AND VENUS
1.0
I f~RDC STD
I 0.8
.............. 0.6 ~o:
~~
10.75-1I.75p wiz CLEAR
------rI
i ~ 0.2 -.......--
HIGH CLOUD
I I
1075-11.751''.
I
0
z
I
\
'j-ARDC STD
OA
...I N
~12
1-- 10: I ........ 'i~ I 1 1 L. ____ -~ ---- , \
14-16p
\
1 1
\
\ \
I
I 0.5
,
. oI. I . ., + 0.5
+1.0
, +1.5'
Figure 2 EARTH HORIZON PROFILES IN VARIOUS SPECTRAL REGIONS
A
C
CONICAL SCAN
RADIOMETRIC BALANCE
B EDGE TRACKER Figure 3 BASIC SENSING TECHNIQUES
382
is a function of the density and composition of the atmosphere. The moon, possessing no significant atmosphere, has an infinite gradient; and the ultimate accuracy is limited only by terrain irregularities. In a practical system, however, the detector noise would probably be a limiting factor because of the low temperature and emission from the dark edge. The uncertainty caused by the fuzziness of the horizon discontinuity of a planet with an atmosphere is a function of altitude. At very low altitudes, the error could be quite large, while at high altitudes, the entire subtense of the atmosphere may become negligible. The earth is the only planet on which horizon sensors have been used to date, and considerable data has now been accumulated. The accuracy of these systems has generally been limited by the size of the detector field of view rather than the horizon gradient. A number of theoretical studies (Ref. 2) have been made of the gradient to be expected in different spectral regions, some results of which are shown in Fig. 2. The spectral regions of greatest interest are the C02 absorption band at 14 - 16 microns. the water vapor bands extending from 20 - 35 microns (and beyond), and the so-called transparent atmospheric "window" between 8 - 12 microns. It will be seen that the horizon gradient is quite sharp in the atmospheric window, but the range of signal level is greatest here because of cloud cover. Operation in the 14 - 16 micron CO~ absorption band masks clouds and w~ll present a very uniform signal, but the gradient will be more gradual, extending over almost 1° from an altitude of 400 miles. Accuracy at this altitude is probably limited to about .05 degrees by the indistinctness of the horizon itself • Much less information is available regarding Mars and Venus. The atmosphere of Mars is fairly tenuous with few clouds and will probably not diffuse the horizon significantly even at low altitudes. Venus has a very dense atmosphere of unknown depth; however, the top of the cloud deck appears to be quite distinct and of uniform temp-
erature. Therefore, a sharp gradient should exist in a spectral region where the clouds are opaque and the atmosphere above transparent. One of the most troublesome problems in horizon sensing is caused by large variations in temperature or radiance over the surface. Some difficulties of this nature were experienced with early earth horizon sensors because of unexpectedly cold cloud tops associated with large storm areas. Spectral filtering and modifications of the electronic processing systems were required to achieve reliable operation for all weather conditions. The situation is much worse with the moon as indicated in Figure 1. The radiant emission from the sun-baked side is 100 times greater than that from the dark side, and there are additional small "hot" and "cold" areas caused by sun-illuminated crater edges and shadows. These variations in signal strength may not only produce errors due to saturation and time constant effects but could cause the system to confuse the horizon with the terminator, since the terminator discontinuity is 50 times greater than the dark edge horizon discontinuity. The planet Venus seems to present the easiest object for infrared horizon sensing, but the sharpness of the horizon gradient in various spectral regions, which depends upon the atmospheric structure, is still unknown. Reflected solar radiation will be super-imposed on the thermal self-emission of the sun-illuminated portions of all the planets. This is in general an unwanted disturbance and can be substantially removed by filtering out wavelengths shorter than 8 microns. Such filtering also greatly reduces the signal produced from direct viewing of the sun which could saturate the system or even damage the detector. 11.
DETECTOR CONSIDERATIONS It has been shown that the range of apparent planetary temperatures to be sensed extends from l20 0 K to 380 0 K. Detectors sensitive to such therma1radiation must respond to long wavelengths from 8 to 40 microns. Photoconductive detectors highly sensitive in this
383
Thermal detectors which depend upon a resistance change for signal generation, such as thermistor or metal bo10meters, are more subject to internal temperature effects than thermocouple types. The reason for this is that in order to sense the resistance change a bias current must be used to convert it to a voltage signal. The desired signal thus appears as a very small change superimposed on the much larger bias voltage. Bridge arrangements of detector elements may be used to buck out the large bias voltage, but it is practically impossible to maintain the degree of balance necessary over a wide ambient temperature range. Optical modulation, of course, solves the problem since the d.c. bias voltage is filtered out, therefore these detectors are seldom used without optical modulation. Thermocouple detectors (Ref. 5) do not require biasing and are thus free of this problem. The only voltage appearing is the thermoelectric EMF produced by the radiant heating of the junction, and this is not superimposed on a biasing voltage. Unmodulated optical systems are therefore feasible with thermocouple or thermopile detectors. However, even with these detectors, care must be taken in the design to eliminate spurious signals from internal temperature gradients and self-emission of optical parts. Also thermocouple detectors sufficiently rugged for space applications are slower and less sensitive than thermistor detectors. The elimination of optical chopping and scanning mechanisms is highly desirable for space missions where very long life and low power are necessary. The spurious signals or drifts, inherent in unmodulated systems, make them relatively less accurate than optically scanned systems. Therefore, thermopile detectors are most suitable for low accuracy, long life applications such as antenna pointing.
region have been developed such as copper or zinc doped germanium (Ref. 3) but these all must be cooled to temperatures in the neighborhood of liquid helium. Reliable cooling to these temperatures for long periods of time is difficult in space systems and the most suitable detectors are the thermal types, particularly thermistor and metal bo10meters, and thermocouples. Of these, the thermistor bolometer (Ref. 4) is the most sensitive and has found greatest use to date. These are usually immersed on a germanium lens for increased detectivity. The thermal detectors operate by virtue of the heating effect when incident radiation is absorbed on the sensitive element. The very small temperature change resulting effects some physical parameter such as the resistance which can be read out electrically. An inherent difficulty in the use of such detectors is the identification or separation of the temperature change caused by the desired radiation from the very much larger ambient temperature variations. For example, in a typical horizon sensor, the change in radiation when the field of view scans from space onto the earth increases the detector element temperature about .001°C. The sensor must be designed so that this temperature differential may be detected in spite of ambient temperature variations 10,000 to 100,000 times greater. The most common way of separating the desired temperature change from unwanted internal effects is by opticalmechanical scanning or chopping. This modulates the external radiation signal at a relatively high frequency, and spurious signals produced by the slowly changing ambient temperature can be removed by capacitance coupling to the electronic amplifier. It is important to notice that such modulation must be done on the radiation signal before detection and not on the electrical output. Electronic chopping will modulate both external and internal signals and is of no value in this respect.
Ill.
BASIC SENSING TECHNIQUES The infrared emission characteristics of the planets and suitable detectors for
384
sensing this radiation have been described. Such detectors must be employed in an optical arrangement to provide information from which the coordinates of the center of the planetary disc can be determined to the degree of accuracy desired. A complicating factor is that for most planetary objects the radiant emission will be highly nonuniform over the surface, and for high accuracy, the system must be independent of the radiance level. A number of systems have been developed for this purpose. These can all be shown to be versions of three general categories which we shall designate as follows: (1) Wide Angle Scanning Systems. (2) Edge Tracking Systems. (3) Radiometric Balance Systems. The detector field of view and scan pattern associated with these techniques are shown in Figure 3. In the conical scan, which is typical of the wide angle scanning systems, the instantaneous detector field is relatively small and is caused to scan through a large cone whose apex angle may be as much as 180 0 , although an apex angle between 50 0 to 120 0 is more usual. The detector signal generated will be an approximately rectangular waveshape repetitive at the scan frequency. This waveshape is usually limited in some fashion to eliminate amplitude dependence, and then position information is derived by a phase or pulse width comparison technique. Two sensors are used to generate pitch and roll attitude information. Most horizon sensor systems flown to date have been of the conical scan type because it possesses a number of very significant advantages. It has excellent acquisition capability, attained without additional search modes because of the wide scan angle. The attitude information is derived from time characteristics of an amplitude limited waveshape and is therefore insensitive to radiance variations over the surface of the planet, which as we have seen can be very large in some cases. Another advantage is that, because of the wide field scanned it is
certain that some portion of the scan will leave the planet and view space. This provides an absolute zero radiance level against which any portion of any planet ~ give a positive contrast. Use can be made of this reference in setting limiting levels so as to prevent the system from confusing radiance discontinuities on the planetary surface, such as may be produced by the terminator or clouds, with the true horizon. This can be a serious problem with edge tracking systems. The primary disadvantage of the conical scan sensor is the need for high speed rotating elements which present life and lubrication difficulties in spaceborne applications. Some sensors have been developed which combine the function of the two scanners by precessing the basic conical scan to produce a rosette or epicyclic pattern (Ref. 6). These may be considered essentially versions of the conical scan system and the same general remarks apply. The basic concept of the edge tracking system is shown in Figure 3B. A small detector field of view is caused to lock onto the radiance discontinuity at the edge of the planetary disc. This is usually accomplished by oscillating the detector field through a small angle normal to the horizon edge and moving the entire sensor or the oscillating field until a balanced waveshape is obtained. Multiple sensor heads may be used, in which case at least three are necessary, or a single oscillating field may be caused to trace around the edge. In either case, the horizon position is determined by reading out the position of the center of the oscillating field with respect to spacecraft coordinates, i.e., angles 9 1 , 92' 93' and 94' in Figure 3B. For equivalent optics and detectors, this type of system will have a better signal-to-noise ratio than the wide angle scanning types because of narrower electronic bandwidth. It leads to a much more complicated system however because separate search and edge tracking movements must be provided on each sensing head along with precision position readouts.
385
65'
Figure 4 TIROS HORIZON CROSSING INDICATOR
ORBITAL PLANE
SP1N DIRECTION
OP.0(":~~
~
..--p~ ",':: SENSOR FI ELD OF VIEW
FIELDS OF VIEWS OF SENSORS
SP1NAXIS
-_
- "-
"-M \...--- CAMERA
~'
"
FIELD
p"""
~'
FiCj1ure 5A
FiQur. 5 B
SPIN STABILIZED SATELLITE GEOMETRY
Figure 6
TELEMETERED HORIZON CROSSING INDICATOR RECORDS
386
A high price is paid for this narrow bandwidth and consequent greater sensitivity. Without the wide angle scan the previously mentioned space reference is lost, and there is no convenient way for the system to sense whether the signal is from the true horizon or some other discontinuity such as a cloud edge or terminator which can be much greater than the horizon signal. Various devices such as auxiliary detectors may be employed to prevent locking on false edges, but they further complicate the system. The edge tracking type of system would appear best suited for application where spectral filtering to an atmospheric absorption band can eliminate radiance variations over the planetary disc. An example for earth horizon sensing would be to filter to the narrow C02 absorption band at 15 microns. The increased detectivity of this type system would compensate for the large reduction in signal caused by the spectral filtering. The former two systems employ optical modulation by mechanical means. It should be pointed out, however, that a stationary array of detectors could be used whose outputs are electronically sampled. For example, in a conical scan system, instead of mechanically causing a small detector field to scan over a wide circle, a stationary array of detector elements can be placed to view the same circle, and the array sequentially sampled electronically. Thermopile detectors are particularly well suited for this technique, and a system of this type is being developed. The radiometric balance type sensor is a non-scanning system and operates by comparing the radiation received from opposite portions of the planet. Very wide fields of view are used to achieve acquisition and also to average radiance variations over the surface. A typical arrangement of detector fields is shown in Figure 3C. Four wide angle stationary fields are employed designated a, b, c, and d; and attitude information is obtained by the difference in radiant power received from opposite fields, i.e., Pc - Pa and Pb - Pd. This is obviously only
correct if the planet is uniformly radiant, and therefore this system is primarily suited for use with planets of uniform radiance such as Venus or where only moderately accurate pointing information is necessary. The great virtue of this system is its extreme simplicity and consequent high reliability. By using thermopile detectors no moving parts are necessary and very long life can be achieved. Nonuniform radiance effects can sometimes be minimized by spectral filtering. In the remainder of this paper a number of operational sensors employing these principles, which have been developed and used in space missions are described. IV. "TIROS"HORIZON CROSSING INDICATORS In a spin stabilized satellite the motion of the vehicle itself can be used to generate a wide angle conical scan. This is the technique used in the TIROS weather satellite series, and permits a very simple sensor design. A photograph of a TIROS sensor is shown in Figure 4. It consists of a sma:l infrared telescope and germanium immersed thermistor detector having a field of view of 1 degree square with a transistor amplifier. The aperture of the objective lens of the telescope is 16 mm. The entire sensor is approximately 1 X I X 12 inches. As the rotation of the satellite causes the small field of view of the sensor to cross the horizon a steep positive or negative pulse is developed depending upon the direction. These horizon crossing pulses are used for a variety of purposes. The period between successive horizon crossings in the same direction (space to earth) gives the spin rate of the satellite. The ratio of the time between on and off horizon crossings to the spin period is a measure of the inclination of the spin axis to the local vertical. The horizon lion" pulse has been used to trigger other instruments such as cameras at appropriate times. For example in a II wheel" type mode where the spin axis is normal to the orbital plane as shown in
387
DETECTOR FIELD
OF VIEW -
__
EARTH
Figure 7 CONICAL SCAN GEOMETRY
...
(PROJECT MERCURY)
~557INCHES
I-=OTOR "r"r>tH-SCANNING PRISM
Figure 8
\
~\
J
CROSS SECTIONAL VIEW OF MERCURY SCANNER
. .\
.J
Figure 9
...
CONICAL SCAN SENSOR
388
Figure Sa, if a camera and horizon crossing indicator are both pointed radially but displaced angularly by half the earth subtense (
v.
''MERCURY'' CONICAL SCAN SENSOR
For an attitude controlled vehicle the scan motion must be accomplished internally. A series of very successful sensors have been developed employing a conical scanning principle wherein the projected image of a thermistor detector is caused to scan over a wide circle, or cone in space. Typical of these is the scanner used on the Project Mercury vehicle, the first American manned spacecraft. The scan geometry is shown in Fig.7. The instantaneous field of view of the detector is 2° X 8° and is caused to scan over a wide circle with a cone apex angle of 110° at 30 revolutions per second. A rectangular waveshape of
this frequency is produced by the detector because of the temperature difference between the planet and the cold space background. Another square wave reference signal of the same frequency is generated internally from the rotating scanning mechanism. By phase comparison of the radiation signal with the internal reference the attitude with respect to the conical scan axis can be determined. A pair of such sensors displaced 90° with respect to each other as shown in Fig. 7 can give attitude information about two orthogonal axis, usually designated as pitch and roll. A cross sectional view of the scanner is shown in Fig. 8. The conical scan is produced by a germanium prism mounted on a hollow barrel which is driven through gearing by a small motor. The internal phase reference signal is developed by a semicircular iron vane on the rotating assembly and a stationary magnetic pick up. The detector is a germanium immersed thermistor detector where the immersion lens also serves as the objective lens of the telescope. The unit is filled with dry nitrogen and hermetically sealed to prevent evaporation of lubricants. A germanium front window is used with a multilayer interference filter blocking all radiation of wavelength shorter than 8 microns. This greatly reduces direct and reflected solar radiation. A photograph of one of these sensors with the front and rear covers removed is shown in Fig. 9. In some applications it is inconvenient to mount the roll sensor on the front or aft ends of the space craft where it can have an unobstructed circular scan. An alternate mode of operation is shown in Fig. 10 where the two sensors are pointed 180° apart. In this mode attitude information about the roll axis is obtained by comparing the duration of the earth waveshapes generated by the two sensors while pitch can be determined from either one by the phase comparison method previously described. The pulse phase or pulse width processirg described requires a clean rectangular waveshape of constant amplitude. This is usually produced by
389
Fi9Ure 10 ALTERNATE MOOE OF CONICAL SCAN
COLD CLOUO
(
- - - - EARTH
A
-
DETECTOR NOISE Vn
l_
->-THRESHOLD B
SPACE
T-
HORIZON) CLOUD SIGNAL Vc OUTPUT WITH ~THRESHOLD
cg:~
CLOUD ERROR
A
OUTPUT WITH
~THRESHOLD
SET AT LEVEL B
Figure 11 CONICAL SCAN WAVESHAPE
-20
~
:I:
~-3 0 Vl
w w
I'
.-/
ii:
a:
Cl W 0
..
-" 0
..... -, _
/
~
' .... '... , -
,_ ..... '"
-..c;
/'
V/
PHOTOS
--- -
--....
--"
~
./" t-,
V/
-
,~/
"....,
'''SCANNER
-'
z
Cl
0
Vl
/
a:
j/
0
a: -' a: -' w 0 a:
0
r---...
/t"I-' '\k?" I
",~
~/
-10 1.38.00
f---<:: PHOTOS
1:39.00
/
/,
1.40:00
1:41:00
~
SCAN~
r""_'
1.42:00
~
\...V7
1:43.00
TIME AFTER LAUNCH HOURS: MINUTES: SECONDS
1.44:00
r-..
/,' "'\
""
L?
1:46:00
Figure 12 COMPARISON OF HORIZON SCANNER READOUT WITH PERISCOPE PHOTOS.
390
an on-off threshold circuit as shown in Fig. 11 which gives zero volts if the input signal is less than the threshold and a fixed voltage if it is greater. It is necessary that this threshold be established well above the detector noise level during the space scan but still below the minimum earth signal that may be expected. It is evident from the figure that if a cold cloud causes the earth signal to drop below the threshold, a "gap" will be produced in the processing waveshape which will cause serious errors in the attitude indication. In the first horizon sensors of this type such errors did occur because the apparent radiation temperature of some cloud tops particularly those associated with large storm areas was considerably colder than expected. This situation was corrected by improving the sensitivity and lowering the threshold level. The apparent radiation temperature of the earth depends upon the meteorological conditions at the point viewed and the spectral region in which the observation is made. In spectral regions where atmospheric constituents are absorbant the surface radiation will not be received and the signal will consist of emission from the absorbing constituents primarily water vapor and carbon dioxide, which will in general be colder than the surface. In atmospheric windows surface radiation will be received unless there is cloud cover in which case the apparent radiation temperature will be that of the cloud tops. The greatest variation in signal will therefore occur in the atmospheric windows, primarily the 8 12 micron region. Flight data from horizon scanners and radiometers show radiation temperatures ranging between l70 0 K and 300 0 K in the 8 - 12 micron window, which produces a variation of ten to one in detector signal, Such large variations in signal level complicate the electronic processing and it is desirable to reduce the dynamic range of signal as much as possible. This can be done by spectral filtering to
eliminate atmospheric windows. The 14.5 - 15.5 micron C02 regions absorbtion band appears to be an ideal region in which to operate. (Ref. 7) Carbon dioxide is well mixed in the upper atmosphere and will abscure most clouds. In this spectral region the apparent earth temperature should be fairly uniform at about the temperature of the stratosphere, approx. 220 o K. (Ref. 8) Of course operation in such a narrow spectral band considerably reduces the radiation received and requires increased optical gain. The Mercury capsule contained a wide field periscope which viewed the entire hemisphere below the spacecraft. This periscope had a graticule to indicate the attitude of the spacecraft with respect to the horizon. In one of the early unmanned orbital tests a motion picture camera was installed on the periscope and the pitch and roll outputs of the horizon scanners were als,o recorded. This provided an excellent independent check on the performance of the horizon scanners. Figure 12 shows a sample of the horizon scanner roll indication with the roll attitude measured from the periscope photographs. A consistant offset of about 2° is evident which was probably due to a boresight error. Applying thi.s correction it was found that the horizon scanner accuracy was about ± 1/2°, VI. "OGO"EDGE TRACKING SENSOR (Re£. 9) The horizon sensing system used with the Orbiting Geophysical Observatory (Project OGO) is a good example of an edge tracking system. A cross sectional view of the basic sensor head is shown in Fig. 13. A germanium immersed thermistor bolometer at the focal point of a fixed telescope views the earth by a reflection from a mirror mounted on the rotor of an electro-magnetic actuator called a posito~ The latter is essentially a permanent magnet type torquer with a set of auxiliary coils excited at several kilocycles to accurately read out the rotor position in a manner similar to a resolve~ It permits the mirror to be positioned at any point over a 45° angle and also super-imposes a small rapid sinusoidal oscillation or "dither" at a frequency
391
, 4 " (PRODUCTION UNITl--j
-------
ELECTRONICS El AZIMUTH DRIVE SYSTEM
SIGNAL PREAMPlIFIERS
TELESCOPE
POSITOR WINDOW
Figure 13 OGO EDGE TRACKING SENSOR
r
4 SECTION
6._1l:~ ELECTRONICS 2:
OPICAL
F f-_ _-1_f-_'L_T_E_R_-+
SILICON DETECTOR
_.1-
----r
THEMOPllE
SUN SENSOR
DETECTORS
OBJECTIVE LENS
s· OIA.
EARTH PRESENCE DETECTOR
OFFSET
HEAT SOuRCE
AlUMINUM SHEll FOR HEAT PATH
FOCAL SURFACE
CONTROL
Figure 14 APOLLO ANTENNA POINTING SENSOR
y
,\~
A
.A
c
,"
~
X
B
0
y
FiQure 15 FIELDS OF VIEW OF APOLLO SENSOR
392
of 30 cps and a peak to peak amplitude of 3°. The rotor is mounted to the frame by a pair of flexure pivots to eliminate lubrication and wear problems. The telescope aperture is 2 cm and the instantaneous field of view of the detector is 1° X 1° which can be positioned over a 90° angle by the Positor. The sensor has two modes - search and track. In the search mode the instantaneous field of the telescope is slowly scanned over a 90° arc. When the horizon is reached a 30 cps signal will be generated on the detector by the small 30 cps dither oscillation superimposed on the Positor mirror. The position of the horizon within the limits of the dither oscillation amplitude is indicated by the 2nd. harmonic content of the detector signal, which goes to zero when the horizon is centered within the dither oscillation. The Positor mirror is servoed by adjusting the average Positor-mirror angle so that the 2nd. harmonic vanishes. The angle to the horizon is then read out from the average Positor-mirror position by the read out windings which have a precision of ± 0.1°. The complete system is comprised of four such sensor heads oriented 90° apart. These measure the angles from a reference axis to four points equally spaced around the horizon from which the attitude of the reference axis with respect to the earth is determined. Actually only three heads are required, the fourth giving redundancy in case of a failure. Also if the sun appears within the field of view of any of the four sensor heads that head is removed from the computation without any loss of performance. Pairs of heads are packaged together in a single casting, the entire system consisting of two such dual tracker head castings and an electronic package. Its total weight is 13.2 lbs. and it consumes 8.5 watts of power (average). VII.
RADIATION BALANCE SENSORS (ReLlO) A typical radiation balance type of sensor is shown in Fig. 14 which shows an instrument developed for keeping a communications antenna pointed toward
the earth. An infrared lens projects an image of the earth onto a curved focal plane which is divided into four adjacent squares by pyramidal condensing light pipes. Each light pipe condenses the radiation in a 15° X 15° field of view onto a thermopile detector. (Ref. 11) The fields of view delineated by the apertures of the pyramidal light condensors are shown in Fig. 15. Diagonally opposite detectors are connected in series opposition, (A-C) and (B-D), such that when equal radiation is incident on the pair of detectors, their outputs cancel. When a radiation imbalance occurs a positive or negative output signal is produced. The four detectors provide error indications about a pair of orthogonal axis, XX and ¥Y. The instrument is designed to operate at altitudes between 6000 miles and the lunar orbit over which the earth subtense ranges from 35° to 2°. It will be seen that this type of sensor is extremely simple requiring no scanning mechanisms or other moving parts and consequently has very high reliability and long life. Its accuracy however, depends upon uniformity of radiance over the earths surface since the error indication is derived by the difference in radiation received from opposite sectors of the earth. For this reason optical filtering is employed to restrict the spectral region of sensitivity to the C02 & H20 absorbtion bands between 14 and 25 u. Theoretical studies and TIROS flight data (Ref. 7 & 8) show the earth radiance to be more uniform in this region. If the sun should appear within any field, a large error would result. To prevent this an auxiliary array of four silicon sun sensors is provided having identical fields of view with the thermopile infrared detectors. If a sun signal is received by one of these it will automatically disconnect its corresponding thermopile detector and replace its output with the average of the two adjacent thermopile detectors. For example referring to Fig. 15, if the sun should appear in Field A, the A sun sensor would disconnect thermopile A and substitute one half of the summed outputs
393
Figure 16 THERMOPILE MOSAIC DETECTOR
....................................
/
/
/
//
\
I
.>--~-..::c,
........
--...... __ "}.r::IMAGINARY DODECAHEDRON
I \
//
I
(
I
\ )
I
------i\ ...-l-
I I
I I
I
\ \
I
l
\
I I I
/
,---" -.~/ "
I
\
"1........ ,........ '........ .....
,
/ / /'
I
\) y
-
.......
I
\
/'/'/'
/'
__'V / '
Figure 18 MICROMETEOROID MEASUREMENT CAPSULE SENSOR HEAD
Figure 17 ORIENTATION OF M MC SENSOR FIELDS OF VIEW
394
of detectors D and B. At null this will be the same as the signal from detector A without the sun in the field. Field effect transistors are used for switching. To prevent the sun from deactivating two detectors simultaneously, the edges of the pyramidal light condensors are separated from each other by 1/2° (the angular subtense of the sun). The detectors used with this sensor are vacuum evaporated thermopile mosaics made up of 308 active junctions arranged in a square array 5 mm on a side. A photograph of one of these detectors is shown in Fig. 16. A unique radiation balance sensing system has been used on the Micrometeoroid Measurement Capsule of Project Pegasus. This is an unstabilized spacecraft which very slowly tumbles in a random manner and the problem was to read out its attitude with respect to the earth upon ground interrogation. The attitude readout system employed consists of six identical sensor heads, each having two narrow fields of view pointing in opposite directions along the same optic axis. The fields of view are 2° X 2°. On the space vehicle the sensor heads are placed in such a manner that their optical axis are normal to the surfaces of an imaginary dodecahedron encompassing the vehicle as shown in Fig. 17. This produces an angular separation between fields of 63° 26'. The presence or absence of earth signal within the twelve fields of view is used to establish the orientation of the vehicle relative to the earth. The system is only suitable for use at lower altitudes such that at least one sensor field will always view the earth. When the earth subtense is known and about 130° (a1t 625 km) a single interrogation will give an attitude readout accuracy of ± 10° about each axes. By successive interrogations, the vehicle tumbling motion can be determined and the accuracy improved to about ± 1°. A photograph of one of the six sensor heads is shown in Fig. 18. Two inch diameter germanium objective
lenses are mounted at opposite ends of an optical tube focussing radiation onto a pair of thermopile detectors at the center of the tube. The thermopiles are connected in series opposition so that a positive, negative, or zero signal will result depending upon whether the earth is seen in one field, the other, or neither. In conclusion it may be stated that infrared horizon sensors have, and are, being effectively used on a wide variety of earth orbiting space missions and it is expected that they will find additional use in the future on orbiting missions to the moon and other planets. REFERENCES 1.
2.
3. 4. 5.
6.
7.
8.
395
Kuiper, G.p. and B.M. Middlehurst (Editors): Planets and Satellites; The Solar System Ill. University of Chicago Press, 1961. Hanel, R.A., W.R. Bandeen, and B.J. Conrath: "The Infrared Horizon of the Planet Earth", Journal of the Atmospheric Sciences, Vol. 20, No. 2, pp. 73-86, March,1963. Potter, R.F. and W.L. Eisenman: Applied Optics 1, p. 567, 1962. DeWaard, R. and E.M. Wormser: Proc. IRE, Vol. 47, No. 9, pp. 1508-13, September, 1959. Astheimer, R.W. and S. Weiner: "Solid Backed Evaporated Thermopile Radiation Detectors": Applied Optics Vol. 3, p. 493, April 1964. Wormser, E.M. and M.H. Arck: Proceedings of ARS Guidance Control and Navigation Conference, August 7-9, 1961. B.J. Conrath: "Earth Scan Analog Signal Relationships in the TIROS Radiation Experiment and their Application to the Problem of Horizon Sensing", NASA Technical Note D-134l, Washington, D.C. W.R. Bandeen, M. Halev and I. Strange "A Radiation Climatology in the Visible and Infrared from TIROS Meteorological Satellites", NASA Report X-65l-64-2l8 Goddard Space Flight Center Greenbelt, Md.
9.
"Infrared Horizon Sensors for Precision Attitude Measurement", ATL-D-I072 published by Advanced Technology Laboratories, Mountain View, California. 10. Falbel, G. and E.A. Kallet: "Infrared Horizon Sensors Having No Moving Parts And Using Evaporated Thermopile Detectors", PROC. IRIS May 1963. 11. Williamson, D.E.: "Cone Channel Condenser Optics", Journal of the Optical Society of America, Vol. 42, No. 10, October 1952.
396
INTRODUCTION It is often necessary to establish the attitude of orbiting and interplanetary vehicles with respect to some external coordinate system. This is usually accomplished by sensing the position of some combination of the sun, stars, or planets. If the object is at a distance such that it appears as a point, the sensor is called a star tracker, while if the planet is close enough so that a definite disc is perceptible, it is usually called a horizon sensor. In the latter case, the direction toward the center of the disc is desired. This paper will consider the problem of determining the horizon position by means of the thermal self-emission of the planets. In order to sense the position of a planet by its thermal self-emission reliably, and with some degree of accuracy, three properties of the planet are desirable. First, it must be warm enough so that a detectable thermal discontinuity exists between the horizon and the space background. Second, for a
Robert W. Astheimer Barnes Engineering Company Stamford, Connecticut (U.S.A.) good accuracy, this discontinuity or gradient should be very sharp; and lastly, system problems are simplified if the radiance is reasonably uniform over the surface of the planet. The last condition can sometimes be attained fairly satisfactorily by selecting the spectral region of operation. A plot of the apparent temperature distribution along a diametral scan ofdE Earth, the Moon, Mars, and Venus is shown in Figure 1 (Ref. 1). A logarithmic radiance scale for the total thermal emission is also shown since it is the radiance rather than the temperature which produces the received signal. E?ch object is assumed at half phase with the terminator bisecting the disc in order to illustrate the difference in si~ nal level between the dark and sun illuminated sides. This figure is a great over-simplification but is u::teful for comparing the gross radiant characteristics of the four objects. It must be pointed out that these characteristics may be very different in restricted spectral regions, particularly in the absorption bands of atmospheric constituents; however, restricting the spectral region can only decrease the apparent radiance, although it may also reduce the dynamic range of signal that must be accepted. A minimum detectable radiance level is shown at 0.1 milliwatts/cm 2 -ster assuming an immersed thermistor detector with a 2-inch diameter collecting aperture, 20% optical efficiency, a 1 0 x 1 0 field of view, a 250 cps bandwidth, and a signal 10 times rms noise. These are realistic parameters for a system as will be shown. It is seen that the horizon of all planets is readily detectable with such a system. We conclude then that all these planets have enough thermal emission to make infrared horizon sensors of reasonable aperture practical. The accuracy with which the horizon can be determined will depend upon the steepness of the horizon gradient \mich
381
100.0 50.0
400' MOON
10.0
~ t;;
:::1
III Il'-
EARTH 7
I
1.0
~
05
cti I 50' ~
is
lr
0
I
~
MOONJ
...J
UJ I-
'CLOUD
I I
UJ
<..> Z et
200'~
r"'
I
~<..>
250'
-
VENUS
...J::lE
300'
,
5.0
...IN
350'
I 20'
z
f:!
I::E lr
g
DARK SIDE
:~
LIGHT SIDE
0.1
MINIMUM DETECTABLE RADIANCE
Figure I TEMPERATURE PROFILES OF EARTH, MOON, MARS AND VENUS
1.0
I f~RDC STD
I 0.8
.............. 0.6 ~o:
~~
10.75-1I.75p wiz CLEAR
------rI
i ~ 0.2 -.......--
HIGH CLOUD
I I
1075-11.751''.
I
0
z
I
\
'j-ARDC STD
OA
...I N
~12
1-- 10: I ........ 'i~ I 1 1 L. ____ -~ ---- , \
14-16p
\
1 1
\
\ \
I
I 0.5
,
. oI. I . ., + 0.5
+1.0
, +1.5'
Figure 2 EARTH HORIZON PROFILES IN VARIOUS SPECTRAL REGIONS
A
C
CONICAL SCAN
RADIOMETRIC BALANCE
B EDGE TRACKER Figure 3 BASIC SENSING TECHNIQUES
382
is a function of the density and composition of the atmosphere. The moon, possessing no significant atmosphere, has an infinite gradient; and the ultimate accuracy is limited only by terrain irregularities. In a practical system, however, the detector noise would probably be a limiting factor because of the low temperature and emission from the dark edge. The uncertainty caused by the fuzziness of the horizon discontinuity of a planet with an atmosphere is a function of altitude. At very low altitudes, the error could be quite large, while at high altitudes, the entire subtense of the atmosphere may become negligible. The earth is the only planet on which horizon sensors have been used to date, and considerable data has now been accumulated. The accuracy of these systems has generally been limited by the size of the detector field of view rather than the horizon gradient. A number of theoretical studies (Ref. 2) have been made of the gradient to be expected in different spectral regions, some results of which are shown in Fig. 2. The spectral regions of greatest interest are the C02 absorption band at 14 - 16 microns. the water vapor bands extending from 20 - 35 microns (and beyond), and the so-called transparent atmospheric "window" between 8 - 12 microns. It will be seen that the horizon gradient is quite sharp in the atmospheric window, but the range of signal level is greatest here because of cloud cover. Operation in the 14 - 16 micron CO~ absorption band masks clouds and w~ll present a very uniform signal, but the gradient will be more gradual, extending over almost 1° from an altitude of 400 miles. Accuracy at this altitude is probably limited to about .05 degrees by the indistinctness of the horizon itself • Much less information is available regarding Mars and Venus. The atmosphere of Mars is fairly tenuous with few clouds and will probably not diffuse the horizon significantly even at low altitudes. Venus has a very dense atmosphere of unknown depth; however, the top of the cloud deck appears to be quite distinct and of uniform temp-
erature. Therefore, a sharp gradient should exist in a spectral region where the clouds are opaque and the atmosphere above transparent. One of the most troublesome problems in horizon sensing is caused by large variations in temperature or radiance over the surface. Some difficulties of this nature were experienced with early earth horizon sensors because of unexpectedly cold cloud tops associated with large storm areas. Spectral filtering and modifications of the electronic processing systems were required to achieve reliable operation for all weather conditions. The situation is much worse with the moon as indicated in Figure 1. The radiant emission from the sun-baked side is 100 times greater than that from the dark side, and there are additional small "hot" and "cold" areas caused by sun-illuminated crater edges and shadows. These variations in signal strength may not only produce errors due to saturation and time constant effects but could cause the system to confuse the horizon with the terminator, since the terminator discontinuity is 50 times greater than the dark edge horizon discontinuity. The planet Venus seems to present the easiest object for infrared horizon sensing, but the sharpness of the horizon gradient in various spectral regions, which depends upon the atmospheric structure, is still unknown. Reflected solar radiation will be super-imposed on the thermal self-emission of the sun-illuminated portions of all the planets. This is in general an unwanted disturbance and can be substantially removed by filtering out wavelengths shorter than 8 microns. Such filtering also greatly reduces the signal produced from direct viewing of the sun which could saturate the system or even damage the detector. 11.
DETECTOR CONSIDERATIONS It has been shown that the range of apparent planetary temperatures to be sensed extends from l20 0 K to 380 0 K. Detectors sensitive to such therma1radiation must respond to long wavelengths from 8 to 40 microns. Photoconductive detectors highly sensitive in this
383
Thermal detectors which depend upon a resistance change for signal generation, such as thermistor or metal bo10meters, are more subject to internal temperature effects than thermocouple types. The reason for this is that in order to sense the resistance change a bias current must be used to convert it to a voltage signal. The desired signal thus appears as a very small change superimposed on the much larger bias voltage. Bridge arrangements of detector elements may be used to buck out the large bias voltage, but it is practically impossible to maintain the degree of balance necessary over a wide ambient temperature range. Optical modulation, of course, solves the problem since the d.c. bias voltage is filtered out, therefore these detectors are seldom used without optical modulation. Thermocouple detectors (Ref. 5) do not require biasing and are thus free of this problem. The only voltage appearing is the thermoelectric EMF produced by the radiant heating of the junction, and this is not superimposed on a biasing voltage. Unmodulated optical systems are therefore feasible with thermocouple or thermopile detectors. However, even with these detectors, care must be taken in the design to eliminate spurious signals from internal temperature gradients and self-emission of optical parts. Also thermocouple detectors sufficiently rugged for space applications are slower and less sensitive than thermistor detectors. The elimination of optical chopping and scanning mechanisms is highly desirable for space missions where very long life and low power are necessary. The spurious signals or drifts, inherent in unmodulated systems, make them relatively less accurate than optically scanned systems. Therefore, thermopile detectors are most suitable for low accuracy, long life applications such as antenna pointing.
region have been developed such as copper or zinc doped germanium (Ref. 3) but these all must be cooled to temperatures in the neighborhood of liquid helium. Reliable cooling to these temperatures for long periods of time is difficult in space systems and the most suitable detectors are the thermal types, particularly thermistor and metal bo10meters, and thermocouples. Of these, the thermistor bolometer (Ref. 4) is the most sensitive and has found greatest use to date. These are usually immersed on a germanium lens for increased detectivity. The thermal detectors operate by virtue of the heating effect when incident radiation is absorbed on the sensitive element. The very small temperature change resulting effects some physical parameter such as the resistance which can be read out electrically. An inherent difficulty in the use of such detectors is the identification or separation of the temperature change caused by the desired radiation from the very much larger ambient temperature variations. For example, in a typical horizon sensor, the change in radiation when the field of view scans from space onto the earth increases the detector element temperature about .001°C. The sensor must be designed so that this temperature differential may be detected in spite of ambient temperature variations 10,000 to 100,000 times greater. The most common way of separating the desired temperature change from unwanted internal effects is by opticalmechanical scanning or chopping. This modulates the external radiation signal at a relatively high frequency, and spurious signals produced by the slowly changing ambient temperature can be removed by capacitance coupling to the electronic amplifier. It is important to notice that such modulation must be done on the radiation signal before detection and not on the electrical output. Electronic chopping will modulate both external and internal signals and is of no value in this respect.
Ill.
BASIC SENSING TECHNIQUES The infrared emission characteristics of the planets and suitable detectors for
384
sensing this radiation have been described. Such detectors must be employed in an optical arrangement to provide information from which the coordinates of the center of the planetary disc can be determined to the degree of accuracy desired. A complicating factor is that for most planetary objects the radiant emission will be highly nonuniform over the surface, and for high accuracy, the system must be independent of the radiance level. A number of systems have been developed for this purpose. These can all be shown to be versions of three general categories which we shall designate as follows: (1) Wide Angle Scanning Systems. (2) Edge Tracking Systems. (3) Radiometric Balance Systems. The detector field of view and scan pattern associated with these techniques are shown in Figure 3. In the conical scan, which is typical of the wide angle scanning systems, the instantaneous detector field is relatively small and is caused to scan through a large cone whose apex angle may be as much as 180 0 , although an apex angle between 50 0 to 120 0 is more usual. The detector signal generated will be an approximately rectangular waveshape repetitive at the scan frequency. This waveshape is usually limited in some fashion to eliminate amplitude dependence, and then position information is derived by a phase or pulse width comparison technique. Two sensors are used to generate pitch and roll attitude information. Most horizon sensor systems flown to date have been of the conical scan type because it possesses a number of very significant advantages. It has excellent acquisition capability, attained without additional search modes because of the wide scan angle. The attitude information is derived from time characteristics of an amplitude limited waveshape and is therefore insensitive to radiance variations over the surface of the planet, which as we have seen can be very large in some cases. Another advantage is that, because of the wide field scanned it is
certain that some portion of the scan will leave the planet and view space. This provides an absolute zero radiance level against which any portion of any planet ~ give a positive contrast. Use can be made of this reference in setting limiting levels so as to prevent the system from confusing radiance discontinuities on the planetary surface, such as may be produced by the terminator or clouds, with the true horizon. This can be a serious problem with edge tracking systems. The primary disadvantage of the conical scan sensor is the need for high speed rotating elements which present life and lubrication difficulties in spaceborne applications. Some sensors have been developed which combine the function of the two scanners by precessing the basic conical scan to produce a rosette or epicyclic pattern (Ref. 6). These may be considered essentially versions of the conical scan system and the same general remarks apply. The basic concept of the edge tracking system is shown in Figure 3B. A small detector field of view is caused to lock onto the radiance discontinuity at the edge of the planetary disc. This is usually accomplished by oscillating the detector field through a small angle normal to the horizon edge and moving the entire sensor or the oscillating field until a balanced waveshape is obtained. Multiple sensor heads may be used, in which case at least three are necessary, or a single oscillating field may be caused to trace around the edge. In either case, the horizon position is determined by reading out the position of the center of the oscillating field with respect to spacecraft coordinates, i.e., angles 9 1 , 92' 93' and 94' in Figure 3B. For equivalent optics and detectors, this type of system will have a better signal-to-noise ratio than the wide angle scanning types because of narrower electronic bandwidth. It leads to a much more complicated system however because separate search and edge tracking movements must be provided on each sensing head along with precision position readouts.
385
65'
Figure 4 TIROS HORIZON CROSSING INDICATOR
ORBITAL PLANE
SP1N DIRECTION
OP.0(":~~
~
..--p~ ",':: SENSOR FI ELD OF VIEW
FIELDS OF VIEWS OF SENSORS
SP1NAXIS
-_
- "-
"-M \...--- CAMERA
~'
"
FIELD
p"""
~'
FiCj1ure 5A
FiQur. 5 B
SPIN STABILIZED SATELLITE GEOMETRY
Figure 6
TELEMETERED HORIZON CROSSING INDICATOR RECORDS
386
A high price is paid for this narrow bandwidth and consequent greater sensitivity. Without the wide angle scan the previously mentioned space reference is lost, and there is no convenient way for the system to sense whether the signal is from the true horizon or some other discontinuity such as a cloud edge or terminator which can be much greater than the horizon signal. Various devices such as auxiliary detectors may be employed to prevent locking on false edges, but they further complicate the system. The edge tracking type of system would appear best suited for application where spectral filtering to an atmospheric absorption band can eliminate radiance variations over the planetary disc. An example for earth horizon sensing would be to filter to the narrow C02 absorption band at 15 microns. The increased detectivity of this type system would compensate for the large reduction in signal caused by the spectral filtering. The former two systems employ optical modulation by mechanical means. It should be pointed out, however, that a stationary array of detectors could be used whose outputs are electronically sampled. For example, in a conical scan system, instead of mechanically causing a small detector field to scan over a wide circle, a stationary array of detector elements can be placed to view the same circle, and the array sequentially sampled electronically. Thermopile detectors are particularly well suited for this technique, and a system of this type is being developed. The radiometric balance type sensor is a non-scanning system and operates by comparing the radiation received from opposite portions of the planet. Very wide fields of view are used to achieve acquisition and also to average radiance variations over the surface. A typical arrangement of detector fields is shown in Figure 3C. Four wide angle stationary fields are employed designated a, b, c, and d; and attitude information is obtained by the difference in radiant power received from opposite fields, i.e., Pc - Pa and Pb - Pd. This is obviously only
correct if the planet is uniformly radiant, and therefore this system is primarily suited for use with planets of uniform radiance such as Venus or where only moderately accurate pointing information is necessary. The great virtue of this system is its extreme simplicity and consequent high reliability. By using thermopile detectors no moving parts are necessary and very long life can be achieved. Nonuniform radiance effects can sometimes be minimized by spectral filtering. In the remainder of this paper a number of operational sensors employing these principles, which have been developed and used in space missions are described. IV. "TIROS"HORIZON CROSSING INDICATORS In a spin stabilized satellite the motion of the vehicle itself can be used to generate a wide angle conical scan. This is the technique used in the TIROS weather satellite series, and permits a very simple sensor design. A photograph of a TIROS sensor is shown in Figure 4. It consists of a sma:l infrared telescope and germanium immersed thermistor detector having a field of view of 1 degree square with a transistor amplifier. The aperture of the objective lens of the telescope is 16 mm. The entire sensor is approximately 1 X I X 12 inches. As the rotation of the satellite causes the small field of view of the sensor to cross the horizon a steep positive or negative pulse is developed depending upon the direction. These horizon crossing pulses are used for a variety of purposes. The period between successive horizon crossings in the same direction (space to earth) gives the spin rate of the satellite. The ratio of the time between on and off horizon crossings to the spin period is a measure of the inclination of the spin axis to the local vertical. The horizon lion" pulse has been used to trigger other instruments such as cameras at appropriate times. For example in a II wheel" type mode where the spin axis is normal to the orbital plane as shown in
387
DETECTOR FIELD
OF VIEW -
__
EARTH
Figure 7 CONICAL SCAN GEOMETRY
...
(PROJECT MERCURY)
~557INCHES
I-=OTOR "r"r>tH-SCANNING PRISM
Figure 8
\
~\
J
CROSS SECTIONAL VIEW OF MERCURY SCANNER
. .\
.J
Figure 9
...
CONICAL SCAN SENSOR
388
Figure Sa, if a camera and horizon crossing indicator are both pointed radially but displaced angularly by half the earth subtense (
v.
''MERCURY'' CONICAL SCAN SENSOR
For an attitude controlled vehicle the scan motion must be accomplished internally. A series of very successful sensors have been developed employing a conical scanning principle wherein the projected image of a thermistor detector is caused to scan over a wide circle, or cone in space. Typical of these is the scanner used on the Project Mercury vehicle, the first American manned spacecraft. The scan geometry is shown in Fig.7. The instantaneous field of view of the detector is 2° X 8° and is caused to scan over a wide circle with a cone apex angle of 110° at 30 revolutions per second. A rectangular waveshape of
this frequency is produced by the detector because of the temperature difference between the planet and the cold space background. Another square wave reference signal of the same frequency is generated internally from the rotating scanning mechanism. By phase comparison of the radiation signal with the internal reference the attitude with respect to the conical scan axis can be determined. A pair of such sensors displaced 90° with respect to each other as shown in Fig. 7 can give attitude information about two orthogonal axis, usually designated as pitch and roll. A cross sectional view of the scanner is shown in Fig. 8. The conical scan is produced by a germanium prism mounted on a hollow barrel which is driven through gearing by a small motor. The internal phase reference signal is developed by a semicircular iron vane on the rotating assembly and a stationary magnetic pick up. The detector is a germanium immersed thermistor detector where the immersion lens also serves as the objective lens of the telescope. The unit is filled with dry nitrogen and hermetically sealed to prevent evaporation of lubricants. A germanium front window is used with a multilayer interference filter blocking all radiation of wavelength shorter than 8 microns. This greatly reduces direct and reflected solar radiation. A photograph of one of these sensors with the front and rear covers removed is shown in Fig. 9. In some applications it is inconvenient to mount the roll sensor on the front or aft ends of the space craft where it can have an unobstructed circular scan. An alternate mode of operation is shown in Fig. 10 where the two sensors are pointed 180° apart. In this mode attitude information about the roll axis is obtained by comparing the duration of the earth waveshapes generated by the two sensors while pitch can be determined from either one by the phase comparison method previously described. The pulse phase or pulse width processirg described requires a clean rectangular waveshape of constant amplitude. This is usually produced by
389
Fi9Ure 10 ALTERNATE MOOE OF CONICAL SCAN
COLD CLOUO
(
- - - - EARTH
A
-
DETECTOR NOISE Vn
l_
->-THRESHOLD B
SPACE
T-
HORIZON) CLOUD SIGNAL Vc OUTPUT WITH ~THRESHOLD
cg:~
CLOUD ERROR
A
OUTPUT WITH
~THRESHOLD
SET AT LEVEL B
Figure 11 CONICAL SCAN WAVESHAPE
-20
~
:I:
~-3 0 Vl
w w
I'
.-/
ii:
a:
Cl W 0
..
-" 0
..... -, _
/
~
' .... '... , -
,_ ..... '"
-..c;
/'
V/
PHOTOS
--- -
--....
--"
~
./" t-,
V/
-
,~/
"....,
'''SCANNER
-'
z
Cl
0
Vl
/
a:
j/
0
a: -' a: -' w 0 a:
0
r---...
/t"I-' '\k?" I
",~
~/
-10 1.38.00
f---<:: PHOTOS
1:39.00
/
/,
1.40:00
1:41:00
~
SCAN~
r""_'
1.42:00
~
\...V7
1:43.00
TIME AFTER LAUNCH HOURS: MINUTES: SECONDS
1.44:00
r-..
/,' "'\
""
L?
1:46:00
Figure 12 COMPARISON OF HORIZON SCANNER READOUT WITH PERISCOPE PHOTOS.
390
an on-off threshold circuit as shown in Fig. 11 which gives zero volts if the input signal is less than the threshold and a fixed voltage if it is greater. It is necessary that this threshold be established well above the detector noise level during the space scan but still below the minimum earth signal that may be expected. It is evident from the figure that if a cold cloud causes the earth signal to drop below the threshold, a "gap" will be produced in the processing waveshape which will cause serious errors in the attitude indication. In the first horizon sensors of this type such errors did occur because the apparent radiation temperature of some cloud tops particularly those associated with large storm areas was considerably colder than expected. This situation was corrected by improving the sensitivity and lowering the threshold level. The apparent radiation temperature of the earth depends upon the meteorological conditions at the point viewed and the spectral region in which the observation is made. In spectral regions where atmospheric constituents are absorbant the surface radiation will not be received and the signal will consist of emission from the absorbing constituents primarily water vapor and carbon dioxide, which will in general be colder than the surface. In atmospheric windows surface radiation will be received unless there is cloud cover in which case the apparent radiation temperature will be that of the cloud tops. The greatest variation in signal will therefore occur in the atmospheric windows, primarily the 8 12 micron region. Flight data from horizon scanners and radiometers show radiation temperatures ranging between l70 0 K and 300 0 K in the 8 - 12 micron window, which produces a variation of ten to one in detector signal, Such large variations in signal level complicate the electronic processing and it is desirable to reduce the dynamic range of signal as much as possible. This can be done by spectral filtering to
eliminate atmospheric windows. The 14.5 - 15.5 micron C02 regions absorbtion band appears to be an ideal region in which to operate. (Ref. 7) Carbon dioxide is well mixed in the upper atmosphere and will abscure most clouds. In this spectral region the apparent earth temperature should be fairly uniform at about the temperature of the stratosphere, approx. 220 o K. (Ref. 8) Of course operation in such a narrow spectral band considerably reduces the radiation received and requires increased optical gain. The Mercury capsule contained a wide field periscope which viewed the entire hemisphere below the spacecraft. This periscope had a graticule to indicate the attitude of the spacecraft with respect to the horizon. In one of the early unmanned orbital tests a motion picture camera was installed on the periscope and the pitch and roll outputs of the horizon scanners were als,o recorded. This provided an excellent independent check on the performance of the horizon scanners. Figure 12 shows a sample of the horizon scanner roll indication with the roll attitude measured from the periscope photographs. A consistant offset of about 2° is evident which was probably due to a boresight error. Applying thi.s correction it was found that the horizon scanner accuracy was about ± 1/2°, VI. "OGO"EDGE TRACKING SENSOR (Re£. 9) The horizon sensing system used with the Orbiting Geophysical Observatory (Project OGO) is a good example of an edge tracking system. A cross sectional view of the basic sensor head is shown in Fig. 13. A germanium immersed thermistor bolometer at the focal point of a fixed telescope views the earth by a reflection from a mirror mounted on the rotor of an electro-magnetic actuator called a posito~ The latter is essentially a permanent magnet type torquer with a set of auxiliary coils excited at several kilocycles to accurately read out the rotor position in a manner similar to a resolve~ It permits the mirror to be positioned at any point over a 45° angle and also super-imposes a small rapid sinusoidal oscillation or "dither" at a frequency
391
, 4 " (PRODUCTION UNITl--j
-------
ELECTRONICS El AZIMUTH DRIVE SYSTEM
SIGNAL PREAMPlIFIERS
TELESCOPE
POSITOR WINDOW
Figure 13 OGO EDGE TRACKING SENSOR
r
4 SECTION
6._1l:~ ELECTRONICS 2:
OPICAL
F f-_ _-1_f-_'L_T_E_R_-+
SILICON DETECTOR
_.1-
----r
THEMOPllE
SUN SENSOR
DETECTORS
OBJECTIVE LENS
s· OIA.
EARTH PRESENCE DETECTOR
OFFSET
HEAT SOuRCE
AlUMINUM SHEll FOR HEAT PATH
FOCAL SURFACE
CONTROL
Figure 14 APOLLO ANTENNA POINTING SENSOR
y
,\~
A
.A
c
,"
~
X
B
0
y
FiQure 15 FIELDS OF VIEW OF APOLLO SENSOR
392
of 30 cps and a peak to peak amplitude of 3°. The rotor is mounted to the frame by a pair of flexure pivots to eliminate lubrication and wear problems. The telescope aperture is 2 cm and the instantaneous field of view of the detector is 1° X 1° which can be positioned over a 90° angle by the Positor. The sensor has two modes - search and track. In the search mode the instantaneous field of the telescope is slowly scanned over a 90° arc. When the horizon is reached a 30 cps signal will be generated on the detector by the small 30 cps dither oscillation superimposed on the Positor mirror. The position of the horizon within the limits of the dither oscillation amplitude is indicated by the 2nd. harmonic content of the detector signal, which goes to zero when the horizon is centered within the dither oscillation. The Positor mirror is servoed by adjusting the average Positor-mirror angle so that the 2nd. harmonic vanishes. The angle to the horizon is then read out from the average Positor-mirror position by the read out windings which have a precision of ± 0.1°. The complete system is comprised of four such sensor heads oriented 90° apart. These measure the angles from a reference axis to four points equally spaced around the horizon from which the attitude of the reference axis with respect to the earth is determined. Actually only three heads are required, the fourth giving redundancy in case of a failure. Also if the sun appears within the field of view of any of the four sensor heads that head is removed from the computation without any loss of performance. Pairs of heads are packaged together in a single casting, the entire system consisting of two such dual tracker head castings and an electronic package. Its total weight is 13.2 lbs. and it consumes 8.5 watts of power (average). VII.
RADIATION BALANCE SENSORS (ReLlO) A typical radiation balance type of sensor is shown in Fig. 14 which shows an instrument developed for keeping a communications antenna pointed toward
the earth. An infrared lens projects an image of the earth onto a curved focal plane which is divided into four adjacent squares by pyramidal condensing light pipes. Each light pipe condenses the radiation in a 15° X 15° field of view onto a thermopile detector. (Ref. 11) The fields of view delineated by the apertures of the pyramidal light condensors are shown in Fig. 15. Diagonally opposite detectors are connected in series opposition, (A-C) and (B-D), such that when equal radiation is incident on the pair of detectors, their outputs cancel. When a radiation imbalance occurs a positive or negative output signal is produced. The four detectors provide error indications about a pair of orthogonal axis, XX and ¥Y. The instrument is designed to operate at altitudes between 6000 miles and the lunar orbit over which the earth subtense ranges from 35° to 2°. It will be seen that this type of sensor is extremely simple requiring no scanning mechanisms or other moving parts and consequently has very high reliability and long life. Its accuracy however, depends upon uniformity of radiance over the earths surface since the error indication is derived by the difference in radiation received from opposite sectors of the earth. For this reason optical filtering is employed to restrict the spectral region of sensitivity to the C02 & H20 absorbtion bands between 14 and 25 u. Theoretical studies and TIROS flight data (Ref. 7 & 8) show the earth radiance to be more uniform in this region. If the sun should appear within any field, a large error would result. To prevent this an auxiliary array of four silicon sun sensors is provided having identical fields of view with the thermopile infrared detectors. If a sun signal is received by one of these it will automatically disconnect its corresponding thermopile detector and replace its output with the average of the two adjacent thermopile detectors. For example referring to Fig. 15, if the sun should appear in Field A, the A sun sensor would disconnect thermopile A and substitute one half of the summed outputs
393
Figure 16 THERMOPILE MOSAIC DETECTOR
....................................
/
/
/
//
\
I
.>--~-..::c,
........
--...... __ "}.r::IMAGINARY DODECAHEDRON
I \
//
I
(
I
\ )
I
------i\ ...-l-
I I
I I
I
\ \
I
l
\
I I I
/
,---" -.~/ "
I
\
"1........ ,........ '........ .....
,
/ / /'
I
\) y
-
.......
I
\
/'/'/'
/'
__'V / '
Figure 18 MICROMETEOROID MEASUREMENT CAPSULE SENSOR HEAD
Figure 17 ORIENTATION OF M MC SENSOR FIELDS OF VIEW
394
of detectors D and B. At null this will be the same as the signal from detector A without the sun in the field. Field effect transistors are used for switching. To prevent the sun from deactivating two detectors simultaneously, the edges of the pyramidal light condensors are separated from each other by 1/2° (the angular subtense of the sun). The detectors used with this sensor are vacuum evaporated thermopile mosaics made up of 308 active junctions arranged in a square array 5 mm on a side. A photograph of one of these detectors is shown in Fig. 16. A unique radiation balance sensing system has been used on the Micrometeoroid Measurement Capsule of Project Pegasus. This is an unstabilized spacecraft which very slowly tumbles in a random manner and the problem was to read out its attitude with respect to the earth upon ground interrogation. The attitude readout system employed consists of six identical sensor heads, each having two narrow fields of view pointing in opposite directions along the same optic axis. The fields of view are 2° X 2°. On the space vehicle the sensor heads are placed in such a manner that their optical axis are normal to the surfaces of an imaginary dodecahedron encompassing the vehicle as shown in Fig. 17. This produces an angular separation between fields of 63° 26'. The presence or absence of earth signal within the twelve fields of view is used to establish the orientation of the vehicle relative to the earth. The system is only suitable for use at lower altitudes such that at least one sensor field will always view the earth. When the earth subtense is known and about 130° (a1t 625 km) a single interrogation will give an attitude readout accuracy of ± 10° about each axes. By successive interrogations, the vehicle tumbling motion can be determined and the accuracy improved to about ± 1°. A photograph of one of the six sensor heads is shown in Fig. 18. Two inch diameter germanium objective
lenses are mounted at opposite ends of an optical tube focussing radiation onto a pair of thermopile detectors at the center of the tube. The thermopiles are connected in series opposition so that a positive, negative, or zero signal will result depending upon whether the earth is seen in one field, the other, or neither. In conclusion it may be stated that infrared horizon sensors have, and are, being effectively used on a wide variety of earth orbiting space missions and it is expected that they will find additional use in the future on orbiting missions to the moon and other planets. REFERENCES 1.
2.
3. 4. 5.
6.
7.
8.
395
Kuiper, G.p. and B.M. Middlehurst (Editors): Planets and Satellites; The Solar System Ill. University of Chicago Press, 1961. Hanel, R.A., W.R. Bandeen, and B.J. Conrath: "The Infrared Horizon of the Planet Earth", Journal of the Atmospheric Sciences, Vol. 20, No. 2, pp. 73-86, March,1963. Potter, R.F. and W.L. Eisenman: Applied Optics 1, p. 567, 1962. DeWaard, R. and E.M. Wormser: Proc. IRE, Vol. 47, No. 9, pp. 1508-13, September, 1959. Astheimer, R.W. and S. Weiner: "Solid Backed Evaporated Thermopile Radiation Detectors": Applied Optics Vol. 3, p. 493, April 1964. Wormser, E.M. and M.H. Arck: Proceedings of ARS Guidance Control and Navigation Conference, August 7-9, 1961. B.J. Conrath: "Earth Scan Analog Signal Relationships in the TIROS Radiation Experiment and their Application to the Problem of Horizon Sensing", NASA Technical Note D-134l, Washington, D.C. W.R. Bandeen, M. Halev and I. Strange "A Radiation Climatology in the Visible and Infrared from TIROS Meteorological Satellites", NASA Report X-65l-64-2l8 Goddard Space Flight Center Greenbelt, Md.
9.
"Infrared Horizon Sensors for Precision Attitude Measurement", ATL-D-I072 published by Advanced Technology Laboratories, Mountain View, California. 10. Falbel, G. and E.A. Kallet: "Infrared Horizon Sensors Having No Moving Parts And Using Evaporated Thermopile Detectors", PROC. IRIS May 1963. 11. Williamson, D.E.: "Cone Channel Condenser Optics", Journal of the Optical Society of America, Vol. 42, No. 10, October 1952.
396