Você está na página 1de 21

FRANK J.

SELLER RESEARCH LABORATO2Y

SRL-TR-72-0008

MAY 1972

""

BASE MOTION ISOLATION OF A 2-AXIS BEAM DEFLECTOR

Capt Gary C Comfort

D, DC
J VUN 9 1972

PROJECT 7904
APPROVED FOR PUBLIC RELEASE:
DISTR_ jUTIOtN UNL0IMTED.

AIR FORCE SYSTEMS COMMAND

NHT:D STATUS AIR F


D "-''-" '

UN-T

- 6-.

UNCLASSIFIED
Security Classification

DOCUMENT CONTROL DATA R & D


(Security lasfllication of title, body of abstract and indexing annotation must be entered when the overall report Is claseltied) 1. ORIGINATING ACTIVITY (Corporate author) I2a. REPORT SECIJRITY CLASSIFICATION

Frank J. Seiler Research Laboratory (AFSC)


USAF Academy, Colorado
3; REPORT TITLE

808402"

G~oup

OUPclassified

Base Motion Isolation of a Two-Axis Beam Deflector


4. OESCRIPTIVE NOTES (TYpe of report and Inclusive dates)

Scientific Report
5. AU THORM(First name, middle initial, last name)

Gary C. Comfort, Capt, USAF


6. REPORT OATC 7a. TOTAL NO. OF PAGES

7b. NO, OF REFS

May 1972
So. CONTRACT OR GRANT NO.

18
90. ORIGINATOR'S REPORT NUMBER(S)

b. PROJECTNO.

7904-00-40

SRL-TR-72-0008
9b. OTHER REPORT NO(S) (Any other numbers that mey be asdlined this report)

c.

DRS 61102F

d. BPAC 681304
10. DISTRIBUTION STATEMCEIT

DDC #AD

Approved for public release; distribution unlimited.


1I. )'JPPLE MCNTARY NOTES 12. SPONSORING MILIVARY ACTIVITY

Frank J. Seiler Research Laboratory (AFSC)

I USAF Academy, Colorado 80840


13. ABSTRACT

The base motion isolation characteristics of an optical beam deflector with a two-axis gimbal support are investigated. The particular configuration of the bean' deflector system considered utilizes rate integrating gyros to provide a rate inner control loop with an optical tracker providing an outer position control loop. Base motion isolation is partially achieved by the gimbal support assumed to be frictionless. H(qever, input beam motion relative to the optical surface uf the beam deflector is shown to produce large excursions of the output beam. As a result, a feed-forward control loop is added to measure and compensate for input beam motion, i.e., base motion. ,'Every

effort is directed to use optimistic estimates of beam deflector performance _i# .der to develop a measure of the smallest pointing error possible with such a two-axis beam deflector configuration. The base motion spectrum used is intended to approximate that of a large aircraft in normal flight.

FORM DD ,Nov 651473

.UNCLASSIFED
Security Classification

UNCLASSIFIED
Security Classifiation
i4. KEY WORDS ROLl WT ROLM KsT ROLE" WT LINK A L.INK LINK C

Optical Pointing Beam Deflectors Heliostats Airborne Heliostats Base Motion Isolation of a Heliostat

UNCLASSIFIED
Security Classification

BASE MOTION ISOLATION OF A TWO-AXIS BEAM DFLECTOR

Capt Gary C. Comfort Research Associate Frank J. Seiler Research Laboratory United States Air Force Academy, Colorado 80840 U.S.A. ABSTRACT The base motion isolation characteristics of an optical beam deflector with a two-axis gimbal support are investigated. The particular configuration of the beam deflector system considered utilizes rate integrating gyros to provide a rate inner control loop with an optical tracker providing an outer position control loop. 6ase motion isolation is partially achieved by the gimbal support assumed to be frictionless. However, input beam motion relative to the optical surface of the beam deflector is shown to produce large excursions of the output beam. As a result, a feed-forward control loop is added to measure and compensate for input beam motion, i.e., base motion. Every effort is directed to use optimistic estimates of beam deflector performance in order to develop a measure of the smallest pointing error possible with such a two-axis beam deflector configuration. The base motion spectrum used is intended to approximata that of a large aircraft in normal flight.

Figure 1.

-Beam Deflector

1. INTRODUCTION Recently applications-have arisen which involve tie airborne transmission and/or reception of information with optical uavelengths. Hany of these applications require a precise optical pointing system to transmit or collect the optical energy. Several configurations of optical surfaces are conceivable for such a purpose. One of the simplest concepts to conceive is that of t"'optical beam deflector as depicted in Figure 1. In order to provide the capability of directing the optical beam in any direction independent of aircraft altitude, it is necessary to mount the beam deflector in gimbals. The particular gimbal ccnfiguration depicted In Figure 2 allows for motion of the beam deflector about two axes. This is the minimum number of gimbals needed to allow motion in both azimuth and elevation.

Fig:ure 2.

Two-Axis Beam Deflector

Depending upon the specific application of the optical information, the beam deflector may be used either to transmit optical energy from the aircraft to a target or to collect optical energy from an external target source into the aircraft. For some applications two-way transmission of optical energy might be required. For purposes of clarity only, the source of optical energy will be considered to be aboard the aircraft and the beam deflector will have

the purpose of directing this energy upon an


external target. Thus the input beam will be considered to be the optical energy transmitted from the airborne source to the beam deflector while the output beam will be the optical energy path from the beam deflector to the external target. For applications in which the role of source and target are interchanged the following investigation applies fully with only a change in nomenclature. The investigation of base motion isolation capability of the particular beam deflector configuration considered herein is intended to: a) determine the smallest pointing error which could conceivably be obtained from such a configured sys.tem operating within an aircraft wiO the base motion, spectrum assumed, and b) serve as a guide to considering the eifi.cts of base motion upon the performance of othi- airborne optical. pointing systems. e

S~e
II. B1EAM DEFLECTOR CONTROL the beam several in In order to discuss quantitatively ccnzrol and performance of the two-axis def!%ctor, it is necessary to introduce coordinate systems. These are depicted Figure 3 as: I - fixed in inertial space V - fixed to the vehicle with coordinate axes i, J, k O - lixed to the outer or azimuth gimbal with coordinate axes n, e, k M - fixed to the beam deflector with coordinate axes r, e, d L - line-of-sight coordinates with coordinate axes 1, e, p. Figure 3. Coordinate Axes

The purpose of the beam deflector control system is to direct the output beam along the desired direction. This is accomplished by controlling the elevation angle c and the azimuth angle n where n is measured between the base and the outer gimbal and c is measured between the outer gimbal and the beam deflector surface as shown in Figure 4.

IP

The angular velocity of the output direction about the e axis is thus

WL
e E

= 2wM - wo
e e

(1)

t Since the direction to be controlled is that of the output beam, it is desirable to measure the motion of this output direction with a high bandwidth sensor. Unfortunately, the output direction is only a direction in space; it does not correspond to any direction in the physical hardware. Thus there is no way of mounting a sensor on this output beam direc-

tion to measure its motion.*


d

Instead motion

kThe

sensors can only be attached to the gimbals and the vehicle itself. The elevation control loop of Figure 5a postulates the use of a rate integrating gyroscope mounted on the inner gimbal with its input direction along the e axis. azimuth control loop of Figure 5b utilizes a rate integrating gyroscope mounted on the outer gimbal with its input direction along thek axis. The rate integrating gyros provide a relatively high bandwidth rate-feedback loop about the elevation and azimuth gimbals.
1.

Inasmuch as the measurement of primary interest is the orientation of the output bepm with respect to the target, an optical tracker is incorporated in the control system. The

tracker is oriented so that the center of its


field of view always is coincident with the output beam direction. Thus the tracker measures angles between the output beam direction and the desired target direction. The tracker thus provides an outer position-feedback

/I

loop.
Figure 4. Controlled Angles

The beam deflector control can be accomplished by two control loops, elevation and azimuth. These basic features of these loops are block diagrammed in Figures 5a and 5b.

The beam deflector is positioned by gimbal torque motors. The commands are input to the torque generatoi of the rate integrating gyroscope. The output angle of the gyroscope produces the electrical input to the gimbal torque motor. Figure 5 depicts disturbance torques Td and Td e upon the elevation and azimuth control

For the particular configuration of beam deflector considered herein, it is assumed that the input beam source is mounted on the outer gimbal coincident with that gimbal's axis of rotational freedom. For such a configuration, an angular velocity of the outer gimbal and hence the input beam, about the elevation axis, W0 I produces a corresponding angular velocity oe of opposite sign of the output beam direction, This is accounted for in Figure 5a by the disturbance angular velocity, -w0' Se A rotation of the mi~ror surface with respect to the input direction about the e axis is easily seen to produce a rotation of twice that magnitude. of the output beam direction,

a loops respectively. These torques are dynamic reaction torques arising from the rotation of
* Other mechanizations of the two-axes beam deflector are possible in which a slaved platform is driven by the angular velocity of the mirror with a two-to-one gear ratio. The slaved platform body axes then correspond with those of the line-of-sight frame. Measurements of the motion of the slaved platform made by sensors which can be physically attached to it, can thus be used as measurements of the motion of the output beam direction. Such a system has been described elsewhere 1 and is not considered in this investigation.

-3-

T'd

wSfe

Tracker

6yro G

bMotor

Inertia

+ T

KG(s)

Figure 5a.

Basic Elevat'ion Loop

AeqTBk+Arin 2 C+AdCOS 2 C (see AWpendsi

A)

cos2c -

C/2)

Motor Ineti
Figure 5b. Basic Azimuth Loop

the gimbal frames with rques to been prespace. These dynamic respect have inertial sentt~d elsewhere 2 Apendix A contains ae the elevation the azimuth gimbal5 including derivation of and equations of motion for both
the dynamic reaction torques.

AeJe

"-K 132000

A(s)

- (s__+i_ 1)

(-s_.+ 67

1 1) )

(8 4 1) (-1

III. BASIC ELEVATION CONTROL LOOP DESIG EVALUATION

AND

In order to track a target with a constant line-of-sight velocity without a steady state tracking error, a type II control loop is desired. This is obtained by choosing the tracker dynamics as

Figure 5a presents a functional block diagram of a basid elevation control loop. In order to evaluate the pointing performance of
such a control loop the following representative

(27
c(s) =

+ 1)
8

parameter values were used

-4-

The overall loop gain can be adjusted by setting the tracker gain, Kc . In order to provide a basis for this choice, a root locus of the control loop with parameter values as given above is presented as Figure 6. To minimize the effect of the disturbances produced by the dynamic reaction torque and by the base motion, a large value for Ioop gain is desired. However, as evidenced by Figure 6, too large a value for loop gain will produce an underdamped response which would lead to unacceptangular sleas. able settling following commanded Thus a compromise value of loop gain is needed. The closed loop poles indicated on Figure 6 result from the choice of tracker gain, Kc, of 500. The basic control loop design is thu5 completed. To evaluate the effectiveness of the basic control loop design in rejecting undesired disturbances, one must consider the two sources of disturbance: the dynamic torque, Td I and the e input beam motion, Oe Both of these disturbances can be written in terms of the base motion wnere W

w0

0 e e

( it

WV)

is the component of base-angular

Vi velocity along the i direction. wV is the component of base angular

velocity a.ong the j direction. W is L the component of line of sight

angular velocity along the p direction. Tne angular velocities of the base and line of sight are predictable only as random variables. Thus these may be described statistically in terms of their power spectral density (PSD) be the PMS pointing error the analysis will functions. The result of angle, . Since 0 is the output of a liut.ar e eRMS system having several random disturbance inputs assumed to be statistically independent, N 2 0 (W) = _iw) 0

disturbances atd the required tracking rate.


functional form eT Tde ( W2 , 2V 2p WV WP L W~p p

In
where

i-l (we) is the performance function

relating the ith independent disturbance, Xi, to the pointing error, 0e.

ISO
100

Tne R4S pointing error is then obtained as


0

{iS

*2 eM

dw

50 0 160 SO -50

In order to carry out the above evaluation procedure, it is necessary to: a. Write the disturbance inputs in terms of the basic variables of base motion and pointing rates. b. Determine the PSD of each of these disturbance inputs assuming independence. c. Determine the performance function relating each of these disturbance inputs to the pointing error angle, ee. Stepti a end b above are carried out in Appendix*B. Step c is readily accomplished noting fl:om Figure 5a that

SISO

-i00 15

Figure 6.

Elevation Root Locus

-5-

I.I

0e -(s d '-a and

1' Kc Gc 3bance + c c cc

G+

As evidenced by Figure,5, the pointing error arises due .to both dynamic torque disturand also due to the angular velocity of the input beam, (, produced by the base e motion. In brder to evaluaie the relative importance of each of these disturbances, the pointing error resulting from the dynamic torque disturbance alone-is evaluated using thd above procedures. The RMS pointing 6rror caused by the dynamic torque alone is found to be

e wo 0 e

1 2 KG F~ cc 2 KK GK

KG2 + GI

)
S

+
2K

+1

(byidynamic torque) - 1.8 prad

The PSD chosen to represent aircraft angular velocity about each axis is given as (0.1)_(14002) 0v 4Wv (W2 + 1.52)(W2 + 14002) (rad/seC) 2 ( ) (2 hz

i is From the above, it in readil concluded that the overwhelming cause of the pointing error determined above is base motion disturbance. For this reabon, the-dynamic torque disturbance effects will not be considered further. For mostprecision pointing applications, the RMS pointing error angle of aboue 1 milliradian determined:above isinot acceptable. In order to reduce the pointing error, base motion compensation is evaluated., c IV. FEED-FORWARD COMPENSATION FOR BASE MOTION , In order to reduce the pointing error caused by the motion of the input beam, i.e., by base motion, a feed-foreard compensation schene is considered. This scheme proposes to add a compensation signal, Xc, as a command torque to the.rate integrating gyro in the elevation control scheme innpr loop. The compensation command must be chbsea sucil that the disturbance of inpht beam motion, -wO , shown,
eI

This base motion PSD is plotted as Figure 7. Using the procedure outlined above the' PSD of resulting pointing error for the basic elevation control loop is determined. This PSD is plotted as Figure 8. The RMS pointing *error angle is evaluated to be 0 eRMs s 2 ='
I

= 942.8 urad

c4 -2

"4r
-6 -1 0 Log f Figure 7. u r ad 22 1 (hz) 2

in Figure 5a is effectively cancelded. Consider the portion of twe elevation cohtrol loop reproduced from Figure 5a as Figure 9. In order to determine the proper cqmpensa I tion signal, )", the performance fundtion of the inner lqop must be iAvestigated. qsing the dynamics postplated previously', the olen loop performance function of this ifiner loop is M Pointing I3rror:2000 s_32_ 1)7 + PFoL(s) + 1)

Base Motion PSD

S
-10 -12
0

' rrr,
-1
Log w Figure 8. (rad/sec) Pointing Error PSD -6

402s + 1)

1
The resultant closed loop performance function for this inner loop is

S!

KG s)

It

a0.

Fur

9. Elevaoi

Loop

lnne

s th aboe Inasuch prfO~ancefuncion


%1 PFCL(S) Figur

asFiuch10 feed-forward compensation isiga hosen


gerralePDs F.Eea ionurner 10oo-pon frtepeeti sasmdta Notthat

nau

as the abov

c f

( + 1)'

(" + 1):
=

feed-for-ward compensation s shown sur1

o) Usn

th c ompesate

contrllop~
niaral

of~

+12)) (42(120) ( (402)(3f00 + P11L(s)

U~6 ersuingth that

compnsteol loop

nludingf th a

foritheipresentitliscomslete

II " P)Ls ee The disturbance caused by input beam motion, -( II can be eliminated by tne addition of a e

1.8 Prd caused valbe in by the dynamic torque disturbance remaining. R1 e pointing error aurmn eI of f 0;0i

10C

" Tracker h CYro Motor

TC

Inert in

Figure l10.

Ehevaton Loop with Feed Forward

-7-

?.thoug.. vhe above shows that complete Ahu .newhich compensation i1 thboretically possible, one must conside: the practical feasibility of (;eLermining the locations of poles and zeroes of an actual hardwav inner loop in order to construct Gcomp(s). In reality, 'ne could not determine the locations of singulirities to five decimel places as assumed 4save. The approximate frequency response of -the actual inner loop could be determined over the frequencies of interest by exhaustive testing using known inprts. The approximate locations of the singularitles could then bo determined analytically to match the measured frequency response in a satisfactory manner. The proximity of the singularities determined in such a manner to those of the actual hardware control loop would depend upon many factors including the accuracy with which the exper!mental measurements are obtained and the 0.i& of the investigator in choosing an analytiLal function with the correct number and fors oE poles and zeroes. As a very optimisti.c estimate of the accuracy of such a procedure, the following "realistic" compensation filtzr is

error is attributable to the imprecision with G om(a) given by Equation (3) approxicomp mates the inverse of the closed inner loop function PFL(s). s The base motion isolation performance of the beam deflector, elevation loop will be further degraded for a perfect measurement of input beam motion, w0 I is not possible. In e reality, such a quantity must be measured by a physical sensor which will add both dynamics and noise. In order to provide a highly accurate measurement of wO p a rate integrating gyro e will be used. By providing a constant gain feedback loop around such a gyro, the electronic analog of a mechanical rate gyro is created without the mechanical problems usually associaoe with rate gyros. The output of a rate integrating gyro is characterized by 3 Ag I g" (A

assumed:
0.5(& .+I) -a )Comp (a..1 2 "153(*)

cImp

(+

) (3

+ 1) 2

where
A , output angle of the gyro 1 g Tg . gyro characteristic time c I :
-

The'additional poles in the filter are required to keep the gain at high frequency finite. Using G

comp

(s) as given by Equation Z3) in

gyro spin angular momentum

the block diagraw of Figure 10, the pointing error PSD was again determined and is plotted as Figure 11. The RMS pointt:ig error angle

- gyro damping

WIA = angular velocity along the input axis WOA - angular velocity along the output axis

was determined to be
a
- 13.1 urad

MS
It must be emphasized that the above error results even though a perfect measurement of the input beam motion, w , is assumed. The e urad 2

(iC -commanded 4D
(v)w 9

angular velocity

gyro drift rate uncertainty (i.e.,noise)

Since the gyro is to be used in a feedback mode, A will remain very small. In addition, for simplicity of analysis, cross-axis ccupling %ill'be neglected for the present. Thus the rate integrating gyro equation is approxim.tely

II
14 McnS Pointing Error: Lo 13.1 .(grad

vdrad

Q-16. -1
Figure 11.

A Iss

1
Log w

2
(rad/spc)

To use the gyro as an electronic equivalenr of a conventional ra~e gyro, the command angular
velocity is made proportional to An. Figurrm 12

Pointing Error PSD With Feed Forward -8-

CC

Figure 12.

Gyro Block Diagram

Note that the output is obtained by a final multiplication by gain Kf to insure a unity steady state gain. The measurement loop output, C(s),
related to the desired input we(a), by

comp~snsarion filter might optimistically be choser. as GcoMp(S)

is 0"5(I+ 0(d0_'190 + 1
s2 %20.7)s,, s2 2(0.4)sj

T5_32 +

1_5_3

Y 94-2

1794-

SC(s) Ggyro(s)

(s) *e

1c
+t

0+

K s f e

sKf

where again additional poles have been added to provide finite high frequency gain, i.e., a
physically realizable filter.

The purameters for the rate Integrating

Using the above filter, the overall block diagram of the elevation axis control loop is shown in Figure 13. With G (s) given by Equation (5) the Comp diagram of Figure 13 was used to determine the pointing error PSD for the elevation axis loop. Assuming no gyro noise in the rate measurement loop, the RMS pointing error
was determined as

representative and correspond to those for a Kearfott King Series miniature rate integrating gyro model C70 2519 001 4 . The parameter values are s= 15 cg 9control
C
*

S1block

0.006 seconds

The choice of feedback gain, Kf about this rate integrating gyro is of necessity a compromise. A high gain produces a desired increase in natural frequency but at the expense of i decrease in damping ratio. A compromise L oice of Kf a is made. This 15 choice zesult in a natural frequency of 193.6 rad/see for the second order dynamics of the rate measurement with a damping ratio of 0.43. Since the designer is aware *that the rate measurement process involves second order dynamics, he can attempt to modify his compensation filter, Gco p(s), to partially compensate fcr the measurement dynamics ds well as tile dynamics of the inner control loop. Again, sincethdynamics of the yo m Agasureent s in c e th e dy n am ic s of th e gnnere asu re men gy ro me
tus

0 eLMS

= 29.2 iirad

Thus it is apparent that the ability of the feed-forwaid signal to compensate for input beam motion is sonewhat degraded by the addition of measurement dynamics. It should be noted, however, that the measurement dynamics are a function of the specific sensor selected. Rate integrating gyroscopes are available with characteristic times, T , an order of magnitude smaller than that used in the above calculations 5. The usefulness of such additional response is determined by the required pointing accuracy for a specific mission. used to measure input beam motion, w 0
d In addition in dynamics, oano actual Ssensor t me ur to t aim ,w c nb
,

loop are not known exactly, the feed-forwae

can be

W0 ro

iC

Gcomp~s

KC C(s) Tracke,
C!

G)1 Gyro Motor Inertia

Figure 13.

Overall Elevation Loop rotation of the pointing direction about the p axis results not only from rctations about the k axis which are controlled but also from rotations about the n aris. Inasmuch as the controlled direction has no gimbal isolation from the base about the n axis, rotations of the base about this axis produce rotational disturbances of the line-of-sight. In a mpnnei analogous to that described in the preceding sections, a feed-forward path can be added to reduce the effect of base motion. Such a loop is shown in Figure 15. The angular acceleration of the base about the n axis is taeasured and providea as a command torque to the azimuth torque motor. The performance function for the angular accelerometer corresponds to that measured for a Systron-Donner Model 4590-F-l-AG angular accelerometer 7 As with the elevation control loop, a filter is added to compensate for the measured dynamics of the control loop and of the angular accelerometer. Again, the matching of the fiiter to the actual dynamics is chosen to be an optimistic estimate of that which could be attained. Using the base motion PSD of Equation (2) and the above control loop, the PSD of azimuth pointing error about the p axis is determinad. The RHS value of this pointing error produced by base notion Is found to be 8 p 14.2 Pirad 42r

e:tpected to add measurement noise. Adequate test measuremtents upon a representative integrating gyro to determine its noise output across the frequency spectrum of interest are very scarce. A noise model PSD for a Honeywell gyro model CG159CI has been reported as 6 0(f) 210 + [fen/hr)2 f-_1(1+f) hz

Assuming the gyro noise to be uncorrelated

with all other disturbacesl the performance of the ewas again evaluated using the above gyro noise PSD. The RMS pointing error caused by gyro noise alone was determined to be

eS

(Syronoise) - 0.19 urad

Recently, information on the noise PSD's of several high quality gyroscopes has been published 5. The noise is presentee in terms of equivalent input angle in arcsec 2 . Across Sh? a spectrum of 0.01 hz to 100.00 hz, several of the titsted gyros are shown to possess RS uncertainties of less than 0.05 arcsec of input angle. This is consistent with the RM pointing error determined above and indicates that short term gyro noise is an unlikely source of significant error in the operation of precision optical directors. V. AZIWT~t Loop DESIGN The basic azimuth loop of Figure 5b is redrawn as Figure 14. It is seen that the
-

10

'2

[Tda

/"Onsin

S-G

c(s) c
Tracker

"
Gr

K~
oo nri

co+s -

2 Aeq Bk+Arsin 2+C.+\cos c (see Appendix A)

s - cos(2c - -/2) Figure 14. Basic Azimuth Loop

0n0

Trackers

_Aeq=Bk+Arsin 2 C+AdCOS 2 (see Appendix A)

cos (2 c - w/2)

Figure 15.

Azimuth Loop With Feed Forward

VI.

ADDITIONAL CONSIDERATIONS

gimbal non-rigidity.
4 As indicated by Equation( ),an angular gyro case about its output axis velocity of the will produce an output signal from the gyro which is not desired. In the control loop presented for eltvation axis controlt it is assumed
-

In an attempt to.assess the minimum pointing error which might be attainable with an airborne two-axisbeam deflector of the design considered, several complications have been neglected. These include gyro output ayis cross-coupling, gimbal bearing friction, and

11

that a single degree of freedom, rate integrating gyro is affixed to the bean deflector with its, input axis aligned along the e axis. The azimuth control loop assumes a similar gyro mounted on the outer gimbal, with its input axis along the k direction. In order to avoid output axis coupling between the loops, the output axis of this azimuth gyro can be aligned with the n direction. On the other hand, the output axis of the elevation Syro must lie in a plane containing the k direction. Hence controlled angular velocities of the azimuth loop about the k axis will produce an erroneous output of the elevation gyro due to its output axis coupling. This coupling is diagrammed in figure 16. Although such coupling provides an unwanted disturbance, it does not alter the loop stability.

the natural frequencies and stability of the loops. Each of these neglected effects can be expected to increase the pcinting errore. The amount of degradation of loop performance could depend upon the specific mechanical design of both the beam-deflector an' its gimbals. VII. RESULTS

For the control loop design and base motion spectrum considered in this investigation, the RMS pointing error for the cases considered is summarized as follows: ELEVATION LOOP

e
1. 2. No feed-forward compensation Feed-forward compensation*. Perfect mcastreme' of w0 e Feed-forward compensation*. Gyro2nd order dynamics in measuring w0

in prad 942.8 13.1

3. CYro Cross Coupling

29.2

4.

Feed-forward compensation. Gyro 2nd order dynamics* and gyro noise in mcaturing wO

29.2 na

;,

Azimnuth Loope
'1

~~~Figure

16.

Gyro Cross Coupling Between Loops"

e
6 RMS
Feed forward compensation.* Accelerometer 2nd order dynamics

in virad

in measuring 40
n In addition angular velocities of the base about the n direction provide erroneous outputs from both the azimuth and elevation loop gyros due to output axis coupling, The effect of output axis coupling can be reduced by compensation. For example, recognizing that angular velocities about the n and k axes will couple to the elevation control loop gyro, one can measure these angular velocities with appropriate gyros. Knowing the characteristic time of the elevation gyro, a compensating command torque can be. provided to compensate for the unwanted output axis coupling. In addition to the effect of gyro cross-axis coupling, the problems of gimbal bearing friction and gimbal non-rigidity have not been included in this analysis. These effects could alter -12-

14.2

VIII. CO4CLUSIONS 1. The use of feed-foivard compensation can significantly reduce the pointing error caused by base motion of an airborne two-axis beam deflector. 2. The resulting pointing error depends upon the degrei with which hardware dynamics can be measure't and matched.

* Feed forward compeneation includes "optimistic" matching of control loop dynamics.


** Effect of gyro dynamics can be ,educed by choosing

a gyro with a smaller characteristic

time.

3. Assuming.-the use of a high quality gyro, noise in the gyro used to measure base motion for the feed-forward compensation does not significantly increase pointing errors. Appendix A Derivation of Dynamic Eauations This section is intended as a detailed derivation of the dynamic equations of the two-axis-gi-,-.balled beam deflector as depicted ir. Figure 2.

axes and that the gimbals are balanced about their respective axes of rotation. Inner Gimbal Dynamics .-. E.Torques on the inner gimbal INERTIAL aLout its c.m. Applying the law of Coriolis di (tu
=

(A.1)

d dt INERTIAL

---

d1
" M

+
+ ;x(JM M) (A.2)

(Jg')

The following definitions are used in


the development: Coordinate Frames

.- 1 d

+(
M nnH

7 dt

Writing coordinates in the I frame

V frame:
0 frame: H frame:

itj,k coordinates fixed to the aircraft


n,e,k coordinates fired to the o u t e r g ima l b r,e,d coordinates fixed to the JAe

. [r
. 0

~
0

A
0

Ad

I
(A.3)

mirror, i.e., to the inner gimbal


INERTIAL frame: inertial frame with unspecified coordinates axes -

-O
r H e M

Notation Superscript indicates the coordinate frame in which a vector is coordinated. Subscript indicates the coordinate frame with regard to which d,.rivatives are considered. Variebles n Mirror azimuth angle - M Mirror elevation angle WVio U Vil WV k - Aircraft angular rates about i,j,k axes, respectively. about ni,e~k axes, gimbal angular respectively. WM tor B, XJ H~ r AruH r d=

Hr

d e d

Yn' Ue' tk

rates "Outer re

A4e

Ad

ul
aV

W d

- Inner gimbal (mirror) angular rates about r,e,d axes, respectively.

(AdA (A-A

B, Bk " Principal moments of inertia for outer gimbal about n,e,k axes, respectively. for inner gimbal (mirror) about red axes, respectively.

(A.4)

Ar, Aet Ad - Principal moments of inertia

(Ae-A'r

Ne

It

is assumed that gimbal axes are principal


- 13-

'7I

Suostituting (A.3) and (A.4) Ar Torques-

into (A.2) H
MH

0
:e 0

0 (ANe3) Bk

+ (Ad -Ae)

" (A.5)

Ae Le + (Ar-Ad)

e
Ad + (A -Ad e

r d
nHd WO e LOkj
'O

Taking the-component of (A.5) along the e axis yiaIds (neglecting bearing friction) Te -l e "A e+ (A-rAd) wHd wMr (A.6)

(A.14)

however, due to gimbnl constraint:

"W . o
r n

cos c -W
k
sin W On W
c + w0

sin c
Cos C

(A.7)
1io_1
(A.8)

1)0

"Wd "H On
W - (W +

ft

"0

(A.15)

sin c cos c (cos


2

c - sin 2 c)

(A.9)

.Substituting (A.9) Aet e_(Ard

into (A.6) yields 2 2 1sin 2 c

n BnwO BeOe

Ok Bk

"-(Ar-Ad)
rquation (A.1O)

wO

0k cos 2 r

(A.10)

or

describes the dynamics of the o n n (Be-Bn) - -Torques the on its c.m. on WO e (k.6)

as a function of outer giubal angular velocities. Outur Gimbal Dynamics

o
I R-Labout

outer gimbal

(A.11)

Substituting (A.15) and (A.16) into (A.12)

INERTIAL

[
n JO T'orqku

o + (Bk-*e) "o
e -0k U + (11B + (Be -Bn)
W0

The torques on the outer gimbal are pro-

vided by the azimuth torque motor, reaction

from the elevation torque motor, and by gimbal restraint from the base and from the inner gim oal. _Ok Td --di INERTIAL 0 (A.12) Writing coordinates in the G frame

Torques
IOk

e Bk

oBr (AB17)
WO

Consider the torques on the outer gimbal along the k axis. These are: a) The k compotent of the torque THO exerted by the inner gimbal on the outer gimbal. b) Tk, the azimuth control torque. Now

-14

torque exerted by -Tom where T%tis the

Thus

the cuter gimbal on: the inner gimbal.


giv

i., i

-Py Equation (&.5) in H coordinates. Thus the comionent of .4I0 along the k

A d-j dt"

d-j0 + dt + d = "-

,w

" 0)

(A.25)

Writing (A.25) in-M frame components

axis is:

*I K

-AA Cos c +(A Md

C -Ae)WH~ Cost

ea

;dlM . Af1 4
0

W -W
.
0

(.6

+AW Mr sin c + (AdAe) wMe6d sln c Substituting (A.18) into (A.17)


O (B -Be)w

Substituting (A.22) into (A.26) yields


(A.19) (A.sn8)M
-

+ Tk

Oncose-

sine

ne
4)s e AtAe) wH d (rd e+

nk

(A.27)

Now

Now

[.Oi e .rCss H4 Arco Ae'Ke A&COS IM


;M dt, H A.0

e H

u 0 -u)

-H 0 + Cans in

Coe

(A.28) osCO

dHJ

0
w. cosc - wo sine

Oe-Me0
w.eW Onsine + W kCos

SSubstituting
yields

(A.28) and (A.27) into (A.23)

o I-1c0;11 dtWO(A.21)

e ncosc - oksint + w. ""


IK UH M n sn

sine +

0 cosc ek
(A.29)

cost
e -yk +iOlcrlyit

From (A.21)

""0 Cos
d Id dl

C-

sin c ksin4

1
( 2

"

"
n

W,^ ~ ~+ Wa eu

cost 00 nA( e
"

(3 .30
Wn"sine

cos sine + wM U OB

0O sin C + O Cos C

Substituting (A.29) yields

and (A.30)

into (A.19)

Tt

"t

x D Io') WH (A.23)

Now,

obviously

~+(H -;0) 14'0

(A.24)
-s5-

I
B w Again rearranging (A.34) produces (BAsin2C + Ad O2ek Tk (iB e WO0w +Tk inn +Ar0 cose sinc n '+A
,

A- i2 + Ar wwsin2C
ne

rO w
,,

0n rOke 0 kSint cose-A rwMe (4~ sin2c-A r'meWO sin cot **r~w~ ine sn~e- w~ ks nc cost W. 0 ostA

(Bna~-,or 2 e + Acos20) -B +Asin d:. 0W e


-(A

e~o
n

He -A cos 2 c + Adcos 2 e)w nWM 0 2 w

sAnc cosc - Ad nd ccost A

C,)s2c + Adwo w0 cos 2 c e n


swM '0 C09cr

sin cos + kde -A ) scn c(A der)O

-dwOwk

ee

- (Ad-Ar)

sin c cos c w0 ek

+ AdWH

snc cos

+ Adco

(A.35) Now,

+Arw

w cosc -Aw

(wdSint + w a

cost)
(A.31)O

Me r
Now

(_

(A.36) 0 k.

Wd

iO sCn

+WOCos cs n

(A.32)

"W

WO cos c - wOksin e

(A.33)

yubstitin8 (A.32) and (A.33) into (A.31) (+A


2 r sin e + Ados 2

) L o2 Tk wO sin2 C
e nc

Id

vi -V}

(A.37)

4(Bn-H)w
n

w +Ar
e n

cosr senc + Aro

coOSr+(An + ArWO wO sin e cos c - Arw1ewO sin2e Arw w0 sin c cos e - Ad4onSin C c? ILI ;J -W' sinn + csn (A.38)

2 + AdWO ewn cos c - AdcWOewksin e cos

"

AdWM wO

2 M coss C + Ad Ne 0ksin c cos c

Now
-

+ AdWM
e

2 (W sin c + W0 sin c cos 0 0

d- " v

-W 0

(A.39)

+ ArW

2 (W0 cos c wHe n

kOSine cos e) (\34)

Again, obviously

o"
Thus

v + tw.o

'VV
I

AA

dt

dt

dt

-W

(A.41)

16' \
l6~

dt~

~z

+Odt (WOW)

Writing (A.41) in 0 frame components 0 + JAI ;v 0ad "10-JAI(A.42)B /

Equation (A.47) describes the dynamics of the beam deflector about the k axis in terms of base motion and motion of the beam deflector

about the k and e axes.

Substituting (A.38) into (A.42) yields WV, Cos n + WVi sin n d sin rl + V Cos n (A.43)

Disturbance Input Relationships The disturbance inputs to the elevation loop, Td and e ,0 and those to the azimuth loop, Td and w0 n can be written in terms of

I -l

base motion and required line-of-sight rates noting:


kOJkby

L
Nown (W " V )0 x X

N0 -- 0 0 =(A.44) e 0 sinn+wV coon k0 W0 O e k -" 0 wk kI WV k

= WVi cos q + WVi sin n


-

(B.1) (8.2)

WVi sin n + WV coo n


VIV

n 0 WV cOsn+wV sinn i J

(A.10) and (A.47) results in all disturbances


expressed in terms of base motion rates and required line-of-si%ht tracking rates. In order to determine the !SD of the pointing error, it is necessary to know the PSD's of the various disturbance inputs. This can be readily accomplished knowing the PSD of each component of base motion and required tracking rates if independence is assumed amongst the various angular velocity componentR. As an example of the determination of PSD consider the disturbance input term wV WV The PSD for wV Equation (2). Assu.ing independence, it shown that VO sin n + wV i Substituting (A.45) and (A.46) yields is directly

Substitution of (B.1) and (B.2) into

-WV

Substituting (A.44) and (A.43) into (A.40) produces


n

I cis n + ij

sin n V WV
W+

(A.45) cos

+ WVi W sin n V

WV W
i Also k

sin

I1,

cosn
k Pj

i or wV

is giva,- by

-v-R"n cos
into (A.35)

(A.46)

(T)' WVW

() WVCr) WVC

(B.3)

i J i V where Rz(T) is the autocorrelation of the variable z. Taking the inverse Fourier transform of Equation (2) yields

(Bk + Ar sin2 c + Ad cos 2 c) L0 k T


2 + (B -B 4-A sin c + A cos n e+r d C) 2

n'O nk e0
neI

rl
n+
wVIWV ksinn

000I
(B.4)

(Ad-Ar) sin c cos cV IV cos

n+

V1 siu

V WVkcos n - 2w V Oksin n + 2W VjwOk cos n j

Substituting (B.4) into (B.3) and taking the Fourier transform yields

(A.47)
-

17

0.01

.6667

.6674

(B.5)

4.

Lichtenstein, P.3., Gyros. Platforms,

j
I

The PSD for other disturkance inputs similarly developed.5.Tnae,.,oegbrg

Accelerometers, Technical Information for the Engineer, Kearfott Division, General Precision, Inc., June, 1963. 5. Truncale, A., Koenigsberg, W., and 11arrisR., ,an arsg, Spectral Density Measurements of Gyro Noise, MIT Draper Laboratory, Report E-2641, Cambridge, Mass.; February 1972. 6. Weinstock, H., Design of a Precision Tilt and Vibration Isolation System, NASA TR R-281 NASA Electronics Research Center, Cambridge, Mass., March 1968. Results of Tests on Systron-Doller Angular AcceleroSeter performed by General Electric, Space Systems Organiat1on, Vally y Forge

REFERENCES 1. Whitaker, J.L., and Burdin, C., A Gyro Stabilized Heliostat for Airborne Astronomy, ISA Transactions, Vol 5, No 2, 1966. Ree, A.K., Stabilization of Precision Electrooptical Pointing and Tracking Systems, IEEE Transactions on Aerospace and Electronic Systems, Vol AES-5, No 5,

2.

7.

September 1969.
3. Wrigley, W., Single-Degree-of-FreedomGyroscopes, Report R-375, MIT Draper Laboratory, Cambridge, Mass.; July, 1962.

Space Center, Pa., 10 July 1970.

18 -

Você também pode gostar