~1 iNS 9 MAR1944 O RESPONSE CHARACTERISTICS OF THE VALVE CONTROLLED HYDRAULIC SERVO by Lewis D. Lipschutz Submitted in Partial Fulfillment of the Requirements for the Degree of Master of Science in Mechanical Engineering from the Massachusetts Institute of Technology October 1943 Signature of Author: Signature Redacted Signature Redacted Certified by: Thesis Supervisor Signature Redacted Chairman, -De p&tment, Committee on Graduate Students CRA Table of Contents Page Acknowledgment......... ... ............................. Introduction ................. 1 3 ....................................... Description of Servo Investigated, and Notation Description of the Valve Controlled Servo ................ 4 6 Diagram of the Valve Controlled Servo .................... .............................................. Notation. 7 Limitations on Servo Investigated ........................ Assumptions Made About Servo Investigated ................ 9 11 Results, and Suggestions for Further Work .................... Family of Theoretical Response Curves Discussion of the Response Curves ........................ 12 Comparison of Empirical and Theoretical Response Curves .. Discussion of the Agreement between the Theoretical 15 13 Operation of the Apparatus ............................... 16 17 Suggestions for Further Work ............................. 18 and Empirical Responses......... ................... Dimensional Analysis and Solution of the Differential Equation of Motion Derivation of the Equation of Motion ..................... Dimensionless Analysis .................................. Series Solution of the Differential Equation.......... Step by Step Solution of the Differential Equation ....... Stages of Operation....................................... 20 25 29 31 34 Description of the Apparatus Description of the Construction, and Test Procedure ...... Photographs and Drawings of the Apparatus................ 38 41 Appendix Sample Calculations ......................... ............. Successive Differentiations of the Equation of Motion .... Data Used in Plotting Theoretical Response Curves ........ Sample Tapes Showing Empirical Response Curves ........... Calculations and Data for the "Comparison Curves" ........ 2 3G5~I 45 50 54 57 58 Introduction The only analytical work previously done upon the valve controlled hydraulic servo has been by Professor C. E. Grosser of M.I.T. In his article, "Designing Hydraulic Servo Circuits", Machine Design, January 1942, he described the valve controlled servo and derived its differential equation of motion. This article was, essentially, the starting point for the following thesis. In "Apply- ing Hydraulic Servo Circuits in Machines", Machine Design, April 1942, the differential equation was solved and the response of a given servo was calculated. This second article is largely superseded by this thesis wherein a better solution has been found for the differential equation. The valve controlled type of servo enjoys a wide use, but its design has always been based upon experience and upon trial and error. There was a definite need of a design chart which would pre- dict the response of a given servo under given conditions, a chart which would indicate the effect of changing the various dimensions of the servo. The purpose of this investigation was twofold; for one, to develop the design chart mentioned above. This was to be accomplished by a dimensional analysis of the servo, a complete solution of the differential equation of motion, and a presentation of the resulting curves in terms of dimensionless parameters. The second object of the thesis was to build a valve controlled servo and check the validity of the response curves which were mathematically calculated. It should be noted that the name valve controlled servo is 2 used in this thesis to denote the servo which is sometimes called a pressure controlled servo. It was thought that the former name was more descriptive of the nature of the control. MMd 3 Acknowledgment The author wishes to express his sincere appreciation of the help received from Professor C. E. Grosser, faculty advisor to this thesis project. In the midst of his own pressing work he very patiently discussed the many problems which arose during the investigation, and was instrumental in obtaining most of the material and parts used in the construction of the servo. The machine shop work was done with the help and advice of C. A. Bulfinch, R. J. Bowley, and C. W. Christiansen, instructors in the Machine Tool Laboratory. 4 Description of the Valve Controlled Servo For a complete description of the valve controlled hydraulic servo see the magazine articles mentioned in the introduction. Only a brief description of this type of servo will here be given. The primary elements of the servo motor are a source of high pressure oil, a four-way valve, and a hydraulic ram. When the system is in equilibrium, the valve lands cover both orifices of the valve and there is no tendency for the main piston to move. Now, if we wished to move the load a certain distance to the left as quickly as possible, we would suddenly move the signal shaft to the left and then hold it stationary in its new position. The diagram on page 6 shows the servo just after this step signal has been imposed. The valve spool is displaced to the left, putting the right hand end of the cylinder in communication with the source of high pressure oil, and the left hand end of the cylinder in communication with the reservoir which is at atmospheric pressure. The difference in pressure causes the piston, with its load, to move to the left. This motion of the output shaft, now that the signal shaft is being held stationary, moves the valve spool back toward its neutral position, closing the the valve ports, and throttling the oil flow to and from the cylinder. When the valve spool again covers both orifices there is no longer any pressure difference between the two ends of the cylinder and equilibrium is established at the new position of the load. As just described, the follow-up was without overshoot, but under some conditions, of course, overshoot and oscillation occur before equilibrium is reached. 'I DESCRIPTION OF SERVO INVESTIGATED and NOTATION 5 In order to avoid negative pressures and possible sucking of air into the system, both ends of the cylinder are connected through check valves to the body of oil in the oil reservoir. In order to protect the cylinder from excessive pressures, both ends of the cylinder are equipped with relief valves. To avoid complication, these two refinements were not shown on the diagram. We will investigate the response characteristics of the valve controlled servo in the simple form just described, but the results will be applicable, through simple and obvious transformations, to the cases in which this servo is used to control the rotation of a mass. In such cases the servo may remain as described with the simple addition of racks and pinions to achieve a limited rotation, or else the hydraulic ram may be replaced by a fluid motor and the translating four-way valve replaced by a rotating one. The results to be obtained will still apply provided that the limitations later to be stipulated, such as leaklessness and inelasticity, are met. 6 MYDRAULIC RAM //////H ////// OUTP FOUR WAY VALVE II -a1,I -- L- . TM~ I -~ ~A LOW PRESSVRE ~ P% RE SERV0Fl PM HIGH PRESSURE -- Pi ACCOtPAL AT DIAGRAM OF VALVE CONTROLLED HYDRAULIC SERVO Eq. 7 Notation t = Time x = Displacement of output shaft in response to signal x -dx = - x= = dx = dt R = Velocity of output shaft = Acceleration of output shaft Ratio of the displacement of the output shaft to the dis- placement of the valve spool (see diagram on preceeding page) e = Magnitude of the step signal measured in terms of displacement of the valve spool S = Re = Magnitude of the step signal imposed, measured in terms of displacement of the output shaft v, = Average velocity of liquid in the connecting lines M = Total effective mass of load, servo motor, and accelerated liquid F = Load force; always in direction opposing motion of the load A = Area of ram piston a, = Connecting pipe and passage area ao = Valve opening orifice area (This quantity is a variable, varying with movement of the output shaft.) = a = a, f = Pipe friction factor; a function of Reynolds number. (a0 )max See Limitations on page 9. Approximately equal to 0.02. L = Total length of connecting pipes and passages d = Diameter of connecting pipes and passages n = Number of entrance and exit losses in circuit K = Average loss factor for entrance and exit losses in the circuit. Approximately equal to 0.75 8 C = Coeffi cient of discharge at rectangular valve orifices. mately equal to 0.6. pH = High p ressure pL= Low pressure p =PH ~ I See page 18. L Mass density of the liquid Dimensionless Groups: T = J X = 1 -- x = R A MSa X=dX dT 2 2 -dT Dimensionless time = Fraction of step signal yet to be recovered Design parameter Approxi- 9 Limitations on Servo Investigated In order to reduce the number of variables in the differential equation of motion, certain limitations must be imposed upon the construction of the servo to be investigated. Only if this is done, can the equation be written in terms of three dimensionless parameters, as is necessary if the results are to be presented as a single family of curves. The limitations, with some comments on their importance, are as follows: (1) F = 0: This means that only a mass loading will be The load force, load friction, and friction in the servo considered. motor itself are all to be negligible. As far as the load is concerned, it need only be mounted on ball or roller bearings to meet this restriction. As far as the hydraulic ram is concerned, leaklessness should now be attained by using close-fitting, carefully aligned, machined surfaces rather than piston rings and piston rod packing. The experi- mental servo described later in this report was constructed with the purpose in mind of eliminating leakage and yet keeping friction to a minimum. (2) pressure: Relief valve pressure equal to three times the operating This is not a serious limitation, for three times the operat- ing pressure is a reasonable safety limit. Besides this, most actual servos will have enough leakage so that the high pressures theoretically reached, will not be reached in practice, and the relief valves will have no occasion to operate. (3) No step signal beyond full orifice opening of the valve: 10 In other words, e is never to be greater than the length of the valve ports. This, too, is no serious restriction, for a step signal which fully opens the valve ports is quite large enough to be a severe test of the servo's response characteristics. (4) The valve orifices: (a) Square valve orifices: This is a rather serious limitation in that it excludes such other shapes as the annular orifice, and the rectangular orifice with tapering ends. In spite of this standardizing on a square orifice however, the effect of varying the size of the orifice, and varying other design dimensions will still be apparent in the results. (b) a, = (ao)max; that is, the area of the connecting piping equals the area of the valve orifices: limitation, but only reasonable design. This is hardly a A slight departure from this equality would not materially affect the response of a servo, and in such an event, the common area to be used in calculations should be taken as the orifice area. (c) Balanced valve spool attained by splitting orifices into two equal areas, diametrically opposed: This means that what has previously been referred to as the valve orifice will now refer to two square ports, diametrically opposed; and the sum of the areas of these two ports is to equal the area of the line. 11 Assumptions Made About Servo to be Investigated In addition to the limitations just listed, two assumptions must be made before deriving the differential equation of motion for the servo. It is assumed that the apparatus is leakless and inelastic. In practice, both of these assumptions can be met quite closely. If rubber pipe lines are not used, or used only in short lengths, and if no gas is entrapped in the circuit, then the apparatus will be practically inelastic. To approach leakless operation it is necessary to machine the piston and cylinder of both the valve and hydraulic ram to very close tolerances. The effect of leakage will not be entered into analytically, but qualitatively it may be said that leakage would not affect the initial response very greatly, but would allow more overshoot than a leakless valve and ram. RESULTS and SUGGESTIONS FOR FURTHER INVESTIGATIONS T 1*11 1>1 I. -A ki g kWid I... I 41 ~- ~~~ ~~ ~ ~ ~ ~ ~~~111 1' Jil - *p ~ q, ,q,~~f~~. ~ I, 4k ii jj- II ,,Y,,I 11' FP 7T II .4-2 I, -"1- - 1 6 TL~2 litkri 'HT, 1 jl 1 -1-i-v ................ [n.~ ?7t 1 L i . . .. . . . . . ft 121: Lll 11 .. . . . -4- . 11dl_ i 4-44-11 1 1 11 4 44+ 41 7I~IV t i ___ ~iA~t~T1 -Tlj I 1:2 7.1 424 ___T]I IT I7 77bd 1~%~/ ~V~ V1 114+ 4;, LII ;t LI ii. . I I - * * +4~t2 4IT K~~vrr 1 ii '1'~~ I. ii 1 LET 1T_ I IE lEE_ _. lan 1:11 L I I . iJIL L i L-1 1LLL _T1 Li :71 rTIht i-f LLL~i K .1 74 Ti~ 4 i 22. I -j-~ 1- Iij' : +4' 411.--ILI' 11 . mka _T_ I. I-- F- hW~1WV i 1+1 I T I. 7 F--- M~&;~sL4 I ~ 19 V J L 1-i * -~ I - ~1~ .1 1'* 'iT: I:," 77 +Jr I I~~~- 2 1 -4~- 1 *-7 7-- 7112 .2~ 13 Dimensionless Response Curves The curves on the preceding page are the most important results of this thesis. In fact, their obtainment was the first object of the investigation; the second object being an empirical verification of them. Dimensionless displacement is plotted as a function of dimen- sionless time for various values of the dimensionless design parameter. The absence of overshoot in some curves, and the rapid extinction of overshoot where it does occur, indicates that this type of servo has a strong inherent damping. not slow the response. Unlike other servos, this damping does It is of considerable interest to note that the design of this servo need not be a compromise between fast response and small overshoot, for these qualities occur together in the valve controlled servo. Noting that Ir - R 2 A3 MaS , the effect of each of these variables on the shape of the response curve can be determined. Increasing the load, orifice area, or step displacement of the load decreases 1 linearly and leads to curves with higher overshoot. Increasing the lever ratio between the output shaft and the valve shaft, or increasing the cylinder area of the ram results in an increase in IT proportional to R2 and A3 respectively, and rapidly leads to curves with less overshoot or none at all. The fact that R, a, and A appear in both the time parameter, T, and the design parameter, 7 , obscures the effect of these variables upon the time of response. Their variations not only change the time scale, but also necessitate a change from curve to curve. This makes it 14 impossible to generalize accurately upon the effects of changing R, a, and A; however, in specific applications the effect of changing these variables can be determined. in which R, a, A, For example, a design may be considered , M, S are chosen in such a way as to make I= 0.018, and it is desired to know the effect of changing A to some new value, or changing A, a, and R to new values. (The effect of varying M and S is apparent at a glance and need not be discussed here.) The. dimensional response curve for Ir = 0.018 is translated into a plot of displacement as a function of actual time (or time if the pressure is unknown). jpressure Then, for the new values of R, a, and A the new value of It is calculated and the dimensionless curve corresponding to this value of Ir is also translated into terms of actual time. The two dimensional curves are compared to determine the effects of the change in design. Thus, the dimensionless curves are used to obtain dimensional curves, and it is these latter plots which are used in the comparison of performances. If the value of IV falls between the few values used in the dimensionless family of curves, interpolation must be used. - I --- 74 ~------.-----.-.--.-.-- ----- .---- 7Ti2 F7~7fl v-k' 7 77. I AIM ~AJMA]9ThQI YA~A~S'oF ~: ALV -24- L . *I I -1MLi< AMOvKI1T -- 77- --- rpjt 4-i 1: -i- _ LV1f i 1 1 1 4T, 1 owi:+ I 4 ~ V- Ir I Al IEAqi ~~ 4 F 4 N~#!Lis ~2 _7 -i g-4 690~J. INF +454 -4..~4 41: .7.I.vTT2I~FLr7.f-~plT--T +~4 8E ~~-j I Z N___ 7 V_ fqAL HE' . - I '~.LJ__ * 4. ~+:I177: A~34LS NfA Itit L4 {- I 4 A K I. . 77.7i~T T::K I.: 7f/ - 1-----:VI -i.I .ES I~p~ + 712.t.L. - / 4... *l. .4 ~-4-- ..- L VTE 727777177 .7 I ~Li I -- I .1 fTy-i I + -.-.1 1. .1~~~~ *4 - t 4-- + - - r~~-.--4--.--I----. 'I 16 Comparison of Theoretical and Empirical Response Curves Two theoretical curves were checked; those corresponding to 1T = 0.128 and to IT = 0.018. It must be remembered that the two empirical curves represent a large spread in the operating conditions. For 7T = 0.128, the load was 97 pounds and the lever ratio 13.1; for IT = 0.018, the load was 354 pounds and the lever ratio 6.73. In the general trend towards overshooting the theory checks excellently with the empirical curves. In time required for 95 per cent recovery the theoretical curves are roughly 50 per cent faster than the obtained curves. This discrepancy must not be judged too harshly, in fact, it constitutes as good an ably be expected or hoped for. agreement as could reason- It is enough of an agreement to warrant a considerable degree of confidence in the theory used and the dimensionless family of response curves arrived at. In the overshooting curves, the disparity between the rates of extinction is quite evident. The oscillation of the actual servo is far more prolonged than would be predicted by the theory; in fact, two full cycles were omitted from the empirical curve in order to fit it on the paper. This prolonged oscillation should not be considered as part of the response, but rather as an instability, for it was found that when the value of Ir was decreased slightly below 0.018 the servo was truly unstable, and vibrated without ever stopping at the equilibbium position. It was this instability which made it impossible to check the response curve for IT = 0.010. 17 Operation of the Apparatus The design of the servo (see page 38) was found to be quite satisfactory. The hydraulic ram operated so well that several of its design features are hereby recommended for consideration in similar applications: The mounting of the moving cylinder upon roller bearing wheels; the "floating" piston shaft; the self-aligning cylinder head bushings. Insofar as leakage out of the system is a criterion, the assumption of leakless operation was closely met. There was no deter- mination made of the leakage across the valve lands or the main piston, but since these parts were machined to the same tolerances as the cylinder head bushings, which were leakless, it may be assumed that the internal leakage was slight. The friction forces present when the hydraulic ram was under pressure were not determined. With no internal pressure, however, the friction forces were actually negligible. The servo was unsatisfactory in that it was unstable when its dimensions were adjusted to produce a 7' of less than about 0.015. This instability must go unexplained in this thesis, although several hypotheses are offered on the immediately following pages. 18 Suggestions for Further Work 1. Introduction of leakage and elasticity, and the stability of operation: In the construction of the servo, leakage and elasticity were kept to a minimum, and in the derivation of the differential equation of motion the leakage and elasticity were assumed to equal zero. These restrictions might be removed, and an attempt made to express and solve the equation of motion taking leakage and elasticity into account. Such an investigation would probably explain the instability of the servo constructed for this thesis and establish a criterion for stable operation. 2. Value of orifice coefficient: There is reason to suspect that the divergence of the actual from the theoretical response curves was due to the value of c, which was taken as 0.6. The value of c might be redeter- mined for the sharp edged orifice as such, or else an average value of c might be determined such that the empirical and theoretical curves coincide more closely. 3. Removal of limitations: The two most serious limitations were the exclusion of friction and load force, F0 , and the standardization upon square valve ports. It would be of interest to investigate the effect of friction or load force, and the effect of differently shaped valve ports upon the theoretical and actual response curves. The servo on which the empirical curves were taken 19 should be tested under pressure to find the friction force present. If this friction force were appreciable it would partially explain the fact that the actual response times were greater than calculated times. 4. Effect of neglected quantities: In developing the differential equation of motion, the term, (f D + nK - 2 ), was neglected. c The first two quantities represent pressure drops in the line and fittings and the third quantity arises from the expression for the orifice pressure drop. Theoretical response curves could be calculated by the same method used in this thesis, but taking these quantities into account. It may be found that their inclusion alters the curves appreciably, perhaps altering them towards a closer agreement with the empirical curves. DIMENSIONAL ANALYSIS and SOLUTION OF THE DIFFERENTIAL EQUATION OF MOTION 20 The Equation of Motion in Response to a Step Signal = Pressure difference maintained by the pump p q v-21 f 2 L = - Pressure drop in circuit due to pipe wall friction d = nK 2 Pressure drop due to n enlargements or contractions in the circuit a1 2 Iv12 2 ( - Pressure drop through the two valve orifices = 1) Equation of motion for the load: = Force Mass x Acceleration a I AL 2 d 2 1) a0 2 2 a -F=Md I dt2 (1) Since the system was assumed to be leakless: vla, -= A dt = A dx Sa, dt Substituting this valve of v, into equation (1), and dividing through by M: - ( a nK + + (p A -F) (=7 C2a02 d--2 x = pA-- da M A3 - 2a F = 0; dx 2 (--) t a, = (a )m L (fd- + n2K 2 Ca = a: 2 a2 ca a ) Since, in the Limitations, (2) 21 Now another simplifying step must be made; the term (f ! + nK - L-) in equation 2 will be neglected. D C2 This is tantamount to assuming that the pressure drops due to pipe wall friction and due to enlargements and contractions in the line are negligible compared to the pressure drops through the valve orifices. There are several justifications for this very drastic step: (1) form. It is necessary if the results are to be in a usable We will finally get the differential equation in terms of three parameters, and response to a step signal may then be plotted as a single family of curves. If the term (f % + nK -- ) were retained, then four parameters would be necessary in the final equation and the results would have to be given in the form of a number of families of curves. (2) The fact that (f L + nK) is positive and (c) is nega- tive, somewhat lessens the magnitude of the neglected quantity. (3) Reasonable values might be: L = 10 feet = 120 inches d = 3/8 inches n= 6 In the Notation f, K, and c were given as: f = 0.02 K = 0.75 c = 0.6 With these values: - =6.40 +4.50 - 5.55 ( + C. 2 = 5.35 22 If the step signal were of such magnitude as to fully open the valve orifices: 2 2 a2 c2 c2 a2 2 2. o.36 5.55 Thus, at the start of the response the pressure drop neglected is of about the same magnitude as the pressure drop retained, but as 2 a--2 2) increases, until the servo responds and a0 decreases, the term (0 ~ c 2a 0 at fifty per cent recovery the term retained is four times the neglected term. If the step signal had been smaller, and only opened the valve halfway, then the retained pressure drop would initially be four times the neglected quantity, and at fifty per cent recovery, would be 16 times the neglected pressure drop. We see that the neglected term may be of importance as the response starts, but rapidly becomes insignificant as the orifice closes and most of the pressure drop occurs at the orifices. Now, luckily, at the start of the response, when the neglected pressure drop may be a considerable part of the total pressure drop, the total pressure drop itself is negligible compared to the applied pressure. equation (2), (f L + nK - ) may be large compared to ( it is so only at the start of the response, when () recovery, is small. Referring to ), but , the velocity of By the time the servo has reached a high velocity of recovery, making the pressure drop term significant in comparison with the -- term, the constant term which was neglected will actually be negligible. 23 Neglecting the term (f I + nK ta 2()) a D A3 d2 x 2Ma ), equation (2) becomes: - dt c ao Or: A3 d2 X pA dt2 M Mc dx21 2 a2( dt Now, let h be the length of the sides of the square ports. Then: (ao)max = a 2h2 h =7 The initial displacement of the valve spool was, e = after the step signal was applied. S just U, When the output shaft has moved a distance x, the valve spool will have recovered a distance X and the R displacement of the valve spool from its neutral position will be (- - ) = 0 R R R w The orifice area uncovered by the valve spool equals twice the height of the ports times the valve displacement: 0 = 2h\2 R And: 2 a Substituting this value of a 0 dx dt2 pA - M AR - 2 2Mc a dx (.) dt into equation (3): 1 (S- x) 2 For later use, note that the first term on the right hand (4) 24 side of this equation is proportional to the pressure difference maintained by the pump. The second term is proportional, by the same constant of proportionality, to the sum of the two orifice pressure drops; that is, the pressure drop of the oil being supplied to the cylinder through one orifice, plus the equal pressure drop of the oil leaving the cylinder through the other orifice. 25 Dimensionless Analysis R, and S. , Equation (4) contains nine parameters; x, t, p, A, a, M, It would be a hopeless task to show graphically the relation- ship between x and t for all possible combinations of the other parameters. is, rewrite the equation in terms of non-dimensional displacement, . There is one very obvious and very helpful simplification, however; that Dividing both sides of equation (4) by S, we get: Sx ~ MS 2 ~ ( dt MaS c X2 (1- (5) ) 2 da2 , etc. are all dimensionor zIt is true that Z A, or S A q M less displacements, and it may be thought that they might just as well .x, be used instead of the term -. S For the present, the choosing of- may be justified by the fact that it is more meaningful than any of the x is the displacement of the output shaft, and S, above alternatives. the step signal, is the final displacement, so S is the fractional recovery. groups. No such simple meaning can be assigned to any of the other will be shown that the choice is not only convenient, Later, it but necessary. d X dt2 = -, equation (5) becomes: 1 TA R pA --MaS 2c2 MS -- d2 2 (---) X2 dt ( - Letting X = 1 - Or: d2X -- dt 2 A 3R2 1 2c0 2 MaS dX 2 1 (-) dt i- x2 pA iMS Another simplification is to let the dimensionless group, SA3R2, be represented by the single letter n . Mas 2c is not included 26 in this group because c, the orifice pressure drop coefficient, is The equation now becomes: assumed to be constant. d2X = dX 2 1 dt dt 2 pA (6) - - -I -=- X MS Each term of the equation now has the dimensions of (time)-2. If we divided the equation through by A, and let T' = dimensionless the equation could be written as: x d--2X Ir dX 22 (-X 2c dT' = -g dT2 - 1(7a) ) time = t The above step is perfectly logical, and equation (7a) was almost used as the final form for the equation; it was thought undesirable, however, to have the step signal, s, appear in all three of the dimensionIn order to eliminate s from the time parameter, multiply less parameters. d 2X _ 2c dt i or the last term of equation (6) by MaS: (dx)2 1 pA MaS dt X MS jA3R2 Or: dXlt dt 2 2c2 1_ (f)2 WA2 R2 X2 dt , and let T Now, divide through by = - dX = dimensionless time: d- --- (dX) d~a 2cadT 1 (7b) X It appears possible to simplify this equation still further and express it in terms of only two parameters; dimensionless time and S2X -- dT = -- 2c Dividing equation (7b) through bylV: (-) dT - dimensionless displacement. T 27 Or: d2-4I ~2 d* - =- (----) dT2 2c2 Letting X' -- dT dT2 S 1 dX1 2c dT 1 -a- R TA3R- (1-) - it d x' X 2 1 X2 According to this, then, a single curve of Xt versus T will show the response of servos with all combinations of A, a, R, p, etc. If equation (8) were integrated, just as equation (7b) will later be integrated, the plot of (1- x') versus T would look like the plot of ~ .0 ------ iT = 1. - ------- X0 AR Solving for x at (1-X') = 1: A3R Ma 2 Solving for x at (1- X') = 0: - - (1- X) versus T for 28 ~A 3 R Ma - x = S 2 It is apparent that at (1- X') = 1, any value of x other than S is meaningless; and at (1- X') = 0, any value of x other than must I2 equal mu s t equal S ( MaS Ma unity) if the plot is to have any meaning. Equation (8), then, does zero is meaningless. That is, Ta not describe servos with any combination of dimensions, but only those SA3R2 1. To get a meaningful plot for the more general MaS case in which I AR2\ 1, equation (7b) must be used. MaS The obtaining of equation (8) was perfectly sound mathematicalin which ly, but the physical phenomenon was lost sight of. The only dimension- less displacement parameter which does have a physical meaning is I, S the fractional recovery. 29 Solution of the Differential Equation A solution must now be found for the equation 2 dX 2 (dX) 2c dT 2 (7b) 1 K X2 - dT All attempts to obtain a formal solution were unsuccessful, but two solutions were found which permitted the complete mapping of the above equation. The two solutions are a Taylor series approxima- tion, and a step by step solution. The series solution will be described first: Let To, Xo, and ( ) be the values of the dimensionless time, dT displacement, and velocity at the start of the interval under consideration. Then, assume that the curve of X as a function of T can be X = Ao + A, (T - TO) + A 2 (T - To) 2n + ... + An (T - TO) + approximated by a series of the form: The coefficients of this series are: n An n dX n n :dT T=T0 d n0 The series can now be written as: dX X =X, + (fT)o0 (T T) + 1 d 2X (-j-) (TO) 2 + 1 d3 X (=) 3 T)+... The successive derivatives of X are obtained by successive differentiations of equation (7b). the Appendix. These differentiations are shown in There, it can be seen that the derivatives are very long and cumbersome expressions, so that the obtaining of the coefficients for equation (9) requires a great deal of arithmetic. The evaluation of dl the derivatives at T = T0 becomes vastly simplified, however, if (-)0 is zero. The series solution will only be used, therefore, for the intervals 30 This restriction of operation in which the initial velocity is zero. is possible only because the step by step solution works well in the Before the step by step solution was other regions of operation. discovered, it was thought that the series solution would have to be used for all stages of operation, in spite of the difficulty in evaluating the derivatives when the initial velocity was not zero. The values of the derivatives at T = T , if () = 0: ddT (d2X) d4 X dT4O 2T 3 2c 2X0 2 ( =6X V4 24- 16 dT 60 2c 2 X0 3 (d 8 ) 0 = 2c 2X 0 5 2 2c X* d10 X J ( 540 -888 + 272 ) 2c2 X ( )21 0 2cX0 20,160-57,696 ( )+41,088 )2 ( ITF 6 = 2c 2X 2c2 X0 - 7,936 (-A ) (-) dT 10 0 1 2c ~ 2c2 X0 J Substituting these values of the derivatives into equation (9): X = X -r (T - TO)2 + [24 - 16 + i274 720 2c2X 03 1 g5 40,320 2c2X04 22 (T - T IT 4 (T - TO)6 2c2X0 F40 -888 L + 27 2c2XO 2 ] 8 (T-T) 2c2Xo0 )r-+41,9088 ( 1r)2 20,9160 - 57,696 ( IT6 + 2 2 +,2,1 6 95 2c X 0 31628,800 22c X05 [ (2c X0) 31 -7,936 , + (T - TO) (2 csQ L6 113.4 - 485 (-2cj) + 583 2 - 254 ( + 35.4 (;r) 4 C 0 (T - TO) ......... (10) This is the series solution, and now the step by step solution will be discussed. As in the series solution, To, Xo, and Xo will be the time, displacement, and velocity at the start of the interval over which the step is being taken; and Tl, X1 , and X1 will denote the values of these quantities at the end of the step. Equation (7b) can be written: dT 2c di diX di dT dX dT dX X But: Now, equation (11) becomes: 2c dX X Or: dX = X d X 2c X Now, for the short interval under consideration, let X - constant = X = X +X2 2 *. (12) Equation (12) then becomes: = a - -- 32 1 Xd - dX- 2'2 2c X This equation is integrable as it stands (see A Short Table of Integrals, B. 0. Pierce, integral number 53). Integrating both sides of the equation: Lx CP X 0 - l (2 2c X )I xo Putting in the limits: -X)=AX= (X c V STr Dividing through by C 7K -ln X -.. 2c X X0 2 _ 2c2'X2 and using each side of the equation as an exponent for e, the base of natural logarithms: j1 2 - 2czf2 IV e - - 2c2 X 2 Solving for X1 2 X, X0 2 _ - -s 2c ZXF + 2c 2X 2 (13) c 2P e This is the step by step solution. To, X0 , and io are all known. At the start of a step, A value of X, is chosen such that (Xj - Xo) = AX is not too large (in the calculations, AX was always chosen smaller than ). 4 Now, knowning X, CiX, Xo, c, and IT the velocity at the end of the step, i, can be calculated from equation (13). The time taken in going from X0 to X, is: (Tj - T.) = AT = AX Average Velocity =- 2 AX . X, + X0 . 2X)J I 33 It has just been shown how the final values of time, displacement, and velocity are determined. For the next step, these values are considered the initial values. Use of both the series solution and the step by step solution is demonstrated in the sample calculations in the Appendix. 34 Stages of Operation The equation of motion, dX( dT )2 2c2 (7b) x as it stands does not apply throughout the entire response. The open- ing of the check valves and of the relief valves must be taken into account by suitable changes in the equation of motion. equation of motion may take four different forms. In all, the The different stages of operation, their duration, the form of the differential equation, and the step by step solution for each stage will now be given. But first, the significance of the terms in equation (7b) must be recalled. In spite of the intervening manipulations, the terms have the same significance as the terms of equation (4). the discussion following equation (4)): 2c2 X 1 2X 2c x Sum of pressure drops Applied pressure And: 2 = 1 Pressure drop through one orifice 4c x Applied pressure Stage I Relief valves and check valves closed Differential Equation d 2 dT 2c 2(X (See 35 Step by Step Solution 2c2 - io xl Z-2 + 2c X Duration: Until the check valve opens. This occurs when the orifice pressure drop equals the applied pressure (see equation (14)): 12 4c () 4C = 1.44 Stage II Relief valves closed, check valves open Differential Equation The operating pressure has no effect, and one half of the pressure drop term must be omitted: d~ dT IV (X) 4c 2 X 2 Step by Step Solution xXX 1 - - e 2cX Duration Until the relief valve opens. This occurs when the orifice pressure drop equals three times the operating pressure (see Limitations, number 2). (-) x = 12c2 = 4.32 36 Stage III Relief valve and check valve open Differential Equation Pressure on one side of the cylinder is zero, and on the other side, three times the operating pressure: dT Step by Step Solution This solution is unnecessary for the simple case of constant deceleration. Duration Until the relief valve closes. This occurs when the orifice pressure drop equals twice the operating pressure (the total back pressure equals three times the operating pressure): Sc = (x) = 2.88 Stage IV Relief valves closed, check valve open, servo overshooting Differential Equation Only one of the orifice pressure drops is effective, and the applied pressure now opposes the motion of the piston, so the sign of the third term must be changed: dT S=-(5() 2 4c X +I7\ Step by Step Solution 2 2 2 io2 + 2c 2 e C 2X2 - - 37 Duration Until the velocity is zero: x=0 F DESCRIPTION OF APPARATUS 38 The Construction of the Servo At the beginning of this report, the valve controlled servo, in general, was described and a simplified diagram of its construction shown. The apparatus used in this project was built according to that description, but since many problems arise in translating the simple diagram into physical reality, some discussion of the construction of the servo is warranted. The design of the hydraulic ram is shown on page 41. Instead of using a moving piston, the piston is anchored and the cylinder, on ball bearing wheels, is allowed to move along steel tracks. In order to ensure an equal distribution of load on the four wheels, one pair of wheels is mounted on a pivoted axle. This gives the cylinder a suspension with three points of support, rather than four. shaft is ground, chromium plated, and lapped. The steel It bears on long cast iron bushings (internally ground) in the cylinder heads, and on a hardened steel bushing within the cylinder. The heavy cylinder heads receive the cylinder tube in lead-filled circular slots, and are then drawn together by four stay bolts which run from head to head. The servo is loaded by bolting lead plates to the side of the cylinder, fastening them with studs set in the cylinder heads. With the high accelerations encountered in the operation of the servo, it is important that the center of gravity of the load be very close to the center line of the shaft if the ram is not to leave the tracks. The self-aligning feature of the cylinder head bushings is one of the most interesting points in the design. The portion of the cylinder heads within the cylinder are ground to a smooth plane surface. 39 The cast iron bushings are turned and polished with a spherical seat, and hard babbitt is cast around this spherical seat. The babbitt cast- ing is then trimmed down and a plane surface turned on the face opposite from the spherical cavity. Now, the bushing can pivot in the babbitt block, while the babbitt block can slide on the cylinder head. Three adjustable compression springs keep the surfaces in contact when there is no internal pressure present to do so. The bushings are made self- aligning in order that they will not bind on the shaft when the cylinder heads are deformed or displaced by the internal pressure and the forces exerted by the load. The drawing shows provision for fixing one of the babbitt blocks in place, thus attaining two point support for the piston shaft. In the actual construction, however, this feature was omitted without any harmful effect. The steel bushing in the cylinder offers sufficient support for the shaft, since the anchored end of the shaft is free to move vertically and horizontally in the slotted anchor plate. The four way valve spool and the valve bore are hardened (and ground) in order to resist the wire drawing effect of the oil. The valve spool is not of one piece, but consists of two hardened washers bolted to a shaft and separated by a spacer. The valve bore is made in a steel sleeve which is pressed into a cast iron valve body. Where the square valve ports are located annular grooves are turned in the valve sleeve, extending almost to the inner bore. These grooves serve several purposes: They locate the valve ports; they leave just a small thickness of metal to be filed out in making the square ports; they connect the two diamet- rically opposed ports comprising each orifice. In the four way valve and in the hydraulic ram all clearances 40 between sliding parts are held to less than two ten-thousandths of an inch on the diameters. It was found that a gas filled accumulator in which the gas and oil were in contact was unsatisfactory. Under pressure, the gas dissolves in the oil and then bubbles out when the pressure drops. For this reason, an accumulator is used in which the oil is separated from the gas by a neoprene diaphragm. In using the apparatus to obtain response curves, the step signal is imposed by quickly moving the signal shaft to a pre-set stop. A pencil, attached directly to the ram, traces its displacement on a tape moving at a known velocity. Samples of these records are shown in the Appendix. The maximum orifice area, a, and the cylinder area, A, are fixed, but it is possible to vary p, M, S, and R, the operating pressure, load mass, step signal, and lever ratio respectively. The last three variables are chosen so as to give the desired value of the design parameter, 7 = R2A 3 MaS TtIL HIYDRAULIC R~A TC~iNG LOAJD =J~iInM ~4iZ~L -~- -.. 42 The Apparatus Assembled on a Base Plate 43 The Hydraulic Ram, Unloaded 44 Piston Shaft and Cylinder Heads APPENDIX 45 SAMPLE CALCULATIONSCv ThE. CALCUL-.ATION3 AS AN EAAMPLE) W"EN ' Y = 0. 0 1 w ILI, NOW DE KaROPue-.-i ve FOR rtHfE ESPONSE STA&E I Ar THE srART oF ?He. K~sorsE, THE sERIES SOLUTION WELL. rHE 3EiEs 50LUT 1,0j EqVATION (,o) Is : (T--T.)3 1L All - 16 + I.3 -'T ] rS [C + + - + . =I. I . WoRKS 1r &M -S77 lk- 616 Ic 7-2: + 2iL (;cIl -0o.719 + 7 I-S Y1 G<1 . +~ Lt*7 WrEN T :Tr Tl-.) 3 (.-3 Ve- Ir )1+ ( T) 0 1::o -n- : -o" I 56' T'+ DlPFERENTIATINM. t r7T' WlTH o' T + q.63 IO ft: EC.OM7ES RE3PECT T6 T + -Z.754 (0 I- 4+).' Irsk) - .2 -. I r7 . .+ t: + OS -- '-~ .0 co . 8+ 63 +.000 0.68000717 '.o of .t~a fap~ I~*? P6x'" 7- + . 3 -1;)L+(11.1a -A + . 6*08o ~i')g~(~b~O 10 +1.oooo IO~s5f+ +. VOit a + . 0o 14 -a. r T it o' +(t. IA . 0001 + - . + 1.1 1'1T ?1 oo a-4 +. 0000 70 + 7 4 x -''.1)+(-4 I 7 T ) SOL-?ION - 3- .( . C : TtlE SEfIES O -4 46 I-.72. -.lg oil ,,',),(, 3, a a..,'t)+(a .ia 00I37 + .000712.. +.0041.+. s'.s3)+ R.iu)(.4A + -OOOI9 *( .000%.e .+.eO0s1fy...eoo5, 4 - +. + - ria 1 .000112. .13W+ AS T IS INCKCASED) THE SEIIES BECOMCS LESS QVtCK.LY THE LIMIt of Ftfe VSEFULtNFSs of TtWE 6E!IFE5 SOLUTOPO CONVERCENT. rmE STEP SoLUTION pI ITIP MV6S 6Eutt X + - Now AT Tea. tAC.HED *SOVEI HAS OEeN 1+.72. .9 0 ~.Wh1ff p. p. T X. i. X. 0 o' -. 9~4. (4~~ .a. .A~o Ub~t ~ .6V17 040 .,I., ow .J#. .9. -. oao , .oao '3. 37'w' r~j .10 ,iiji -09'I.:j I..,'. i, -. oree .0,,, STASE .,6,o .7t ,.I (. -:,.30 .o~b4,1 .0019g *0.90( .0 1+It 1.041,E I 2.2. -.9 2.I ,..'. -. eIi 49., i e ---- y .1.4 -:,. .0, .Dolt Ar -. 311 I.e's. .01 &2t -9uq . rfos ,il . 1 W H E N().. sti 1.015' -.,130 -V*913 -. It .0302, -. -. 1.41 i ... (iA1 1. a I .01313L .oi3 -'.010 0 I. I .zIe .aAe -.. 36 92. iv I U.. IE.-ONS I I I I STAGE IT 3ri7 rmE STEP i SOLUTION PoR 47 STAQE E. is: -AI e* a .va . . I3.-TO e .. ai as .q .62 .O(W .01S 1.-S3 .665a . i -00 IS.16i DS,6 .bsg- .'2aIf .o75- .801 -. .031( o.essa .0sr- -IS' 01 6t% *4O'fe .01370 -0341 *5 -- ,17 .1370 .oS* . 04 6 ..5s i3.ill .6e'jql 1113 .*,4A3 1.o42 Ollat -........ STAGE II DIPPERENTIAL. E-QI%?i0N .=h 3Tr = .03 INITIAL. C.ONDITIONS: T, : 13 1% Xe,:.099 THE tr Tr(a) + ( T2~r ~+ I: AT I.e I I-* too T EQUATION DIFFEMENTIAL 1. (ar) +I .os(ir) -. is4 .OI(A 1 't.1 -.oi vt 8ta 1 I.rl &-e??-es 3.V1 -ett r9 -. o 9e -. s ,1t STAG =R N3s WHEN DR oP3 ro 2. vs . INtEffATING , I'l -. 1170 - 117 -. 1io -071 1.0of -01311 -. fl4j .IWO W"MR=.2 STAGEF 11 IENDS THE oj .1113 .00,17 .,1 3. O +. l Z= STAGE DIFFERENrIAL 4(22 V(, T EQVATION: +-a - -- 48 &t.g I +. . 061+. STEp 8t STEP 30LUTION' + xel tc,4 2C -ax c 1-.AT-2 yIAI e~ 1is +- I. 1. -. 0I .o 1 0 t 1,I . 1 6I .041 16'. 31.03 .t I, AT -*e717 t30 . cIDV 2. 1.047 -. 010 -. l -- I 4. -. ofl .*207S 16.17. -. 16 -. 030 -. 11 .. 313 -.. 't-O (- NoW THAT DIF'FER.NTIAL EQUATION %et44 FIRST TERM BE NE.LECrTD, THE AS IS .61+." SMAL.. AS RIGHT H~ANo or, 3(00 .0111s, .070 AT TMC TOP OF THE PAGE ON THE AND . .0. -.. 'I't~ . Wt I IOlt * ,eac i . 20 ,0 SIDE EQUATION WRITtEN . 00 fo .o4 A CvLANCE SHOW5 OF THE 116*3 AT THAT EQUATION THE THE MAI AS: STu |twErccpT iNG, AND Pvrrlwc, rH E INITiAL lid 'ntoidT)-.ego .. r)r IL i20 -. aLo .rag :s1 10 AT: Z W"FN .* AT.:a, AT:Z4 t:-.3% ): THe 1. Oat .. o?71 '46 q -. X FovoTrm STA GtF is lEpED, NOW T'AT X CONMDITION3: (of I p M7j 49 THE FINEC.-EVING SAMIPL AS FAIN As =0 I Ilf E CIvERING IN FACT to. A STEP A14 0 OF THE SIGrNAL, . :.390 THE OrCATION SARvO HAVE 8EEN PoiN? wnlr_ rHe ov~gqsmooriNGr MAS rrti 5APAr srACvES C ALCUtLATlON MAI WIrt1 SFRVO AS MAVE P453e3 ACrAIN JVST #0W BE TD1E I1ITIAL BEEN CoHSIPERCO C CPtRAIEO $ToppD, rHaOu6H AA O rmE ILIJVS1rRA7O. A5 PESPOPIN& ~ONDITh. ZO -LICE SSIVE- DIFFERENTIATIONS 50 -- E THI BASIC EQUATION TME IDIFFLfE?.7ALL P-QUAro1 OFmoTito is Tr jx "74- A A~xF ;V 'A.~V amat L~~a i'~4rI (T) %Aa i Ra \4jT1 opoMES: /1 [z' *A I ~(~3 This E94tm kN 4 A~Li.1ill3J 12 IL IA (9/A,- t' +Z*( qxitAtj]1 44, A Xq tdX [4 4 q (-[ -4 -re A () [ T) (A)/kL * .]. 3 I L a. + (.([r 141 1' + + 4 rrj -X 00 + 4 AL V +r *0 A l.*j*a ~ KL+ + ' LX -X' A,' II ot'T .+ 2)'[ 10. P 70* 4 -...L * 4T 4 4 + 2.S N 1 .11 4].ugJFT94. +Le ] 4Pr"Al~ y or~ rri... esArcCT it; ?P1E NwW * 5 1 id ' . ~ 4 ro ii PIFEetrtAriom3 ~ip V-7)0 11S rpfd 06C wirnotJT m-cLFcCrEP SECCAL'IE E4Ch TIERM 1 rple itoweK cFIslr CcI"rAINIGotL WrfIr IS 4f 7EWAS E %romsoos: F (~)ARE IN LFsts r.IAN T?1f- + EQ~UAL Iw rfTO cK CK tOwfft SIMLIFILt' P I~yA'7-ivi Li Is -- 'iN TMIESE WMILL HEKEAFTER I 4, . -rrffS It IN~ W'fIc" six, WILL 7-PIF vrfIC" TARM3 1 94eAt& *.isremvmiSPo -Xv.k' -T 6 ANYt rnm TEAa rsiAM wtArtE MAY1 :Si FROVOCES A TietM ZERO. THAN 'rH4 KESUL1S. ?Wjjt.PrpI VeltvArive. ALI. 4REATER +f)AAa[ V1'T) Tte FINAL. ALL Td6RMI PIRIV4-Ivf. oft GftE~rfZ TO DIFfEE141IAED) To A rawti ,IVfC AISF_ To TnOE AfoaecTNG, SaLveMri t ?t- CG"1AL VM To A POWFt( cervflNINp() 4 Tl1JE TwCFIrpEAvie jL IN rKE LEIG,.TiI DiEAWATIVI POWER OF 5f( r) 13 EQUAL *( oA %otlE4TEA 7ro A,. I ~4 T'/ P4CG.4.CY 4LL V FIVE.: at~ L& 1,~o ~a~i. + 7 OI -l ~ riifAJ 144 i t~eso - amiA ~~i iso'k Lj .[A.p~ Prwcl kar) 1'1 L +~~j [' er 00 [_,8_ __!U__ 'S I.Af 14 lit Eq~'l. OEK To lati 40 + EQVAL 4"1111 Toi A ',-o N9eGrLErr OR G-IEhAll + 7SIV 4~AIL TE9,.- rKAP4 TrnRg k; rr) + f xI I *1160 - C70616 #Wu? - ~22L~~J Ea13 5 It 'I 1 ?)S + w, off -'t - 7,9 34 K v Irl , * &A 1 22 * A feltz S3.37toq' toIgo+ 1MSAittO o t +113 I C40 3%6 I X + 1 416I h I 3s tfl *3ii'i4 5, *141050 0) a *1s'L' 1 I ?(& .+.z A* 00 oo -a.30+4., 17 V-s-i O VDI IZ to #1 +. to Ist -* i3toe~AL~ rerlTrl DeRlvArpvvE IS powefA M~ I alo 1;t I f AL i+ IN' TplE tHJAN _ XI 13 1% #6to*I L-I G5AEArzot L I V'T) L~(ca ke +k 1- #,%% I .i/tI\' k*+v,#O0+ '176 Is I- 2140"k' xvI] +Al *.~~ V' 44r/ 4 Ii-a.1.s +q1* rrIA X6 1' C qT- 52 ?HE i'4 wHItiN ?6lDbMS I 'ii'. +4 a 53 at s qA 1n , iN rug 60 -rW.L NeGLECT GAEATAR A*I LI, Jo : o 600- \4T/+ 3sf ,76 3a.jg7 1V* THAN LEO07:t ALL reAMS -Z.t?,4/6 )-I I. rme +17gL + + P+S3,t -ir;79L l,t PEm otivArIvE, IN vmt 1H TW O *1 a+t, . (a i- OF OEAIVATIVIL, NEG.LECTi i l74,71 f36 I + 3Ce 7E EAMS ,-Aj* coNTAliss tUG ( . POWER ELEVEP47i * :M THE 54 Data Used in Plotting Response Curves Stage 7t= 800 T 1-X X .100 .200 .300 .400 .500 .550 .600 .650 .700 .740 .780 .810 -. 805 -. 720 -. 636 -. 551 -. 466 -. 404 -. 361 -. 318 -. 276 -. 238 -. 204 -. 171 -. 148 -. 127 -. 111 -. 093 -. 076 -. 0594 -. 0466 -. 0382 -. 0296 -. 0238 -. 0204 -. 0170 -. 0140 -. 0115 -. 0094 -. 0077 -. 0060 -. 0047 Stage It = 0.128 T 1-X 2.50 3.00 .387 .544 .628 X 0 .248 .409 .556 .725 .822 .937 1.068 1.215 1.383 1.539 1.760 1.919 2.105 2.251 2.419 2.625 2.862 3.158 3.347 3.583 3.878 4.028 4.209 4.423 4.617 4.853 5.045 5.281 5.575 5.763 .840 .860 .880 .900 .920 .940 .950 .960 .970 .974 .978 .982 .985 .988 .990 .992 .994 .995 3.25 II I 3.375 3.50 3.60 3.90 4.225 4.635 4.80 5.00 5.27 5.496 6.120 6.404 .669 .709 .735 .825 .900 .950 .960 .968 .975 .980 .990 .992 -. -. -. -. -. -. -. -. -. -. -. -. -. -. -. 296 328 232 331 324 316 266 183 075 048 030 025 019 013 002 -9 55 Data Used in Plotting Response Curves = Stage II T 6.000 6.143 .179 .699 .730 6.281 .760 6.419 6.559 6.564 6.573 .790 6.801 6.903 7.010 7.027 ii~i IV 1 - X 3.000 6.702 7.427 7.981 7.991 8.058 8.156 8.301 8.490 8.786 9.131 9.919 10.262 10.451 10.648 10.793 10.960 11.109 11.291 11.441 11.623 11.849 12.139 12.326 I7 0.040 .820 = .018 x Stage T 1- X -. 119 -. 217 -. 218 -.218 -. 216 -. 213 -. 212 -. 212 I 4-08 8.16 9.79 9.95 II 10.09 .150 .594 .844 .869 .890 10.21 .910 10.35 1.930 1.000 1.050 1.060 .821 .823 -. 206 .850 -. 200 .870 -. 1927 .890 -. 1823 .910 -. 1800 .913 .975 -. 1320 1.030 -. 0656 1.031 -. 0642 1.035 -. 0554 1.040 -. 0463 1.046 -. 0366 1.052 -. 0268 1.058 -. 0138 0 1.060 .0264 1.050 .0319 1.040 .0317 1.034 III IV 10.90 11.39 11.52 11.72 11-79 12.10 12.41 13.00 13.81 15-44 I 1.180 1.150 1.120 18.91 19.26 1.100 1.080 1.070 1.060 1.050 1.040 1.034 1.028 1.024 19.80 .0291 1.017 1.014 .0183 .0146 20.39 1.012 .0121 1.010 1.008 1.006 1.005 .0099 .0078 .0060 .0047 20.63 20.74 20.88 .0219 1.100 1.120 1.150 1.180 1.205 17.93 18.55 19.42 19.61 20.01 II 1.080 17.11 1.028 1.024 1.020 .0260 1.070 20.14 20.28 20.50 21.01 21.19 21.34 1.020 1.016 1.013 1.010 1.008 1.006 1.005 x -. -. -. -. -. -. -. 0735 1431 1589 1552 1523 1488 1430 -. 1131 -. 0858 -. 0801 -. 0755 -.0716 -. 0637 -. 0569 -. 0452 -. 0300 0 .0300 .0435 .0527 .0570 .0596 .0600 .0505 .0499 .0476 .0435 .0402 .0374 .0338 .0289 .0242 .0182 .0134 .0080 .0053 56 Data Used in Plotting Response Curves 7r Stage T 5. 10. 12. 12.31 12.56 12.80 13.05 13.21 13.37 II 13.52 13.70 13.83 III IV 13.91 13.92 14./+2 14.92 14.975 15.105 15.297 15.497 15.778 16.074 16.552 17.260 19.260 21.260 22.650 25.380 26.572 27.499 28.280 28.805 II .125 .498 .714 .920 -. 1193 .935 .944 .945 .999 1.045 1.049 1.060 1.075 1.090 -. 1170 -. 1150 -. 1148 .750 1.052 30.542 1.040 1.034 32.832 33.396 1.080 1.060 1.028 1.027 1.003 .986 .985 .982 .978 .974 .968 .964 Stage I II -. 0998 -.0848 -. 0832 -. 0797 -. 0765 -. 0731 1.110 -.0694 1.130 -. 0656 1.160 -.0600 1.200 -. 0530 1.286 -. 0330 1.332 -. 0130 0 1.340 .0283 1.300 1.260 .0388 1.220 .0475 1.180 .0548 1.150 .0594 30.352 32.358 -.0499 -. 0990 .780 .810 .840 .860 .880 .898 1.070 31.925 32.122 X -. 1167 -. 1191 -. 1209 -. 1220 -. 1229 -.1230 -. 1228 -. 1215 30.070 30.224 31.764 31.794 IV 1 - X 1.120 1.100 31.264 7r = .010 (Continued) .010 29.295 29.609 29.917 30.642 30.746 30.764 III = -0411 .0261 .0252 .0220 .0186 .0154 .0100 .0042 1-X 33.820 34.844 35.355 35.633 35.894 36.089 36.295 36.369 36.531 36.671 36.840 36.982 .963 .968 .974 .978 7r = Stage I T 10.6 21.2 28.2 29.7 30.4 30.7 30.8 II III .0628 .0645 .0654 .0652 .0646 .0635 .0611 .0592 .0566 .0561 T IV 31.1 31.2 31.6 31.8 32.0 32.2 32.2 32.3 32.6 32.7 32.8 33.0 33.1 33.3 33.5 33.9 34.3 34.8 35.8 42.9 50.0 58.6 .982 .985 .988 .989 .991 .993 .994 .995 -. -. -. -. x 0 0098 0137 0151 0156 -. 0152 -. 0139 -. 0132 -. 0115 -. 0099 -. 0079 -. 0063 .002 1 - X .1125 .4500 .4987 .881 .923 .945 .949 .966 .974 .990 1.006 x -. 0212 -. 0424 -. 0566 -. 0595 -.0601 -.0602 1.120 -. 0601 -. 0595 -. 0586 -. 0570 -. 0553 -. 0536 -. 0531 -. 0576 -. 0573 -.0459 -. 0555 -. 0551 -. 0546 -. 0544 -. 0535 -. 0529 -. 0514 1.140 -. 0502 1.170 1.210 -.0488 -. 0459 1.484 1.658 1.735' -. 0022 -. 0012 0 1.022 1.026 1.029 1.033 1.053 1.058 1.063 1.070 1.080 1.090 1.100 57 \41 ---F- - I *1----------0Q'0 1 Sample Tapes Showing Response of Actual Servo to a Step Signal -A 58 Curves" Data for the "Corison (Empirical versus Theoretical Responses) Fixed Servo Dimensions: 2 (1.5 Cylinder Area = A Orifice Area = a = 2 (}) 2 ) 0.982 inches 2 = 0.125 inches2 Oil Density (Unavis #47) = 0.868 x density of water 8.13 x 10-5 lb. see. 2 inches~ 4 Variable Parameters: I= Pressure = p 0.128 500 Step Signal = S - = 0.251 386 3.20 Lever Ratio = R 13.10 Mass = M IT= 0.018 500 psi lb. sec. = 0.916 inc _ inch 386 2 1.68 inches 6.73 Speed of Tape: 6.57 inches sec. Several curves were made for each value of IT. Since the curves of each set were almost exact duplicates, there was no need of any averaging process, and the following values were taken from two tape records selected at random. The Empirical Curves _K = 0.018 V = 0.128 Displacement of the servo Displacement of the tape (x) (y) 0 .08 .38 .76 1.22 1.60 1.97 2.27 2.59 2.99 3.11 3.12 3.11 3.12 3.14 3.19 3.20 =S 0 .06 .16 .24 .32 .39 .48 .55 .64 .74 .81 .89 1.05 1.20 1.34 1.51 1.68 x S _ y _ S6.57 = see. 0 .025 .119 .238 .381 .500 .615 .710 .810 .935 .972 .975 .972 .975 .981 .997 1.000 0 .0091 .0243 .0364 .0486 .0593 .0729 .0835 .0965 .1123 .1230 .1352 .1595 .1823 .204 .230 .255 Displacement Displacement x of the servo of the tape S _y T 6.57 = sec. (x) (y) 0 .08 .32 .72 1.12 1.53 1.68 =S 1.77 1.86 1.88 1.82 1.74 1.68=3 1.59 1.51 1.52 1.58 1.68 =S 1.76 1.80 1.74 1.68=S 1.64 1.60 1.63 1.65 1.67 0 .08 .24 .41 .56 .69 .75 .78 .84 .91 .98 1.04 1.08 1.14 1.21 1.32 1.41 1.50 1.61 1.76 1.92 2.03 0 .048 .190 .428 .666 .910 1.000 1.052 1.108 1.118 1.082 1.033 1.000 .946 .898 .905 .940 1.000 1.048 1.070 1.033 1.000 2.10 .975 2.28 2.38 2.43 2.48 .952 .970 .981 .994 0 .0122 .0365 .0624 .0851 .1050 .1140 .1185 .1278 .1382 .1490 .1580 .1642 .1733 .1840 .201 .214 .228 .244 .268 .292 .308 .319 .348 .362 .370 .378 U' '0 60 The Theoretical Curves Dimensionless Time = T = W= t = TR A TR 0.982 )813 x 05 = 0.00102 TR xX S T and (1 -X) are tabulated earlier in the Appendix: IT= 0.018 1Y= 0.128 t = (0.00102) (13.10) T = 0.01335 T x 1 - X t = 0.01335 T .100 .387 .544 .0167 .0334 .0400 .0434 .0451 .0467 .0480 .628 .669 .709 .735 .825 .900 .0521 .950 .960 .968 .975 .980 .990 .992 .0620 .0565 t = (.00102) (6.73) T = 0.00686 T =1 - X .150 .594 .844 .890 .930 1.000 1.080 1.120 1.205 1.150 .0705 1.000 1.050 1.034 .0734 1.020 .0817 1.010 1.005 .0640 .0667 .0855 t = 0.00686 T .0280 .0560 .0671 .0692 .0710 .0748 .0809 .0852 .1060 .1230 .1300 .1360 .1380 .1410 .1435 .1462