CA1115866A - Method and apparatus for producing a speed pattern for an elevator car or similar vehicle - Google Patents

Abstract

ABSTRACT
A system for producing a velocity pattern for a vehicle such as an elevator car which must be brought smoothly to a stop accurately aligned with a desired one of a number of stations.
The distance between the vehicle and the station at which a stop is required (target distance) is expressed in digital form, and is processed electronically to produce a velocity pattern which results in constant deceleration of the vehicle during the entire approach except for an initial build-up to, and a final tapering off from, the constant deceleration. A single adjustment for deceleration rate automatically adjusts the slowdown distance to the correct value to suit the selected deceleration rate.
The calculation is done repetitively at high speed to minimize any lag between the diminishing target distance and the corresponding velocity pattern. Further adjustments permit control of the final tapering off from constant deceleration, with the slowdown distance automatically adjusted to suite these adjustments also.

Description

86~
This application is a division of our copending application Serial No. 288,471 filed October 6, 1977 and entitled METHOD AND APPARATUS FOR PRODUCING A SPEED PATTER~ FOR AN ELEVATOR
CAR OR SIMILAR VEHICL`E.
The present invention relates to vehicle control systems generally and, more particularly, to a system (i.e. method and apparatus) for producing a velocity pattern for an elevator car or similar vehicle which must be brought smoothly to a stop -accurately aligned with a desired one of a number of stations.
While t~e present invention is primarily intended for, and has been developed for use with elevator control systems, its principles are not limited to the elevator environment. As will be understood from the discussion below, the present invention has general applicability whenever it is necessary to produce a speed pattern signal for a vehicle, such as an elevator car, railroad car or any other movable conveyance, that is arranged to move between two or more stopping positions or stations .
In transportation ~ystems with automatic control, a vehicle must be capable of ctopping at an accurately aligned position coinciding with any one of various stations where pa~sengers or good~ can enter or leave the vehicle. A stopping sequence may be initiated after the vehicle has attained its maximum speed, or it may be initiated while the vehicle is still accelerating towards maximum gpeed. This latter case is particularly common in passenger elevators where the distance required to accelerate to full speed and then come to a stop is greater than the distance between adjacent stations.
It is obviously desirable to delay any stopping sequence as long as possible so as to not waste time running at a lower speed than necessary, and to avoid creeping towards the sta~ion at low speed. It is equally obvious that if a stopping sequence is delayed too long, an excessive rate of deceleration 1~15866 lS required which could be beyond the capability of the equipment or too uncomfortable for the passengers.
~he usual method for achieving good stopping perform-ance for such transportation systems is to have the velocity of the vehicle controlled in accordance with the distance, which can be conveniently called the "target distance", between the vehicle and the station at which a stop is being made. When the vehicle i8 far enough away from this station, ma~imum speed is permitted, but as the vehicle approaches closer and closer to the station, progressively lower speeds are called ~or. When the vehicle is running at maximum speed without a selected station stop, or if the selected gtop is farther away than the stopping distance of the vehicle, the target distance may be set at a predetermined value that is equal to or greater than this stopping distance.
In the case of an elevator car travelling in a hoistway, the target di~tance may ~e defined as the distance between the car and the so-called "target floor", which is a well-recognized concept in the elevator art and is described, for example, in U.S. patent No. 3,407,9~5 issued to John A. Gingrich on October 29, 1968. It will be appreciated that the target distance may be calculated in many wayg, hDwever, depending upon the particular transportation system involved.
For an ideal slowdown, where the vehicle has constant deceleration during most of th~ stopping sequence, the relation-ship between velocity and target distance is unfortunately not linear. Instead, since the distance a vehicle travels is proportional to the integral of its velocity over time, the velocity must be made proportional to the square root of a distance which is equal to the target distance less a constant.
This constant is required to compensate for the rounding-off at the end of the stopping sequence where the deceleration is reduced to zero. The maintaining of the constant deceleration

- 2 -i~l5~66 _ight down to zero velocity would, of course, theoretically result in an infinite rate of change of acceleration (sometime called "bump" or "jerk"~ at the instant of stopping: a condition which would be objectionable to passengers.
During the final round-off, the relationship between velocity and target distance is no longer a square root. Ideally the velocity should be parabolic with respect to time and this require~ the velocity to be made proportional to the cube root of the target distance squared. ~owever, good results can be obtained by making the velocity directly proportional to the target distance. This relataonship theoretically produces exponential rounding-off, but in practice the response of the ~ystem is such as to make it quite accepta~le and not much different than the ideal parabolic rounding-off.
For elevators, a common practice for many years has been to reproduce the car motion, to some suitable scale, on a mechanical device known as a selector. Examples of such selectors can be found in the U.S. patents No. 2,074,578 to E.D. Dunn et al and No. 2,657,765 to C. Savage. On these selectors, the target distance iB represented by the displacement between two parts of the selector. Electrical contacts operated by this displacement are caused to operate in sequence as the target distance decreases during a stopping sequence. These contact usually feed relays whose con~acts control the magnitude of field current in the generator of a motor-generator set, and the generator armature supplies direct current to the hoist motor. The motor speed iB
thus reduced in a series of steps. By adjusting the values of target distance at which the steps occur, and by adjusting the corresponding reduction in generator field current, any desired relationship between velocity and target distance can be obtained.
When the mechanical selector is replaced by a solid state sy~tem, the target distance i~ obtained electronically and is usually expressed in digital form by low voltage signals on 1~5866 ~ne or more wires. One system for determining the target distance electronically is disclosed in our co-pending patent application No. 281,842 filed June 30, 1977; another s~stem is disclosed in the U.S. patent No. 3,773,146 issued to Dixon, Jr.
et al on November 20, 1973.
Further circuitg are required to establish a velocity pa~tern, preferably an analog signal, which has the correct relationship to the target distance. Various speed control ~cheme~ are available to force the motor speed to obey the dic-tation from an analog signal. An example of such a system is disclosed in U.S. patent No. 3,706,017 issued to John A. Gingrich on December 12, 1972. Alternatively, the velocity pattern could ~e expressed by a series of relays, driven by the solid state circuit output, which correspond to the relays which would other-wise be driven from the target distance as determined by a mechanical selector.
In accordance with the present invention there is provided a novel and improved method and apparatus for repetit-ively calculating the square root of distance for the purpose of controlling the velocity of a vehicle as it appreaoche~ a station at which a ~top i8 required. This system uses a repeating calculating sequence to process the target distance. At the beginning of a calculating sequence, during a first step or step No. 1, the target distance is loaded into a register. At each succeeding step of the calculation, a specific amount i8 sub-tracted from the register so that the amount in the register decreases. The speed pattern is made proportional to the number of ~teps required to ~ring the contents of the register down to zero. The calculation is repeated many times per second.
If the specific amount which is substracted each time i~ a constant, the relationship between the speed pattern and the target distance is linear. This occurs for a small portion of the calculating sequence, starting at step No. 2 and continuinq lilS~66 to step x-l, where "x" is predetermined integer. This is to allow for the final rounding-off of deceleration where the deceleration is reduced to zero.
If the specific amount which is subtracted each time is increased by a constant amount at each step of the calculation, the speed pattern is proportional to the square root of the target distance. This procedure is carried out beginning with step x, during which an amount q is subtracted from the register. At step x+l the amount q+d is subtracted, and at step x+2 the amount q+2d i5 subtracted. This continues, in a similar manner, ~or the remainder of the calculation.
If suitable values of q and d are chosen, the number of steps of the calculation which are required to reduce the content~ of the register to zero can be used to determine how many speed relays should be energized, to obtain relay operation ~imilar to that obtained from a mechanical selector. For example, 10 speed relay~ may be used, with all ten energized for maximum speed. When the target di~tance has initially decreased below the slowdown distance, only nine steps of calculation are required to bring the contents of the register to zero, and one of the speed relay~ iB thereby de-energized. Further reduction in target distance results in only eight steps of calculation being required, and this causes a second speed relay to be de-energized so that only eight remain energized. This process continues in a similar manner until all of the relays are de-energized when the car reaches floor level where the target distance is zero.
This system automatically causes the speed relays to be de-energized at equally spaced intervals of time provided that the velocity is reduced by the same amount at each step.
This corresponds with what has ~een considered the best practice on mechanical selectors. An important advantage of the solid l~lS~66 state system according to the present invention, as compared to mechanical selectors, is the ability to increase or decrease the slowdown distance by changing the values of q and d. These values can be programmed by small toggle switches, for example.
Corresponding adjustments on mechanical selectors are consider-ably more difficult.
Generally, while both q and d may be made adjustable, it i8 actually the ratio of these two values which is important.
Therefore, a fixed value of d may be chosen which results in a reasonable number of steps for the entire calculation. Once the value of d is chosen, q can be adjusted to vary the ratio of g to d.
Basically, the correct value of q is determined by the amount of error (between desired speed and actual speed) which is required to produce the constant deceleration which exists for most of the slowdown. In the aforementioned U.S.
patent No. 3,706,017 entitled "Motor Speed Control", the slider 194 on resistor 190 in Fig. 9 can be adjusted to vary the amount of error required to produce the required amount of deceleration.
Al~o, ~lider 524 on resistor 520 in Fig. 11 accomplishes the same result in a different manner. This adjustment is made to obtain a suitable smoothness for the speed control.
Changes in this adjugtment (the sliders ~4 and 524) require corresponding changes in the value of q if the constant deceleration i~ to be achieved. The adjus~ment is generally not criticalt and has negligible effect at higher sp~eds. However, if q is set too low, the deceleration rate will increase toward the end of slowdown, just before the taper-off begins.
Similarly, if ~ is set too high, the deceleration rate will decrease toward the end of deceleration, before the taper-off begins. This may be an advantage, in some cases. The effect of armature reaction in the hoist motor usually shows during a down slowdown with a heavily loaded car, just about where the 1 ~ ~ 6 ~
taper-off begins, and if there is a pro~lem, it may be solved by using a higher setting of the quantity q.
In a preferred embodiment of this invention, very small values of q and d are chosen, so that the number of steps of calculation is high. A maximum of ~56 steps, for example, is quite suitable. Then, the speed pattern (which i8 proport-ional to the number of stepg required to reduce the contents of the register to zero) has so many possible values that it is alst equivalent to an analog signal. It is, in fact, expressed in the form of a repeating pulse W~Dse length is proportional to the velocity pattern. Thi~ pulse goes high near the beginning of a calculating sequence, and goes low when the contents of the register have been reduced to zero. It is convenient to refer to thi~ pulse as signal SR, for "Square Root".
In accordance with the present invention there is also provided a novel and improved apparatus for converting the pulse width-modulated signal SR, representing the square root of distance, into an analog signal for use as a speed pattern for a vehicle. More particularly, thi~ analog computing system converts the width of each pulse on signal SR into a steady voltage which is proportional to this width. The steady analog voltage becom~s the input to a speed regulating system, to dictate the desired ~peed of the vehicle.
The principal advantage of the present system of digital-to-analog conversion is to cause a single adjusting device, such as a potentiometer, to simultaneously adjust the deceleration rate of a vehicle and its slowdown distance so that whatever deceleration rate i8 chosen, deceleration automatically commences at the correct instant to assure that the selected deceleration rate will prevail during st of the stopping sequence, with a suitable rounding-off at the end of the stopping sequence which i8 also adjustable. The versatility of this system of digital-to-analog conversion will ~e illustrated by the disclosure of a ~1S866 method for automatically reducing the deceleration rate of an elevator and simultaneously increasing the slowdown distance, when it is travelling down with a heavy load. A similar reduction in acceleration rate is also descri~ed for a heavily loaded car travelling up.
An operational amplifier is used in the analog computing system as an integrator and it is convenient to refer to its output as the "working signal n . At step 1 of each calcu-lating sequence, the working signal is caused to assume an adjusta~le voltage proportional to the landing speed of the vehicle. Then, for ~teps 2 through x-l, an adjustable current i~ applied to the integrator to cause the working signal to increase at an adjustable rate. For step x and all succeeding step~, a second adjustable current causes the working signal to increase at a different rate.
When the contents of the calculating register have been reduced to zero, and the pulse on signal SR goes low, the value of the working signal i~ held in a sample-hold circuit until the similar ~ampling point is reached in the next calculation.
Thu~ the voltage on the sample-hold circuit becomes the final out-put of the calculating ~ystem, and is used as an input to a speed regulating system. The calculation is repeated at a sufficiently high frequency that the sample-hold circuit changes its voltage by only a small amount at each sampling point.
The deceleration rate is adju~ted by the second of the two adjustable currents mentioned previously. If this current is adjusted to a low value, a longer time is required for the working signal to reach a value corresponding to full speed, and - this corresponds to a greater target distance. If this current is adju~ted to a high value, a short tim~ is required for the working signal to reach a value corresponding to full speed, and thi~ corresponds to a short target distance. Thus a single adjustment controls the deceleration rate and also adjusts the 111~66 slowdown distance to suit this deceleration rate.
Adjustment of the first of the two previously mentioned currents controls the constant of proportionality between the velocity pattern and target distance which exists during the final rounding-off at the end of a stopping sequence.
The system according to the present invention is particularly suitable for use with a speed control system such as disclosed in the aforementioned U.S. patent No. 3,706,017.
In thi~ system the actual velocity of the vehicle lags behind the velocity pattern ~y an amount proportional to the accelera-tion of the ve~icle. This eliminateg the need for rounding-off at the commencement of a stopping sequence.
If a stopping sequence commences after the vehicle ha~ reached top speed, the relationship between the velocity pattern and target d$~tance can immediately assume the square root pattern which i8 required for the latter part of the ~low-~own sequence. Althoug~ the velocity pattern will then start to decrease a~ruptly at the commencement of slowdown, the actual velocity cannot decrease a~ruptly, and a smooth transition from constant speed to constant deceleration is inherently achieved.
Similarly, if a stopping se~uence commences while the vehicle is still accelerating towards top speed, a velocity pattern which is proportional to the square root of distance is precisely what i8 required for the speed regulating system of the aforementioned patent No. 3,7Q6,017 to smoothly control the transition from acceleration to deceleration.
For a more complete understanding of th~ present invention, reference may ~e had to the following detailed description taken in conjunc~ion with the accompanying figures of the drawings in w~ich:
Fig. 1 illustrates graphically the computing procedure used in the present invention;
Figs. 2A and 2B illustrate graphically a velocity ~attern, on a time base, which can be obtained with this computing procedure, and the corresponding vehicle velocity;
Fig. 3 illustrates graphically a further extension of this computing procedure used in t~e present invention;
Figs. 4, 5, and 6, taken together, show a circuit for calculating the square root of the target distance, or a portion thereof, Fig. 7A shows some of the voltage waveforms associated with the circuit of Fig. 4;
Fig. 7B shows a program chart containing the calculation algorithm carried out by the circuit of Figs. 4, 5 and 6.
Fig. 8 show~ a circuit for energizing speed relays in accordance with one preferred embodiment of this invention;
Figs. 9A, 9B, and 9C illustrate graphically a method for producing an analog speed pattern in accordance with another preferred em~odiment of thi~ invention;
Figs. 10, 11, and 12, taken together, show a circuit for producing an analog speed pattern in accordance with the method of Figs. 9A, 9B and 9C; and Fig. 13 ~hows miscellaneous circuits to fit in with a typical elevator system.
The basic computing procedure for producing a velocity pattern is illustrated by a graph in Fig. 1 where a solid line show~ the relationship between velocity and time for a major portion of a stopping sequence for a vehicle. In this figure it i8 the straight line portion of the graph which is of particular interest, but a typical rounding-off to zero acceleration is also illustrated. The initial part of the stopping sequence is not shown.
~ vertical dotted line is drawn at time tQ which corresponds with the end of the straight line portion. Further Yertical dotted lines are drawn at time tl, t2~ t3 etc. such that they represent events which are equally spaced apart in time.

l~lS866 The area under the solid line is proportional to the distance travelled, and thu~ the area below the solid line between the vertical dotted lines at t1 and to is proportional to the distance travelled between these two times. This distance will be given the value q. The distance travelled after to will be given the value r.
The next area, between t2 and tl is greater than the preceding area by an amount which will be given the value d.
A horizontal dotted line vl is drawn to separate this area into an upper portion with area q and a lower portion with area d.
Similar lines v2, v3, v4, etc. can be drawn as shown in Fig. 1.
Because of the constant deceleration, as a result of the ~traight line relating velocity to time, each succeeding area, progressing from right to left, is greater than the preceding ~rea by the amount d. The successive areas are q, q+d, q~2d, q+3d, q~4d etc. as shown in Fig. 1.
In Fig. 1, the flow of real time is directed from left to right. The flow of time during the computing sequence i8 from right to left on the graph, and is at a much higher speed BO that during any one computing sequence, the flow of real time i8 very small and real time can be considered to have stopped at a ~pecific value during said computing sequence.
Basically, the computing sequence involves the successive substraction of quantitie~ rCpreferably in several equal steps), q, q~d, q+2d, q+3d etc. from an initial quantity equal to the target di~tance. Each subtraction is considered to be one step of the computing sequence, and the number of steps required to compietely deplete this initial quantity can, in one embodiment of thi~ invention, be used to determine the number of generator field controlling relays which should be energized.
If, for example, real time has progressed to somewhere between t5 and t4, and if it is assumed that the quantity r is ~ubtracted in three equal ~teps, æeven stepæ of the computing 1~15866 sequence will not quite deplete the initial ~uantity, but eight steps will more than deplete the initial quantity. Thus the actual switching points, where the number of steps required to deplete the initial quantity changes, occur at times t6, t5, t4, t3, etc. which are equally spaced in real time. The speed control system must cause equal reductions in speed at each such ~tep to achieve the constant deceleration.
Fig. 2A shows, by a solid line, a speed pattern which might be produced in this way by a series of steps produced by relay~ and smoothed by ramp smoothing as disclosed in the afore-mentioned U.S. patent No. 3,706,017. The actual velocity is shown by a dotted line in Fig. 2A for a stopping sequence which commences while the velocity is steady at full speed.
Fig. 2B shows, in a similar manner, a stopping sequence which commences while the vehicle is still accelerating towards top speed. The speed pattern is still computed in the same way, but initially the velocity is low, and thus the time required for the pattern to decrease by one step is much greater than it would be if the vehicle were decelerating from top ~peed as in Fig. 2A. As the vehicle accelerate~, the increasing velocity cau~es the steps in the speed pattern to occur at more frequent interval~. The dotted line shows the actual velocity.
When, in Fig. 2B, the decreasing speed pattern approaches the ri~ing vehicle velocity, the full acceleration is succeeded by a gradual reduction in acceleration as the error between actual speed and pattern speed decreases. At the instant the two curves cro~s, zero acceleration is called for. Then, the decreasing pattern velocity becomes lower than the actual velocity, and the acceleration becomes gradually more negative until constant deceleration is achieved. The smoothness with which this tran-~ition ta~es place is adjustable by means described in the afore-mentioned U.S. patent No. 3,706,017.
In Fig. 3 a small portion of the straight line portion lilS866 ~f the graph of velocity versus time is expanded on the time axi~ to show, in a manner similar to Fig. 1, the result of making ~ and d have much smaller values. The area between time ta and tb in Fig. 3 is assumed to be equal to the height of one cell on a loop stick as descri~ed in our co-pending patent application Serial No. 281,842, filed June 30, 1977. This height i~ typically 2 1/2 inches.
From Fig. 3 it can be seen that although the computing procedure has a number of steps between tb and ta, wi~h corres-ponding tiny steps in speed pattern, only the speed patternscorresponding to tb and ta are actually used since the target distance which is loaded initially into the calculation can only as~ume values which are multiples of 2 1/2 inches or whatever height i8 used for the cells in the loop stick. The quantities q and d, however can have fractional values.
Generally, the total number of steps of calculations which are allowed should be as high as possible to approach equivalence with an analog ~ignal. ~owever, if there are too many steps, the calculation will require too much time to complete, and it will not be pos~ible to repeat the calculation at a high enough frequency to keep it up to date with the decreasing target di~tance. A suitable value to avoid these two extremes is 256 steps, but fewer or more could be used successfully.
With a maxiumum of 256 steps, and with cells of appro-ximately 2 1/2 inches, a reasonable portion of the stopping sequence ha~ a step of speed pattern occurring every 2 1/2 inches.
HoweYer, when the target distance i8 large, the reduction in speed pattern required for each 2 1/2 inches of travel is quite small, and because there are only 256 aiscrete values of velocity pattern,frequently no reduction in speed pattern will occur after 2 1/2 inches of travel, and several cells of 2 1/2 inches will hsve to be pacsed before th speed pattern can change from one of its 256 discrete values to the next.

~66 This lack of a step in speed pa~tern for each 2 1/2 inches of travel has no degradation on the accuracy of the resulting performance of the vehicle, for the steps are still very much more closely spaced in time than with conventional systems.
The important part of the stopping sequence is closer to the end where each ~ucceeding reduction of 2 1/2 inches in target distance result6 in larger and larger reductions in speed pattern. Here, a reductioh every 2 1/2 inches is closer to current practice, but i8 still generally superior to the best of current mechanical selector systems. With ramp smoothing of the speed pattern, there is no need to seek finer steps, although the basic principle of this invention places no restriction on the use of finer steps of measurement of target distance.
Figs. 4, 5 and 6 show a circuit for creating a signal SR
which is generally proportional to the square root of distance (except for the rounding-off at the end of a stopping sequence~.
This circuit operates in accordance with the algorithm illustrated in Fig. 7B. In Fig. 4 a basic clock consists of a mono~table multivibrator ~MM~ 12, MM 13 and AND gate 16. The output BC of this basic clock is a square wave which is high for a period of time determined by MM 13 and which is low for a period of time determined by MM 125 the~e times could be equal. A frequency of about 250 kHz would be appropriate for this clock.
During the entry of data, when the target distance is sequentially loaded into the system, the basic clock is cau~ed to assume a different frequency, generally slower, to match the frequency of the read-out ~equence of the calculator which repetitively calculate~ the target distance. Such a calculator i8 disclosed in our co-pending patent application Serial No. 281,842 filed June 15, 1977. Output RO, in this co-pending patent application, is high during the read-out period, and output CXD happens to be suitable for driving the basic clock during this period. Similar signals would be required from any ll~S866 .ther system for determining target distance.
While input RO is high~ the output nRO of inverter 14 i8 low and this holds the reset input nR of MM 12 low so that its output is high. This enables the pulses on input CKD to be gated through INVERTER 17, NAND gate ~5 and AND gate 16 into the input of MM 13. T~e resulting output pulses of MM 13 remain high for the same length of time as during normal clocking, but the pulaes are spaced further apart in time, if necessary, to follow the slower clocking of input CKD.
The basic clock signal BC is applied to an 18-step counter consisting of U/D binary counter 18, gate 19 and clocked flip-flop (FFl 20. If counter 18 is CMOS type 402~, input B
must ~e connected to the positive supply line P10 to produce binary counting rather than binary-coded decimal counting. Also, ~carry-in" input nCI muat be connected to the negative supply line OV.
A four-digit binary counter such as counter 18 in Fig. 4 i~ itself capable of only 16 stepg, but 18 are ~teps obtained by ~tepping it bac~ward one step at a particular count, 80 as to repeat two of the 16 steps. The output KX of FF 20 distinguishes between the two original and the two repeat steps 80 that 18 identifiable step~ are obtained.
The input U of counter 18 is connected to signal nB10 which i~ high for 17 of the 18 stepg, to cause up counting; for one of the 18 steps it is low to cause down counting.
The two least significant outputs of counter 18 are decoded by decoder 24A which could be one half of a dual decoder such as the CMOS type 4555. The two most significant output~ of counter 18 are decoded by decoder 24~ which could be the other half of the dual decoder. If type 4555 is used, the enable input nE must be held low in each half by connection to the negative supply line OV, as shown on Fig. 4.
The tLme relationship of the eight decoded signals KLO, 1~ 1 58S6 ;Ll, KL2, KL3, KMO, KMl, KM2, and KM3 are shown in Fig. 7A along with other signals produced by the circuits of Fig. 4. A
flip-flop consisting of NAND gates, 26, 27, 28 and 29 is used to create signal KF which is used to identify a portion of the 18 step~ for the purpose to ~e described later.
The 18 repeating steps obtained from the circuits of Fig. 4 are used to control the sequential processing of 18-bit binary numbers. Ten of these 18 steps are used to directly represent distances expressed in units of about 2 1/2 inches;
one of these ten digits is a polarity bit and this leaves 9 bits which can represent up to 511 units, or about 106 feet. The remaining 8 digits are used to express fractions of the basic unit of 2 1/2 inches. The least significant digit represents 1/256th of a unit.
Slowdown distances of more than 106 feet are required only for speeds over 180~ ~eet per minute, and thus nine bits are sufficient for most elevator application~. For other applications, the basic unit of 2 1/2 inches could be increased, or additional binary bits could be added to the calculation withDut altering the basic idea.
At the instant when signal R~ goes high, MM 11 i~
triggered and its resulting output pulse ROA is used to set the counter 18 and the flip-flop 20 to the appropriate state to identify the least significant digit of the target distance, ignoring the eight fractional digits. The load inputs Ll, L3, ~nd ~4 of counter 18 a~e connected to OV and load input L2 is connected to P10 to achieve the desired loading of the counter when ROA is high.
After the initial read-out period, the 18 bits are sequentially processed starting with the least significant bit and proceeding through to the most significant bit blO; when bit blO is being processed, NAND gate 21 makes signal nB10 low.
Signal nB10 i~ applied to the clock input CK of binary i~15866 _ounter 22 in Fig. 5, to cause it to advance one step when signal nB10 goes high after the most significant bit has been processed. Thus counter 22 counts the steps of the calculation, where each step has 18 parts, one part for each bit. The purpose of counter 22 is to identify step x which is adjustable by switches SW5, SW6, and SW7. During step x, all inputs to NAND gate 30 are high, and its output is low.
Counter 22 is loaded to a specific count when signal ROA
goes high at the ~eginning of a data entry sequence when RO is high. Load inputs Ll, L2 and L4 of counter 22 are connected to OV, and load input L3 is connected to P10. Thus, counter 22 is initially loaded to a count of four.
If switches SW~, SW6 and SW7 are all open, three of the input~ to yate 30 are held high by registors R7, R8 and R9, and four ~teps are required before counter 22 has advanced to a count of eight to make signal S8 high. This corresponds to xs5 and is the minimum setting. ~ith signal S8 high, all inputs to gate 30 are high to make its output low.
If switch SW 5 i~ closed, and switches SW6 and SW7 are open, signal Sl will hold one of the inputs of gate 30 low at a count of eight on counter 22 and it i8 not until the count of nine thnt all inputs to gate 3Q are high. Thus with switch SW5 only closed, x~6. By similar reasoning, x=7 when switch SW6 only i~ closed.
By closing appropriate combinations of switches SW5, SW6 and SW7, x can have any value ~etween 5 and 12 inclusive. Thi~
corresponds to distances of 10 inches through 27.5 inche~ in 2.5 inche6 steps, as~uming 2.5 inches for each unit of distance.
This adjustment determines the distance over which the final rounding-off will occur.
At step x, the output of gate 30 is low, and this sets a flip-flop consisting of gates 31 and 32 which was previously reset during the RO period when nRO was low. When this flip-flop 1~
S set, signal SXM goes high and ~ignal nSXM goes low. These signals serve to distinguish between steps 1 through x-l, when SXM is low, and step x and all succeeding steps when SXM is high.
A clocked flip-flop 23 is clocked by signal nB10 and uses signal SXM for its data input. This causes signal ASX to go high for step x+l and all succeeding steps. Flip-flop 23 i8 re~et by signal R~A. Signal naSX is an inverted form ofsignal ASX .
Thus, signals SXM and ASX divide the steps into three part~:
(1) step~ 1 to x-l when bot~ signals are low;
C2~ step x when SXM is high but ASX is still low;

(3) step x+l and up, w~en both signals are high.
After step 12, counter 22 steps to a count of zero and repeats its counting but has no further effect on signals SXM
and ASX until after the next data entry sequence when signal R0 i~ high.
A further group of ~witches SWl, SW2, SW3 and SW4 is provided as ~hown on Fig. 4, to create a 4-bit binary number expressed in serial form on ~ignal MQ. The signal M~ is only u~éd during ~he period when XF is high, and during this period the output~ of decoder 24A go high, one at a time, in the order RL2, KL3, KL0 and KLl.
At each of tXese four steps, signal Mg will be high only if the corresponding switch i8 closed, via OR gate 25. The

4-bit number thereby created on signal MQ is equivalent to the amoDt q which is adjusta~le and can assume any one of 15 different values. The 16th condition with all four switcheæ
open, mu~t not be used.
In the above ~ummary of thig invention, the calculating procedure iB described a~ the subtraction of increasing amounts q+d, q+2d, q+3d etc. at steps x~l, x~2, x+3, etc. In Figs. 5 and 6, the opposite procedure is u~ed to facilitate the use of ~erial adders. The target distance is loaded into a shift register 36 as a neqative number and the increasing numbers q, q+d, q+2d etc. are created in shift register 35 and are added to the contents of shift register 36 at each step at the cal-culation. When the contents of shift register 36 change from negative to positive, the output signal SR goes low. Thi~
procedure has the same final result as the procedure described in the summary.
The circuit of Fig. 5 uses a CMOS triple serial adder 34 of type 4038, this device has internal carry and thereby avoids the need for further external circuits. Other devices could, of course be used with suita~le modifications to Figs. 5 and 6.
Similarly, the two 18-stage shift registers 35 and 36 can be CMOS type 4006 or any other suitable type. If type 4006 is used, the four internal sections must be externally connected together to get 18 stages; this is not shown on Fig. 6 but would be apparent to anyone skilled in the art.
During step 1, RO is high and the target distance enters a~ a ~equential 10-bit binary number on input nTGD. One section 34A of triple serial adder 34 is connected to convert the target distance into a negative number. The connection of input nB of adder 34A to the negative supply line OV is equivalent to adding binary number 1 111 111 111 which is interpreted as -1 by the adder. The output nS of adder 34A is already inverted by the characteri~tics of type 4038 and thus inverting input nI of adder 34A is connected to PlO to prevent further inversion. The combination of subtracting 1 and inverting causes the target distance to be converted from a positive number into a negative number in the f ~;liar 2'8 complement form.
During step 1, the target distance, as a negative number, is entered into shift register 36 via NAND gate 41, AND gate 45, serial adders 34B and 34C and AND gate 40. During this step, all other inputs to gate 45 are kept high becau~e inputs SXM, ASX and ~1~6S
RO, into NAND gate~ 42, 43 and 44, respectively, are all low.
Also during step 1, no data enters the other input nA of adder 34B since signal ASX is low to hold the output S12 of NAND
gate 38 high. Similarly signal nRO is low to hold the output of NAND gate 39 high to prevent the entry of data into the other input nB of adder 34C.
During steps 2 through x-l, data on signal nSS2 enters input nA of adder 34C via gates 44 and 45 and adder 34B. This data con~ists of one positive pulse at each step, timed to occur when signals KMO and KL3 are both high as controlled by gate 44.
This timing agrees with the time during which a particular bit of the 18-bit sequence i8 procesged: the bit whose weight is equivalent to one cell of 2.5 inchec. This occurs-, of course, after the eight fractional bits have been processed.
A circuit is provided in Fig. 5 to assure that any fractional bits left over in shift register 36 are cleared during the early part of step 2. This circuit consists of inverter 46 and clocked flip-~lop 47. ~t the end of the RO period, when signal nRO goes high, the Q output of flip-flop 47 goes low because data input D is connected to OV. The first time the output of gate 44 goes low, the flip-flop is set via inverter 46 ~o that it~ output Q is h~gh. This output ~signal nCSR) is applied to AND gate 40 to force the input of shift register 36 low during this period. This period i~ when fractional bits are processed.
During steps 2 through x-l, inputs RO, SXM and ASX to gates 41, 42 and 43 respectively, are low; this makes their outputs high 80 that data enter~ gate 45 only from gate 44. Thi~
- data which enters input nA of adder 34C i8 added to the present content~ of shift register 36 which is routed through gate 39 which has its other two inputs nRO and nEOC both hiqh. The resulting sum is inverted by adder 34C ~y input nI of adder 34C
being connected to OV) 80 that it i~ now hiqh to indicate 1 and ow to indicate o. This sum is entered back into shift register 36 via gate 40. Thus the contents of the register are changed, one bit at a time from the previous value to the new one.
Thus it can be seen that during each step between step 2 and step x-l inclusive, the negative number in shift register 36 i6 reduced to a negative number of lesser magnitude by the amount of one cell of 2.5 inches. If the target distance is less than the amount allotted to the rounding-off distance, the contents of shift register 36 will change from negative to positive sometime prior to step x, and obviously the number of steps required to do thi~ is directly proportional to the target distance. The detection of when the contents of shift register 36 become positive is accomplishRd in clocked flip-flop 52 which i~ arranged to clock the output of shift register 36 at the end of each step when the most significant bit blO is available on the output of the shift register.
During step x, data enterg input nA of adder 34C via gates 42 and 45, and adder 34B. The data which enters i~ the amount q a's adjusted by switches in Fig. 4 as previously described.
The outputs of gate~ 41, 43, and 44 are held high during ~tep x by inputs RO, ASX and nSXM being low. The only difference between ~tep x-l and ~tep x i~ that the amount q is added, rather than one cell, into shift register 36. Although shift register 35 ha~ not yet been used in the calculation, it is receiving at its input the same data which i~ presented to input nA of adder 34C.
Thu-, at the end of step x, the amount q has been entered into shift register 35.
During ~tep x~ ignal ASX is newly high 80 that the contents of shift register 35 are applied to input nA of adder 34B via inverter 37 and NAND gate 38. Input nB of adder 34B
receives one pulse at the beginning of step x~l via gates 43 and 45. This one pulse is timed to represent the least significant fraction, and this is equivalent to giving the least po~sible l~lS8~6 value to the quantity d. The output nS of adder 34B is thus the sum q+d.
At each succeeding ~tep, the contents of shift register 35 are increased by the amount d in a manner similar to the description in the preceding paragraph. At the ~ame time, this increasing number from shift register 35 is added to the contents of shift register 36; thu~ the amount added in steps x+l, x~2, x+3 et~.is q+d, q+2d, q~3d etc.
At the end of each step, when nBl~ goes high, clocked flip-flop 52 has its clock input CR made high by signal nBla 80 that the most significant bit no~ available on the output of shift register 36, and applied to data input D of flip-~lop 52, causes output SR of flip-flop 36 to assume a state corresponding to the polarity of the binary num~er in shift register 36.
During step 1, when RO is high the flip-fop 52 is reset 80 that SR is low. At the end of step 2, SR becomes high unless the target distance is one unit or lesg. Signal SR remains high until the contents of shift register 36 have become positive.
Thus the length of time during ~hich signal SR is high is proportional to one less than the number of steps of calculation required to deplete the target distance by decreasing its neg~tive value in a serie~- of ateps.
A further clocked flip-flop 53 uses t~e inverted output nSR of flip-flop 52 to create signal nEOC which goes low when nSR goe~ high to signify the end of the calculation. It remains in this state until step 1 when signal RO ~ets flip-flop 52.
Signal nEOC is applied to gates 38 and 3~ to prevent any further increases in the contents of ~hift registers 35 and 36 while the circuit awaits th start of the next calculation, wh~n RO goes high again a~ an up-dated target distance is entered into the system.
The data input for clocked flip-flop 52 i8 obtained from the output, rather than the input of ~hift register 36. This is i~iS866 ~one to make the pulses on signal SR disappear not only when the target distance i~ zero, but also when the target distance i~ only one unit. This is to assure that in circuits to be described later in connection with Fig. 10, the velocity pattern will assume an adjustable value equivalent to landing speed for the last 2 inches of travel.
A further clocked flip-flop 48 uses the output of adder 34C a~ its data input; the resulting output nLV is low when the target distance is zero, indicating that the car is level at the target floor. Here, level is taken to mean within 1/2 inch of exact alignment. If the target distance is not zero, indicating that the car is mor~ than 1/2 inch from the target floor, signal nLV is high.
Another clocked flip-flop 5a can be provided to create ~ignal ZN which may be u~eful. The zone loops descri~ed in our co-pending patent application Serial No. 281,842 filed June 15, 1977, can be used to create signals which indicate when the car i8 within a specified distance of any floor. Signal ZN, however, iB high only when the car ig close to the tar~et floor. The use of signal~ S8 and nASX through gate 49, as the clock input CZ of flip-flop 50, makes the signal ZN high when the car iB
closer than 10 inches to the target floor. This signal can be used, for example, to limit the power opening of the elevator door~ to a region extending 10 inches above and below the target floor.
The baRic cloc~ signal BC i8 applied to the nCX inputs of both shift registers and the triple serial adder. This is required on the shift regifiter~ to specify when thé data should be shifted. Thi~ input i~ required on the adders to clock the internal carry latch. Also, input nCR (carry reset~ of the adders i8 connected to the output of OR gate 51. This is to separate each step of the calculation 80 that th~re is no carry-over from the st significant bit into the least significant 1~1S866 bit. This nCR input is low when n~l0 is low except during the RO period.
Fig. 8 shows a circuit for using signal SR to control the deenergization of conventional speed relays during slowdown.
Two 8-bit addressable latches 59 and 60 are used to control up to 16 speed relays. Any of the 16 outputs of these two 8-bit latches can be amplified, by circuits not shown here, to drive relays.
The circuit of Fig. 8 require~ signals nB10, ROA and SR
as inputs; the-~e signals have already been described in connection with Figs. 4, 5 and 6. A counter 55 is operated in a manner similar to counter 22 on Fig~ 5 and thus it counts the steps of the calculation. Counter 55, however, is loaded to zero when ROA
goes high, because its load inputs Ll, L2, L3 and L4 are connected to the negative supply line OV. Also, the carry-in input nCI of counter 55 i8 obtained by inverting the carry-out input nCO by inverter 54; this causes the counter to stop at the maximum count of 15 (last Qtep of sixteen num~ered from 0 to 15) and ignore further pulses on signal nB10 until counter 55 is reset when ROA

goes briefly high at the start of the next calculation.
During the first eight steps of the calculation, output Q4 of counter 55 i8 low and thus the output of inverter 56 is high. This causes the WRITE DISABLE input WD of 8-bit latch 59 to be held continuously high via OR gate 57 while the similar input of 8-bit latch 60 follows the pulses on signal nB10 via OR gate 58.
During the second eight steps of the calculation, output Q4 of counter 55 is high and thus the output of inverter 56 is low. This causes the WRITE DISABLE input of 8-bit latch 60 to be held continuously high via OR gate 58 while the similar input of 8-bit latch 59 follows the pulses on signal nB10 via OR gate 57. At the last of these second eight steps, the counter 55 stops responding to the nBlQ pulses, and holds this lllS~66 state until the next calculation starts.
During both of these eight steps, the Ql, Q2 and Q3 outputs of counter 55 are progressively addressing the eight latches in each of the two 8-bit latches 5g and 60. The infor-mation contained in signal SR is being written into these latches, and after the process is complete, each of the sixteen latches holds a condition representing the state of signal SR
at the time of the writing. Depending on the length of the pulse on signal SR, more or fewer latches will be holding the high state. Each of the sixteen latches has its state applied to a corresponding output Q~, Ql, Q2 etc. of either of the two 8-bit latches 59 and 6Q.
Thus during slowdown, as the length of the pulse on signal SR decreases, fewer latches have high outputs, and fewer speed relays are energized. The dropping-out of these speed relays will generally occur at equal intervals of time except for the last ones where the constant deceleration ceases and the deceleration tapers off.
The circuit of Fig. 8 is disclosed here to show approximately what is required to achieve the result~ shown in Figs. 2A and 2B. ~his, however, is not the preferred embodiment of thi~ invention. The values of q and d obtained from the circuits of Figs. 4, 5 and 6 are suitable for the preferred embodiment where an analog speed pattern i8 produced by circuits yet to be described in connection with Figs. 9, 10, 11 and 12.
For the circuit of Fig. 8, larger values of q and d would be required, and although thia is not shown on the figures, it involve~ simple changes which would be obvious to anyone skilled in the art. For steps 1 through x~l, it would be appropriate, although not necessary, to retain the amount subtracted at each step, which is one cell; thi~ would tend to drop out a speed relay every 2 1/2 inches during the rounding-off near floor level.
If coarser ~teps are desired, some of the outputs of the 8-bit iii~866 latch 60 could be left unused.
Fig. 9A illustrates the manner in which the "working signal n WS is caused to operate for the purpose of creating a "speed pattern~ SP which can be used to dictate the speed of the vehicle. The lower graph shows the waveform of the voltage on signal WS, on a time base, for one complete calculation. The length of time required for one complete calculation might be approximately 16 milliseconds, and possi~ly 256 steps of cal-culation might occur in this time.
Fig. 9A also includes the waveforms of five other pertinent signals. During the data entry period, when signal R0 is high, the working signal WS is brought down rapidly to a low value corresponding to an adjustable landing speed. Ths rate at which Rignal WS is brought down must be sufficiently high to a~sure that it reaches the desired value before the end of the R0 pulse regardless of the starting point.
For the duration of ~R pulse, the working signal rises, but not beyond the value corres~onding to an adjustable top speed.
Two adjustable rates are used for this rise:
1. Steps 1 through x-l when signal nSXM is high.
2. Steps x and higher, when signal nSXM is low.
Two additional signals SRA and SRB are derived from SR.
Generally SRA and SRB coincide with SR, but when the working signal has reached a value corresponding to the adjustable top speed, SRA and SRB are made low regardless o signal SR.
In ~ig. 9A, a fiituation is depicted which occurs when the target distance is in excess of the slowdown distance, which can be defined as the di~tance required to bring the vehicle to - a stop from the adjusted top speed using the deceleration rate determined by the adjusted slope of the working signal. In an elevator, this corresponds to the situation existing when the car is running at top speed and the target floor keeps advancing, because no stops are yet required, so as to always be ahead by lil5~66 at least the slowdown distance.
Under these conditions, signals SRA and SRB are terminated early by the working signal reaching top speed, while signal SR
remains high for a longer period. A monostable multivibrator is used to create signal nALT which goes low for a brief period immediately at the cessation of the pulse on signal SRA. When signal nALT reverts to the high state at the end of its timing period, the state of signal SR is clocked into a flip-flop.
If signal SR is high at the termination of the nALT
pulse, as shown in Fig. 9A, the clocked flip-flop gets set to a state which indicates that the target distance is at least somewhat greater than the slowdown distance. At each new calculation, the flip-flop receives the same data at the clocking point, and hence maintainæ its state, provided that the target distance, as expressed by the duration of signal SR, does not decrease too much.
As the car continues-to move at full speed, however, the point at which signal SR goes low in each calculation travels to the left in Fig. 9A until it reaches a point indicated by the dotted lines. Now, at the termination of the nALT pulse, signal SR is low. This causes the clocked flip-flop to be set to its other state.
This other state of the flip-flop is used to indicate when another advance of the target floor should be made, if no stop is required at the present target floor. The purpose of the nALT timing is to start the advancing se~uence slightly early to assure that the car does not start to decrease its speed for the old target floor due to any ~light delay in the advancing of the target floor.
After an advance of the target floor, the duration of pulse SR immediately aæsumes a larger value at the next calculation, and then decreases as the car continues to move nearer to thiæ new target floor.

ll~S866 The arrow 61 in Fig. 9A indicates the value which the speed pattern should have to dictate top speed. A continuous comparison is made between the rapidly changing working signal and the relatively slowly changing speed pattern. If the working signal becomes higher than the speed pattern prior to the arrow in Fig. 9A, it is an indication that the speed pattern is too low. It is then caused to rise at an ad~ustable rate until the arrow is reached. This generally does not bring the speed pattern up to the desired value, but the process is repeated at each new calculation. As the speed pattern nears agreement with the value of the working signal at the arrow, the time during which the speed pattern is caused to rise becomes shorter because the working signal exceeds the higher speed pattern later. This gives a desirable rounding-off of the rising speed pattern as it nears top speed. This effect also occurs when a stopping sequence is called for while the car is still accelerating.
The proceeding ~equence, which causes the speed pattern to increase in value, occurs mostly during the accelerating portion of a trip. During slowdown, a different process is required to bring the speed pattern down as the target distance decreases, while the vehicle approaches and stops at a floor.
Fig. 9B shows the waveform of the working signal when the target distance is less than the slowdown distance. This situation exists during slowdown, and can also exist during acceleration. Compared with Fig. 9A, the working signal in Fig. 9B does not reach as high a value because it is limited by the duration of the pulse on signal SR. Signals SRA and SRB
are now identical with SR.
The dotted lines in Fig. ~B show the situation which exi~ts when the target distance is much smaller; ~ow, the working signal rises to a lesser value, and can be brought down in a shorter time from this value when R0 goes high.
The arrows 62 and 63 in Fig. 9B show the points where 1i~5866 the value of the working signal corresponds to the value which the speed pattern should have. If the speed pattern is too low, the procedure already described causes it to be broug~ up to the correct level.
If, however, the speed pattern is too high, the procedure for bringing it down starts at the arrows and continues until the speed pattern has been reduced until it equals the working signal which retains the desired value for some time after the arrows.
It is desirable to place some restriction on the length of time during which this pulling down of the speed pattern can occur.
Otherwise, any failure of components in the calculation of the target distance could cause a too severe reduction in speed pattern.
Generally, when there is no such fault, the speed pattern quickly follows the descending value of the working signal as sampled at the arrows.
When the car is close to being level at the target floor, pre~erably within one cell away, signal SR has no pulses. When this occurs, the working signal follows a waveform as shown in Fig. 9C which is no longer in synchronism with signal RO. Now, the working signal oscillates between two levels corresponding to up landing speed and down landing speed which are adjustable. A
signal FT is created which gives a flat top to the WS waveform in order to permit the use of the same system as before for bringing the speed into alignment with the working signal as sampled at the points indicated by the arrows 64. Signal SRB is caused to change its state when the working signal arrives at a value corresponding to the adjusted up and down landing speeds.
Although there is no need to set the up landing speed different from the down landing speed, it is desirable to have separate adjustments in order to make up and down landing speeds equal in spite of inaccuracies and offsets in various opexational amplifiers.

It is desirable to use one polarity of speed pattern for one direction of travel, and the opposite polarity for the other direction. For the purpose of this embodiment, the negative polarity will be used for up travel and the positive polarity for down, for the speed pattern. The reverse could just as well have been used. The comparison between speed pattern and working fiignal is easiest to make, via operational amplifiers, if they are of opposite polarity. Thus, in Figs. 9A, 9B and 9C the positive polarity which is shown for the working signal repre-sents the up direction; for the down direction, the polarities would be opposite.
The circuits shown in Fig. 10 are mainly for creating signals nWN, WR, WD and WU which are required to cause the working signal to perform as previously described and illustrated in Fig. 9. Other useful signal6 such as SRA, nSRB and nTSM are also created.
The principal inputs are SR, RO and nSXM from Figs. 5 and 6, but further inputs are required. Signals U3 and D3 are obtained from Fig. 13 which will be described later; signal V3 is high either because the target floor is above, and thus the car should go up, or because the car i8 being controlled by constant pressure pushbuttons for maintenance or inspection and the up direction is selected. Signal D3 is similar, but for the down direction. Signal nIN is low for normal operation, and i8 high only for constant pressure operation during main-tenance or inspection. Signal nTFB is low when the target floor is bel~ the car; it is high when the target floor is above the car and also when the car is level at the target floor.
Inputs LSD, LSV and TS are o~tained from Fig. 11 and indicate when the working signal is within several adjustable values:
1. TS is high whenever the working signal exceeds an adjustable value corresponding to top speed in either 1~15866 direction;
2. LSU is high whenever the working signal is more posiiive than the positive adjustable value corresponding to up landing speed; and 3. LSD is high whenever the working signal is more negative than the negative adjustable value corresponding to down landing speed.
A retriggerable monostable multivibrator (MM) 82 is retriggered whenever signal SR goes from low to high, and the timing of this MM must be somewhat greater than the time to complete one calculation. Thus as long as there are pulses on signal SR, the output SRP of MM 82 remains high. When landing speed is required, signal SR remains continuously low and the output of MM 82 quickly returns to the low state and remains there until there are again pulses on signal SR.
A flip-flop consisting of NOR gates 78 and 79 is used to create signal nTSM. The purpose of this flip-flop i8 to detect when signal TS first goes high to indicate that the wor~ing signal has reached a value corresponding to top ~peed. Signal nTSM goes low whenever TS is high, but in addition, while signal nSXM is low (for most of each calculation) nTSM maintains its low condition after TS goes high in case TS goes low again. This could occur if the integrator which creates the working signal has a tendency to drift slightly when it should be maintaining a steady output. The 78-79 flip-flop is reset when nSXM goes high at the beginning of a calculation.
Signal SRA is derive~ from AND gate 73. In normal operation, nIN is low and the output of NAND gate 72 i8 high.
Thus signal SRA is equivalent to signal SR until nTSM goes low;
then it is low regardless of signal SR. This agrees with Fig. 9A.
Inverted signal nSRA i8 ohtained from inverter 96.
When there are pulses on signal SR, MM 82 is continually being retriggered, and its output SRP is continuously high. The 11~5866 output nSRP of inverter 93 is then low, and this makes the output of NAND gate 95 high. AND gate 97 then passes the inverted signal nSRA through to make nSRB equivalent to nSRA.
When there are no pulses on signal SR, nSRP becomes high, and SRA remains low. Now, signal nSRB is obtained from EXCLUSIVE-OR 94, routed through gates 95 and 97. This condition occurs when landing speed is being dictated, and signal nSRB must be controlled in a suitable manner to obtain the operation illus-trated in Fig. 9C.
A flip-flop consisting of NAND gates 87 and 88 is used to assist in achieving the operation illustrated in Fig. 9C.
Input nlN is low for this operation, and thus the outputs of NAND gates 75 and 76 are high. NAND gates 84 and 85 acts as inverters to invert signals LSU and LSD and apply them to the flip-flop consisting of gates 87 and 88.
When the working signal rises ahnve equivalence with the adjusted up landing speed, LSU goes high, the output of gate 84 goes low, and flip-flop 87-88 is set to a state where signal X is high and signal nX i8 low. Later it will be explained that this causes the working signal to stop rising and either immediately or after a delay start to ramp down ~i.e. become more negative~.
Similarly, when the working signal falls below equivalence with the adjusted down levelling speed, flip-flop 87-88 i6 6et to the oppo~ite condition, where signal X is low and signal nX is high, via gate 85 from input LSD. This stops the ramping down of the working signal and causes it to ramp up either immediately or after a delay.
For the up direction, as illustrated in Fig. 9C, input U3 is high and EXCLUSIVE-OR gate 94 inverts signal X. For the down direction input U3 is low and gate 94 passes signal X without inversion. The purpos~ of this circuit is to make the end of the pulse on signal SRB occur at the end of the ramp-up part of the 11158~6 waveform of the working signal for up travel and to make the end of said pulse occur at the end of the ramp-down part for down travel. It is the end of the SRB pulse which determines when the working signal has the value which the speed pattern should assume and hold until the end of the succeeding pulse on SRB.
Outputs WD and WU from Fig. la are used to control analog circuits in Fig. 11 which cause the working signal to ramp down or to ramp up. Output WD is high to cause a ramp-down; output WV is high to cause a ramp-up. When WD and WU are both low, the working signal holds its previous level.
The fast ramp down when RO goes high, as shown on Figs. 9A and 9B, is o~tained by output ~D going high through gates 83 and 89. All three inputs to NAND gate 83 are high and it~ low output makes the output of gate 89 high. This condition remains until the working signal has ramped down to a value corre~ponding to up landing speed, when input LSU goes low.
If the working _ignal has a negative value when RO goes high, it is gate~ 86 and 92 which operate since input LSU is low and input LSD i8 high. This makes WU high instead of WD, to ramp the working Rignal up until it reaches equivalence with down landing speed, and input LSD goes low.
During either of the two preceding operations where the working signal is brought quickly down to a value corresponding with landing speed, signal RO and SRP are both high and thu~
the output WR of AND gate 81 is high. This causes circuits on Fig. 11 to force the working signal to change its voltage at the maxLmum rate.
While æignal SRA is high, the working signal is caused to rise by output WU going high for the up direction via gates 74 and 92. Similarly, for the down direction WD i~ made high via gates 77 and 89. Thi~ action continues until SRA goes low either because the working signal has reached equivalence with illS866 the adjusted top speed or because the pulse on signal SR has ended.
Initially, during this operation, signal nSXM is high and SRP is also high; this makes output nWN of AND gate 80 high and this causes circuits on Fig. 11 to select from one of two adjustable rates for the ramping action of the working signal.
During and after step x, nSXM is low and this makes nWN low to select the other of the two adjustable rates.
Fig. 10 also contains a circuit for controlling output nAL which is required for the purpose of indicating to other circuits, which are not part of this invention, when the target floor has advanced far enough to permit full speed. For example, in our co-pending patent application Serial No. 281,842, filed June 15, 1977, an input nAL is required. The operation of this circuit has already been partially described in connection with Fig. 9A. At the end of each pulse on signal SRA, output nALT of MM 67 goes briefly low and when it goes high again the clock input CX of clocked flip-flop 68 goes high and thus the informa-tion on its data input D, which is signal SR, is remembered internally and available in inverted form on output nQ of the flip-flop, which i named nAL.
If the target floor has advanced far enough ahead to permit full speed, signal SR will be high when flip-flop 68 is clocked, and nAL will be low to indicate that no further advances of the target floor are presently needed. If the target floor has not advanced far enough to permit full speed, signal SR will be low when flip-flop 68 i8 clocked, and nAL will be high. Other circuits such as, for example, those on Fig. 12A of our co-pending patent application Serial No. 281,842, filed June 15, 1977, will then cause the target floor to advance to a new floor unless a stop i~ required at the present target floor.
Normally, EXCLUSIVE-OR gate 65 has no important effect on flip-flop 68 since U3 and nTFB are normally both high together ~6 or both low together except briefly when starting and stopping, and then the state of nAL is of no importance. Gate 65 is provided, however, to recognize a condition which can occur if the target floor is below an upbound car, or above a downbound car. This situation can arise when the car is being operated at low speed via constant pressure pushbuttons for maintenance or inspection purposes.
It is desirable to have the target floor advanced during such constant pressure operation in a similar manner to normal operation. This allows automatic slowing down as the car approaches either terminal floor. The circuits described here are suitable for such operation. Alternatively, the system described in our co-pending patent application Serial No. 281,842, filed June 15, 1977, could be used; it causes the target floor to be that floor which is closest to the car, during constant pressure operation.
Assuminq that the target floor advances during constant pressure operation in a manner similar to normal full speed operation, and that these ad~ances occur only when the car approaches the point where the target distance ls equal to the slowdown distance xequired for the slow speed at which the car moves durlng thl~ type of operation, then it can be seen that lf the car is stopped just after a target floor advance, and is then moved in the opposite direction, the target floor may be further away than the slowdown distance, although in the wrong direction.
The circuit consisting of MM 67 and FF 68 would, without EXCLUSIVE-O~ gate 65, fail to cause target floor advancing under these conditions and the car could move farther and farther from the target floor. Gate 65, however, recognizes this condition and resets FF 68 to cause output nAL to go high to cause notching, and thus get the target floor ahead of the car in the direction of travel.

lllS866 While the car is being operated by constant pressure push~uttons for inspection or maintenance, the speed must be held to a low value, typically somewhere around 60 to 100 feet per minute. This is accomplished in Fig. 10 by MM 70 whose timing is preferabiy made adjustable by components not shown on Fig. 10.
In normal operation the output of MM 70 is not used since input nIN is low, and is connected to all gates, 72, 75 and 76, whlch use the output of MM 70 or the inverted output of inverter 71. When nIN is high, however, signal SRA has an additional factor via the output of gate 72 which causes signal SRA to cease even earlier than before, when the working signal has reached a value corresponding with the desired low speed. In-creasing or decreasing the timing of MM 70 causes an increase or decrease in this speed.
If the car approaches a terminal floor, the decreasing target distance causes the length of the pulses on signal SR to reduce, and thereby slow down the car automatically. If the car is operated during construction before the equipment for measuring target distance is installed, a circuit involving inverter 66 and gates 69, 75 and 76 causes the working signal to operate in a similar manner to landing speed, as shown on Fig. 9C, but with the working signal ramping up or down for a longer period, deter-mined by MM 70, in the direction of travel.
This causes the speed to be adjustable by MM 70, as before, but without the automatic reduction in speed during the approach to a terminal floor.
The purpose of gate 90 is to make the flat top to the waveform of the working signal as shown in Fig. 9C for the up direction. Gate 91 causes a similar flat bottom to the waveform for the down direetion which is not illustrated. This applies only to landing speed or constant pressure operation without target distance measurement, as determined by MM 82. The flat top is needed to provide sufficient time to bring the speed pattern into agreement with the appropriate sampling point of the waveform of the working signal before it ramps up or down to the other level.
Fig. 11 contains an integrator consisting of operational amplifier 102, capacitor Cl and optional resistor R43 which, if suitably sized, may reduce the zero-offset error of the operational amplifier. The output of this integrator is the working signal WS.
Three resistors R31, R32, and R33 can be selected by analog switches 107, 108 and 109 to supply current to the integrator. Two analog switches 106 and 110 control the polarity of the current. These analog switches must be capa~le of handling both polarities; a negative supply line M10 is shown to provide minus 10 volts with respect to grounded supply line OV, while the P10 line provides plus 1~ volt~ with respect to ground. The logic inputs to these analog switches operate between ground ~OV~
and plus 10 volts only.
Devices are available which permit positive and negative switching of analog signals under the command of logic inputs which are referenced to ground, but they may be limited to plus 5 volts on the logic inputs. The LF13202 device manu~actured by National Semiconductor Corp., for example, has this limitat$on but is otherwi~e suitable. Therefore resistors R26 through R30 are provided on Fig. 11 to limit the current on the inputs to the analog switches to a value in agreement with the manu~acturer's specifications. If other deviGes are used which do not have this limitation, these resi~tors can ~e eliminated.
Also, the aforementioned device has limitations on hDw close the analog signal can go to the plus and minus power supp~y voltages. For this reason, voltage dividers are used, such as R37 and R38, to assure that the analog voltages cannot go too close to the power supply lines P10 and M10. These voltage dividers serve an additional purpose in limiting the current ~ 15866 through the analog switches if switches 106 and 110 are falsely turned on at the same time.
When signal WD goes high, it causes analog switch 106 to turn on to flow current into the integrator; this makes its output WS ramp down. Similarly, when signal WV goes high, analog switch 110 turns on to flow current out of the integrator which causes its output WS to ramp up.
If signal WR is hi~h, analog switch 107 selects resistor R31 whose resistance is low er.ough to obtain the desired fast ramping of the output of the integrator. If signal WR is low, signal nWN selects either resistor R32, via gate 104 and analog ~witch 108 (which occurs when nWN is high) or resistor R33 via inverter 103 and analog switch lQ9.
When analog switch lQ8 is turned on, the current to the integrator is adju~table by potentiometer R34 which affects roundin~-off at the end of a stopping sequence. When analog switch 109 is turned on, the current into the integrator is adjustable by potentiometer R35 which determines the deceleration rate and the corresponding slowdown distance, as previously described.
Input LLM to Fig. 11 is normally low, and transistor Ql i~ turned off. If desired, transistor Ql can be turned on by making input LLM high, to connect resistor R41 in parallel with resistor R38 to thereby select a lower deceleration rate for the down direction only. A correspondingly greater slowdown distance automatically occurs. This might ~e used when the car is loaded heavily, where the gentler slowdown reduces the currents handled by the motor and M.G. set.
An operational amplifier 99 is used to control signal TS.
When the working signal WS rises to a value more positive than equivalence with top ~peed up, current flows from WS through diode D5 and forces the R+" input o$ the operational amplifier 99 to have a more positive voltage thAn its n _ n input. It~ output then goes up close to the P10 supply line, and output TS goes high. The voltage at which this occurs is adjustable by poten-tiometer R12, since diode D4 is reverse biased at this time and thus is not conducting.
Similarly, when the voltage on WS goes down to a value more negative than equivalence with top speed down, the n - " input of the operational amplifier 99 is made more negative, via diode D4, than the I+~ input and again the output switches up close to the P10 supply line. The voltage at which this occurs is adjustable by potentiometer R15, since diode D5 is not conducting at this time.
Thus, the top speed is separately adjustahle for up and down by potentiometers R12 and R15, respectively. Separate adju~tments are shown in Fig. 11 for up and down because it is easier to do so; normally they would be set to equal values.
However, if there were any advantage in making them unequal, it could be done, and the slowdown distances would automatically be suitably different for the two directions.
When the voltage on WS i8 between these two adjustable settings, current flows from P10 through resistor Rll, part of potentiometer R12, diodes D4 and D5, part of potentiometer R15, and through resistor R16 to M10. This makes the two input~ of the operational amplifier 99 separated in voltage by two diode drops, or about 1.5 volts, in a polarity which makes its output switch down close to the M10 level. Resistor R17 limits the current into gate 78 on Fig. 10 which, if of the CMOS type, has internal diode protection. Otherwise, gate 78 might be damaged by excessive current.
Two comparators, 100 and 101 are used to control signals LSU and LSD. Comparator 100 has its n_n input connected to potentiometer Rl9 which is part of a voltage divider between supply lines P10 and OV. When the working signal becomes more positive than the reference determined by potentiometer Rl9, the ~utput of co~parator 100 goes high to make signal LSU high.
Resistor R18 has a much higher resistance than potentiometer Rl9, so that a relatively small positive voltage on WS causes LSU to go high.
Similarly, signal LSD does high if the working signal goes more negative than the slightly negative value adjustable by potentiometer R20. Thus potentiometer Rl9 adjusts the up landing speed and potentiometer R20 adjusts the down landing speed.
It is assumed that comparators 100 and 101 (and similarly comparators 116 and 117 on Fig. 12~ are of type similar to quad comparator LM2901 manufactured by National Semiconductor Corp.
This type has active pull-down ,(to minus 10 volts in Figs. 11 and 12) but no internal pull-up. Thus resistors R22 and R24 are supplied to act as pull-up resistors. Resistors R23 and R25 are also supplied to limit the current in gates 83, 84, 85 and 86 on Fig. 10 when a comparator switches its output down to minus 10 volts.
In Fig. 12, an integrator consisting of operational amplifier 114, capacitor C2 and resistors R51 and ~52, is used to control the speed pattern SP. The capacitor C2 has much higher capacitance than capacitor Cl in Fig. 11 because the speed pattern operates in real time, and must not change too rapidly, while the working signal must operate at high speed.
The speed pattern SP i8 caused to ramp down when signal BU goes high, and to ramp up when BD goes high. Th~ polarity of SP is negative for up and positive for down; this is opposite to the working signal.
~ When signal BU is high, analog switch lllis turned on, and this applies a positive voltage to the integrator to make its output ramp down to more negative values. When signal BD is high, analog switch 112 is turned on, and this applies a negative voltage to the integrator to make its output ramp up to more 11~5866 positive values.
Two rates of ramp on the output SP of the integrator are possible. When analog switch 113 is turned on, the maximum rate, determined by resistor R51, is obtained; when analog switch 113 is turned off, an additional resistor R50 is inserted to give a lower, adjustable rate to the ramp on signal SP. Signal BS, which will be descri~ed later, controls analog switch 113.
An operational amplifier 115 is used as a comparator to continuously compare the rapidly changing working signal WS
with the more 810wly changing speed pattern SP. Resistors R53 and R54 can have any suitable ratio of resistances, but for the purpose of this description they are assumed to be equal. Then, whenever the working signal is hlgher Ci.e. more positive~ than what the speed pattern would be if of opposite polarity, output nWH of operational amplifier 115 i5 low. Diode D8 clamps the output to a small negative value, with respect to grounded supply line OV, rather than permit nW~ to ~ecome close to minus la volts, which might damage gates 118, 122, 124 and 125. Similarly whenever the working signal is lower Ci.e. more negative~ than what the speed pattern would ~e if of opposite polarity, output nW~ i8 high and diode D8 has no effect. The ~peed pattern is opposite in polarity to the working signal for these comparisons.
The circuit for causin~ the speed pattern SP to become more negative to match the sampling point of the working signal, for the up direction, includes inverter 118, NAND gates 119 and 120, and analog switch 111. If, during the time signal SRB
~which is inverted by inverter 98 from signal nSRB~ is high, the working signal WS becomes more positive than the speed pattern, signal nWH goes low and the output WH of inverter 118 goes high.
Thi~ causes the output of gate 119 to go low and this causes the output BU of gate 1~0 to go high to make analog switch 111 turn on.
This makes the speed pattern ramp down at the lower rate, adjustable by resistor R50, to a more negative value which dictates a higher speed in the up direction. Typically the rate at which the speed pattern changes is such that it does not reach complete agreement with the sa~pling point of the working signal before the pulse ends on signal SRB. Thus the same process is required again on subsequent calculations and as the speed pattern approaches agreement with the sampling point of the working signal, smaller times are available for causing the speed pattern to ramp down because signal nWH goes low later in the rise of the working signal, nearer to the end of the pulse on signal SRB. This causes the speed pattern to approach its final value exponentially and thus to round off the speed pattern as it approaches top speed or inspection speed, or as it rounds off during a single floor run. This effect can be varied via adjustable resistor R50.
A further restriction on the amount of change on the speed pattern SP for the up direction is obtained from a comparator 116. A voltage divider consi~ting of resistors R56 and R57 establishes a voltage reference on the "~'~ input of the comparator.
This is compared with an adjustable portion of ths voltage obtained from input AP. Resistor R60 adjusts this portion.
Input AP is derived from the speed control system; it ~hould be proportional to the rate of change of generator voltage, if an M.G. set is used, or the acceleration of the vehicle if a suitable method is available for making such measurement. If a speed control system based on the aforementioned ~.S. patent 3,706,017 is u~ed, the voltage E of Figs. 1 and 9 of this patent i8 precisely what is required to be applied to input AP
of Fig. 12 of the present patent.
If the voltage on i~put AP rises above a level adjustable by resistor R60, the output of comparator 116 goes low, toward the minus 10 volt supply and resistors R64 and R65 attenuate this to give a logic low signal, near the grounded supply line OV, - ~2 -i~6s to one input of gate 119. This terminates any ramping down of the speed pattern SP.
During upward acceleration, the action of this circuit is to shorten the period in which the speed pattern ramps down slightly at each calculation, to achieve a substantially constant acceleration. Near the end of the acceleration, exponential rounding takes over to limit the repetitive changes to the speed pattern.
This control of the upward acceleration rate, achieved by the circuits of Fig. 12, i8 an alternative to performing similar operations in the speed control system itself. The advantage of doing it as part of t~e solid state selector is that it permits a reduction in up acceleration under the same load conditions as were used to limit the down slow~own, as described for transistor Ql in Fig. 11.
Transistor Q2, in F;g. 12 is assumed to be turned of~ for normal operation, as just descri~ed. When lifting heavy loads !
however, transistor Q2 can be turned on by signal LLM to thereby connect resistor R62, which could ~e adjustable, in parallel with resi~tor R57. This lower~ the reference voltage for comparator 116, and automatically causes a lesger rate of acceleration in the up direction only.
For the down direction, inverter 129 and NAND gates 124 and 123 are used to control the ramping up of the speed pattern to more positive values, to dictate higher down speed. Now, signal nWH is used instead of WH because it is only when the descending working signal has gone more negative than the speed pattern that signal BD should go high to make SP go more positive.
Inverter 129 has its output nU3 high for the down direction, and also when zero speed i8 required; signal U3 is low in ~oth cases.
Comparator 117 operate~ for the down direction in a manner similar to what comparator 116 does for the up direction.
Its reference voltage is determined by a voltage divider consisting l~lSB66 of resistors R58 and R59. It is assumed that the voltage on input AP is positive for up acceleration and negative ~or down acceleration.
The preceding description applies to the increasing of the speed pattern (i.e. away from zero~ which occurs when the speed pattern is lower than the sampling point of the ~orking signal. The circuit for performing the opposite operation, in which the ~peed pattern i6 lowered, involves EXCL~SIVE-OR
gate 125, clocked flip-flop 126, MM 127, and NAND gates 121 and 122 in addition to gates 12~ and 123 which are used for both operations.
Basically, flip-flop 126 is clocked at the end of each pulse on signal SRB, when nSR~ goes ~igh. Since the data input D of flip-flop 126 is connected to the positive supply line P10, the output Q of flip-flop 126 goes high at the instant of clocking to make signal BS high. However, the reset input R, if high at the instant of clocking, will prevent this from happening and BS will remain low.
Gate 125 is arranged to recognize when it is necessary to lower the voltage on the speed pattern SP. If, at the instant of clocking flip-flop 126, the working signal does not exceed the speed pattern, signal nWH is high for the up direction when signal U3 is high or signal nWH i~ low for the down direction when signal U3 is low. Since gate 125 performs the EXCLUSIVE-OR
function, either of these conditions causes a low output which permits signal BS to go high when flip-flop 126 is clocked. When the working signal does not exceed the speed pattern, it is an indication that the speed pattern should be lowered.
The polarity of ~ignal nWH indicates which way the speed pattern must go in order to get into agreement with the working signal which remains for a time at the desired level as illus-trated in Figs. 9A, 9B and ~C. If nWH is high, gate 122 has a low output to make BD high; if nWH is low, WH is high and gate 121 makes BU high. In either case, this causes the speed pattern to ramp in the correct direction to reach agreement with the working signal.
Analog switch 113 is turned on when signal BS is high to make the speed pattern move at a considerably greater rate, because resistor ~50 is now shorted out, compared with the previously described operation of raising the speed pattern. It is desirable to permit a considerahly greater rate here because during slowdown the speed pattern should generally follow the dictation of target distance only, with no possibility of built-in time delays preventing the proper reductions in speed which are required to stop the vehicle correctly with no over-shooting.
Generally, it is expected that at each calculation, the speed pattern will have sufficient time to come completely into agree-ment with the working signal.
~owever, the time available for bringing the speed pattern into agreement with the working signal varies from a minimum at high speed to a maximum at low speed. If the rate at which that speed pattern can move is made high enough for full speed by the choice of resistor R51 and capacitor C2, a failure of the target floor calculating system, or a failure of the square root calculating system, could cause excessive deceleration as the ~peed pattern approaches low values.
Therefore, a time limit is placed on this operation of lowering the speed pattern; MM127 establishes a time window in which such lowering can occur. At the samR time that flip-flop is clocked, MM 127 is triggered and its output FT becomes high for a suitable length of time. At the end of that time, signal FT goes low and causes both signals BU and BD to be low via gates 120, 121, 122 and 123.
Signal FT is inverted by inverter 128 to create signal nFT which is required on Fig. 10 to hold the flat top to the waveform of the working signal, as shown on Fig. 9C, for landing lllS866 speed and also for inspection speed.
The circuits so far described do nothing about pulling the speed pattern down completely to zero, as required to finally bring the vehicle to a stop. Instead, the speed pattern is, for convenience, allowed to hold the value corresponding to down landing speed, when zero ~peed is required. This is preferable to allowing it to drift to an unpredictable level which might interfere with the next trip. The speed control system is assumed to have an independent means for regulating the speed to zero when required. This can be most simply achieved by disconnecting, with a contact or an analog switch, the SP input to the speed regulator.
Fig. 13 shows some miscellaneous circuits which may be required to use the system of this invention to control a typical elevator. Some communi.cation is required between the circuits which.are disclosed in this and our co-pending patent application Serial No. 281,842 filed June 15, 1977, and the rest of the elevator circuits. The interface circuits have been kept to a minimum in thi6 disclosure, and consist of inputoe GUl, GDl, BK and nIN with a further optional load measuring switch LS, all of which are shown in Fig. 13.
Input nIN is assumed to be low for normal full-speed operation, and high.when special low-speed constant pressure operation i8 required for inspection or maintenance.
Input GUl is assumed to be high only when the doors are closed in preparation for an up run, and during the up run until the target floor coincides with a floor where a stop i8 required.
Input GDl is similar to GUl, ~ut for the down direction.
Inputs GUl and GDl are both low when it is required that the vehicle slow down and stop at the present target floor.
The preceding description of inputs GUl and GDl applies to normal full-speed operation, when input nIN is low. When nIN
is high, inputs GUl and GDl, wh n both low, cause a brake stop 1~5866 from the low speed associated with inspection and maintenance.
Either GUl or GDl can be made high to cause the car to start and run at low speed in the selected direction, on constant-pressure operation.
Input BK is assumed to be high either when the brake is actually lifted, or when the brake is about to be energized and all of the usual safety circuits are closed to permit the car to move, even though the brake has not yet been energized. The main purpose of input BK is to assure that, in normal operation, when an emergency stop is made, the speed pattern starts out from zero when the car 6tarts again, rather than having the speed pattern remain at a value determined by the target distance.
Outputs GU and GD are generally equivalent to inputs GUl and GDl, and are intended to ~e connected to similarly named inputs on Fig. 12A of our co-pending patent application Serial No. 281,842 filed June lS, 1977. However, gates 131 and 132 can be used to hold GU or GD low, in spite of GU1 and GDl, when the target floor coincides with the highest or lowest floor served.
If the target floor coincides with the highest floor served, inputs XT and YT are assumed to be both high to make the output of gate 131 low, and to thus make output GU low. Input XT
can be connected to any one of outputs Xl throu~h X8 of Fig. 12B
of our co-pending patent application Serial No. 281,842 filed June 15, 1977, and input XT can be connected to any one of outputs Yl through Y5 to achieve this.
Similarly, inputs XB and YB can be arranged to be both high when the target floor coincides with the lowest floor served. This as~ures that in spite of failures in other circuits, the target floor cannot advance above the highest floor served or below the lowest floor served.
Diodes D9 and D10 may be used to lower signal nACP for redundancy at terminal floors, in addition to the cards of Fig. 13 in our co-pending patent application Serial No. 281,842 filed June 15, 1977 as used a~ terminal floors. This permits these two cards to be removed without losing the ability to stop the target floor at terminal floors; otherwise, with said cards out, the target floor might step into a state representing a floor not served by this elevator.
Outputs U2 and D2 are intended to be amplified by relay drivers to control the usual up and down relays or contactors which are required by most elevators. Contacts on these relays might, for example, energize the brake and strengthen the motor field. These relays, of course, would be also controlled by the various safety circuits to prevent operation when it is not safe.
Alternati~ely, outputs U2 and D2 could enter into solid state circuits.
When input nIN is high, for low-speed constant-pressure operation, gates 135 and 144 are active to control gates 136 and 143, and outputs U2 and D2 are equivalent to inputs GUl and GDl respectively.
When input nIN is low for normal operation, the output of inverter 134 is high and gates 137 and 142 take over the control of gates 136 and 143. Now, inputs GUl and GDl have no direct control over outputs U2 and D2. Instead, outputs GU and GD determine the direction in which the target ~loor advances, and this influences input nTFB which comes from Fig. 14B in our co-pending patent application Serial No. 281,842 ~iled June 15, 1977.
When the tar~et floor is below the car, input nTF~ i8 low and the output of inverter 141 is high. This permits gate 142 to lower one input of gate 143 to make its output D2 high.
~hus the down direction i~ called ~or when the target floor i~
below.
When the car is level at the target floor, signal nLV, from Fig. 6 i8 low and this forces both outputs U2 and D2 to be low. When the target ~loor is above the car, inputs nTFB and i~lS866 nLV are both high and the output of gate 137 goes low to make output U7 high.
It is very important that outputs U2 and D2 be controlled from nTFB, rather than from inputs GUl and GDl, during normal operation. Inputs GUl and GDl are low during slowdown, but U2 or D2 must remain high until the car stops. Also, the use of inputs nLV and TFB causes outpus U2 and D2 to control relevelling operations if required due to, for example, cable stretch as the load moves onto or off of the car.
Outputs U3 and D3, which are required on Figs. 10 and 12, are obtained from AND gates 138 and 140 so t~at they can be high only when input BK is ~igh. When either U3 or D3 is high, the output of OR gate 139 is high. Thus wheneve~ the car starts, flip-flop 145 is clocked and the state of its input D is remem~ered until the next clocking.
A load switch LS can be provided, if desired, it is assumed to be closed whenever the load in the car exceeds some predetermined amount, such as 80% of rated load, for example.
If, when the car starta, load switch LS is closed, output LLM
goes high and remains high regardless of further opening or closing of switch LS during acceleration and deceleration, until the next 6tart. If, when the car starts, load switch LS is open, output LLM i~ low and remains low regardless of any closing of switch LS
during acceleration or deceleration. As previously described in connection with Figs. 11 and 12, signal LLM, when high, causes a longer slowdown distance to automatically occur for the down direction, with a corresponding reduction in the deceleration rate, and also a reduction in the up acceleration rate.
This optional feature is often desirable on installations where the equipment is not quite capable of top performance at full load, where the motor current is at a maximum during up acceleration and down deceleration. With conventional mechanical selectors it would be quite difficult and expensive to obtain ~5866 this feature. The method which is frequently used with conventional selectors is to not allow the car to operate in either direction if loaded beyond its rated capacity, or to permit it to run at reduced speed in both directions, if loaded beyond an adjustable amount.
The above-disclosed embodiments have been described generally for systems in which a motor-generator set is used to control the speed of an elevator via a direct current motor.
The invention i8 in no way limited to such speed control systems, however. The calculation of the square root of target distance, and the production of an analog speed pattern based thereon is basic to any type of speed control, including s~stems employing SCR's in place of an M.G. set or a variable frequency a.c. motor control. The concepts di~closed herein also apply to any auto-matic transport system, not just to elevators.
Thus there is pro~ided, in accordance with the present invention, a novel and improved method and apparatus for produ-cing a speed pattern signal for an elevator car or similar vehicle which is movable between a plurality of stopping positions.
The system operates to repeatedly calculate the square root of the target distance TD b~tween the pre~ent position of the vehicle and a selected ~topping position. The result of each calculation i8 a dîgital pulse having a pul~e-width substantially proportional to the number of calculating steps. This pulse may be utilized to energize conventional motor speed relays, to obtain relay operation similar to that of a mechanical selector in an elevator system. In a preferred embodiment of the invention, however, the pul~e is converted to a steady analog Yoltage which may be applied a~ a speed pattern to an analog 6peed regulating system of th~ vehicle. Conveniently, the digital-to-analog conversion apparatus provides one adjusting device, such a~ a potentiometer, for simultaneouæly adjusting the deceleration rate of the vehicle and its ~lowdown distance, and another

- 5~ -illS866 adjusting device, such as potentiometer, for adjusting the round-off deceleration and distance at the end of the stopping sequence.
It will be understood that the above-described embodiments are merely exemplary and that persons skilled in the art may make many variations and modifications thereto without departing from the spirit and scope of the present invention. All such variations and modifications are intended to be included within the scope of the invention as defined in the appended claims.

Claims (17)

THE EMBODIMENTS OF THE INVENTION IN WHICH AN EXCLUSIVE
PROPERTY OR PRIVILEGE IS CLAIMED ARE DEFINED AS FOLLOWS:

1. Apparatus for controlling the speed of a vehicle movable between a plurality of stopping positions, said apparatus comprising, in combination:
(a) means for generating a digital pulse (SR) having a pulse-width substantially proportional to the speed pattern of the vehicle;
(b) first signal source means for producing a constant first signal of adjustable magnitude;
(c) first switch means, connected to said first signal source means and responsive to said pulse generating means, for passing said first signal during at least a portion of the duration of said digital pulse;
(d) signal integrator means, connected to said first switch means, for producing a working signal proportional to the integral of said first signal;
(e) sample and hold means, connected to said signal integrator means, for producing an output signal proportional to the maximum value attained by said working signal during the continuance of said digital pulse; and (f) means for controlling the vehicle speed in accordance with said output signal.

2. The apparatus defined in claim 1, wherein said vehicle is an elevator car and said stopping positions are floors.

3. The apparatus defined in claim 1, wherein said pulse generating means include means for repetitively generating said digital pulse, whereby said signal integrator means repetitively integrates said first signal.

4. The apparatus defined in claim 1, wherein said first signal source means includes a potentiometer for adjusting said constant first signal.

5. The apparatus defined in claim 1, wherein said first signal source means includes a current source and said constant first signal is a constant first current.

6. The apparatus defined in claim 1, further comprising second signal source means for producing a constant second signal of adjustable magnitude; and second switch means, connec-ted between said second signal source means and said signal integrator means and responsive to said pulse generating means, for passing said second signal to said signal integrator means during a portion of the duration of said digital pulse.

7. The apparatus defined in claim 6, wherein said first switch means passes said first signal during a first portion and said second switch means passes said second signal during a second portion of the duration of said digital pulse, and wherein said first and second portions of said pulse are mutually exclusive.

8. The apparatus defined in claim 7, wherein said second portion precedes said first portion in time.

9. The apparatus defined in claim 6, wherein said second signal source means includes a potentiometer for adjusting said constant second signal.

10. The apparatus defined in claim 6, wherein said second signal source means includes a current source and said constant second signal is a constant second current.

11. The apparatus defined in claim 1, further comprising means, connected to said signal integrator means and responsive to said pulse generating means, for setting said working signal to a prescribed value proportional to the landing speed of the vehicle at the commencement of said digital pulse.

12. The apparatus defined in claim 11, wherein said prescribed value is adjustable.

13. The apparatus defined in claim 1, wherein said vehicle is movable in opposite directions between said stopping positions;
and wherein said first signal is positive for one direction and negative for the opposite direction of motion.

14. The apparatus defined in claim 6, wherein said vehicle is movable in opposite directions between said stopping positions;
and wherein said first and second signals are positive for one direction and negative for the opposite direction of motion.

15. The apparatus defined in claim 11, wherein said vehicle is movable in opposite directions between said stopping positions;
and wherein said working signal setting means includes means for setting said working signal to a positive prescribed value proportional to the landing speed for one direction and to a negative prescribed value proportional to the landing speed for the opposite direction of motion.

16. The apparatus defined in claim 15, wherein said positive prescribed value and said negative prescribed value are independently adjustable.

17. The apparatus defined in claim 15, wherein said working signal setting means includes means for causing said working signal to oscillate between said positive and negative prescribed values when said vehicle is at a standstill at a stopping position.