FR2591752A1 - SPEED AND PHASE MEASUREMENT SYSTEM FOR AIRCRAFT HELICOS - Google Patents

Abstract

Techniques for measuring the instantaneous speed and phase of propellers are disclosed. The system of the invention, applied to contra-rotating propellers 15, 18 comprises toothed wheels 30, 27 coupled to the respective propellers, the teeth of which generate signals as they pass in front of sensors 36, 33. The speed of rotation of each propeller on the basis of the time intervals between successive signals and it is possible to perform the speed calculation several times during a single revolution of a propeller. Error coefficients are calculated in order to correct defects in the uniformity over time of the signals obtained, resulting from a possible lack of uniformity in the spacing of the teeth. Application to turboprop engines with contra-rotating propellers. (CF DRAWING IN BOPI)

Description

The present invention relates to sensors for

  rotational speed for aircraft propellers, and it relates more particularly to a system that measures the speed

  rotation of the two propellers in a pair of contra-rotating propellers

  presses. The system also measures the phase relationship

between the propellers.

  The speed of rotation of a

  aircraft propeller 3 (figure 4) while climbing on the shaft of the

  a toothed wheel 6 (which will be called a "detection wheel")

  Each tooth (or marker) 9 produces a signal in a magnetic pick-up coil 12 as the tooth passes the coil, and additional circuits (not shown) process the signals.

  following way to measure the speed of rotation.

  If the toothed wheel 6 has eight teeth, the circuit measures the frequency with which each tooth passes in front of the tooth.

  coil, and he deduces from this frequency the rotation speed

  tion. For example, if the frequency is 160 Hz (ie

  say if it passes 160 teeth per second), and if there are eight teeth per revolution, the rotation speed deduced is 1200

  rpm (that is 1200 = 60 x 160/3).

  A disadvantage of this technique is that it provides only an average speed over the time interval necessary for the passage of several teeth in front of the coil. Accelerations and decelerations of the helix during this interval are not detected. In addition, this

  technique does not provide information on the posi-

  instantaneous movements of the propeller blades. For exemple,

Z591752

  it may be desirable to know the precise moment at which the blade on the propeller was at the "1 o'clock position

minutes".

  The object of the invention is to provide a new and improved speed measurement system for aircraft propellers.

  Another object of the invention is to provide a

  velocity measurement system for a pair of aircraft propellers

  contrarotative which measures the speeds of each propeller.

  A further object of the invention is to provide real-time data concerning rotational positions

  snapshots of aircraft propellers.

  The rest of the description refers to the drawings

  which respectively represent: FIG. 1: a form of the invention associated with a

pair of counter-rotating propellers.

  Figure 2: a schematic representation of a

pair of counter-rotating propellers.

  Figure 3: two arches 125 and 135 which are also

shown in Figure 2.

  Figure 4: a sensor of the prior art for measuring

sure of the speed of a propeller.

  Figures 5 and 5A: changes in the posi-

  the teeth 45 of the wheel 30 of FIG.

  FIG. 6: a sequence of pulses that the counter 57 of FIG. 1 produces, as well as pulses of

  loadingindicated in Figure 1.

  Figures 7 and 8: sequential diagrams that are used for the explanation of the asynchronous operation of the

  counter 57 and the microprocessor 66 of FIG.

  Figures 9 and 10: graphical recordings of simulations that compare the operation of a form of the invention using error coefficients in the calculation

  speed, with another form that does not use it.

  Figures 11 and 12: two schematic configurations

teeth 45 in FIG.

  In one form of the invention, a clock recorded

  be the actual instants at which propeller blades pass through a reference point. From these real moments, one can calculate almost instantaneously the speed of rotation

of the propeller.

  FIG. 1 represents a pair of air propellers

  nave 15 and 18. These propellers rotate in opposite directions

  tell me arrows 21 and 24, so they are called counter-rotating propellers. Detection wheels 27 and 30 are attached to the propellers and also rotate in opposite directions. Magnetic pickup coils 33 and 36, known in the art, produce on load lines 39

  and 42 signals (here called load signals-

  45). The Fan Speed Sensor, reference 726452, can be used for such a coil.

  marketed by Electro Corp., Sarasota, Florida.

  Loading lines 39 and 42 are connected

  to the load inputs of latch networks 48 and 51.

  The charging signals thus cause the latch networks 48 and 51 to load data present on the data bus 54. The data bus 54 carries the output signal of a 16-bit counter 57. The 16-bit counter It counts from the binary number zero to the binary number 216-1 (which is commonly called 64 K and which corresponds to the decimal number 65535), at the rate of 2 MHz, and this counter is used as a clock. Thus, the counter 57 changes state 2 million times per second, in sequence, from the decimal number 0 to the decimal number 65535, and it starts again.

  then to zero (it starts over again) and continues.

  The signals of the sbrties 60 and 63 of the networks of

  latches 48 and 51 are appended to a microprocessor 66 de-

  signed by the symbol pP. The data bus 54 also applies

  information to the microprocessor 66. Thus, the two latch networks 48 and 51 and the microprocessor 66 receive input signals from the counter 57, and

  they have access to a real time signal. The microproces-

  66 is programmed in accordance with the described flowchart

  the eight steps shown in Table 1 below.

  after. The table is followed by a detailed description of

  each step. The reader is now invited to

  this description, which follows the Table step by step

1.

TABLE 1

  1. Calculation of the time required for a complete turn.

  Time (full revolution) = AT (1) + T (2) + AT (3) + t T (4) + tT (5) + AT (6) + AT (7) + AT (8) 2. Calculation of error coefficient for the tooth opposite the

  common tooth. m and n are indexes.

  m = (n + 4) modulo 8 error (m) = AT (m) time (full revolution) / 8 3. In a sub-speed condition (less than 340.9 rpm), the coefficients are reset has a. They will gradually converge to their correct values when the under-speed condition ends. A sub-speed condition exists when a latch network does not change state for eight consecutive readings. (We call

  this the "rule of the eight executions").

4. Calculation of the speed.

  Speed ERROR (n) x (60 s / min) x (2,000,000 increments / sec) 2 ViesAT (n) 8 teeth / turn

  5. Selection of good sensors. (A1 and A2 designate two cap-

  mounted on a propeller. B1 and B2 designate two

  mounted on the other helix.) 5.1 IF the absolute value of (sensor A1 - sensor A2) is less than; -n orend: speed = (A1 sensor + A2 sensor) and restores indicators to indicate that the two sensors of the front propeller are good; OTHERWISE, the following operations are performed: 5.2 IF both sensor B1 and sensor B2 are good: 5.2. 1 If the absolute value of (sensor A1 - rear propeller speed) is less than the value

  absolute of (sensor A2 - U ?: se of the propeller ar-

  the value A1 for the speed of the forward propeller and we set the indicators that the sensor before A1 is good and that

the front sensor A2 is bad.

  5.2.2 ELSE we take for the speed of the front propeller

  the value indicated by sensor A2 and we

  The number of testers indicating that the sensor

  A1 is bad and the A2 sensor is good.

  5.3 If neither sensor B1 nor sensor B2 is good,

  take the one that indicates the lower value.

  5.4 Repeat 5.1-5.3 for the other propeller, replacing

  A1 by B1, A2 by B2, B1 by A1 and B2 by A2.

  6. Control to determine if sensors provide

  no value when the engine is running.

  IF the speed of the turbine is greater than 10,000 rpm AND the absolute value of (not before - not before programmed) is less than 3.0 degrees AND the absolute value of (step back - step back programmed) is less than 3.0 degrees, the following actions are performed: ú

  IF (front sensor A z35D rpm), the indi-

  indicating that the front sensor A is de-

defective.

  SI (front sensor B f350 rpm), the indi-

  This indicates that the front sensor B is defective.

  IF (rear sensor at 4350 rpm), the in-

  This indicator indicates that the rear sensor A is defective.

  IF (rear sensor B e35D rpm), the in-

  This indicator indicates that the rear sensor B is defective. 7. Calculation of the phase (instant of loading) loading moment} angle I of the front flip-flops J of the rear flip-flops J of phase = x45 (instant of loading last instant of

  flip-flops before - loading

  The flowchart is written on the assumption that each propeller 15 and 18 in the figure has eight blades, correspondingly eight teeth on each

  detection wheel 27 and 30. However, to facilitate the illus-

  each propeller is shown with only four blades

  LEMENT. Step 1 is a summation by which the total duration of a turn of a helix is calculated. We perform this calculation

  for each propeller. The calculation is done in the following way

  boasts. As indicated above, when a tooth 45 passes

  the detector coil 36, the signal appearing on the line 42 causes the lattice 51 to be loaded with the number present at the moment on the bus 54. In fact,

  the network of latches 51 is loaded with the exact time to the-

  the tooth 45 is passed in front of the detector coil 36.

  The fact that the counter 54 counts from zero up to 64 K and then starts again at zero does not affect this

  principle, as will be explained later.

  quently. In addition, we will explain in connection with the step

  2 the exact definition of what is meant by "passage"

the detector coil 36.

  The microprocessor 66 continuously reads the

  n of each of the latch networks 48 and 51 and places the real time data in a Random Access Network (RAM) 70. A sub-network of the MEV is indicated by four boxes 73

  for blades 1-4 of propeller 15, and a similar subsystem

  re 75 is indicated for the helix 18.

  The boxes in sub-networks 73 and 75 represent

  In fact, they are memory positions in MEV. Each box corresponds to a propeller blade. The usual sequence of operation would be: a tooth passes, which changes

  the number contained in the latch network 51o The micropro-

  the processor 66 reads the contents of the latch network 51 and

  register in the SRM 77, in the subnet 73, the number he just read. A next tooth passes before] abdim 36, oe

  which again changes the number in the latch network 51.

  The microprocessor 66 reads again the contents of the network of flip-flops 51 and then records in another MEV 79 the number just read, and so on, which leads

  the recording of the actual moments of appearance of

  loading machines. This amounts to recording the actual moments of the passage of the teeth, which is equivalent to recording the actual instants at which blades pass at a predetermined point, such as a point 82. This last statement

  is true because the relative geometrical configurations

  of the propeller 15 and the gearwheel 30 are known to

  in advance, according to the construction of the propeller system.

  As an example, for the clock rate of 2 MHz indicated above, for an eight-tooth wheel and for

  a constant propeller speed of 1200 rpm, at a given moment

  the numbers contained in the MEV for the propeller 15

  would be those indicated, that is "t = 9000", etc. The reading

  Note that all successive numbers differ by 12500, which is the number of counter increments during the 0.00625 second interval between tooth passes.

  The microprocessor 66 also records from the same

  in the subnet 75 of the data relating to the other helix 18. The speed of execution of the microprocessor 66 (ie for example 1 million assembler operations per second) is so much greater than the rate of the loading signals which change the data on the data bus 54 (ie for example 160 changes per second for an 8-tooth wheel running at 1200 rpm), that the microprocessor has no difficulty in reading and recording the contents of the two data networks. flip-flops between loading operations of the flip-flops. The actual imps information regarding blade passes at specific points, which is recorded in

  the MEV 70, allows the microprocessor 66 to calculate the inter-

  time intervals (AT) between successive passes of the blades at specific points. The interval is the difference between the times of real time that are recorded for two

  successive blades, as indicated by the AT symbol on the

  1, near boxes 77 and 79. In this example, T is

equal to 12500.

  At T (1) designates the time interval (AT) between the passage of the tooth n 8 and the passage of the tooth n 1. AT (2) designates the time interval between the passages of the tooth nl and of tooth # 2, and so on. So, step n 1

  calculates the total time interval for a single round of

than pale.

  Step n 2 calculates an error coefficient. We

  explain a reason for using the erosion coefficient

  with reference to Figure 5. During manufacture

  gear wheels 6 of FIGS. 1 and 5, it is almost inevitably

  that a tooth 9A in FIG. 5 is not exactly placed

  at its intended position, and this tooth may (1) be moved to position 85 indicated in dashed lines, or (2)

  to be oversized, as indicated by the lines in

  88, or (3) be undersized, as indicated by the dotted lines 89. In these three cases, the faces 90 may be offset from the predicted positions 93,

  with an offset of up to 0.1, indicated by

  e8. Thus, the signals produced by the detector coil 12 will in fact appear at different times from those

  that would be obtained if the faces 90 were in their posi-

  foreseen. As a result, the time intervals

  between the teeth carrying the faces 90 (indicated by AT (1) and tT (2)) will be different from the time intervals measured between the teeth 9C and 9D (i.e., aT (3) and

  AT (4)), even if the toothed wheel 6 rotates at a constant speed

  aunt. In the absence of correction, the data contained in the latch network 51 of FIG.

  that the wheel 30, and therefore the propeller to which the wheel is

  set, undergoes acceleration followed by deceleration,

  that the time interval AT (1) is less than

time-time AT (4).

  In addition, even if the gearwheel 30 was made

  in a perfect way, inconsistencies in the uniformity

  the material of the wheel may induce inconsistencies in the loading signals. One reason for this is that the coil 12 is triggered by a

  reluctance change in the region 95. The composi-

  wheel 30 as well as the geometrical configuration of

  the wheel intervene jointly in the change of

  luctance. It is to the change of reluctance given that the

  12 reacts to determine the passage of a tooth.

  Step 2 corrects differences in composition and configuration.

  geometric ration by means of an error coefficient.

  As noted in step 2, m and n are indexes. ("Modulo 8" means that the highest number that is used is 8, so that if n = 6, m is not equal to 10 but to 2. 9 becomes 1, 10 becomes 2, 11 becomes 3 and 12

  becomes 4. We could consider a pendulum as an Appa-

  providing a "modulo 12" indication. The highest number that is used is 12. If we add 4 hours at a given time equal to 11 hours, we do not get 15 hours, but 3 hours. The "8" in modulo 8 "refers to the 8 teeth.) For example, when m = 1, we have an = 15 and, therefore, with 8 teeth, we calculate an error coefficient for the tooth opposite the tooth which triggers the harassment

  current of latch networks 48 and 51 in FIG.

  This is important during acceleration and

  will explain it in more detail at the end of the description.

  tion. The error coefficient is calculated by means of the equation indicated in step 2. The equation has the effect of normalizing the time interval for the opposite tooth, with respect to the eighth of the corresponding time interval. in one complete turn. For example, if the gear wheel was made of

  perfect way, and with a perfect material, and if the

  the propeller speed was constant, all the ATs in step 1 would be identical. If the duration of a complete turn was 8 units, each A4T would be equal to 1 unit, and the coefficient

  of error in step 2 would be equal to unity.

  However, if the time interval AT (1) in FIG. 5A was equal to 3/4 units, and if the time interval AT (2) was equal to 1 + 1/4 units, the error coefficient for tooth 1 would be 3/4, according to the equation given in step 2 (3/4 = 8 x 3/4). (The T's are referenced to lines 99 passing through the centers of the teeth rather than the faces, for the sake of clarity.) The error coefficient is the ratio between the real time interval T (1) in Fig. 5A and an ideal time interval.A T (ID) at constant speed. We would obtain AT (ID) in the case of a perfect geometrical configuration and a composition

  perfect. We estimate AT (ID) by dividing by 8 the duration

  responding to a complete turn, as shown in Figure 5A. The error coefficients are used in step 4, but the microprocessor 66 first determines in step 3

  if there is a sub-speed condition. Such a condition

  under-speed is the idling motor operation. Another occurs during startup. If the

  sub-speed condition exists, all the coefficients of error

  are reset to unity. One reason for resetting the coefficients to unity is that it is not necessary to have high precision for

  the propeller speed measurements at such a low speed.

  In addition, the startup seems a logical moment to set

  variables, such as error coefficients, to

  their nominal, like unity. In addition, the range of

  64 K of meter 57 imposes a limit on

  the lowest rate that can be measured. The error coefficients are therefore useless at speeds below the limit, because the speed is not calculated. The following is a brief digressions referring first to Figure 6, which considers some of the difficulties

for measuring low speeds.

  FIG. 6 shows a pulse train 101 which is produced by the counter 57 in FIG. 1. Although the output signal of the counter 57 is actually a binary number which is constantly changing, it may be considered in FIG.

  this explanation is equivalent to the train of impulse

  101 of Figure 6, in which the successive impulses

  sives are separated by 1/2000000 s, as shown. If the pulse 103 which appears on the loading line 42, in correspondence with the passage of a tooth 45 in FIG. 1, is separated from the next pulse 105 by a distance

  which is greater than or equal to 64 K x 1/2000000 s, the micropro-

  cessor 66 can not distinguish pulse 105 from an impulse

  impulse 107 which appears exactly a period Tcycle plus

  early. The two pulses 105 and 107 have the same data.

  real time to the latch networks 48 and 51, on the

  54. The speed which is calculated on the basis of the

  103 and 107 would be the same as that calculated on the basis of

  pulses 103 and 105, although pulse 105 represents

  feels a lower real speed.

  Another way to express this is to say

  loading pulses 103 and 105 must be separated

  with a distance of less than 64 K pulses from the

  in train 101, to allow correct calculation of the

  speed. In the case of 8 teeth, a counter of 64 K and a clock rate of 2 MHz, the lowest rotational speed that is measurable is 228.7 rpm, which results from next calculation: Tcycle in Fig. 6 = 536 cycle increments, 328 second (1) 2,000,000 cycle increments - 0, ff328 - (1) second

  Tcycle is the maximum time interval between two teeth.

  For an eight-toothed wheel, Tcycle is 228.7 rpm.

  2287 1 cycle 1 revolution 60 seconds (2) 2287 = 0.0328 seconds x 8 cycles x 60 minutes (2)

  This limitation could be eliminated by using a

  more than 16 bits, such as a 32-bit counter or a larger counter that starts the counting cycle less often, which would increase the time interval Tcycle in Figure 6, but this would impose an increased cost, as well as possible difficulties in finding material

existing correspondent.

  This limit imposed on the measurement speed that we have just assumed assumes that the data in the SRM 70 are continuously updated. However, if the update is not continuous, but periodic, we obtain a different limit. The different limit mainly results from the fact that the 2 MHz clock that operates the counter 57 can be asynchronous with respect to the clock that operates the microprocessor 66, as will now be explained. There is a larger control system (the

  "main control system", not shown) for the

  and the aircraft with which propellers 15 and 18 are used. The reader does not have to worry about the main control system, but he only needs to know that one more

  computer program (the "main program") of

  the main control system must be executed, from the

  start to finish, every 10 milliseconds (ms). Au-

  in other words, the main program is repeated every 10 ms, as indicated by arrows such as those indicated

  at 150 in FIG. 7. The arrows 150 indicate the start

  rages of the main program. The requirement of 10 ms is imposed

  by factors that are not related to the invention.

  The program in Table 1 (the "program of

  se ") is executed inside the main program at 4

  1st, 8th and 10th ms during each execution of the program.

  me main. We can consider the speed program

  me a subprogram of the main program. The executions of the speed program are indicated by lines 155 on

  Figure 7. The time required for the execution of the pro-

  gram of speed is short, for example 50 microsecon-

  of, a microsecond being the fraction 1/1 000 000 of a

  second. This duration is so much shorter than

* Values of 10 ms (ie 10/1000 seconds) between

  margrades 150 of the main program, that the execution time of the speed program can not be drawn to the scale in Figure 7. The execution time is too short. exe-

  would take place for example in the interval of 50

  croseconds extending between the instants 49,975 and 50,025 shown in the figure. Such a duration would probably be

  invisible to the naked eye with the scale used.

  Therefore, the speed program is executed

  4th, 8th and 10th milliseconds during each ex-

  implementation of the main program. Executions of the program of

  are so fast that they can be considered as

  The scale of Figure 7 can also be considered as instantaneous with respect to Tcycle, which is 32.8 ms. Each run of the speed program puts

  the SRM 70 in FIG. 7 as explained above.

  We will now consider the asynchronous aspect of the counter

57 and the microprocessor 66.

  Four Tcycle intervals are shown and cnm-

  start at 0, 4, 8 and 10 ms. The counter 57 in FIG.

  can start from scratch (that is, start a new cy-

  cle) at any of these points, or at any point

  only between them. So, in a sense, the 57 meter and the micro-

  processor 66 are asynchronous: the start time 150 for the main program does not necessarily coincide with the start of Tcycle, and the start instant 150 has no known fixed relationship with the beginning of T cycle In

  this asynchronous situation, the analysis made by the invent-

  has led to the following conclusion: except in the case of an exception indicated later, there must be

  following relation between executions 155 of the pro-

speed grape.

  PREMIERE will be called the execution of the speed program which takes place immediately before a load is produced. LAST is the execution of the program 155

  speed that occurs after the next load. The

  quence is as follows: PREMIERE takes place, then a

  loading takes place, zero, one or more

  The program's speed medals then take place, then a second load takes place, and finally LAST takes place. In FIG. 1, PREMIERE can be the execution 155 at 5 ms, the loading can take place at point 157, and LAST would be

  then the execution taking place at 28 ms.

  The inventors have reached the conclusion that PREMIERE and LAST must take place during the

  same Tcycle interval, to safely eliminate the

  considered in relation to Figure 6. In other words, if PREMIERE and LAST are not included in the same

  Tcycle interval, it is not certain that the numbers

  held in the latch networks 48 and 51 provide data from which a velocity of

  precisely. Figure 8 illustrates this problem. The loading signals 157A and 157 load a number, for example

  3935, in the latch network 51 of FIG. 1. Then, in one case, a loading signal

  157C in FIG. 8, distant from more than one

  Tcycle value, load in the latch network 51 a second number, for example 5986. In another case, a loading signal 157D can load in the latch network 51

  an identical number (5986) because the counter 57 has

  started a new cycle in point 159. Thus, the speed program would see the same number (5986 in both cases), but this number represents extremely different TAs,

  as shown in Figure 8. The requirement that PRE-

  MINE and LAST must occur in the same interval Tcycle eliminates this error which is caused by different ATs.

  We can now look for what is the speed

  the smallest helix that can be measured in the circumstances just indicated, ie a Tcycle interval of 32.8 ms, asynchronous repetition of the main program every 10 ms, and an execution of

  speed program at the 4th, 8th and 10th millisecon-

  in each repetition of the main program. We get an answer to the question by moving

  Tcycle forwards and backwards between the four posi-

  Tcycle (1) -Tcycle (4) shown in FIG.

  look for the Tcycle position that gives the smallest name-

  run of the speed program between PREMIERE and LAST. For example, if we consider that the execution at 0 ms is included in Tcycle (2), and if this execution is

  the one called PREMIERE, the execution called DER-

  NIERE takes place at 30 ms. Intermediate executions of the

  speed gram occur at 4, 8, 10, 14, 18, 20, 24 and 28

  ms, for a total of 8 intermediate executions. In application

  for an analysis similar to the remaining Tcycle, we get

  the data shown in Table 2.

TABLE 2

FIRST

appears at

LAST

  appears in Interim Executions of speed program to Number of intermediate executions of speed program 1) 2) 3) 4) 0 ms 4 ms 8 ms ms 3) (after modification) 8 ms ms 34 ms ms ms 38 ms

4, 8, 10, 14, 18, 20, 24,

28 ms

8, 10, 14, 18, 20, 24,

28, 30 ms

, 14, 18, 20, 24, 28,

, 34, 38 ms

14, 18, 20, 24, 28, 30,

34, 38 ms

, 14, 18, 20, 24, 28,

  , 34 ms minimum Table 2 indicates that the lowest number of intermediate executions in the speed program is 8, in the last column on the right. Therefore, if the data contained in the latch network 51, which is read during execution of the velocity program, changes over an interval of eight executions, or less, of the velocity program, it is considered that FIRST and LAST appear together. during the same interval Tcycle. If the data in the latch network 51 remains unchanged for more than eight consecutive executions, 155, of the speed program,

  PREMIERE and LAST are considered to be out-

  Tcycle and, therefore, the two cycles of loading signals may have occurred outside the same Tcycle interval. The reader will note that the limit of eight unchanged readings of the latch network has the effect of modifying line 3 in Table 2. If the actual Tcycle interval that appears is the Tcycle (3) interval in Figure 7, LAST actually appears at 38 ms and not at 40 ms as at line 3, because a change in the contents of the network of flip-flops

  that occurs after 38 ms is not used, although it

  Moreover, the characteristics of a DER-

  NIERE, according to the rule of the eight executions of step 3 of the

  Table 1. This modification of line 3 is a consequence

  this of the asynchronous character. Although LAST appears at 40 ms with Tcycle (3), it is not known that Tcycle (3) is actually the Tcycle interval that appears. The interval in question could be Tcycle (1). Thus, any execution of the speed program following eight executions with data

  unchanged in the flip-flop network is in fact ignored.

  We can easily calculate the minimum speed that

  can always be measured according to the rule of eight

  once the rule has been established. This speed is

  linked to the smallest AT which could appear between two

  loading machines separated by eight intermediate

  from the program. This TA is the difference between the first

  intermediate implementation of the program and the last, ie

  say 28 - 4 = 24 ms for case 1 in Table 2. From Table 2, the minimum is 22 ms (case 2). By performing the calculation in the same way as in Equation 2, for a

  eight-tooth wheel, we get a speed of 340.9 rpm.

  1000 ms / s x 60 s / min 3409 8 teeth / turn x 22 ms If there is no underspeed condition,

  according to the rule of the eight executions, step 4 calculates

  following the speed present. As the expression between

  parentheses, the speed is adjusted by ERROR (m), for

  nier account for possible errors in the positioning of the teeth, shown in Figures 5 and 6. It will be assumed as an example that the complete duration of a lathe is 160000

  increments (that is 1/8 turn for 20000 increments) ,.

  but that the time intervals dT (1) and AT (2) on the

  5A correspond to 15000 and

  ments. The error coefficients in step 2 for teeth

  1 and 2 will be 3/4 and 1 + 1/4 respectively. Thus, at

  For example, the real speed calculated on the basis of AT (1) will be 314 60 x 2,000,000 750 rpm = 15,00 x 8 Thus, although the time interval actually measured is

  15 000 increments instead of 20 000 increments, the coef-

  error ficients allow you to calculate the actual speed

  of the propeller under steady state conditions.

  However, during acceleration and deceleration

  rations, the speed calculated in step 4 will be slightly different

  Feel the real speed. The difference will depend on

  the relative difference between the acceleration rate of the

  and the speed of calculation of the microprocessor 66 or, in

  simpler terms, the number of times that step 4 is ac-

  Complicated in a second, compared to the rate of acceleration of the propellers. The inventors performed a simulation in which step 4 was run at a rate of 300 times per second, while the propellers were accelerated with a maximum acceleration rate of 393 rpm. Fig. 9 is a graphical representation of the measured propeller speed (CSPD) and the measurement error (TERR), both expressed in rpm. Measurement error is low

  and never exceeds 1 rpm. By way of comparison, the

  Figure 10 shows the same simulation with all the coefficients

  set at 1 (that is, by removing the calculation

  done in step 2). Errors exceed 10 rpm. this is

  presented in order to demonstrate the effectiveness of the coefficients

errors.

  In the foregoing explanation, it has been assumed that single pickup coils 33 and 36 are used in FIG. 1 for each of the gears 27 and 30. However, it may be desirable to provide second coils or spare coils. , 110 and 113, as well as networks of reserve flip-flops 115. The inventors point out here

  that by using the reserve sensors 110 and 113,

  now four speeds: step 5 is executed for each of the four sensors. The sensors (ie

  the coils 36 and 113) for the propeller 15 will be called captain

  A1 and A2 in Table 1 and, similarly, the sensors relating to the propeller 18 will be the sensors B1 and

  B2. Step 5 checks that the sensors are working properly.

  The expression "sensor Ai" is an abbreviation of the phrase the speed calculated from the sensor Ai. "Step 5.1 determines whether the speeds indicated by the two sensors

  for a given helix are sufficiently similar; it is-

  ie in this case if they do not differ from each other by more than 40 rpm. If they are similar, the speed is taken equal to the average of the two speeds, and an indicator relative to each sensor is restored to indicate that both sensors are good. An indicator may be any type of memory device, such as a

memory position in MEV.

  If the difference between speeds falls to the

  Below the range of 40 rpm, step 5.2 is executed. Step 5.2 first determines whether the two speeds of the other helix (the back propeller in this example) are "good", based on step 5.1; that is, if they do not differ by more than 40 rpm. Steps 5.2.1 and 5.2.2 indicate in a more detailed form the following determination: among the sensors A1 and A2 (for the front propeller 15) which is the one that deviates the most from the speed (for example the speed ") which is indicated by the other propeller's detection system? (The reverse speed is the calculated speed for the rear propeller at

  step 5.1). The sensor with the smallest gap is

  stunned as the right sensor. If step 5.2 indicates that none of the sensors B1 and B2 are "good" (that is, the "backward speed" is not a reliable criterion), the step

  5.3 is executed. Step 5.3 determines which sensor indicates which

  than the lowest speed. The sensor indicating the

  lowest rate is chosen because the inventors

  consider that in the event of sensor failure, it is preferable

  to place the propellers 15 and 18 in an overspeed condition only in a sub-speed condition. Choosing the sensor with the lowest speed

  makes the speed control equipment of helicopters

  these (which we do not describe here) that propellers spin more

  slowly than the normal, and the equipment thus attempts to

  celery propellers, which places them in an overspeed condition.

  Step 6 consists of a double check. A pan-

  common of all four sensors, such as an electrical failure of the excitation circuit (not shown) can lead the step 5 to establish for the

  four sensor indicators that signal a "good" status.

  Step 6 prevents this. The sentence starting with "SI" at the beginning defines three conditions. (1) The turbine speed must exceed 10,000 rpm. (The turbine speed is referred to as the speed of the fast turbine of a gas turbine engine that can drive the propellers.) (2) The deviation of the actual pitch of the propeller from the programmed pitch must be less than a 3 and, similarly (3) the pitch deviation of the helix 18 must be less than 3. The existence of these conditions

  indicates that the propeller system operates under conditions

  corresponding to a certain power level. Under these conditions of pitch and engine, it is highly unlikely that either propeller rotates at less than 350 rpm. Therefore, if a measured value of 350 rpm is obtained, it is considered that the sensor providing this value is defective, and an indicator is accordingly set up.

  So far, only the detection of

  hostess. However, in a counter-rotating propeller system, it may also be desirable to detect the phase angle between the propellers. The phase will be defined with reference to FIG. 2. FIG. 2 schematically represents an axial view of two coaxial helices. The blades of one of the propellers are indicated by squares 120 and the blades of the other helix are indicated by circles 123. The phase angle is defined as the angle 125 between a blade of a helix

  and the blade closest to the other helix, in hori-

  lodged, but measured at the moment at which the blade 123 is at a

  predetermined position, such as the "noon" position which is

  presented. The actual angle 125 will obviously change constantly as the counter-rotating blades move one

  towards the other. However, when measured at the moment

  determined that we have just indicated, if the propellers rotate with constant and identical speeds, the phase angle

will be a measurable constant.

  The phase angle in fact describes the points of intersection

  in the space of the propeller blades. For example, if the blades 123A and 120A move at identical speeds, they will intersect approximately in the region 130. For acoustic and other reasons, it is sometimes desirable to be able to move this crossing point, for example

  by moving the region 130 to the region 133 in Figure 2.

  According to the invention, the phase angle is measured in step 7. It is considered that step 7 is understood

  itself. Step 7 corresponds to the calculation of the

  two time intervals. We can illustrate the

  arches 125 and 135 in FIG. 3. The arc 125 represents the duration that the blade 120A sets in FIG.

  move from point 137 to point 139. Similarly,

  arc 135 (also shown in FIG.

  shows the time interval that the blade 123A moves from point 141 to point 144. The ratio of the two

arcs (or angles) is the phase.

  The reason this ratio indicates the phase angle is that it gives the relative position of the blade 120 of Figure 2 with respect to the blade 123A when

  the latter is at a predetermined position, as the po-

  "noon" session which is represented. When the blade 120A is closer to the blade 123A (the angle 125 is lower), the phase calculated in step 7 will be lower. The reverse is

also true.

  The phase angle measured in step 7 expresses

  As the percentage is higher, the blade 120A in FIG. 2 is close to point 144 when the blade 123A is in the "midi" position. Therefore, the phase angle indicates the relative position of the blade 120A when the blade 123A

is at the "noon" position.

  In the invention just described, the speed

  rotation of an aircraft propeller is calculated a large number of

  many times a turn. One aspect of the invention is a tachometer. For example, at 1200 rpm, one turn takes 50 ms (0.050 s). In the time conditions of FIG. 7, 16 executions of the speed program take place between 0 and 50 ms, this last value being included; the speed is thus calculated 16 times per revolution. The invention can calculate

  this speed in a pair of counter-rotating aircraft propellers

  tives. In addition, the invention also calculates the phase angle of the counter-rotating propellers at the same rate of 16 times per revolution. The invention thus provides the computer and the

  pilot of the aircraft operating data that are dis-

  almost simultaneously with the measurement of TAs.

  In the introduction to the description

  it may be desirable to know the moment at which the blade n 1 was at the position "1 h, 30 min". This can be accomplished by adding a counter (not shown) that counts the cycles of the counter 57. A second set of memory locations, in addition to the VAM 70, can be used to record the data taken from the second counter. The microprocessor 66 would then read the contents of the second counter each time the contents of the flip-flop networks 48 and 50 are read, and record the two values of the counters in a pair of memory positions for the considered tooth 45. By

  example, a reading of 5 in the cycle counter at

  For example, if a network of latches contains a value of 12,000, it would indicate that the tooth that caused the loading of the value 12,000 in the latch network passed past the coil 36 at a time corresponding to 164.006 ms in time.

  real (164,006 = 5 x 32.8 ms + (12,000 / 2,000,000) x 32.8 ms).

  An invention has been described in which

  the T's using a magnetic pickup coil 36

  shown in FIG. 1. Another form would use an optical sensor, known in the art, to detect

passages of the teeth.

  An important aspect of the invention can be explained with reference to FIGS. 5 and 5A. As noted above,

  constant speed, a deviation of one tooth from its position

  The expected deviation results in a difference in TA measured against the ideal or corrected TA. This difference is used to calculate a coef-

  are in error in step 2 above. Then, later

  When re-measuring a time interval based on the tooth that has a deviation, the actual velocity can be calculated using both the time interval.

  measured (that is, uncorrected) and the error coefficient.

  In a sense, the corrected or ideal & T is regenerated from the

At T measured.

  The invention operates as indicated in the following example. It will be assumed that the squares 170 in Fig. 12 are in fact the teeth (i.e. the marks) in Fig. 1. It will further be assumed that the angle between all the pairs of neighboring marks is 45, but that the mark 170B is moved so that the angle between the marks B and 170C is 200, while the angle between the marks 170A and 170B is 70. The total angle between the reference 170A and the reference 170C is therefore 900. It will also be assumed that

  the duration of a constant speed lap is eight seconds.

  With these assumptions, AT (ID) is equal to one second and each

  measured T will be equal to one second, with two exceptions.

  The markers 170B and 170C have an AT of 20/45 x 1 second, while the markers 170A and 170B have an AT of 70/45 x 1 second. The error coefficients match the measured T's to the ideal T's. The error coefficients make it possible to calculate the real angles between the marks, in at least two ways: (1) the error coefficient for the angle of is equal to 20/45. It is known that the ideal AT represents an angle of 45 and, therefore, the real angle is equal to the product of the error coefficient by the ideal angle: 20/45 x = 20. (2) The complete set of landmarks describes a circle, which corresponds to 360. The AT for the interval of 200 was

  20/45 s. Because the complete circle represents eight se-

  for the angle: (20/45) x (360/8), ie 20.

  The knowledge of the angle between the marks allows

  calculate velocity based on TAs. In the example below

  above, the AT between the marks 170B and 170C, equal to 20/45 s,

  allows you to calculate the speed: 20/45 s for 20 'of

  corresponding to 1/45 s for 1, ie 360/45 (that is, eight seconds) for the complete set of

  on 360, which is in agreement with the speed considered.

  Therefore, once the coef-

  ficients of error, we know the actual angular positions of the landmarks. Then, we can deduce the speed from a

  This is true, in principle, even with a

  the ruler, as shown in Figure 12. Knowing the error coefficients, which contain data relating to the angular separation of the markers, the angular velocity of the helix can be calculated based solely on the known AT. . A complete lap is not necessary, and we can actually calculate the speed

  several times during a round, which gives an informa-

extremely recent

  It can be considered that the invention develops and records

  maintains a set of data, including TAs and time

  elapsed for a complete turn, from which it can be calculated

  move the positions of the marks. We create a kind of

  model of the detection wheel, in the SRM.

  In the foregoing explanation, only the effect of the positions of the marks on the T's has been envisaged. However, as indicated above with reference to FIGS. 5 and 7A, not only the position but also the geometric configuration and the composition in the TAs. Therefore, the TAs that the invention measures do not necessarily have a clear relationship with the teeth or faces 90 (Figure 5). However, the ATs actually have a well-defined relationship with the faces 90. By way of example, AT (1) can be measured in Fig. 5 for the teeth 9A and 9B. _AT (1) does not end on face 93, but on face 90 shown in dotted lines. This does not create any problem since this AT will usually end in the dashed face 90. Thus, the TAs do not establish

  the actual geometric angular separation between ad-

  but they actually establish what will be called

  the angles between the "actual" positions of adjacent teeth

  Centes. The dotted face 90 is one of these positions

"Effective".

  It has been stated previously in relation to the former

  plication of step 2 that the error coefficients are calculated for the blade opposite to that which has just caused the loading of a network of flip-flops. We will explain a

  because of this by taking an example.

  It will be assumed that the last tooth that has passed

  was tooth 8, and that at T (8) has just been calculated.

  Therefore, n = 8. It will also be assumed that a decelerating

  occurs, which is indicated by continual

  increasing, as shown in Table 3.

TABLE 3

  T (1) = 12 100 1 = (n + 1) modulo 8 AT (2) = 12,200 2 = (n + 2) modulo 8 6T (3) = 12,300 3 = (n + 3) modulo 8 T ( 4) = 12,400 4 = (n + 4) modulo 8 AT (5) = 12,500 5 = (n + 5) modulo 8 AT (6) = 12,600 6 = (n + 6) modulo 8 AT (7) ) = 12,700 7 = (n + 7) modulo 8 to T (8) = 12,800 8 = n Therefore, in this example: Time (full revolution) / 8 = 99,600/8 = 12,450

This is the estimated t T (ID).

  As noted above, the error coefficients are in fact a ratio between the real A T and A T (ID). It has also been reported that T (ID) is estimated from eight consecutive ANDs. In addition, the speed is calculated in step 4 from individual ATs, perhaps several times per turn. The inventors have found that, during constant acceleration or constant deceleration, adding 4 or 5 to the index in step 2 gives a more accurate speed calculation from a single AT, giving more accurate error coefficient, by a better estimated i T (ID). this is

  indicated in the right-hand column of Table 3.

Claims (5)

  1. A system for measuring the rotational speed of an aircraft propeller (15, 18) having a detection wheel (27, 30) carrying a plurality of markers (45) which pass to a reference point during rotation , characterized in that it comprises: (a) means (48, 51, 57, 66) for   measure time intervals between successive passages   references (45) to the reference points; (b)   means (66) for determining the angular   be adjacent markers, from time intervals; and (c) means (66) for calculating the speed on the basis of a time interval.   2. System for calculating the rotational speed   an aircraft propeller (15, 18), characterized in that it comprises: (a) a detection wheel (27, 30) coupled to the propeller and carrying a plurality of markers (45); (b) a sensor (33, 36) which produces charging signals by the passage of the markings (45); and (c) correction means   (66) for modifying the measured time interval.   between loading signals, which reduces   the effects of inconsistencies in the position,   position or the geometric configuration of the landmarks, or in   any combination of these features.   3. System designed to compensate for defects in   formed in the temporal characteristics of loading signals, in a tachometer for measuring the speed of rotation of an aircraft propeller (3) and comprising a detection wheel (6) carrying marks (9) which trigger the appearance loading signals in a sensor (12), characterized in that it comprises: (a) means (48, 51, 57, 66) for measuring the time intervals between   charging signals; (b) means (66) for the elaboration   set an ideal time interval; (c) means (66) for   to develop error coefficients by comparing   real time intervals with the ideal time interval;   and (d) means (66) for calculating the speed of rotation.   tation of the propeller on the basis of a time interval and of one of the error coefficients.   4. System for measuring rotational speeds   counter-rotating aircraft propeller (15, 18), characterized in that it comprises: (a) first and second detection wheels (30, 27) respectively fixed to first and second propellers (15, 18), 18), each detection wheel bearing a number of marks (45); (b) first and second sensors (36,33) for generating first and second loading signals by passing markers (45) past the sensors (36,33); (c) clock means (57) for generating a time signal; (d) first and second latch arrays (51, 48) for receiving the time signal and recording the time signal under the effect of the respective load signals; (e) means   memory (73, 75) for recording data obtained from   from the latch arrays, from which the time intervals between consecutive marks (45) on one of the detection wheels (30, 27) can be calculated; and (f) computing means (66) for calculating: (i) the total elapsed time for passing two or more markers (45); (ii) error coefficients to modify the   intervals calculated in (e) to compensate for irregularities   rites in the distribution of marks (45) on a wheel; and (iii) the almost instantaneous speed of the helices (15, 18), based on the time intervals calculated in (e) and the   error coefficients calculated in (f) (ii).   5. System for measuring rotational speeds   propellers in a pair of counter-rotating propellers   (15, 18), characterized in that it comprises: (a) a first   a detection wheel (30) coupled to the first propeller (15) and carrying a first set of marks (45), and a second detection wheel (27) coupled to the second propeller   (18) and carrying a second set of marks (15), the   partition of each set of reference marks not necessarily   uniformly uniform; (b) a first sensor (36) for   deriving a first loading signal due to the passage of marks (45) belonging to the first set of marks, and a second sensor (33) for producing a second loading signal under the effect of the passage of a marker belonging to at the second set of landmarks; (c) a clock (57) for   to provide a time signal; (d) a first network of   means (51) for recording the time signal that exists at the time of the occurrence of one of the first loading signals, and a second latch network (48) for   record the time signal that exists at the time of   displaying one of the second loading signals; (e) processing means (66) for reading the contents of the latches defined in (d), and for storing in a memory array (73, 75) the time signals recorded in the   networks of flip-flops; and (f) calculating means (66) for   to calculate: (i) the total time elapsed for the passage of a selected number of marks (45) on each wheel of   detection (30, 27); (ii) error coefficients that in-   indicate possible inconsistencies in the distribution   markings (45); and (iii) the speeds of the propellers (15, 18), based on the signals recorded in (d) and the   error coefficients calculated in (f) (ii).