Apparatus for registering a number of revolutions.

The invention relates to a device for detecting the speed of a rotary shaft<ul id="ul0001" list-style="none"><li>a) a digital incremental encoder,<ul id="ul0002" list-style="none"><li>- Which generates an angle signal that changes approximately periodically, at least in sections, with the angle of rotation of the rotary shaft, the respective period length of which is given by the angle ranges between approximately equidistant angular positions, which are determined by specifying an angle signal reference value, and its instantaneous amplitude in each case the instantaneous angle of rotation - in relation to a discrete angular position - corresponds to, and</li><li>- which in each case emits a counting pulse when the instantaneous angle signal amplitude assumes the reference value, and</li></ul></li><li>b) a counter which counts the counting pulses emitted within a measuring period for determining the angular ranges passing through.</li></ul>

A tacho dynamogenerator is usually used for an analog detection of the speed of a shaft, e.g. in a speed-controlled machine, the output signal of which is only approximately linear to the speed and is still superimposed by fluctuations caused, for example, by the lamination or contact voltages on the coals . At low speeds, these fluctuations cover the approximately speed-proportional part of the output signal in such a way that exact speed detection is not possible. In addition, there is a reversing error and a long-term drift, which is caused, for example, by the temperature response of the Kbl 2 Rch /17.2.81 generator even at high speeds.

Incremental rotary encoders are known for digital speed detection, in which a pulse disk connected to the rotary shaft, on which markings are attached at equidistant angular intervals, is scanned by a probe. These can be magnetic markings and Hall probes or optical markings and optical sensors. Depending on the angle of rotation, the measurement signal of these probes is initially a periodic analog signal that runs continuously between two extreme values, one extreme value corresponding to a discrete angular position in which the mark and the probe face each other, while the other extreme value corresponds to a discrete angular position in which the probe faces the middle between two marks. In order to obtain a digital signal from this generally approximately sinusoidal to triangular analog signal, the intersection points of the output signal with a fixed reference signal need only be determined by means of a limit value detector. The number of intersections per measurement period then corresponds to the number of marks moved past the probe, ie the number of discrete angular positions that lie within the angular range passed during the measurement period. The angle of rotation of the rotary shaft covered largely results from the product of this number and the angular distance of the discrete angular positions. If the length of the measuring period is fixed, the speed itself is therefore proportional to the number of angular positions scanned.

The digital speed detection therefore only determines the angular range traversed as an integral multiple of that between the equidistant ones. Angular positions lying angular range. Therefore, when specifying a fixed measuring period, the speed is only determined with sufficient accuracy if a sufficient number of passes through the equidistant angular positions take place during this measuring period. At low frequencies, this results in impermissibly long measuring times. But you can also. specify a number of passes and measure the time required for this; In this case too long measuring times result at low speeds, in addition there is an uncertainty due to time measurement at high speeds (very short measuring times). Therefore, the analog acquisition is often combined with the digital acquisition by coupling both a dynamogenerator and a digital rotary encoder to the rotary shaft and using the output signal of the digital rotary encoder to correct the analog rotary speed acquisition. However, this represents a high expenditure in terms of equipment. For mechanical reasons and because of the operational safety, the use of a dynamogenerator is often undesirable. In addition, exact detection of low speeds is still not possible. Rather, the speed of rotation zero, for example when the machine is at a standstill or when the direction of rotation is reversed, is a singular operating state which cannot be determined in this way.

The invention has for its object to provide a device for speed detection that allows a high-precision speed with little effort, especially in the range of low speeds to a standstill.

This object is achieved by starting from a digital incremental rotary encoder of the type mentioned at the outset, the continuous signal of which, in addition to the digital evaluation, is subjected to a second evaluation according to the invention. According to the invention, an output for the continuous angle signal as well as a memory connected to it and a downstream computing stage are provided on this digital rotary encoder. Memory and computing level are designed so that<ul id="ul0003" list-style="none"><li>- At the beginning of the measuring period by means of an. introductory pulse, the current angle of rotation (ϕ (t₁)) corresponding amplitude (U_s(ϕ₁)) read out and stored in the memory until the end of the measurement period,</li><li>- At the end of the measuring period by means of an end pulse the current angle of rotation (ϕ (t₂)) corresponding amplitude is read out,</li><li>- The difference between the two read amplitudes or the instantaneous rotation angle corresponding to the amplitudes is formed and</li><li>- As a correction quantity, either the difference normalized to the maximum amplitude of the angle signal is added to the number of angular ranges passed through</li></ul>or the difference to the product of the number of angular ranges passed and the mean angular distance of the discrete angular positions is added and the result of the addition is tapped as a measured speed value.

Practically all known digital rotary encoders can be used as digital rotary encoders, since the optical or magnetic probes used generally first generate an angle signal which is approximately sinusoidal to triangular in shape. Each discrete angular position corresponds to a zero crossing or another specific value of the angle signal and, for example, the zero crossings are counted during the measurement space to form the digital speed signal. If a reversal of the direction of rotation is to be permitted and recorded, it is common for these encoders to use a second probe. The probe is mounted in such a way that a second angle signal which is out of phase with the first angle signal is generated. In this case, a direction discriminator is provided which, when the direction of rotation is reversed, counts the discrete angular positions passed through with opposite signs.

As in the prior art, the rotary encoder determines the angle of rotation traveled during the measurement period as a whole number multiple of this mean angular distance, regardless of the angle initially run through until the first counted discrete angular position was reached and the angle at the end the last discrete angular position passed through the angle passed through. According to the invention, the integer multiple of the mean discrete angular distance determined by the digital rotary encoder is supplemented by adding a correction variable.

This takes advantage of the fact that the instantaneous amplitude of the angular signal is a measure of the angular distance between the instantaneous position of the shaft and the next discrete angular position, e.g. the angular position belonging to the zero crossing of the angular signal, regardless of whether this discrete angular position is even assumed in the measurement period. While the digital acquisition chronologically records the zero crossings of the angle signal, the position of the shaft between two angular positions corresponding to the zero crossings at the beginning and end of the measurement period is recorded angularly to form the correction variable. The correction value is the difference between the start position of the shaft and the end position. If both angle signals are the same, the correction quantity is zero and the angular range traversed is practically the multiple of the mean angular distance counted by the digital speed sensor. However, for example

the end position closer to the last angular position passed than the initial position, the angular range is correspondingly smaller than this multiple. Accordingly, the difference between the two angle signals is formed and normalized to the mean angular distance, ie dividing it by the mean distance between the discrete angular positions gives a fraction of the mean angular distance by which the number of discrete angular positions determined by the digital speed sensor has to be corrected. In the same way, however, the product of the number and the average distance determined by the digital encoder can also be used, in which case the difference between the two difference angles can then be used directly as the correction variable. In both cases, a signal proportional to the angle passed in the measurement period is generated.

If the measurement period is fixed, the angle passed is proportional to the speed. Specifying the length of the measurement period is often disadvantageous, however, rather the measurement period should be able to be changed, for example for speed controls, in accordance with the required interval between two actual speed recordings. In this case, the angle passed must be divided by the length of the time interval to switch to the actual speed value. A timer is advantageously provided, the start and stop impulses can be freely entered and which emits the impulses that initiate and end the measurement period. The time control unit advantageously contains a timer for determining the length of the measurement period and the computer a divider for forming the quotient of the size and the length of the measurement period resulting from the correction quantity addition. This angle related to the respective measuring period is then independent of the length of the measuring period and proportional to the average speed during the measuring period.

In principle, the output signal can alternately rise and fall monotonically between the discrete angular positions. This is the case if the encoder only generates a probe with an approximately sinusoidal, triangular or trapezoidal output signal. With a sufficient linear gradient, the difference angle between a current position of the rotary shaft and the discrete angular position following (or previous) in a certain direction of rotation is proportional to the difference between the current angular signal and the amplitude of the angular signal in this discrete angular position, with a corresponding sign change to take into account whether the current position corresponds to a rising or falling range of the angle signal.

However, it is advantageous if the rotary encoder contains a plurality of probes with a plurality of output signals which are offset from one another. By means of a switching device, these signal curves are each switched to the output for the angle signal in such a way that a sawtooth-like angle signal is produced which always increases monotonically with the angle of rotation within a period. In the case of an angle signal curve linearized in this way, the difference between the respective values ​​at the beginning or Angle signals measured at the end of the measurement period are used themselves, provided the minima and maxima of the angular signal curve are approximately constant. In the case of a nonlinear angle signal, linearization can also be carried out before the difference is formed.

The maximum amplitude, ie the stroke of the angle signal between two angular positions and thus the slope of the angular signal itself is proportional to the mean distance between the angular positions. As a result of irregularities in the marks on the pulse disks, aging, temperature response, and other influences of the rotary encoder, however, the slope of the angle signal curve and the angular spacing of the discrete angular positions is not a constant. A second memory is advantageously provided, in each of which the stroke of the angle signal between two adjacent discrete angular positions, ie the maximum angle signal amplitude, is read in and stored during the measurement period in order to form a current mean value for the angular distance between the discrete angular positions. The difference between the read out amplitude values ​​normalized to this mean distance is thus independent of these interfering influences.

It is also advantageously provided that the pulses that initiate and end the measurement period are delayed during the actuation of the switching device. Conversely, it is also advantageous to delay the actuation of the switching device during the reading of the angle signals. The time control mechanism already mentioned can be used for this. If the start and stop commands coincide with the changeover process, this prevents the readout from occurring at a point in time when no actuation of the changeover device results in a defined signal at the corresponding angle signal output.

The invention is explained in more detail with reference to a preferred embodiment and 4 figures.

Figure 1 shows the construction of a device according to the invention, Figure 2 shows various signal curves that occur, Figure 3 shows schematically a limit value circuit for forming the digital signal corresponding to the number of angular positions scanned during the measurement period, and a switching device in the device according to Figure 1, FIG. 4 shows the signal curves in a particularly extreme operating case, and FIG. 5 serves to explain the timing mechanism used.

In Figure 1, the digital encoder 1, the counter 2, the memory 3 for the initial amplitude of the angle signal, the arithmetic stage 4, a switching device 5, a second memory 6, which is used as a total memory for the between the two discrete angular positions in the measurement period stroke traveled, a timer 7 and a timer 8 shown.

A pulse disk 10 is attached to the shaft of a speed-controlled machine and bears markings arranged at equidistant intervals, which are scanned by means of a corresponding probe device, for example an illumination device 11 and an optical sensor 12.

First of all, let us assume the simple case that only one direction of rotation of the rotating shaft is possible and should be detected. In this case, a probe device from a single sensor, the output signal (angle signal) of which, for example, a photo voltage U is sufficient_s can be. The dependence of this angle signal on the position of the axis of rotation (angle ϕ) can, for example, be approximately sinusoidal in accordance with curve A in FIG. If N marks are arranged along the circumference of the pulse disk, the period of the curve profile A is 2π / N. The zero crossings shown at π / N, 2π / N occur when the turntable 10 with its markings is positioned in the beam path between the lighting device 11 and the sensor 12 in such a way that the sensor is half. Receives light intensity. For example, these positions can be used as the discrete angular positions. According to the prior art, the zero crossing of curve A must then only be detected by a limit value signaling circuit 13 and fed to counter 2 via a pulse generator 14. In the example shown, the pulse disk 10 is located, for example, at the beginning (time t₁) of the measuring period T in the position ϕ₁ shortly before a zero crossing and at the end (time t₂) of the measuring period in the angular position ϕ₂ shortly after an amplitude pass, an angle was covered during the measurement period T, which is slightly above π / N, corresponding to a speed slightly above 1 / (2NT). In this case, however, the number n of discrete angular positions stored in the counter is n = 2, corresponding to an angle of 2π / N or a speed 1 / (NT). The digital speed measurement itself is initially imprecise.

In order to also be able to detect a reverse direction of rotation, a sensor device with two sensors is used according to the prior art, which are arranged on the pulse disk in such a way that a second signal, designated B in FIG. 2 and out of phase with respect to A, is generated . There is a positive direction of rotation if the angle of rotation y changes progressively from left to right according to FIG. A positive edge of the limit value signal then always coincides with positive values ​​of B, while when the direction of rotation is reversed, curves A and B in FIG. 2 are run from right to left, a positive edge of the limit value signal coinciding with negative values ​​of B. A direction discriminator 16 can therefore be used to detect the direction of rotation. This directional discrimination is, for example in "The numerical control of machine tools", Munich 1971, p. 13. For each discrete angular position passed in the positive direction of rotation, a corresponding pulse signal is then fed to a counter 18 for the positive direction of rotation, and for each discrete angular position passed in the negative direction to a counter 19 for the negative direction of rotation. The two counter readings 18 and 19 are subtracted in an adder 21 at the end of the measuring period and thus form a digital signal n which is approximately proportional to the speed at a constant length T of the measuring period.

A finer digital speed detection is obtained if the discrete angular positions durch given by the zero crossings of curves A and B_O = π / N, 2π / N, ···· further discrete, equidistant angular positions ϕ '_O be placed. This is achieved in the preferred embodiment in that, in addition to curves A and B, the corresponding inverted curves A, B and their intersections with curves A, B are formed (inversion elements 23, 24 in FIG. 3). A limit value signaling circuit 5 'with several limit value indicators 25, 26 forms angular ranges in which B&gt; A, or B&gt;<o>A</o> applies that by the signals U_a and U_b are specified. Switching signals can then be formed via further AND, NAND or NOR gates 27 which, via corresponding switches 31 of the switching device 5, respectively the branch of the curves A, which rises by a zero crossing,<o>A</o>, Federation <o>B</o> connect to the angle signal output 20. This creates the sawtooth-like curve U shown in FIG. 2_s for the angle signal. The control of the switches 31 via the line 42 does not take place directly, but, as indicated in FIG. 3, via the time control unit 7, which is yet to be explained.

This circuit therefore works with twice the number of discrete angular positions (average distance 2π / 2N), which corresponds to digital interpolation with a step size of 0.5 compared to the simpler version. For the through the angular positions ϕ₁, ϕ₂ designated angular range in the example of Figure 2 results as a digital output signal n = 4, corresponding to a speed 4/2 NT. As can be seen, the accuracy of the speed determined for this example has not improved as a result.

Rather, in order to improve the speed detection according to the invention, the angle signal of the digital rotary encoder is brought out via an output 30.

The approximately sawtooth-shaped signal U now present at the output of the switching device 5_s has the amplitudes U at the beginning and at the end of the measuring period_S (ϕ₁) or U_S(ϕ₂) directly as a measure of the angular position_en ϕ '₁, ϕ₂'can be used at the beginning and end of the measurement period. Using the mean distance between the discrete angular positions<o>φ</o> denotes, as can be seen directly from FIG. 2, the much more precise relationship results for the entire angle Δϕ swept within the measurement period<maths id="math0001" num=""><img file="EP0059244A2_D0001.tif" /></maths>

The digital speed n initially determined must therefore be corrected by a correction value<maths id="math0002" num=""><img file="EP0059244A2_D0002.tif" /></maths>or with linearity of U_s(ϕ) and known maximum amplitude U_O:<maths id="math0003" num=""><img file="EP0059244A2_D0003.tif" /></maths>

Getting corrected.

This correction takes place in detail in that the input 30 of the memory 3 is connected to the output 20 of the switching device 5 and via a control line 32 at the beginning of the measuring period (time t₁) to save the current amplitude (U_s(ϕ₁) is released. From this memory 3, U_s(ϕ₁) taken over by the computer 4 at the end of the measurement period and read into the computer memory 33, for example. In the memory 3 can now at the end of the measurement period (time t₂) present amplitude_ude U_s(ϕ₂) can be read in as the initial value for the next measurement period; U_s(ϕ₂) is also read into the arithmetic unit 34 of the computer 4. The one on the subtractor₃₅ difference U formed_S(ϕ₁) - U_S (ϕ₂), if they are already on the average voltage swing of the saw tooth U_S is standardized, can be added at an adder 36 to the output signal n of the digital rotary encoder. The variable obtained by this correction can be tapped directly at an output 37 as a measure of the angle of rotation Δϕ covered during the measurement period. If the length of the measuring period T is constant, a measure of the speed is also obtained.

As already mentioned, however, it is advantageous to adapt the length of the measurement period to the respective requirements. For this purpose, the corrected signal obtained at the adder 36 must be divided by the length of the measurement period (divider 37). This length T is in turn supplied by a timer 8, for example the counter reading at the end of the measuring period of a counter for counting pulses of constant frequency set at the beginning of the measuring period. Setting and reading out this counter 8 is also controlled by the time control unit 7 via the line 32.

The probes used to form curves A, B can deliver somewhat different signal profiles, for example due to inaccurate adjustment, as is shown in a greatly exaggerated manner in FIG. The consequence of this is a fluctuation in the amplitudes of the sawtooth U_SP which is also usually subject to temperature change and aging. Therefore, it is advantageous to use the mean distance<o>φ</o> to determine the corresponding maximum amplitude U of U during the measurement period and to normalize the difference ϕ '₁ - ϕ₂' or. ^U_s (ϕ₁) - U_s(ϕ₂) to use. This can be done by reading out the amplitude value of the angle signal and adding it in the counter 6 each time the switching device 5 is actuated. At the end of the measurement period, this value can be read out and divided by the digital signal n in the divider 40 for averaging. For n = 0, ie in the vicinity of the standstill, the values ​​for n and the memory content 6 determined in the previous measurement period can be used. The divider 41 serves to normalize the amplitudes U_s(ϕ₁) and U_s(ϕ₂).

In this device, the counter 2 and the timer 8 are set and the memory input 3 is opened at the beginning of a measurement period. During the measurement period, triggered by the signals of the limit signal circuit 5 ', the switching device 5 and the memory 6 are actuated and at the end of the measurement period the memories are read out and the arithmetic stage 4 is activated. A time control unit 7 is provided for this time control, the functions of which are to be explained with reference to FIG. 4.

In Figure 4, the limit case is simultaneously shown that during the measurement period, the rotating shaft temporarily stops when the direction of rotation is reversed.

First, assume a constant positive direction of rotation. The angle signal follows the linear part of curve A up to the intersection (time t₀) of curve A with curve B. As a result, the state of signal U changes in the manner already explained in connection with FIG_ao Therefore U_S switched from curve A to curve B. This process does not occur suddenly, rather the switchover is only after a certain time_u completed. The worst case is now assumed that the time t_a, at which the start command to initiate the measuring process is given, falls into the switching process, so that at time t_a there is no defined angle signal. It is now planned to start with the start pulse (time t_a) do not set the start of the measurement period directly, but rather first input this pulse to the time control unit 7. This pulse is first stored there and only processed after the switching process has ended. This can happen, for example, in that the pulses formed by the switching device 14 each trigger a corresponding timing element in the time control unit 7 for delaying the initiating pulse. In Figure 4 with U_u denotes the output signal of the delay element with which the start pulse input for the duration Δ t_u the switchover time is blocked.

The principle of a timing control unit 7 will be explained with reference to FIG. 5. With the changeover pulse (edge ​​of signal U_a) for one of the switches 31, a timer 50 (output signal U) serving as a delay element is first triggered, these time elements for a duration At of a pulse generator to form the initiating pulse (time t₁) Block upstream gate 52 so that the start pulse, which initially triggers another delay element 53 upstream of gate 52, is stored. Thus, a pulse for activating the timer 8 and the memory 3 can only be emitted after the switchover time has elapsed.

Likewise, the switchover pulse in the timing element 50 is also delayed if, during the readout from the limit value detector 5¹ (Signal curve U_a) a switchover was triggered. For this purpose, the gate 50 is followed by a gate 54 with a pulse generator 55, the gate 54 for the gate 53 (output signal U_t) certain readout time remains blocked.

The gate 54 is in an analogous manner during the readout (time t₂) of the final amplitude U_s(ϕ₂) ^Ge-locks in that the corresponding output signal of a timing element, not shown, is also connected to it, this additional timing element being triggered by the stop pulse (time t). This timer is also followed by a gate and a pulse generator so that the delivery of the end pulse can also be delayed if the stop pulse falls into a switchover process. This case is shown in Fig. 4th shown, with the additional assumption that the direction of rotation is changed and the rotary shaft is temporarily at a standstill at time t = o.

After changing the direction of rotation, the angle signal follows the curve <o>B</o> from Fig. 2 and at time t_e is the stop pulse, which in this case with the pulse t₂coincides at the end of the measurement period. During the readout (duration Δ t_a) the comparator 5 'reports the curve intersection of A and B, but now the switching of the angle signal U from the curve <o>B</o> delayed on A until the end of the reading.

As can be seen from FIG. 4, at times t₁ and t₂ practically the same angle signal U in each case_S(ϕ₁) = <o>u</o>_S(ϕ₂), ie the correction quantity zero. Furthermore, since there was no zero crossing of the angle signal U in the measuring period, n = 0, so that the value zero is present at the output 37 for the angle traveled and the value zero is also present at the output 43 of the dividing element 38 for the speed.

The invention thus not only enables more precise speed detection, with disruptive influences such as_' For example, the temperature response and aging of the sensors are largely averaged, but the zero speed is not a singular value and the standstill can also be recorded exactly.