WO1989011080A1

WO1989011080A1 - Position-coded sensor

Info

Publication number: WO1989011080A1
Application number: PCT/EP1989/000357
Authority: WO
Inventors: Günter Schwesig
Original assignee: Siemens Aktiengesellschaft
Priority date: 1988-05-10
Filing date: 1989-04-03
Publication date: 1989-11-16

Abstract

In order to generate a position-coded signal in a position sensor at virtually any given time, two tracks are scanned by two sensors, each of which generates a pair of sinusoidal signals (cos alpha, sin alpha; cos beta, sin beta) or a similar periodic signal with a definite phase relation at any given time, the periods of the phases being in the ratio m/m-1. The desired relative distance is derived from the difference between the two phase relations.

Description

Position-coded encoder

The invention relates to a method for determining the relative distance between two mutually movable parts with the features specified in the preamble of claim 1 and a device suitable therefor (DE-OS 34 16 090).

Measuring systems (with an optically or inductively operating) sensor are used to determine the relative position of two parts that move relative to one another, which emits a signal each time a corresponding mark is moved past it. The signal is often shaped into a pulse with sharp pulse edges by electronic means. In the case of a track consisting of marks arranged equidistantly along the direction of movement, a pulse sequence then arises which breaks down the path covered into path increments and the path covered can be determined by counting the pulses.

By appropriate design of the marks it can also be achieved that the sensor delivers an approximately triangular or sinusoidal output signal. However, this signal is ambiguous within a path increment, i.e. every signal value lying between the extreme values is assumed with two relative positions between detector and mark.

The elimination of this ambiguity and at the same time a distinction of the direction of movement succeeds in that this incremental path measuring system is supplemented by a second way mark track and / or a second sensor in such a way that two signals which are phase-shifted by 90 ° are generated, from their combination a sign for the evaluation of the distance traveled Path increments is determined. EP-AO 059 244 discloses a possibility of obtaining a signal of 4 times the frequency from two such phase-shifted signals of the same period length, which signal is practically sawtooth-shaped. Regardless of whether the sinusoidal, triangular or sawtooth-shaped signal is further processed as a strictly continuous analog signal or as a quasi-constant digital signal, this is - in contrast to the discontinuous signal formed by binary counts - a continuous function of the current relative distance, since there is a signal value belonging to the instantaneous distance at all times, which changes when the instantaneous distance changes and thus has a defined phase position, while the mentioned counting pulses only occur when the mark and sensor are opposed to each other and thus conditions always occur which a change in the instantaneous distance does not cause a change in the binary pulse. From the instantaneous value of a binary signal, the instantaneous phase position of the sensor signal cannot be concluded.

The mentioned, known sawtooth signal already represents a reversible-unique function ("one-two-one-function") of the relative distance within a path increment and thus a signal with a clear phase position. With regard to a measuring range composed of many path increments, in the case of a rotating one Machine, for example, an entire revolution, but this continuous signal is also ambiguous according to the number of path increments. Therefore, in the known arrangement, the electronic formation of counting pulses for the travel increments traveled is additionally required. As a result, the measuring range, which is initially given by the path increment or the period of the sensor signal, can be expanded as desired. Another extension of the measuring range is shown in the aforementioned DE-OS 34 16 090 and shown in Fig. 1 for a rotary machine. A code disk G1 is coupled to its shaft 1 with a code mark R1, which is thus movable relative to a stationary sensor S1 and generates a periodic code signal in the detector, which is a function of the relative angular distance, ie the angle of rotation of shaft 1. In addition to this first measuring system, a second measuring system is provided which contains a reference disk G2 with the reference mark R2 which is coupled to the shaft 1 via a gearbox G and which is therefore movable relative to the sensor S2 (containing a corresponding code detector). It generates a reference signal r2, which is also a function of the relative distance between the shaft 1 and the sensors S1 and S2, but has a different period length compared to the code signal r1 due to the transmission ratio in the gearbox G.

By scanning the two mark systems R1 and R2, two signals are generated as periodic functions of the angle of rotation of shaft 1. This encoder is intended to detect the angle of rotation of shaft 1 within a measuring range which comprises 12 complete revolutions. A measuring range X = 12 is therefore for this shaft rotation angle. 2 ￗ provided. With regard to this measuring range X, the code signal r1, the period of which corresponds to one complete revolution of the shaft, thus has the number of periods 11, the gear G with the gear ratio 12:11, but has the effect that the reference signal r2 contains only 11 periods in this measuring range X, thus has a number of periods different by the value 1.

In order to obtain a distance signal from the reference signal r2 and the code signal rl, which has a clear function of the relative distance within the intended measuring range X. That is, a special evaluation device is required, which initially consists of two way mark tracks with the equidistant way marks W11 ... W1n and W21 ... W2n, which are phase-shifted relative to each other by 90 °, and corresponding way mark detectors in sensor S1. The corresponding incremental path signals w1 and w2 are shaped into pulse sequences in an electronic evaluation unit, just like the code signal r1 and the reference signal r2, and initially allow, in the manner described at the beginning, the path increments covered within a complete shaft rotation, which have a length of 2 ￗ / n, to count. In addition, if there is a pulse of the code signal r1 and the next pulse of the reference signal r2 q pulses of the path signal wl, this means that the axis has a starting position in which the marks R1 and R2 face the sensors S1 and S2, about the angle of rotation q. 2 ￗ is turned further.

To determine the position of shaft 1 within the entire measuring range X = m. 2 ￗ to be determined even if the count number for the path increments has initially been lost due to a fault or is not yet available after the encoder is switched on, it must first be waited until a pulse of the path signal wl and the code signal r1 occur simultaneously and then the pulses of the path signal w1 must be counted until a pulse of the reference signal r2 occurs.

Such long evaluation times are often not favorable for slowly rotating, frequently reversing machines. Another application relates to converter-controlled induction machines in which, for example, the position of the rotor must be known from the start with sufficient accuracy so that when the inverter starts, the current flow through the machine is at the correct angle to the axis of the rotating field and the machine starts up in the desired direction of rotation with the desired torque can. According to the number of pole pairs However, the field distribution of such machines is ambiguous, so that the spatial assignment of the field axis and current flow required for starting the machine is possible with several different positions of the rotor. The measuring range for the position of the rotor in this case is not 2 ￗ, but in accordance with this ambiguity only 2 ￗ / p, however, the encoder should also record the rotor position as precisely as possible in a snapshot or in any case within short evaluation times.

The methods and devices described so far are not suitable for this.

Rather, position-coded encoders have mostly been used so far, which, in addition to a measuring system that breaks down the angle of rotation into small path increments in accordance with the desired resolution and is correspondingly ambiguous, contains further measuring systems that each break down into larger path increments to gradually eliminate ambiguity. These can be sensors with binary output signals or with signals that are a continuous function of the current relative distance.

However, the production of sufficiently precise, multi-track brand discs of this kind is difficult and costly and the corresponding large number of sensor systems means additional effort. This is particularly the case with series products, e.g. Rotary field machines or also in processing machines in which a tool is moved linearly or rotationally relative to a workpiece are not justifiable.

The invention is therefore based on the object of determining the relative position between two parts which are movable relative to one another as simply as possible in the manner of a position-coded snapshot. This object is achieved by a method having the features of claim 1 and an apparatus having the features of claim 2. Advantageous developments of the invention are characterized in the subclaims.

The invention is explained in more detail with reference to three exemplary embodiments and seven figures.

FIG. 1 shows the prior art already explained,

2 and 3 show the application of the invention in a linear displacement, carried out with optical measuring systems, FIG. 4 and FIG. 5 show an application of the invention for rotary displacement, carried out with electromagnetic measuring systems and signals occurring in the process.

Figure 6 and Figure 7 shows a particularly advantageous embodiment in plan and cross-section.

Figures 2 and 3 show schematically two sections through the measuring part of a machine, not shown, with linear movement. A measuring slide 3 is movable along rails 1, which are fastened to a stationary holder 2, and carries the detectors of two optical measuring systems. The beam path of these measuring systems consists of a miniaturized light source 4, 5, a system of lenses 6, 7 and diaphragms 8, 9, which can also carry grids (not shown) for better beam guidance, and leads to corresponding sensors 10, 11. An The holder 2 is also attached to a carrier 12 which carries two tracks 13 and 14 with corresponding optical marks 15, 16 (eg line grids).

The carriage 3, which can be shifted between the starting position 3 'and the end position 3''by the measuring range X, is in a position according to FIG. 3, which is described by the “relative distance” x from the starting position 3 ′ resting on the holder becomes. The along the tracks 13 and 14 laid marks have a different spacing, the marks 15 with the smaller spacing being referred to as "code marks" and the measuring range X in m (at least three, in the example: m = 8 but often a hundred or, depending on the desired measuring accuracy) more) Disassemble "code increments". The reference track 14 divides the measuring range into m − 1 = 7 “reference increments” by means of corresponding, equidistant “reference marks” 16, with m being selected as an integer for better understanding.

Each detector is designed to provide a pair of signals 90 ° out of phase with each other. For this purpose, for example, the code detector 10 scanning the code track 15 of FIG. 2 contains a pair of code sensors 10 'and 10''which are a quarter or three quarters of a code increment apart in the direction of movement. Appropriate design of the code marks 15 and the elements 4, 8 and 10 scanning these marks ensures that the sensor signals u 'and u''have an approximately sinusoidal course depending on the relative distance x. In the case of another, for example triangular, curve shape, the sinusoidal shape can also be forced by electronic means. The signals u 'and u''can be analog signals or correspondingly finely resolved digital signals. At the input of an electronic evaluation device 17, which is a pure computing device, the amplitudes of the two sensor signals can be matched and normalized so that a signal pair u '= cos α = cos (2 ￗ. M. X / X) for the reference signal u u '' = sin α = sin (2 ￗ. m. x / X) arises. Such signals can be understood as orthogonal components of a length which is standardized in a plane and a length which can be described by the angle α. Corresponding standardization elements are known as "vector analyzers" VA standard elements for the control of induction machines. The same applies to the reference detector 11 scanning the reference track 16, which thus contains the sensor pair 11 ', 11''and the signal pair v' = cos β = cos (2 ￗ (m-1) x / X) v 'as the reference signal v '= sin β = sin (2 ￗ (m-1) x / X) returns. The code signal u (ie the signal pair u ', u'') is al so a clear function of a phase angle α, and thus a clear, periodic function of the relative distance x with the period length X / m and the number of periods m in the measuring range X; the relative distance x itself, however, as x = cos (2 ￗ mx / X) in the measuring range X is 2 m times ambiguous. The reference detector also determines the relative distance x as a correspondingly ambiguous function of the “reference angle” β contained in the reference signal v.

This ambiguity is eliminated by a difference generator, which now forms a distance signal which, as a function of the difference angle between code angle and reference angle ε = α-β = 2 ￗ x / X, is uniquely assigned to the relative distance x. This difference angle ε can also be understood as the direction angle of a vector e and can therefore be described by an angle function pair e '= cos ε, e''= sin ε corresponding to orthogonal vector components. In order to process the difference angle between two vectors as corresponding angle functions, a so-called "vector rotator" VD- is used in the control of induction machines, which performs the following operations: cos (α- β) = cos α cosβ + sin α sinα sin (α - β) = -cos α sinβ + sin α cosβ so that the evaluation device 17 by means of the corresponding vector rotator as the distance signal e the pair of angular functions e '= u' v '+ u''v''= cos 2τrχ / X e''=-u' v '' + u '' v ' = sin 2 ￗ x / X returns. In the case of a three-phase machine, this is already the suitable form that allows the machine control to use the relative distance x for the position and speed control of the machine. If a linearized signal is required for the relative distance x, this can be determined from sin e '', the ambiguity of this function being eliminated by the sign of e '.

In order to determine the phase position of the angle signal e with sufficient accuracy, it suffices that the phase positions of the code signal and reference signal are detected with an (m / 2) times lower accuracy. In particular, simple electromagnetic measuring systems can be used.

This is shown in FIGS. 4 and 5 for the preferred application that a three-phase machine (for example a synchronous machine SM) is fed by a converter SR. The machine has a p-fold symmetry, where p is a positive integer, for which p = 3 is assumed in this case. For a given starting position of the rotor, the converter SR must be controlled by the control set CT in such a way that the feed current is only fed into certain windings of the stator reference system, so that the machine starts with a desired torque in the correct direction of movement. The control therefore usually contains a vector rotator VD +, in which an angular addition is ultimately carried out so that the control of the headset CT and the converter SR can search for the windings from the stator windings arranged on the circumference of the stator, the winding axes of which compared to those given by the rotor position Axes of symmetry of the field to the desired starting angle point. The sum of the starting angle and the rotor angle then gives the required control angle in relation to a reference axis of the stationary stator winding system. Because of the p-fold symmetry of the machine, the rotor angle only needs to be determined within a measuring range X = 2 / p, since after each rotation of the rotor around this measuring range the same conditions are again present in the machine.

Teeth of the electromagnetic measuring systems are advantageously equidistantly distributed on the circumference of circular disks 20, 21 which are rigidly coupled to one another and are scanned by corresponding sensors. For a better visual representation, the teeth on the code disk 20 or the reference disk 21 are shown with alternating magnetization (one pair of teeth 22 or 23 is therefore a magnetic mark). The corresponding sensors 24 ', 24' 'of the code detector scanning the code marks are intended to emit a signal pair of approximately sinusoidal signals which are 90 ° out of phase with respect to one another, which is why the sensor 24' 'faces the space between two teeth when the sensor 24' faces a tooth directly . The same applies to the sensors 25 'and 25' 'of the reference detector, but the tooth spacing of the two disks 20, 21 are different.

On the reference disc e.g. R = 100 marks (pairs of teeth) arranged with a corresponding angular distance 2ττ / R, while the code disk 20 bears M = R + p marks to generate the desired p-fold symmetry, in the example M = 103.

If the two disks are rigidly connected to the rotor of the synchronous machine SM, then a complete revolution of the rotor is broken down into M path increments, ie incremental angular ranges of the size T _M = 2 ￗ / M. The standardized output signals u ', u'',v', v '' of the sensors 24 ', 24'',25' and 25 '' are angular functions of the angle of rotation ε with the periods

From these signals the vector rotator then delivers a pair of angular functions depending on the differential angle α -β, for which the following applies: e '= cos (α-β) = cos (M - R). ε e '' = sin (α- β) = sin (M - R). ε Since the number of marks for the two tracks was specified according to the relationship M = R + p, the distance signal e in the measuring range X = 2 ￗ / p thus has a phase position that is proportional to ε and can be measured immediately: e '= cos (p . ε) = cos (2 ￗ. x / X) e '' = sin (2 ￗ. x / X).

FIG. 5 shows the sensor signals u 'and v' as well as the signals e 'and e' 'of the distance signal e and the phase angle ε. of the distance signal e shown. At a certain point in time, at which the phase difference δ prevails between the signals u 'and v' in a snapshot, the angle of rotation ε of the rotor axis is equal to δ / p or (δ / p - 2 ￗ / 3).

Both circularly closed tracks of the arrangement according to FIG. 4 require integers for the number M and R of the marks 22 and 23. As the example with R = 100 shows, however, there is no integer number of periods m = M / p for the measuring range X = 2 ￗ / p and the corresponding number of periods X / T _R = (Mp) / p = (m-1) is not in this measuring range integer. It is only necessary that the two period numbers differ by the value 1.

Accordingly, the zero point ε0 for the measuring range X of the rotor position shown in FIG. 4 does not begin with the phase position α = 0 and β = 0 of the code signal u and the reference signal v, but with corresponding phase positions α ₀ and β ₀ , which are only represented by α ₀ -β ₀ = ε _{0 are} marked. If the distance between the two reference sensors and within the code system the distance between the two code sensors are therefore only adjusted within the reference system, a careful mutual adjustment of the two detectors is no more necessary than a careful adjustment to the desired reference axis of the angle of rotation. Rather, with any arrangement, any angle IQ can be specified as the zero point of the measuring range.

It is sufficient to correspondingly shift the phase angle ε of the distance signal e by means of a corresponding vector rotator VD ₀ , to which the angle functions of this reference angle ε _{0 have been} entered. The distance signal thus obtained then has the phase angle x = ε - ε0, which changes linearly in the measuring range between 0 ° and 360 ".

In some cases it is only necessary to deliver a reference pulse at a certain reference position, ie a certain reference angle ε *. Despite the ambiguity of the angular functions e '= cos ε and e''= sin ε, this can be determined by a corresponding threshold value element SW, which responds, for example, at e''= sin ε *, the ambiguity being removed from the sign sign el can. If, for example, ε * lies in the angular range between 0º and 90º, the threshold value element evaluates the condition e '' = sinε * only for the values ε for which sign e '= 1 applies. Figures 6 and 7 show in top view and cross section a position-coded angle encoder according to a first variant of the invention. On both broad sides of a disk S rotatable about an axis of rotation W, two tracks are arranged along the circumference, which are formed from a homogeneous magnetically conductive material and have a profile with teeth that extend in the radial direction and in the axial direction on opposite sensors are directed. The code track contains m = 100 equidistant teeth along the circumference, while the reference track RS only has n = m - 1 - 99 equidistant teeth. The code track CS is scanned electromagnetically by two code sensors U1 'and U1'', which respond to the variable distance between the sensor and the tooth surface in accordance with the profile being moved past. The two code sensors U1 'and U1''are adjacent, but offset from one another in such a way that they deliver a pair u1 of approximately sinusoidal sensor signals u1', U1 '', which are 90 ° out of phase with each other.

The reference track RS is correspondingly scanned by a pair V, of reference sensors, which are covered in FIG. 1 by the pair U _{1 of} the code sensors U1 ', U2', since they are opposite one another, while in FIG. 2 the two sensors lying next to one another of the two sensor pairs U1 and V1 appear like a single sensor.

The amplitude of the sensor signal pairs u1 and v1 (ie the individual ignale u1 ', u1 ", v1', v1 '') is now falsified by a radial play, which can be caused, for example, by an inaccurate centering on the axis of rotation W. Because the If both sensor pairs are subject to these falsifications at virtually the same time at the same location, this radial play affects all signals equally, and it can also be largely suppressed if the sensors are smaller than the width of the track. In addition, axial play can also occur if, for example, the disk is not exactly perpendicular to the rotating shaft W and therefore "wobbles". Although this affects the two sensors of a pair of sensors practically in the same way, it amplifies the amplitude of a pair of sensor signals while simultaneously reducing the amplitude of the other pair of sensor signals.

It is therefore provided in the variant according to FIGS. 6 and 7 that a pair of compensation sensors is arranged on each broad side diametrically opposite to a pair of sensors, which also scan the track and provide the same sinusoidal output signals as their associated sensors. With an axial play, e.g. the output signals u2 ', u2' 'of the compensation sensors U2', U2'1 are disturbed in the opposite sense to the signals u1 ', u1' 'of the sensors U1', U1 ''. An addition device AD1 therefore forms a signal pair u which consists of the two individual signals u '= u1' + u2 'u "= u1' '+ u2' ', for which it is therefore possible to use u' = cos (m. Ε) , u '' = sin (m. ε).

In the same way, a corresponding compensation sensor pair V2, which is arranged on the side of the reference track RS and is assigned to the reference sensor pair VI, supplies the individual signals v2 ', V2' ', and an addition device AD2 supplies the code signal pair v with the individual signals v '= v1' + v2 '= cos ((m-1) ε) v' '= v1 "+ v2' '= cos ((m-1) ε).

A vector rotator VD- delivers the relationships e '= u'.v' + u1 'accordingly. v '' = cos ε e '' = -u '.v' '+ u ", v' = sin ε. 7 claims 7 figures

Claims

1. A method for determining the relative distance between two mutually movable parts, wherein by scanning two marking systems, two signals are generated as periodic functions of the relative distance with period lengths in the ratio m / (ml), characterized in that continuous signals of the current relative distance are generated as signals and the relative distance from these two signals is determined as a function of the instantaneous phase difference of these signals.

2. Position encoder for determining the relative distance between two moving parts of a machine, with

a) a first measuring system coupled to the movable parts, which contains a code mark system and a code detector movable relative to it for generating a code signal which is periodic as a function of the relative distance,

b) a second measuring system coupled to the movable parts, which contains a reference mark system and a reference detector movable relative thereto for generating a periodic reference signal as a function of the relative distance, the period lengths of the code signal and the reference signal differing, and

c) an evaluation device connected to the reference detector and the code detector for forming a distance signal which is a function of the relative distance within the intended measuring range X,

characterized in that d) the code mark system contains a code track extending along the direction of movement from a multiplicity (at least, however, m = 3) of equidistantly arranged code marks, by scanning the code detector generating a continuous function of the current relative distance with the period length X / m,

e) the reference mark system contains a reference track extending along the direction of movement from another plurality of equidistantly arranged reference marks, the scanning of which causes the code detector to produce a continuous function of the current relative distance with the period length X / (m-1), and

f) the evaluation circuit contains a difference generator for generating a function of the phase difference of the reference signal and the code signal which is reversibly clear at any time in the entire measuring range X.

3. Position encoder according to claim 2 for determining the rotary movement of a rotary machine within the measuring range X = 2 / p, characterized in that the reference track is closed in a circle and R contains equidistant reference marks, the code track is closed in a circle and M = (R + p) contains equidistant code marks.

4. Position sensor according to claim 2 or 3, d a d u r c h g e k e n n e e c h n e t that the two measuring systems are electromagnetic measuring systems with brands that are attached to the circumference of two rigidly coupled disks.

5. Position encoder according to one of claims 2 to 4, characterized in that each measuring system has as detectors a pair of sensors which phase pair a pair by 90º set, approximately sinusoidal sensor signals and that the differeπzbildner delivers as a distance signal a pair by 90º phaseπverstellt, depending on the phase difference sinusoidal individual signals.

6. Position sensor according to one of claims 2 to 5, g e k e n n z e i c h n e t d u r c h an electronic adjustment device, which is entered a reference phase angle and which forms the difference between the phase difference and the reference phase angle.

7. Position sensor according to claim 5, a threshold value controlled by the sign of the one individual signal, to which the other individual signal and a reference value are supplied.