WO2005073673A2

WO2005073673A2 - Displacement sensor and torque sensor

Info

Publication number: WO2005073673A2
Application number: PCT/GB2005/000333
Authority: WO
Inventors: Robert Andrew Pinnock; Roger John Haelden; Gavin Stuart Brown
Original assignee: Trw Limited
Priority date: 2004-01-30
Filing date: 2005-01-31
Publication date: 2005-08-11
Also published as: GB0402013D0; EP1709409A2; WO2005073673A3

Abstract

A torque sensor comprises a light source, an array of radiation detectors, and first and second modulating elements, each having at least one displacement encoding track of alternating first and second modulating regions having different optical characteristics. A resilient connecting means connects the first modulating element to the second modulating elernent. The rnodulating elements carry tracks which modulate the light whereby an optical image pattern is formed on the detectors. The sensor determines both the relative position of the elements and also the absolute position of at least one of the elements by cornbining information from a variable pattern (typically a pattern of changing intensity) produced from a pair of displacement sensing tracks which move relative to one another and an additional pattern - preferably a simple fixed pattern - from an additional encoding track.

Description

DISPLACEMENT SENSOR AND TORQUE SENSOR

This invention relates to an improved displacement sensor and to a torque sensor which incorporates such a displacement sensor.

One known type of displacement sensor is the rotary displacement sensor which measures the relative angular displacement between two rotary elements that are constrained to rotate about a common axis. By providing a link between the two elements, such as a torsion bar, it is possible for such a sensor to provide a measure of torque. The more torque applied the more the torsion bar will twist and so the displacement between the elements can be used to indicate the amount of torque.

An example of such a rotary sensor is known from EP1001256A1, which represents the closest prior art for the present invention. An optical displacement sensor is disclosed which comprises a first and second coaxial disks mounted on to an input and an output shaft respectively. The input and output shafts are connected by a torsion bar and the shafts and two disks are arranged about a common axis of rotation. Each of the two disks has a respective annular track of alternating spokes and slots formed therein, and the disks are sandwiched between a light source and a detector array such that the tracks face are aligned with and face one another. A pattern of light modulated by the tracks is formed on the detector array. The pattern always includes at least two edges from one slot on each disk and the slots on one disk are chosen to be of a different width from those on the other to enable them to be distinguished in the pattern. A processor determines the relative position of the disks from the pattern of light formed on the detector array by detecting the relative position of the edges of a slot from each of the disks. The applicant has appreciated that in some instances it would be advantageous to know the angular position of the elements relative to a fixed datum as well as the relative movement between them. In the case of a steering system, for example, this can be used to indicate the direction in which the wheels are pointing on a vehicle. Such valuable information could be combined with the torque measurement as part of a steering control strategy. .

It has therefore been proposed that in addition to the arrangement of the slots that provides a torque signal, an extra index slot is provided in one of the tracks which is of a different width to the other slots. The index slot replaces one of. the normal slots in one of the tracks and the position of the index slot relative to a datum on the shaft is preset. Typically this datum will be chosen to correspond to the straight ahead position of the steering system. This slot, when detected in the pattern of light incident upon the detector array, provides a measurement of the shaft position at that one point, indicating that the steering is pointing straight ahead. Once detected, knowledge of the actual position of the shaft is maintained by counting the spokes passing through the pattern on the detector array as the elements rotate.

A drawback with such a system is that if the index spoke is not apparent in the pattern of light on the detector array when the system is switched on it may be some time before the elements are rotated far enough for the index spoke to pass the detector array. Until such time no position information is available. Clearly it would be advantageous to provide position information at start up for all positions of the shaft.

An alternative solution to the problem could be found by providing a separate position sensor which is fixed alongside the torque sensor. It is widely known, for example, to provide a rotary encoder which is a sensor for the measurement of the angular position of a rotary member, such as a shaft. These typically comprise a set of tracks of encoding patterns that are detected as the carrier rotates. This is not a favourable solution however. Having more than one sensor increases the size and complexity of the system. It requires extra space which in many applications is not available. This is especially true in an application to measure the torque carried by and position of a steering column shaft of limited length.

According to a first aspect the invention provides a torque sensor comprising a source of optical radiation, an array of radiation detectors, first and second modulating elements each having at least one displacement encoding track of alternating first and second modulating regions, each of the first and second regions having different optical characteristics, a resilient connecting means connecting the first modulating element to the second modulating element which is so constructed and arranged to provide relative displacement between them when a torque is applied and in which the modulating elements are positioned in an optical path from the source to the detector array to modulate the light whereby an optical image pattern corresponding to at least a portion of the displacement encoding tracks is formed on the detectors, and processing means which is adapted to receive first signals from the detector array representing the pattern of radiation falling on each detector of the array and operating to determine the relative positions of the first and second elements by detecting the relative position of features in the image on the detector array caused by the second regions of the first and second elements, the device being further characterised by comprising: at least one additional detector; at least one additional position encoding track comprising a pattern of first and second regions which encodes positional information, the modulating regions having different optical characteristics, the additional encoding track being positioned in an optical path from the source to the detector to modulate the radiation from the source which falls upon the additional detector; and in which the processing means is also adapted to receive at least one second signal from the additional detector indicative of the intensity of the radiation falling on the or each additional detector, the processing means being adapted to determine the angular position of at least one of the elements relative to a known datum from positional information obtained from both the first and second signals.

The invention therefore provides a single sensor, which determines both the relative position of the elements and also the absolute position of at least one of the elements by combining information from a variable pattern (typically a pattern of changing intensity) produced from a pair of displacement sensing tracks which move relative to one another and an additional pattern- preferably a simple fixed pattern- from an additional encoding track. The applicant has appreciated that although the pattern from the relative position encoding tracks does vary, provide the range of relative movement is suitably limited the overall sequence of modulating regions defined by both relative position encoding tracks can be maintained which enables the pattern to be used to encode some, position information. This "dual use" of the relative position encoding tracks reduces the amount of additional information needed to give a measure of absolute position.

By absolute position measurement we mean a measurement of position which is unique for the range of movement of the elements and as such does not repeat, allowing position to be unambiguously determined.

The elements may comprise rotary elements which are constrained to rotate relative to one another about a common axis. It is most preferred that the elements are disks which abut one another and which may be sandwiched between the source and the detector array/additional detector (s). Where they are rotary elements the sensor will provide a measurement of the absolute angular position of at least one of the elements. This can be readily adapted to measure torque by providing a torsion bar or the like to connect the first and second elements.

The resilient connecting means may comprise a torsion bar or quill shaft.

The track of relative displacement encoding regions on each element which modulates the light falling on the detector array may comprise a set of alternating spokes and slots, the spokes corresponding to the first modulating regions and the slots the second modulating regions. The slots on the track of the first element may all differ in size from those of the second disk such that the processor, by identifying a slot forming part of the pattern on the detector array, can identify which element the slot belongs to. Of course, both tracks could carry slots of the same size although the ability to distinguish one element from another would then be lost.

The tracks of the first and second elements may comprise first regions (or second regions) of three or more different widths which are arranged around the tracks to form an alternating pattern of region widths. For example, with three different widths of first region A, B and C the alternating pattern A,B,C,B,A can be provided (or some other pattern) which can be used to provide at least some angular position information which may be combined with the angular position information that is provided by the additional track(s) to give an absolute measurement. The more different widths of first region the more complex a pattern can be provided. Depending on the total number of first regions making up each tracks (which should be the same for both tracks) the pattern will be repeated several times. This pattern may be identified in the pattern of light falling on the detector array to provide some position information. Since the pattern repeats over the complete range of movement, however, this will not be absolute position information. The ambiguity can be removed by the additional position encoding track(s) .

The processing means may determine the relative displacement of the elements from the first signals by identifying the position of the first regions in the pattern of light falling on the array and from this determining the relative position of the two elements.

Preferably one element has a relative position-encoding track which has only first regions of a single width and the other a track of first regions of two different widths. Another arrangement which permits the two discs to be distinguished one from the other using only a pair of adjacent spokes. Hence, an alternative is to have (for example) spoke widths A and C on the first disc, and spoke widths B and C on the second disc, where A,B and C are different widths. Then, any combination such as

BC will enable the discs to be distinguished. It is only necessary to avoid cases such as CC which will not allow the discs to be distinguished. This is described later on in this document when the specific sensor embodiment is discussed. In actual fact, the requirement that a single pair of spokes must be able to distinguish the discs becomes irrelevant if absolute position encoding is achieved, since it will be possible to distinguish the discs from any unique code pattern.

To permit a spoke from each disk to be identified-a necessary requirement if relative position is to be measured- at least five edges (transitions from a first region to a second region) must be visible in the pattern of light on the array at all times. Also, the first regions should be arranged such that two first regions of the same width cannot appear next to one another in the pattern on the detector array. This can be achieved by ensuring that combination AA.BB, CC are avoided and by limiting the maximum relative range of movement of the two elements.

For true absolute position information a resolution of one first modulating region (or second modulating region) from the position-encoding track should be attained. This can be achieved by combining the limited position information that can be obtained from the first signals with additional information from the second signal. The second signal need not in itself provide enough information for absolute position to be determined. As will become clear, it is preferred that it does not as the relaxation of this limitation reduces the complexity of the device.

The additional encoding track may be provided on the first element or the second element, or on a third element which is fixed against movement relative to the first or second elements. Where the elements are disks for example it may be formed on a third disk which is sandwiched together with the first and second disks. More likely the extra code track will form an additional part of one of the existing pair of discs, as described later when the specific embodiment is discussed.

The additional track or tracks may define an encoded pattern of modulating regions which modulate the second signal in a way such that the combination of the second signal and the first signals provides the required position information. If each combination of first signals and second signal are unique over a complete revolution then absolute position information will be provided. If the pattern of first regions defined by the relative position encoding tracks repeats at least N times, then the additional track or track should provide at least N different codes with a different code used for each part of each pattern repeat. The precise code is defined by the arrangement of first and second regions around the additional track or tracks and the location and number of detectors associated with the track (s) . It will be chosen together with the pattern of modulating regions used for the position track.

Consider the case where the position track uses three different width spokes AB and C. This allows a total of 6 different spoke pair combinations to be provided which obey the rules that permit the processing means to identify which disk a spoke is on- meaning identical spokes can not be provided adjacent one another on the two elements. Six combinations allows a track having an integer multiple of six spokes to be provided with the pattern of spokes allowing some position information to be determined. Of course, if it repeats N times it is not possible from this to tell which of the N pattern repeats is modulating the light on the detector array. The additional track should therefore encode N or > N bits of information in order to provide the information needed to overcome this ambiguity and give an absolute position measurement.

Of course, so far we have assumed that due to manufacturing constraints the detector array only images one pair of spokes and so a unique code is needed from the additional track or tracks for each pair. If a larger array is used (or narrower spokes) three or more may be imaged at a time. In such a case, a unique code need only be provided for each triplet (or more for more spokes) which would alter the number of bits needed from the additional tracks for a given sequence of spokes. The device may include more than one detector which receives light modulated by the additional track, the detectors being spaced around the additional track. All of the detectors of the apparatus may comprise different parts of a single detector device for example. It is preferred that they are spaced by the same amount as the spacing from centre to centre of one spoke on the encoding tracks. Providing three detectors would provide three bits of information for any angular position of the elements, which allows for six unique codes. This would satisfy the need to have greater than or equal to six codes needed for absolute position encoding with three modulating region widths.

It is preferred that the transition from one code to the next on the encoding track does not coincide with the transition between spokes on the array. This further constraint in fact doubles the number of codes needed. In the case of a three spoke arrangement a code of at least four bits would therefore be needed. The provision of four extra detectors rather than three would permit this.

A potential ambiguity may arise in such a scheme if more than one bit of the pseudo-random code switches at once. Consider the case where, for example, a 4-bit pseudorandom code sequence goes from 1100 to 1001.

In this sequence, the first bit remains as a 1, the second bit changes from

1 to 0, the third bit remains as 0, and the fourth bit changes from 0 to 1.

Whilst in theory we may consider the transmutation from one 4-bit code to the next as happening instantaneously, in practise (as a result of positioning tolerances, switching sensitivity differences, and so on) either the second bit or the fourth bit will switch a fraction before the other.

Hence, we might in practise get the sequence 1100 - 1000 - 1001 (if the second bit switches a fraction before the fourth bit), or, alternatively, 1100 - 1 101 - 1001 (if the fourth bit switches a fraction before the second bit) . Hence, the potential exists for reading a code sequence such as 1000 or 1101 (the "interstitial" codes in the above example) in the wrong place.

The use of an additional "timing" track allows us to avoid this problem by providing an additional switching signal "well away" from the point in time at which the bits in the main track might be changing. The only potential ambiguity which then remains is that we might be unlucky enough to switch the sensor on when it happens to be positioned at one of the interstitial positions in a transmutation from one true code to the next. Hence, at the moment of switch on, we still cannot be absolutely sure that any particular code sequence (such as 1101) is really the true 1101 position code rather than a transient interstitial code, until we have rotated the sensor enough to pass through one transition of the timing code. This is only a small movement (depending on the resolution of the timing code sequence), but if this is still a problem, we can resolve it by having a complete second pseudorandom code sequence (phase-shifted with respect to the first one) , or by using a an alternative encoding scheme, such as the Gray code scheme, which only switches one bit at a time.

In an alternative, therefore, an additional timing track may be provided which modulates light falling upon a still further detector. The timing track may comprise alternating regions of different optical properties, which cause the output from the further detector to alternate from one state to another as the elements rotate. The timing of the change in state of the further detector should be chosen to be in phase quadrature or even 180 degrees out of alignment with a change in state of the second detectors, the value of the second signals then only being read at the position at which the timing signal has changed. The provision of the pseudo-code track together with the encoding track provides a simple, robust device, which permits both relative position and absolute position to be determined. Since only two tracks are needed then the device is compact.

In an alternative, more than one additional track may be provided which allows a Gray code sequence to be used. In this case, rather than having a number of detectors on a single additional track to give a multi-bit code, one detector can be provided on each additional track with each additional track encoding a single bit of the code. A feature of a Gray code is that only one bit changes on each transition between successive codes; hence any ambiguity that may arise from reading a code at a transition is reduced to two adjacent codes.

Of course, if more than one additional track is provided other forms of coding could be used instead of a Gray code. As with the use of a single additional track the code can be chosen such that when combined with the information from the first signals absolute position can be determined.

It will be appreciated that absolute position sensing need not always provide a measure of position which has the resolution of one modulating region of the relative position encoding tracks. Indeed, any device in which some position information can be gleaned from the tracks encoding relative position together with additional information that enhances the position measurement from at least one additional track may fall within the scope of at least some embodiments of the invention. For example, it may suffice to know the position to an accuracy of say 60 degrees, or 30 degrees or perhaps half a revolution. In this case, an additional track or tracks which define a far simpler code can be provided. In a relatively simple arrangement, a single additional track may be provided which defines a rolling code. A plurality of detectors may be spaced around this additional track to provide a multi-bit second signal with one detector defining each bit.

The first and second elements may be secured to a shaft to measure the position and torque carried by the shaft. They may, for example, be fixed relative to opposite ends of a portion of shaft of reduced section. The reduced section permits the shaft to twist more for a given torque, and hence produce a greater relative deflection between the two elements.

The elements may be sandwiched between the source and the detector array and additional detectors. The light from the source may pass through the elements, which modulate the light providing a transmissive device. In this case the elements preferably comprise a pair of disks. Alternatively, the elements may reflect the light from the source onto the detector array and additional detectors, which provides a reflective device. In this case it is most preferred that the elements comprise cylinders with the tracks provided around the perimeter of the cylinders.

The first modulating regions may have a higher transmission or reflectance than the second regions to give them their different optical characteristics, the regions of lower transmission or lower reflectance defining a spoke as described hereinbefore. Also, more than two types of modulating region could be used. Whereas so far only two regions (first and second are described which are typically slots or spokes that give the maximum contrast in the pattern formed on the detector array) more than two different regions could be used. All that is needed is that the pattern allows the different regions to be distinguished from one another. It will also be understood that the use of the terms first and second regions in relation to the relative position encoding tracks need not imply that the first or second regions of the additional tracks have the same or similar properties. In its broadest sense it simply means that each track is made up from at least two regions with different optical properties.

According to a second aspect provides a torque sensor which includes: a first modulating element and a second modulating element, the first and second elements both having at least one track of modulating regions which is associated with a corresponding track on the other element to define a pair of tracks; the pair of tracks together defining a repeated pattern of modulating regions; and in which the sensor further includes a further track of modulating elements which is fixed relative to one of the elements and comprises a pattern of modulating regions in which each part of the pattern when combined with the pattern defined by the first pair of tracks is unique for each pattern repeat of the pair of first tracks.

The two elements may comprise rotary elements such as a pair of disks which may be adapted to rotate relative to one another in use about a common axis.

In use the combined pattern formed by the pair of first tracks will change with relative movement and so can be observed to provide a measure of the relative angular movement of the elements. Some bits of position information can also be obtained by observing which part of the pattern repeat can be seen. This is not absolute position information because it repeats. However the pattern from the second track can then be used to tell which repeat is being observed and thereby provides absolute position information. Preferably the pattern from the additional track or tracks is fixed in order to make the determination of position simpler. Depending on how the elements are connected together the displacement of the elements indicated from the changing pattern of the first tracks could provide a measure of torque.

The elements may comprise disks which are positioned in register with one another like a stack of placemats, or may comprise cylinders which are arranged concentrically with one "nested" inside the other.

The tracks may each comprise a set of first and second regions which respectively may define spokes with slots there between. The slots may be generally arcuate with the tracks following a circumference of the disk or cylinder.

The first track of the first element may comprise a set of spokes of a single width. The first track of the second element may comprise spokes of the same width as the first track of the first element, or may comprise spokes which are all equal but of a different width to the spokes of the first element.

It is most preferred that the spokes of the first track of the first disk or the second disk are of at least two different widths. The pattern of overlapping spokes in the first tracks on the two elements should then be chosen such that when a portion of the first tracks is viewed together with the pattern of a portion of the regions of the second track of the first element a unique combination of patterns is visible over the complete range of movement of the elements.

By viewed portions we mean viewing two portions of the elements through windows of fixed size and relative position. The first element may carry a third track of modulating regions. Alternatively this may be provided on a part of the second element which is not in register with the first or second tracks on the first element. This third track if provided may include a pattern of spokes which when viewed through a window with the view of the first and second tracks (seen by a detector array of limited size for example) provides a unique pattern. This would allow the view of the first and second tracks to be other than unique on their own, and in effect enables a more complex and higher resolution pattern to be provided.

Of course more tracks could be provided in the same way.

There will now be described, by way of example only, several embodiments of the present invention with reference to the accompanying drawings of which:

Figure 1 is a schematic diagram of an example of a torque sensor according to the present invention; Figure 2 is a part plan view of a first modulating element;

Figure 3 is a part plan view of a second alternative modulating element; Figure 4 is a plan view of a first arrangement comprising two elements of the same type;

Figure 5 is a plan view of a second arrangement comprising two elements of different type; Figure 6 is a plan view of a complete modulating disk which includes an additional pseudo-random track that provides position information; Figure 7 is a plan view of an alternative modulating disk which includes an additional pseudo-random track that provides position information and also a timing track;

Figure 8 is a plan view of a still further alternative modulating disk which includes two additional pseudo-random tracks that provide position information, both additional tracks being identical yet displaced angularly relative to one another;

Figure 9 is a plan view of a different modulating disk which includes a set of concentric additional tracks that define a Gray code providing position information; and

Figure 10 is a final alternative modulating disk which includes an additional low-resolution track for identifying the location of the position of the disk to within a 60-degree sector.

The sensor shown in Figure 1 provides both a measurement of the torque carried by a rotating member, such as a steering column shaft, as well as an absolute measurement of the angular position of the shaft over a complete revolution.

As shown in Figure 1, an input shaft 2 is coupled to an output shaft 4 via a torsion bar of reduced cross sectional area. The torsion bar 6 is coaxial with the input and output shafts 1,2. The torsion bar 6 extends within a recess formed in the output shaft. The torsion bar would be shielded from view but in Figure 1 is illustrated by a broken line so as to illustrate the internal structure of the arrangement. First and second elements 10,12, in this case disks, are carried by the input and output shafts 1,2 respectively.

The disks 10,12 are closely spaced and have a plurality of slots 16,18 in them, which define annular tracks. The slots define first modulating regions and the lands between them define second modulating regions, which are herein referred to as spokes. The tracks of one disk are arranged to overlap the tracks of the other disk. The disks are provided between a light source 20 (which may emit visible or ultraviolet or infrared light) and an array of photodetectors 22 which feed signals to a processing means such as a microprocessor 24. The photodetectors are sensitive to at least some of the wavelengths of radiation emitted by the source. Because the disks are between the source and the detector array the tracks can be positioned such that they modulate light that passes through the slots defined by the tracks to form a light pattern on the detector array.

The light source may be a point source such as a light emitting diode (LED) or may be an array of point sources. A diffuser may be provided for diffusing the light from the light source. The detector array may be a linear array of detectors such as photodiodes, preferably provided in a single integrated package to ensure a fixed spatial relationship between the detectors throughout the life of the device. They maybe contiguous to produce a set of output signals which form an image of the intensity of radiation upon the detector array. A suitable array is provided by the

Texas TSL213 or TSL401 devices which incorporate 64 (or 128 in some cases) light sensitive elements and appropriate readout circuitry to read their outputs in a sequential manner. An alternative would be a CCD array although this is not preferred as it would be more expensive. The pattern of radiation intensity formed on the detector array corresponds to the arrangement of spokes, which overlap the array at any given position of the disks. This pattern can vary as the disks rotate together upon rotation of the input and output shafts. It can also vary when a torque is applied across the torsion bar, which causes relative movement between the two disks and their tracks.

Figure 2 shows the slots and spokes on a portion of a first design of disk envisaged in this invention. Regularly spaced slots are formed in the disk. The slots are shown as being relatively thin and extending in an arc following a circumference of the disk. However other slot shapes can be used. When viewed from the axis of the disk each slot subtends an angle α and each spoke between the slots an angle β .

The first and second disks may have the same track provided thereon. Figure 4 shows a device which uses two disks with tracks of the same design. In a preferred arrangement the track on the first disk is provided with a set of spokes of different widths from the track on the second disk such that one disk has a track of narrow width spokes and the other has a track of wide spokes. The angular spacing between the centres of the spokes on each track is the same. Figure 3 shows an alternative disk track which has slots that subtend an angle γ and spokes which subtend an angle φ. Figure 5 shows a device which uses the different design tracks of Figures 2 and 3.

A further description of the two tracks and the manner in which the pattern falling on a photodetector can be used to measure the relative movement between the two disks can be found in EP1001256A1, the teaching of which is incorporated herein by reference in its entirety. Although in theory any number of spokes can be used in the tracks, in this example 38 spokes are used to define each track. This number is chosen to suit the type of array used.

When designing a set of spokes for a track in a device of this type the followings rule should be applied that at all times at least 5 edges should be visible in the pattern formed on the detector array to allow two spokes (one from each disk) to be identified at any given time. Furthermore, although not essential, it is desirable to provide each disk with different width spokes to enable the position of each disk to be identified.

In addition to the narrow and the wide spokes, one of the tracks also includes a set of spokes of a third, different width. In this case these spokes are of a width intermediate the narrow and wide spokes. The narrow, intermediate and wide spokes may be denoted A, B and C respectively. The wide spokes may be twice the width of the narrow spokes, and the intermediate spokes 1.5 times the width of the narrow spokes. Together these three types of spoke permit six different spoke pairings to be provided (ignoring matching pairs of identical spokes) . The combinations are as follows: AB BA AC CA BC CB.

Matching pairs AA BB and CC should be avoided as this breaks the rule that the position of each disk can be separately identified.

These six pairs can be arranged in a repeating sequence such as ABCBACABCBAC. which repeatedly cycles through all the allowed spoke pair combinations. The sequence in effect provides some positional information if the width of a pair of spokes can be read from the pattern of light on the detector array. By arranging for the total number of slots making up the track to be an exact multiple of the length of the sequence it can be repeated an integral number of times. If not a partial sequence must be used at one point on the track. For a 38-spoke arrangement the following sequence can be used:

ABCBACABCBACABCBACABCBACACABCBACABCBAC

The pattern is a repeating set of 6 ABCBAC patterns with an additional AC and CA pair to bring the total to 38.

In the arrangement of the embodiment in accordance with an aspect of the present invention the index spoke does not therefore appear only once as in the prior torque sensors but instead appears many times around the track to provide some "bits" of position information which can be read at regular intervals as the shaft rotates. The spokes are arranged so that one disk carries only spokes A and C and the other only B and C. The C spokes can then be arranged in such a way that a C spoke only occurs in combination with an A or a B spoke, thereby allowing identification of the discs from the spokes.

The bits of information provided by the tracks on the disks do not, however, provide an absolute position measurement as the pattern repeats within a complete revolution. To overcome this constraint one of the disks is also provided with at least one additional track of modulating regions. The additional track or tracks are also positioned to modulate light from the source (or a separate source) onto at least one additional detector. The modulated light that passes through the additional track or tracks causes the output of the additional detector (s) to vary and the information contained in this varying signal can be combined with the pattern of light that passes through the torque-sensing track to provide information from which the angular position of the shaft can be determined.

It is important to note that the modulated signal produced by the additional tracks does not on its own provide enough information to determine the absolute angular position to within a resolution of one spoke. The information from these tracks must be combined with information from the torque sensor track to do so. This offers the advantage that the number of additional tracks used overall is reduced, which in turn reduces the size and complexity of the device.

The set of tracks for the angular position/torque sensor could have many forms. Several examples of different arrangements are described herein below which have been designed around the constraint that the relative position encoding tracks use three different spoke widths and have 38 spokes around the disk. For other spoke numbers different additional tracks will be needed although the skilled man will readily understand how to modify the examples from this teaching to provide solutions for different numbers of spokes.

When designing a suitable additional code then in principle, for 3 widths of spoke there are 6 unique pairs (ignoring pairs of the same spoke width such as CC) , and so with an additional 3-bit pattern a total of 48 unique position codes could be generated. This would be enough to give a different code to each of the 38 spokes in the main tracks by using one of the 8 possible codes for each of the 6 ABCBAC sub patterns. However, for ease of manufacture it is preferred that the transition from one pseudo code value to the next should not co-incide with a change in torque spoke pair since this would place a huge burden on the manufacturing process to ensure perfect alignment. It is therefore needed that each spoke pair corresponds to two pseudo- values. This effectively doubles the number of values needed from a pseudo-code.

In a preferred arrangement with 38 spokes in the torque track a four-bit code is needed. The code is said to be 4 bit since four "bits" must be read to form the code, giving a total of 16 times 6 = 96 possible codes when combined with the main track.

The coding of a four bit the additional track is relatively straightforward where an integer multiple of 16 (e.g. 32 or 48) . In the present example this is not the case which makes the process of coding more difficult and the resulting pattern of regions more complex.

To form the pattern, the 16 codes that can be provided are initially assigned a number from 1 to 16. These are then arranged in terms of the 38-spoke pattern given above so as to ensure that no additional code pattern repeats with its association with any spoke pair. An example of such a code arrangement used for the 38-spoke torque sensor track is given in Table 1 below (where two additional code numbers are associated with each spoke pair to take account of switching requirements and the spoke pair sequence wraps around from row to row to give the complete sequence) :

TABLE 1 Table 1 tries to represent the combination of position data from the two different "sources" - the spoke pairs (AB, BC, CB, etc.) and the 4-bit additional code. The 4-bit code can provide up to 16 unique codes (llll, 1110, 1100, etc.): for simplicity, these are represented with the numbers 1 to 16 in the table. Within each cell in the table, a pair of codes from the 4-bit sequence are shown (16 / 1, 1 / 2, etc.) This illustrates the fact that, for any spoke pair at any time, one of two of the codes from the 4- bit encoder may be associated with it (because the 4-bit code is arranged to switch when the spoke pattern is well away from a transition) . Thus, a sequence using the combined encoder patterns might be the following: AB/16 - AB/1 - BC/1 - BC/2 - CB/2 - CB/3 - etc. In order to ensure that we can resolve absolutely any position, it is necessary to ensure that no combined encoder pattern repeats (for example, the combined pattern AB/16 must not occur again until the sensor has rotated through an entire 360 ° rotation) . The table shows one way in which this can be achieved for a 38 spoke system. If you look down any of the columns labelled AB, BC, CB, etc. , you can see that in no case does any number 1 to 16 (representing the contribution of the 4-bit encoder) appear in more than one cell in that particular column. Hence, we can achieve absolute position measurement.

We now have to generate a pseudorandom (or whatever) code pattern which obeys both the sequential requirements of the table above, and also the "rules" associated with the particular type of encoding chosen. Taking the case of a pseudorandom bit sequence, there are only certain sequential code transitions which are possible for any particular pseudorandom code sequence. Take the following bit sequence: 111010. The first four bits give the code 1110; shifting one bit along gives 1101; the next one-bit shift gives 1010. It is obvious from the above that, for this particular sequence, we cannot go directly from 1110 to 1010: we have to pass through the code 1101. The sequence of "allowable" transitions is determined by the order of the individual bits in the complete sequence.

Referring to Table 1, it can be seen that code 6 sometimes has to transmute to code 7 (for example, in the fourth cell down in column CB), and sometimes to code 9 (for example, in the second cell down in column AB) . Similarly, code 7 has to transmute to both code 8, and also to code 16 (last cell in column CA). All the other code values (1, 2, 3, 4, 5, 8, 9, 10, 11 , 12, 13, 14, 15 and 16) only ever transmute to a single adjacent code (for example, code 1 only transmutes to code 2), as in a normal pseudorandom code sequence, but the "dual" transmutations put additional constraints on what codes 6 and 7 (and also codes 9 and 16) are allowed to be. If, for example, code 6 is 0010, then code 7 might be 0101 and code 9 0100, since code 6 has to transmute into either one of these values in the sequence set by the table above. There are no other possibilities for code 7 and code 9 (except that they may be interchanged - that is, code 7 could be 0100 and code 9 0101) : code 7 could not, for example, be 0111, because this is not a valid pseudorandom transition from code 6 0010. Hence there are pretty severe constraints put on the allowed values of (in this example) the codes 7 and 9; and there will be similar constraints put on the values for codes 8 and 16, since code 7 has to transmute into both of these (thus, if code 7 is, for example, 0101 , then code 8 could be 1010, and code 16 could be 1011: again, there are no other possibilities) . The aim, therefore, is to find a pseudorandom code sequence which fulfils the requirements of the table above in terms of the order of the code sequence, but which also complies with the "rules" of pseudorandom encoding. The following description is of solutions that employ either a pseudo-random code or a Gray code to solve this encoding problem.

PSEUDO RANDOM CODE

A first embodiment of an additional track of modulating regions that provide the code sequence of Table 1 is shown in Figure 6 of the accompanying drawings. This is a plan view of a disk 60, which shows the annular tracks of slots and inter-slot regions. Other embodiments are shown in Figures 7 to 10 of the accompanying drawings.

In this embodiment a single track 62 is provided which encodes the codes needed for the angular position information and not the torque information. The information from this track together with the information from the torque track provides absolute angular position information.

The additional track comprises a repeating sequence of slots and inter-slot spokes which define a pseudo-random code pattern. To provide the code pattern of Table 1 the pattern of regions defined in Table 2 can be used. Note that a one corresponds to a first modulating region such as a spoke and a zero to a second modulating region such as a slot between spokes.

TABLE 2

From Tables 1 and 2 a full 38 spoke sequence, based on the 16-bit code can be written as:

It is proposed that the same source that illuminates the tracks for the torque reading are used to illuminate the additional track. As the disks rotate the 4 bit code represented by the additional track is read sequentially, one bit at a time, by a single detector such as a photodiode. Thus, the disk must initially rotate through four "transitions" for the code initially to be acquired from the output of the detector. It is proposed that the light source that illuminates the torque track also illuminates the pseduo-code track, although this is not essential to all embodiments.

In a modification of this arrangement, four diodes could be provided which are spaced around the circumference of the pseudo-code track to enable all four bits of the code to be acquired at once rather than in sequence. These should ideally be placed under adjacent bits of the pseudocode. This does, of course, improve performance although the cost is an increase in the number of diodes needed.

It has also been appreciated that an ambiguity can arise in such a system when the code switches to its next value since more than one bit may be changing at once between successive codes. This can be overcome with one more track, which provides timing information and is read by a single photodiode. An embodiment of such a sensor which includes a disk 70 having such a timing track 72 is illustrated in Figure 7 of the accompanying drawings.

This timing track may be configured such that its switching edges are in phase quadrature (or even 180 degrees out of phase) with respect to the switching edges of the pseudocode track. The main code may then be read only when the timing track has just undergone a transition, thereby ensuring there is no ambiguity in the reading of the pseudocode track.

Some ambiguity may still remain even when a timing track is used, especially at the moment of switch on, since the main relative displacement coding tracks could potentially be precisely at a point of transition at that time. This ambiguity will essentially lead to a reduction in confidence in the validity of the position code until a very small turn of the sensor disks has removed the ambiguity. In this case, the ambiguity is purely associated with positional tolerances of the components-discs, detectors, etc- making up the position sensor.

Although the amount of rotation needed to overcome this ambiguity is small, in cases where it may still be considered to be a problem, then it is possible to remove it entirely by having, instead of the timing track, a second fully-coded track the same as the first pseudo-random code track but in phase quadrature or even 180 degrees out of phase. An example of an embodiment of this kind with such a "dual" pseudo-random track is shown in Figure 8 of the accompanying drawing. Both random tracks 82,84 are shown which are each read by two sets 86,88 of 4 photodiodes.

Gray code

In a further embodiment shown in Figure 9 of the accompanying drawings, a disk 90 carrying a Gray code sequence may be used instead of the Pseudo-random code. The Gray code is defined by a set of four additional tracks 92,94,96,98 arranged concentrically around the outside of the relative displacement track on one of the disks. The other disk, as before, is clear in the region of the additional tracks so as not to interfere with the pattern of light that passes through the additional tracks.

The Gray code tracks each comprise a pattern of first and second regions which provide spokes and slots. A single detector is associated with each track and produces a second output signal which is binary, the value depending on whether a slot or a spoke is positioned between the source and the detector. A feature of a Gray code is that only one bit of the pattern changes at once, such that the output from only one detector changes between any successive codes. Hence, any position ambiguity arising from a code being read at the moment of transition is limited to adjacent codes.

For the present example with 38 spokes, the Gray code has to obey the same constraints as the pseudo-code used in the embodiment of Figure 6. That is, that code 6 must not only sequentially precede code pattern 7 but also sequentially precede code pattern 9 also elsewhere on the track. Also, code pattern 7 must sequentially precede code pattern 8 and 16. An example of a suitable four bit code from four tracks and four additional detectors is given in Table 3 with the complete 38-bit code sequence as derived from this given in Table 4.

TABLE 3

TABLE 4

The disadvantage of this arrangement over the pseudo-code example is that more tracks are needed. However, if it is important that ambiguity is not present then this is a more economical solution in terms of detectors needed than the combination of a pseudo code track with a timing tracks (5 needed) or a dual pseudo code (8 needed) .

Rolling code

In some arrangements the angular position sensor may not need to provide absolute angular position with such a high resolution. In an automotive steering application it may, for example, be satisfactory to provide a device which indicates which of six 60-degree segments that shaft is positioned in. This can be provided using a simple rolling code for the additional angular position track. An example of such an arrangement is provided in Figure 10 of the accompanying drawings.

In this embodiment a disk 100 carrying a single additional track 110 which is read by three photodiodes 120,130,140 is provided to give a rolling code. The code sequence is given in Table 5:

TABLE 5

With this code, the basic sequence 111000 repeats, with a "one-bit" phase difference between successive code elements. A single additional code track, using three photodiodes separated by 60 degrees around the track can therefore be used to give an indication of absolute position to within a 60-degree sector.

Claims

1. A torque sensor comprising a source of optical radiation, an array of radiation detectors, first and second modulating elements each having at least one displacement encoding track of alternating first and second modulating regions, each of the first and second regions having different optical characteristics , a resilient connecting means connecting the first modulating element to the second modulating element which is so constructed and arranged to provide relative displacement between them when a torque is applied, and in which the modulating elements are positioned in an optical path from the source to the detector array to modulate the light whereby an optical image pattern corresponding to at least a portion of the displacement encoding tracks is formed on the detectors, and processing means which is adapted to receive first signals from the detector array representing the pattern of radiation falling on each detector of the array and operating to determine the relative positions of the first and second elements by detecting the relative position of features in the image on the detector array caused by the second regions of the first and second elements, the device being further characterised by comprising: at least one additional detector; at least one additional position encoding track comprising a pattern of first and second regions which encodes positional information, the modulating regions having different optical characteristics, the additional encoding track being positioned in an optical path from the source to the detector to modulate the radiation from the source which falls upon the additional detector; and in which the processing means is also adapted to receive at least one second signal from the additional detector indicative of the intensity of the radiation falling on the or each additional detector, the processing means being adapted to determine the angular position of at least one of the elements relative to a known datum from positional information obtained from both the first and second signals.

2. The torque sensor of claim 1 in which the elements are disks which abut one another and which are sandwiched between the source and the detector array/ additional detector(s) and in which the sensor provides a measurement of the absolute angular position of at least one of the elements.

3. The torque sensor of claim 1 or claim 2 in which each track of relative displacement encoding regions on each element which modulates the light falling on the detector array comprises a set of alternating spokes and slots, the spokes corresponding to the first modulating regions and the slots the second modulating regions.

4. The torque sensor of claim 3 in which the slots on the track of the first element all differ in size from those of the second disk such that the processor, by identifying a slot forming part of the pattern on the detector array, can identify which element the slot belongs to.

5. The torque sensor any preceding claim in which tracks of the first and second elements comprise first regions (or second regions) of three or more different widths which are arranged around the tracks to form an alternating pattern of region widths.

6. The torque sensor of any preceding claim in which the processing means determines the relative displacement of the elements from the first signals by identifying the position of the first regions in the pattern of light falling on the array and from this determines the relative position of the two elements.

7. The torque sensor of any preceding claim in which one element has a relative position-encoding track which has only first regions of a single width and the other a track of first regions of two different widths.

8. The torque sensor of any preceding claim in which at least five edges-transitions from a first region to a second region- are visible in the pattern of light on the array at all times and in which the first regions are arranged such that two first regions of the same width cannot appear next to one another in the pattern on the detector array.

9. The torque sensor of any preceding claim in which the additional encoding track is provided on the first element or the second element, or on a third element which is fixed against movement relative to the first or second elements.

10. The torque sensor of any preceding claim in which the additional track or tracks define an encoded pattern of modulating regions which modulate the second signal in a way such that the combination of the second signal and the first signals provides the required position information.

11. The torque sensor of any preceding claim in which the pattern of first regions defined by the relative position encoding tracks repeats at least N times, and the additional track or track provides at least N different codes with a different code used for each part of each pattern repeat.

12. The torque sensor of any preceding claim which includes more than one detector which receives light modulated by the additional track, the detectors being spaced around the additional track.

13. The torque sensor of any preceding claim in which the transition from one code to the next on the encoding track does not coincide with the transition between regions on the array.

14. The torque sensor of any preceding claim in which an additional timing track is provided which modulates light falling upon a still further detector.

15. The torque sensor of claim 14 in which a Gray code sequence is used.

16. The torque sensor of any one of claims 1 to 14 in which the additional track defines a rolling code.

17. A torque sensor which includes: a first modulating element and a second modulating element, the first and second elements both having at least one track of modulating regions which is associated with a corresponding track on the other element to define a pair of tracks; the pair of tracks together defining a repeated pattern of modulating regions; and in which the sensor further includes a further track of modulating elements which is fixed relative to one of the elements and comprises a pattern of modulating regions in which each part of the pattern when combined with the pattern defined by the first pair of tracks is unique for each pattern repeat of the pair of first tracks.

18. A torque sensor according to claim 17 in which the two elements a pair of disks which rotate relative to one another in use about a common axis.

19. A torque sensor according to claim 17 or claim 18 in which the tracks each comprise a set of first and second regions which respectively may define spokes with slots there between.

20. A torque sensor according to claim 19 in which the first track of the first element comprises a set of spokes of a single width and the first track of the second element comprises spokes of the same width as the first track of the first element.

21. A torque sensor according to any one of claims 17 to 20 in which the spokes of the first track of the first disk or the second disk are of at least two different widths and the pattern of overlapping spokes in the first tracks on the two elements are such that when a portion of the first tracks is viewed together with the pattern of a portion of the regions of the second track of the first element a unique combination of patterns is visible over the complete range of movement of the elements.