Method and device for measuring an angle

The invention relates to a method and a device for measuring an angle according to the preambles of claims 1 and 24, respectively.

The angle which two bodies rotatable relative to one another about an axis of rotation form is not initially clearly defined. It can be assumed, however, that in each specific case in which such an angle is to be measured, two radial beams proceeding from the common axis of rotation and each rigidly connected to one or the other body can be determined, the relative angular position of which is continuously monitored and to be measured. Depending on the type of angle measurement method or angle measurement device used, these two radial beams are then clearly determined by the positioning of sensors or reference marks.

Known angle measuring methods, such as those used for. B. DE-A-2 501 373, DE-A-2 649 898 or DE-B-2 237 138 can all be found in common that a marking carrier designed as a flat circular disk is rigid with one of the two bodies is connected so that the common axis of rotation runs through the center of the disk rotating with the body, while on the other of the two bodies a probe scanning this marking carrier is rigidly attached. In these arrangements, for example, one of the two radial beams mentioned above is defined by a "zero marking" attached to the marking carrier and distinguishable from the other markings by a special shape, while the second radial beam is defined by the measuring probe scanning the markings.

In order to be able to measure the angle enclosed by these two radial beams with a high resolution with respect to the axis of rotation, a large number of markings spaced apart from one another in the direction of rotation in the form of radially running lines, for example a line, are provided on the disk forming the marking carrier have different light transmission than the surrounding areas of the marking disc, so that the sensor comprising an electro-optical sensor emits an electrical signal whenever the brightness of a light beam striking the sensor and emanating from a light source arranged on the other side of the pane is changed by the passage of such a marking line. Thus, if one of the two bodies rotates against the other, a sequence of pulse-shaped electrical signals is generated, the evaluation of which provides information about the angular position that the two bodies assume after the rotation has ended.

From GB-A-1 498 829 a method and an apparatus of the type mentioned in the preamble of claim 1 and claim 24 are also known.

The measuring accuracy of all these known methods is determined to a decisive degree by the precision with which the scale marks are arranged on the marking carrier, the following individual factors being of particular importance within the general term "precision": the individual scale marks must be as narrow as possible in the direction of rotation and to one another have boundary edges running as precisely as possible in parallel; in addition, they must run as radially as possible and be as identical as possible to one another. Furthermore, the angular spacing of the scale lines, ie the angles enclosed by the scale lines with respect to the axis of rotation, must agree with the highest degree of accuracy. Since it is not enough if the marker carrier used in the known methods has these properties only at or immediately after its completion, very high requirements must also be placed on the dimensional stability and in particular on the temperature stability of the material used to produce such a marker carrier.

According to the prior art, the general procedure is therefore that special glass panes are used as marking carriers, into which the required number of markings, for example in the form of very precisely positioned and extremely narrow scale lines, is etched, for example, using very complex and high-cost processes. Typically, the line thickness is approximately equal to the line spacing and is often on the order of approximately 10 u. This results in the further problem of a very unfavorable signal-to-noise ratio and a considerable optical and / or electronic circuitry effort has to be made in order to extract the useful signals assigned to the individual markings from the strongly noisy sensor output signal for further processing. In addition, the maximum achievable measurement accuracy is adversely affected by this poor signal-to-noise ratio.

The high manufacturing costs resulting from the problems just mentioned, both for the marker carrier and for the evaluation electronics, mean that until now a very precise and high-resolution angle measurement could only be used in those cases where the use of a with such high manufacturing costs measuring device was economically justifiable.

In contrast, the invention has for its object to provide a method and an apparatus of the type mentioned that enable a highly accurate, if necessary high-resolution angle measurement using a simple, inexpensive to produce marking carrier.

To achieve this object, the invention provides the features set out in claims 1 (method) and 24 (device).

An essential difference of the method according to the invention from the prior art and in particular from GB-A-1 498 829 is that the markings arranged on the marking carrier are not treated anonymously but as individuals.

This already begins with the determination of the so-called measuring angle distance of at least two markings in a zero angle position of the two bodies selected in accordance with the respective measuring situation, which precedes the actual angle measurements. Since in this zero-angle position the measurement lines of gravity of the at least two sensors do not necessarily have to lie on one and the same radial beam, the measuring-angle distance which can be measured in this zero-angle position with the aid of the two sensors settles between the µth and the vth mark together from the angular distance, which enclose the two markings on the marking carrier with each other and the correct angular distance of the measurement centroid lines of the two sensors in the zero angular position. If the two measurement lines of gravity coincide in the zero angular position, the measuring angle distance of two markings µ and v is of course equal to the angular distance actually present on the marking carrier.

The measuring angle distance between two or more markings can be determined in different ways. A particularly preferred method for its determination is described below with reference to claim 7.

The just described distinction between angular distance and measuring angular distance makes it possible, if necessary, to define a separate zero point for each individual angular measurement, which can be selected independently of the zero point of the preceding or following angular measurements.

The only requirement is that for a selected zero angular position of the two bodies, the measuring angle distance defined above is determined for at least one pair of markings with an accuracy which corresponds to the accuracy of the time interval measurement used in the subsequent angle measurements; this at least one measuring angle distance is then stored in such a way that its association with the relevant pair of markings is retained. Is this e.g. B. for the m-th and the n-th marking, it is sufficient to determine the t of the two bodies at any time₁ included angle ϕ (t₁), the time interval Δθ_mn (t₁) to be measured between two signals, one of which is generated by the m-th mark passing one sensor and the other shortly thereafter by the n-th mark passing the other sensor. However, an essential prerequisite for this is that the time interval between the signals that are assigned to the two markings m and n is actually measured, for which the measuring angle distance_I.p_mn was determined and stored in a retrievable manner. Then it can be determined from the measured time interval Δθ_mn, the current angular velocity w (t₁), which is initially assumed to be known here, and the measuring angle distance _I.p_mn between the mth mark and the nth mark for the current angle ϕ (t₁) calculate a first value of the two bodies according to the following equation:<maths id="math0001" num=""><img file="EP0040359B1_D0001.tif" /></maths>

Since each sensor covers a large number of markings, the measuring angle distances are preferably determined for a large number of marking pairs with an accuracy corresponding to the accuracy of the time measurement used and are stored while maintaining the assignment to the respective marking pair. Then, by measuring the time intervals of these further markings, for example between the markings m + 1 and n + 1, m + 2 and n + 2, etc. a large number of further angle measurements according to the corresponding equations<maths id="math0002" num=""><img file="EP0040359B1_D0002.tif" /></maths><maths id="math0003" num=""><img file="EP0040359B1_D0003.tif" /></maths><maths id="math0004" num=""><img file="EP0040359B1_D0004.tif" /></maths>determine, the angular values ​​ϕ (t_r) must agree with each other if the two bodies have not moved against each other in the relevant period.

In contrast to the known methods, in which the individual markings are not treated as individuals but completely anonymously and therefore have to be applied to the marking carrier with the greatest possible precision at precisely defined angular intervals, that with a marker carrier produced at low cost, the angular spacing of the markings can fluctuate within relatively wide limits and the desired accuracy is achieved by accurately determining the measuring angle spacings of the markings of interest before the actual angular measurement, with reference to the markings treated as individuals and then, if these markings recognized as individuals during the angle measurement process have also been used for the mentioned time interval measurements, use the respectively associated stored measurement angle distance value in order to calculate the instantaneous angle between the two bodies according to the above equations.

Before the identification of the individual markings, which plays an important role in this context, is pointed out that the method according to the invention can be carried out regardless of whether the two scanning devices on the marking carrier scan the same markings or two marking groups spatially separated from one another. However, the second of these two cases is preferred, since in the first case the difficulty arises that the relative trajectories of the two scanning devices run very close to one another, so that there may be considerable space problems, especially when angles greater than 360 ° are to be measured . When using two spatially separated marker groups, it is necessary to consider the angle between the two groups in addition to the angular spacing of the individual markers within each group, but this only appears to be a disadvantage compared to the use of a single marker group.

If you do not want to make an extremely high adjustment effort that can hardly be maintained over a long period of time, then even if the two scanning devices scan the same marking group, the two tracks on which this occurs will not be identical to one another, but rather, for example, that Track of the one scanning device have a slightly smaller radius with respect to the axis of rotation than the track of the other scanning device. However, since marking carriers are to be used in which the markings, for example elongated, radially directed strips, have been applied without particularly high technical outlay, it must be assumed that these marking strips not only fluctuate in their angular spacing but also do not run exactly radially. In this case, the marker carrier will assume a slightly different angular position if one and the same marking generates a signal on the sensor of one or on the sensor of the other scanning device. The angular correlation between the signals generated on one sensor on the one hand and the signals generated on the other sensor on the other hand must also be established here, so that ultimately the same situation arises as when using two spatially separated marker groups. In the following, reference will therefore always be made only to the circumstances of the preferred case according to claim 2.

As explained in detail above, it is an essential aspect of the method according to the invention that the markings associated with the sensor signals used for measuring the distance between both the one and the other group are each treated and identified as individuals. The individual markings within the respective group are preferably identified with the aid of asymmetries present in this marking group.

These asymmetries can consist, for example, in that the individual markings differ from one another in terms of their length and / or width or by some other coding in such a way that they are recognizable as individuals. However, the coding effort required for this during the production of the marking carrier and also the decoding effort to be carried out with the aid of corresponding scanning devices during the actual angle measuring process is extremely great.

It is therefore preferred that with the help of asymmetries in each marking group only one marking is identified, which then serves as a zero marking in the identification of the remaining markings of this group by counting the associated sensor signals. In this case, the coding effort required is significantly lower, since it is sufficient, for example, to make the zero marking of each group somewhat longer or shorter than the other, essentially identical markings, and to sense this difference with the aid of a corresponding sensor.

According to a particularly preferred embodiment, the effort required for this can be saved in that for each of the marking groups the time intervals of all the sensor signals which follow one another during at least one revolution are measured and the angular distances of the markings which are not exactly equal and which are determined in the process are evaluated as asymmetry.

The asymmetries required for the identification of the individual markings thus result automatically from the fact that with the aid of a correspondingly precise time measurement method it is always possible to determine the angular distances between the markings, in particular between the respectively immediately successive markings, with such accuracy. that there are no two corresponding measured values ​​for a given angular velocity within the relevant marking group even if the marking carrier is produced with far greater precision with regard to the positioning of the individual markings than is desired and required in the process described here.

It is thus possible to recognize each of the markings within a group as an individual with the help of their angular distance from the immediately preceding and / or from the immediately following marking or also with the aid of the angular distances from two or more preceding markings, once all the angular distances all markings of this group that follow each other in the direction of rotation are measured and stored.

For example, if a certain function is to be triggered when the nth marking of a group passes the associated sensor, this triggering can take place when the first sensor signal occurs after the reappearance of the known angular distance that is only present once in this marking group between the (n-1) th and the (n-2) th mark has been recognized.

Since in the method according to the invention the angular distance between two markings is measured over the time interval between the associated sensor signals and the instantaneous angular velocity of the marking carrier is included in this measurement, it is expedient to identify the nth marking, for example the quotient of the angular distances between the (n-1 ) -th to the (n-2) -th and the (n-2) -th to (n-3) -th marking time intervals, since the instantaneous angular velocity drops out in this quotient formation if the rotational or angular velocity of the marking carrier is constant during the period composed of these two immediately successive time intervals; because of the shortness of this period, this condition will generally be met. Also, the probability that there are different marking sequences on a marking medium within a marking group, for which the quotients formed in the manner just described have the same values ​​as part of the very precise time measurement used, is again significantly less than the probability of the accidental occurrence of two exactly equal angular distances between successive markings.

Even with the particularly preferred identification method just described, it is not necessary to identify all markings individually. Instead, with the help of the existing asymmetries, only a single marking can be identified in each marking group, which then serves as a zero marking in the identification of the remaining markings by counting the associated sensor signals.

It should be expressly pointed out that, in the most general case, it is not necessary to define these zero markings from the outset and as such in the manufacture of the marking carrier. Rather, each of the markings present on the marking carrier can be picked out arbitrarily during the operation of a measuring device operating according to the method according to the invention and used as a zero marking for a subsequent period.

The selection of any marking as a zero marking is preferably carried out with the aid of the rule that the counter used for the periodic counting of the markings, which begins again with each revolution, should always be reset to zero by the marking which is the last of the three immediately successive markings, the two time intervals of which give the smallest or alternatively the largest quotient.

For the sake of simplicity, however, it is also possible to use asymmetries generated "intentionally" in the marking groups to identify at least the zero marking instead of or in addition to the "naturally" existing asymmetries. For example, it can be provided that one of the markings is defined as a zero marking in that its angular distance from one of its immediate neighboring markings differs substantially from the angular distances from all other immediately adjacent markings.

As the equations (1) shown above show, in order to determine the instantaneous angle Wink (t) between the two bodies according to the invention, it is not only necessary for the instantaneous distance measurement AO_", (t) identified markings ki and v as such, but the measuring angle distance winkelµ, ν of these markings must also be known. In principle, it is now possible to carefully measure a marker carrier produced without particularly great effort before using it in the measurement method according to the invention, and to store the individual measuring angle distances between the markings of one group and the markings of the other group in an electronic read-only memory and thus for later use to provide necessary arithmetic operations. However, this would mean that the exact radial course of the individual markings in each group and the long-term constancy of the entire arrangement geometry would have to be extremely demanding.

To avoid these difficulties, for a particularly preferred variant of the angle measuring method according to claim 7, that with the help of the angle measuring device itself for determining at least one measuring angle distance in the zero-angular position of the two bodies, the time interval between a signal caused on one sensor by an identified marking of the one group and one on the other sensor by an identified marking of the other group Signal measured and with the angular velocity of the marking carrier prevailing during the measurement is multiplied. This multiplication provides the measuring angle distance of these two markings belonging to the different groups and thus an angle connection between the two groups.

As already mentioned, this measurement and calculation does not record the true angular distance of the two markings under consideration on the marking carrier, but rather the measuring angle distance of these markings, in which the radial beams traversed between the two sensors at that time, by the device according to the invention, per se alone not measurable zero angle is received. This time interval measurement thus defines an arbitrary zero point of the angle measurement, which can be determined in any desired manner on the basis of the method described and can also be changed again as required. Do you z. B. the time interval measurement in question at a point in time at which it is ensured that the two radial beams passing through the sensors come to coincide exactly with one another, this position of the two bodies whose angular distance is to be measured is defined as the zero angular position. If, on the other hand, the zero angular position of the angle measurement is to be defined by aligning two predeterminable edges of the mutually rotatable bodies, which move towards or away from one another during the rotational movement of the bodies, then the connection described above will be used time interval measurement producing the two marking groups is then carried out when these two body edges are in the desired position.

The connection between the two marking groups relating to the zero point of the angle measurement can also be established by measuring the time intervals for several pairs of marks in the manner described and multiplying them by the instantaneous angular velocity in order to determine the associated measuring angle distances. If the mean value is formed from this large number of measurements, the measuring accuracy can be increased further.

Furthermore, it is advantageous, in accordance with claim 9, in a calibration run, which takes place at least once before the actual angle measurement processes, to first determine the angular distances between the respectively immediately adjacent markings within each marking group in a first step by changing the time intervals between these markings associated sensor signals at least during a full revolution of the marker carrier. The size of the time interval measurement values ​​obtained in this way naturally depends on the prevailing angular velocity or rotational speed of the marking carrier; they must therefore be multiplied by the value of this angular velocity in order to obtain the calibration values ​​required for the subsequent measurements, which are then stored in such a way that their assignment to the individual and re-identifiable markings is retained.

In order to achieve a high resolving power of the angle measurement, according to claim 9 using the angular distances known from the calibration run within the individual marking groups for a multiplicity of pairs of markings, one of which always belongs to one and the other to the other marking group calculates and stores the measuring angle distances related to the angle measurement defined in the manner described above, so that they are then immediately available for the actual angle measurements.

The angular velocity of the marking carrier, which is determined both when determining the measuring angle distances and during the calibration measurements, can be known, for example, by regulating the drive that causes the marking carrier to rotate continuously so that the speed of this rotation is kept constant at a predetermined value. In addition to the circuitry complexity required for this, this has the disadvantage that the calibration values ​​which are an integral part of the final angular measurement can only be determined with the accuracy with which the control device which keeps the rotational speed of the marking carrier works. This is not sufficient for many applications.

It is therefore preferably provided that the angle measuring arrangement itself determines the rotational speed of the marking carrier during the determination of the measuring angle distances or during the calibration runs according to the European patent 0 039 900 based on a simultaneously filed patent application, the content of which is hereby expressly included in the disclosure content of the present description is included.

This can either be done in accordance with claim 10 in that if each of the marking groups is scanned with only a single sensor, the above-described calibration run measurements are carried out over at least two full revolutions of the marking carrier. At least two rows of time interval measured values ​​are then obtained for each marking group, the same place number within each measuring row having measurements always coming from the same pair of markings. If all of the measured values ​​of the various measurement series, which correspond to one another in this way, now coincide within predetermined tolerance limits, it can be concluded that the rotational speed of the marking carrier was constant during the recording of these measurement series, ie for example during two revolutions. If, during these revolutions, at least for one marking, the time intervals of the signals are measured, which from this marking on the first and after the second passing a sensor, ie Thus, when the marking carrier has been rotated by exactly 360 °, an immediate measure is obtained for this constant rotational speed during the calibration run, with which the measured values ​​are then either the series of measurements obtained and temporarily stored during the first full rotation, or the series of measurements that are practically identical Measured values ​​of the second series of measurements can be multiplied to obtain the required calibration values.

The requirement to be met for this method that the angular velocity of the marker carrier must remain constant during at least two full rotations can be met without further ado, since even if the rotational speed of the free-running marker carrier which is not subject to any loads fluctuates frequently, it can be repeated in between There are periods in which no such fluctuations occur. If you want to shorten the time during which the angular velocity of the marking carrier must be constant in order to carry out a calibration run, it is possible, for example, to scan each marking group with two diametrically opposed sensors. In principle, the method runs exactly as described above, but you can already start here after half a revolution with the comparison of the two series of measurements created here at the various sensors, so that the entire calibration run can be completed within a full revolution.

Another possibility for checking the constancy of the angular velocity is that the time intervals of the signals belonging to the markings immediately adjacent to one another are not used for this purpose, but over two full revolutions of the marker carrier for each marker measures the time interval between the signal triggered by the first time this marker passes the sensor and the signal generated by the same marker the next time it passes this sensor and compares all these 360 ​​° time intervals with one another. If there is a match within the specified tolerance limits, one can conclude from this that the rotational speed of the marking carrier is constant during this period and the time intervals measured simultaneously for the adjacent markings can be used as calibration values ​​after a corresponding conversion.

The calibration measurements just described must be carried out at least once before the actual angle measurement operation can be started. In order not to have to make any special demands on the long-term constancy, in particular of the geometry of the measuring arrangement, it is preferred, however, that the calibration run be repeated for constant re-calibration. This can happen between the individual angle measuring processes.

Basically, however, it is not a particular difficulty to obtain measured values ​​practically simultaneously with the angle measurement processes, which can be used for constant re-calibration of the system. So it is z. B. possible to measure the time intervals between all immediately successive sensor signals with each revolution of the marking carrier and to compare them with the corresponding time measurements of the previous revolution. If it turns out that the recorded values ​​do not correspond to one another within predetermined tolerance limits, the old, stored calibration values ​​are initially retained and the deviation that occurs is interpreted as a sign of a momentary change within the measuring system. However, as soon as two successive revolutions of the marking carrier again lead to the same time interval measured values, ie This means that these time interval measured values ​​coincide with one another within predetermined tolerance limits, new calibration values ​​can be formed immediately and stored instead of the old calibration values. This enables the measuring arrangement to react very quickly to any drift phenomena that occur and to eliminate the influence of these drift phenomena.

In previous considerations, it was always assumed that the instantaneous angular velocity of the marking carrier is known and that its value is therefore available for the calculations required, for example, according to equations (1). As already mentioned, this can be achieved by regulating the drive device which drives the marking carrier for its constant rotation to a constant angular velocity, but this leads to the difficulties mentioned, in particular with regard to the accuracy of the instantaneous angular velocity value.

A particular advantage of the method described here can therefore be seen in the fact that the instantaneous angular velocity of the marking carrier can be measured extremely precisely not only during the calibration run but during each revolution with the aid of time interval measured values. If the absolute values ​​of the angular distances of the markings of at least one marking group are known, which is the case after the calibration run has been completed, then the instantaneous angular velocity or The rotational speed of the marking carrier can be determined by measuring the time intervals between the signals generated by these markings on the associated measuring sensor and comparing them with the stored calibration values ​​using the following equation:<maths id="math0005" num=""><img file="EP0040359B1_D0005.tif" /></maths>

Here w (t) mean the instantaneous angular velocity of the marking carrier, ΔT_{Eµ, µ + 1}the time interval determined during the calibration run between the two sensor signals belonging to the identified markings p and µ + 1, ΔT_{µ, µ + 1}(t) the currently measured time interval of the signals belonging to these markings and wE the angular velocity determined during the calibration run.

To determine the instantaneous angular velocity w (t) from the stored calibration values ​​and the instantaneous measured value ΔT_{µ, µ, + 1}(t), therefore, only a simple arithmetic operation is required, which can be carried out electronically without further ado and with little expenditure on equipment.

In the method just described for measuring the instantaneous angular velocity w (t) of the marking carrier, it is of particular importance that the measured values ​​in question always represent only an integration of the differential variable "angular velocity" over the very short period of time associated with the immediately successive markings used Sensor signals is. The mean value of the angular velocity over one or even several full revolutions of the marking carrier is therefore in no way formed, so that the requirement of a constant angular velocity over a long period of time during the calibration run is eliminated. The instantaneous angular velocity w (t) required for the determination of ϕ (t) can therefore be measured with the required accuracy even when the marking carrier is changing its rotational speed.

From the explanations given above, it can be seen that all the quantities required for the calculation of the instantaneous angle ϕ (t) according to equations (1) can be obtained on the basis of the described method by pure time interval measurements between sensor signals.

Since time intervals between two successive electrical signals can be determined electronically very precisely, for example with the aid of a quartz-controlled oscillator circuit, without high technical and costly expense, the angle ϕ (t) sought is obtained with the accuracy of the time measurement used and completely regardless of the accuracy or regularity, the markings scanned by the measuring sensors have been attached to the marking carrier. Compared to the known angle measuring method, the use of a markedly cheaper marker carrier can save considerable manufacturing costs, of which only a very small part is used up again due to the somewhat larger electronic measuring and evaluation effort.

In addition, for all parts of a measuring device working according to the described method, with the exception of the time measuring unit, special measures to ensure long-term constancy can be dispensed with, since, as already mentioned, such a measuring system automatically and automatically is re-calibrated and is therefore able to influence both To completely eliminate short-term fluctuations as well as long-term drift phenomena on the measurement result.

In principle, it is in principle possible to determine the instantaneous angle between the two bodies which can be rotated relative to one another, in order to determine the time interval between two identified markings, one of which belongs to one marking group and the other to the other marking group. However, the method variant according to claim 14 is preferred, which represents a special case of the measuring method described in connection with claim 1 and the equations (1), in that it is assumed that on the one hand there are two spatially separated marking groups, one of which is from the other one and the other is scanned by the other sensor, and that on the other hand the time intervals Δθ required by equation (1)_{µ, ν}(t) are not measured between any identified markings of the one and the other marking group, but rather in such a way that the time interval measurement in individual cases is initiated by the passage of any identified marking, identified here by the counting index μ, but then by the immediately next, am other sensors that occur, the corresponding signal identified by the counting index v and assigned to the other group is ended. The time intervals just mentioned are briefly referred to below as "mixed" time intervals in order to make it clear that they are measured between signals belonging to markings from different groups.

So that the time intervals Δθ_{µ, ν}(t) can be used in a meaningful way to calculate the instantaneous angle cp (t) between the two bodies, the angular velocity m (t) of the marking carrier prevailing during the respective measurement period must be constant in the sense that it is at least at the time , in which the marking p initiating the measurement passes its sensor has the same value as at the time in which the marking v terminating the measurement triggers the associated signal on the other sensor.

Since the instantaneous angular velocity w (t) cannot be determined as an infinitesimally small value at these times, but rather only as an average over the (albeit very small) period of time that passes until two immediately consecutive markings have passed a sensor, the constancy of the angular velocity in the above sense is preferably checked according to claim 15 in that its mean value is measured using one of the measurements of Δθ_{µ, ν}immediately preceding period, namely the time between the passing of the (µ-1) -th and the u-th marking, and over one of the measurement of Δθ_{µ, ν} Immediately following period, namely in the time between the passing of the (µ + 1) th and the (p + 2) th marking at the sensor concerned and comparing these two mean values ​​with each other. If they agree with one another within predefined tolerance limits, it can be concluded with great reliability that the angular velocity during the measurement of Δθ_{µ, ν} has not changed in the above sense.

Since a large number of “mixed” time intervals are preferably measured, for example in order to further increase the measuring accuracy by averaging the angle values ​​obtained in this way, it is expedient to use the angular velocity values, which are anyway measured continuously, for the check just described. If it appears that the angular velocity of the marking medium changes temporarily, the measured values ​​Δθµ, ν obtained during this period are not used to calculate the instantaneous angle between the two bodies, but it is waited until the angular velocity ro (t) of the marking medium is constant again. Since, as already mentioned, this constancy only has to be ensured for extremely short periods of time, namely only as long as four immediately consecutive markings have passed a sensor in order to obtain a further angle measurement, this will generally be the case very quickly .

Before the mean value is formed from several measured angle values, it must be checked whether a rotational movement between the two bodies or another disturbance influencing the measurement result has not occurred during the measurement period. This is achieved by comparing the individual measured angle values ​​with one another and only forming the mean value when they coincide with one another within predetermined tolerance limits.

At this point it should be pointed out that because of the free rotatability of the two bodies relative to one another, angular positions can easily occur in which the time interval to be measured between the signals coming from the two sensors becomes practically arbitrarily short. The time measurement is therefore carried out in such a way that such "near coincidences" or real coincidences are recorded and evaluated in the correct manner. This will be discussed in more detail below. For the resolution of the measurement of the angle ϕ currently enclosed by the two bodies or the radial rays mentioned, this means that this resolution is not determined by the angular distance between the markings adjacent to one another on the marking carrier, but solely by the very high resolution of the time interval measurement used.

A particularly important aspect of the method described here can be seen in the fact that an angle measuring device working according to it can be used as an extremely precisely operating speed sensor without particularly additional technical effort, in particular for very slow rotational speeds of up to a few thousandths of a Hertz and less. If, as described above, if a series of time interval measurements is carried out between the signals assigned to the markings belonging to different marking groups and these measured values ​​are compared with one another, then one which is occurring during this period can be determined. Determine the relative rotation between the two bodies of progressive change in these values ​​with very high accuracy and use this to calculate the angular velocity on which this relative rotation is based.

As an alternative to this, the measurements described above for determining the instantaneous rotational speed of the marking carrier can also be carried out on both sensors and the corresponding angular speed can be determined from a comparison of the different results obtained on the two sensors.

Furthermore, the two methods just described for measuring slow angular velocities can also be combined with one another.

From the statements made so far, it follows that the accuracy of the angle measurement method described depends solely on the accuracy with which the time intervals of interest are measured in detail. Therefore, this time interval measurement is of particular importance in the context of the present method, and it is preferably provided that the pulses of a free-running, quartz-controlled oscillator are counted for the various time interval measurements and that when a signal occurs at one or the other sensor, the one after the rising edge of the next oscillator pulse is reached and further processed.

Instead of the next rising edge, the next falling or simply the next pulse edge can also be used. In any case, this method makes it possible to obtain the time intervals of interest by forming the difference between the count values ​​concerned and multiplying them by the very precisely known and fixed period duration or half the period duration of the oscillator oscillation. However, this only leads to high measuring accuracy if the signals whose time interval is to be determined are so far apart and / or the oscillator oscillates with such a high frequency that there is such a large number of oscillator periods or pulses between the two signals occurs that the fact no longer plays a significant role that with this type of measurement the smallest unit of time, which can no longer be divided, is the length of an entire or half an oscillator period.

Since, as already mentioned, the time interval between two signals, which are emitted by different sensors, can be as small as required, the above requirement can only be met for oscillators that vibrate at extremely high frequencies, in which one can measure two within a single oscillator period or without a large measurement error Half-period events can simply be viewed as "simultaneous".

In order to avoid the effort required for such extremely high-frequency oscillators and still achieve very precise time interval measurements, a preferred variant provides that to determine the temporal position of the sensor signals within the respective oscillation period of the oscillator, each of these signals starts a time / amplitude converter circuit associated with the sensor in question and is stopped by the rising edge of the next oscillator pulse.

If the two signals whose time interval is to be measured are so far separated from one another that one or more oscillator periods occur between them, their time interval is practically determined by three individual measurements, namely a coarse time measurement given by the counting of the oscillator pulses and two using the Time / amplitude converter circuit or circuits carried out fine time measurements determined, the results of which are added up with the correct sign to the total time interval of interest.

If, on the other hand, the two signals are so close together that they fall in the same oscillation period of the oscillator (coarse measurement value - 0), which, with a suitable choice of the oscillator frequency with regard to the rotational speed of the marking carrier and the number of markings contained in a marking group, only for signals coming from different sensors can occur, a separate time / amplitude converter circuit is available for each of these signals, which measures the time interval of this signal from the next rising oscillator edge. The difference between these two time measurement values ​​then results in the sought time interval value with very high accuracy even if it is almost or exactly zero.

Since the usual time / amplitude converter circuits can have drift phenomena over longer periods of time, the output amplitude reached within certain time periods after the start signal changing, it is preferably provided that each time / amplitude converter circuit is repeatedly verified by that it is started by an edge of an oscillator pulse and stopped by a subsequent oscillator pulse edge and that the time measurement value of the time / amplitude converter circuit obtained in this way is compared with the time period lying between these two oscillator pulse edges.

It has already been stated above that the accuracy and long-term constancy of the present angle measuring method only depend on the accuracy and long-term constancy of the time measuring method used to measure the angular distances of interest. Due to the fact that the time / amplitude converter circuit, which by no means has the long-term constancy of a quartz, is also continuously calibrated with the help of the quartz during the time measurement that resolves the oscillation of the quartz-controlled oscillator, the entire angle measuring method obtains the same accuracy and long-term constancy as the quartz, without having to make any special demands on the remaining parts of the device according to the invention in this regard. Since very precisely vibrating crystals are available inexpensively even over very long periods of time, the described method provides an extremely inexpensive and yet extremely accurate protractor.

The usual time-amplitude converter circuits also have the property that the amplitude of their output signal increases linearly with time only after a certain start-up time after each start signal, while a more or less strong non-linearity is present immediately after the start signal. Since the free-running oscillator is in no way synchronized with the sensor signals, it is readily possible that there is such a short period of time between the sensor signal starting the time / amplitude converter circuit and the next oscillator pulse edge used for stopping that the time / amplitude converter circuit works in the non-linear range, which can lead to a falsification of the time measurement value obtained in this way.

To get even more reproducible and precise results here, It is provided that the time / amplitude converter circuit is stopped by the rising edge of the corresponding oscillator pulse with every time interval measurement and each calibration measurement with a predetermined time delay and that the time / amplitude converter circuit is started after each of these measurements by a start signal generated internally in the circuit and by a signal generated simultaneously with this start signal stop signal also subjected to the predetermined time delay is stopped again and the correction value thus obtained is subtracted from the previously obtained time measurement value.

Through these measures, the period of time during which the time / amplitude converter circuit works with each time interval measurement and also with each calibration measurement is first extended by a predetermined time period, which is selected so that it is certainly greater than the time period during which the amplitude of the output signal of the time / amplitude converter circuit increases in a non-linear manner. Immediately after the measured value obtained in this way is temporarily stored, the time / amplitude converter circuit is reset to the initial state and then restarted by an internally generated start signal. Simultaneously with this start signal, a stop signal is also generated internally and supplied to the time / amplitude converter circuit in the same way in which it receives the stop signal originating from the oscillator pulse edge during the actual time measurements.

This stop signal, which is generated simultaneously with the internally generated start signal, therefore also experiences the delay mentioned above, so that the time / amplitude converter circuit, despite the simultaneous generation of the two signals, operates for a period of time which corresponds to this delay (and any runtime differences between the start and stop line) which are present in the same way during the actual time measurement). A correction value is thus generated, with the aid of which the transit time differences and nonlinearities contained in the just preceding time measurement are recorded exactly. This correction value is subtracted from the time interval measurement value or calibration measurement value which is still buffered. The difference represents a measured value as would be obtained with a time / amplitude converter circuit that operates in an ideally linear manner and is controlled without runtime differences and is therefore optimally suitable for further processing. Since the correction values ​​are always obtained immediately after the measurement to be corrected, the influence of long-term drifts that may occur at this point is also eliminated.

Furthermore, it is provided according to a particularly preferred embodiment that ideal marks are abstracted by each measuring sensor from the real markings made on the marking carrier and their time intervals are measured.

The consequence of this is that not only do no particular requirements have to be placed on the long-term dimensional accuracy of the material used to produce the marking carrier and on the accuracy of the angular positioning of the markings made on the marking carrier, but also that the width and the concrete geometric shape these markings and the accuracy of their radial course are not essential, the role influencing the accuracy of the angle measurement play more.

It is of particular importance that the width of the real markings, viewed in the direction of rotation, can be selected to be substantially larger than is possible according to the prior art, without the measuring accuracy of the method being impaired in any way. Rather, the larger width of the markings allows the signal-to-noise ratio at the output of the sensors to be considerably improved, which means that the circuitry required for processing the useful signals can be significantly reduced and the measuring accuracy can be further increased.

The ideal marks are advantageously abstracted from the real markings in that the passage of an electrical signal derived from the signal emitted by the sensor as a mark passes by through a predetermined constant voltage level serves as the ideal mark, a differential photo receiver with at least one downstream differential-forming element is preferably used as the measuring sensor, and the zero crossing of the output signal generated by the differential-forming element as a mark passes the differential photo receiver serves as an ideal mark.

A differential photodiode can be used as the differential photodetector, the light-sensitive surface of which is divided into two halves or four quadrants by one or two dividers.

In the first case, the photodiode is arranged so that its separating web runs approximately radially with respect to the axis of rotation of the marking carrier. For signal processing, the photodiode is followed only by a difference-forming element, preferably a differential amplifier, the two inputs of which are each connected to one of the two halves of the photosensitive surface. When each marking passes, the difference-forming link delivers an output signal, the zero crossing of which defines a kind of "optical center line" of the real marking in an at least short-term, reproducible manner and is therefore excellently suited as an ideal, practically "punctiform" mark over time. It is possible that the "marking lines of gravity" formed in this way change their mutual angular distances slightly over longer periods of time due to drift phenomena, for example the photosensitivity of the photodiode. Due to the constant re-calibration, this remains irrelevant.

In the second case, the photodiode is arranged in such a way that one of the two separating webs again runs approximately in the radial direction, while the other extends approximately tangentially to the path described by the measuring sensor with respect to the marking carrier when it rotates. In order to process the four signals resulting from the passing of a marking in this arrangement, two difference-forming elements are used which are connected in such a way that their two inputs each receive the signals from two quadrants of the photosensitive surface opposite one another with respect to the intersection of the separating webs. In such an arrangement, the signals or signals emitted by the two difference-forming members when a marking passes by whose zero crossings are staggered in time if the edge lines of the marking are not exactly symmetrical to the radially extending separating web of the photodiode. From a change in the time interval between these two signals, each belonging to a marking, the extent of a radial relative displacement that has occurred between the measuring sensor and the marking carrier can be determined and, for example, the system can be re-calibrated if a predetermined tolerance limit is exceeded. The occurrence of eccentricities can also be recognized with this arrangement.

It is advantageously provided that for each output signal emitted by the difference-forming member when a marking passes the differential photoreceptor, a rectangular pulse is generated, the at least one edge of which is at a fixed time interval from the zero crossing of the output signal, and the time intervals of these edges are measured and processed further. The square-wave pulses emitted by the sensor, consisting of a sensor, a difference-forming element and a subsequent buffer circuit, can be designed, for example, in terms of their slope and amplitude so that they are suitable for actuating the commercially available TTL-IC digital modules, which means that the subsequent evaluation circuit can be constructed particularly cost-effectively. However, other discrete or integrated circuit components can also be used for further signal processing. The only thing that is essential is the strict temporal assignment of a clearly detectable part, for example the rising edge of the signal generated by the encoder when a marking passes the measuring sensor, so that the time interval between these signal parts can actually be used as a measure of the angular distance between the markings.

A device suitable for achieving the object on which the invention is based is described by the features summarized in claim 24. The function and meaning of the individual components of this and the device defined by claims 25 and 26 have already been discussed in detail in connection with the description of the method according to the invention. It should only be pointed out here that these components essentially formed by electronic circuits do not necessarily have to be present individually next to one another, but rather can be combined in terms of circuitry. So it is z. B. possible to combine the identification circuit, the control circuit and the comparison circuit with the arithmetic circuits to form a single computer, which in turn or in parallel performs the corresponding functions. Such computers are available, for example, in the form of microprocessors at low prices, so that their use does not in any way conflict with the aim aimed at by the invention, namely the creation of an extremely precisely working and nevertheless inexpensive angle measuring device.

In a preferred embodiment, the time measuring unit comprises a quartz-controlled oscillator, a counter which counts the oscillator periods and two time / amplitude converter circuits with analog / digital converters which can be triggered by the signals emitted by the associated transmitter and which can be stopped by signals emitted by the quartz oscillator.

Since it is necessary to ensure that the individual time measurements are carried out correctly in terms of time and function, the time measuring unit also includes a sequence control, which is either constructed separately or included in the z. B. sequence control of the entire measuring system implemented by a microprocessor can be integrated.

It is essential that a sequence control is available, which takes over the functions mentioned and, if necessary, also ensures that the time / amplitude converter circuits are repeatedly re-calibrated with the help of the oscillator between the time measurements and that both after each time measurement and after each calibration measurement, the correction values ​​already described above are obtained and processed in the required manner.

It is indeed possible to digitize the analog measurement and correction values ​​emitted by each time / amplitude converter circuit and only then to process them further. However, it is preferably provided that they are temporarily stored in analog form with the aid of sample-and-hold circuits, subtracted from one another by a differential amplifier and only then fed to an analog / digital converter.

As already mentioned, the described method allows, instead of the real markings present on the marking carrier, to use ideal, abstracted, ie temporally practically punctiform marks, and to measure and further process their time intervals. This has the direct consequence that, unlike in the prior art, it is no longer necessary to make special demands on the geometric shape of the markings. In particular, the markings no longer have to be made as extremely narrow in the direction of rotation as before; Rather, it is preferred that the markings of each marking group differ in terms of their optical properties from the regions of the marking carrier which adjoin them and run essentially perpendicular to the direction of rotation of the marking carrier, resulting in electrical signals which are particularly easy to process. if the width of each marking strip is larger than the width of the separating web of the differential photodiode and smaller than the width of the differential photodiode. The stripe-shaped design of the markings leads to a significantly improved signal-to-noise ratio at the sensor output, since, owing to the larger marking width, a much stronger optical contrast can be achieved between the markings and the areas of the marking carrier surrounding them.

The probe can be scanned by the sensor either by the incident light method or by the transmitted light method. In any case, care must be taken to ensure that the measuring light beam emanating from the light source strikes the surface of the marking carrier as perpendicularly as possible, which in the case of the incident light method has the consequence that the reflected light returns on the path of the incident light. Since the light source and light receiver cannot be in the same place, a beam splitter, for example a semi-transparent mirror, is used to decouple the reflected light.

In order to achieve the best possible illumination of the photo receiver forming the measuring sensor, it is expedient to arrange the light exit opening of the lighting arrangement as close as possible to the marking carrier. This can be done either by arranging the light source as close as possible to the marking carrier using imaging or condenser optics. In cases where this is not possible, for example due to the space available, it is preferred that a light guide directing the light of the light source in the immediate vicinity of the marking carrier is provided. This light guide can also advantageously be used to homogenize the light coming from the light source. Not only incandescent lamps, but also the light-emitting diodes (LEDs or laser diodes), which are preferably used as light sources, have a structure that means that the light is not radiated point-wise, but over a large area, with the individual points of this surface shining with very different brightness. The light guide arranged between the light source and the marking carrier can perform a condenser function here, ie its end face facing the marking medium appears as a largely homogeneously luminous surface, with the aid of which the light-sensitive surface of the photodiode forming the sensor can be uniformly illuminated.

The invention is described below using exemplary embodiments with reference to the drawing; in this shows:<ul id="ul0001" list-style="none"><li>1 is a schematic representation of a measuring device according to the invention,</li><li>2 shows an enlarged detail from a marker carrier,</li><li>3 is a circuit diagram of an encoder connected downstream of a sensor,</li><li>4 shows a diagram which shows the electrical signals occurring at various points on the transmitter,</li><li>5 shows a schematic block diagram of a preferred time measuring device,</li><li>6 and 7 signal diagrams illustrating the operation of this time measuring device, and</li><li>8 shows a further embodiment of a marking carrier that can be used in the method according to the invention.</li><li>1 shows the two bodies 2 and 3 which can be rotated relative to one another about an axis of rotation 1 in the direction of the arrow F and whose respective instantaneous angular distance is to be measured.</li></ul>

One body 2 carries the marker carrier 5, which is rotatably mounted by means of a shaft 4 aligned with the axis of rotation 1, and which is permanently supported by a drive 6, which is also fastened to the body 2 and which can either be an electric motor or a transmission or other drive device Rotation is driven in the direction of arrow R.

The marking carrier 5 essentially consists of two circular disks 10 and 11 fastened axially spaced apart from one another on the shaft 4, through the respective center of which the shaft 4 runs approximately. Along the edge of each pane 10, 11 are arranged approximately radially extending and spaced from one another in the direction of rotation, strip-shaped markings 12, which differ in the illustrated embodiment in terms of light transmission from the surrounding areas of the respective pane 10 and 11 respectively. The markings 12 arranged on the disk 10 form a first marking group α, which is arranged spatially separated from the second marking group β arranged on the disk 11. It is not necessary for there to be any special, predetermined angle relationship between the marking groups a and β. Since in the context of the present method, the measuring angle distances of the markings 12 belonging to group a from the markings 12 belonging to group β are determined by corresponding time-distance measurements, based on the arbitrarily selectable zero point of the angle measurement, the disks 10 and 11 can be in any mutual angular position on the Shaft 4 be attached.

The rotational speed with which the drive 6 sets the marking carrier 5 in rotation is in no way critical and no special measures have to be taken to stabilize it. Due to the fact that the angles of interest are determined by time interval measurements, the rotational speed of the marking carrier is significantly involved in the measurements, but within the scope of the present method it is measured continuously with a very high degree of accuracy and with a very high temporal resolution, so that both short and long-term fluctuations in the rotational speed can always be detected and taken into account immediately.

On one of the two bodies 2, a first scanning device 14, which scans the pane 10 of the marking carrier 5, is also attached, which essentially consists of a carrier 16, a light source 17 arranged above the one flat side of the pane 10 and a sensor 15 designed as a photodetector , which is arranged on the opposite side of the marking disc 10, that the connecting line between the light source 17 and the photodetector 15 is approximately perpendicular to the surface of the dial 10 carrying the markings 12.

As a result of the rotation of the shaft 4 in the direction of the arrow R, opaque markings 12 and translucent marking gaps run through the measuring light beam in constant alternation, which is modulated in this way.

A light-emitting diode, which can either be an LED or a laser diode, is preferably used as the light source 17.

The opto-electrical sensor 15 outputs an electrical signal via the line 18 to a measuring and evaluation circuit 25 each time one of the markings 12 of the marking group a runs past it. Attached to the second body 3 is a scanning device 19 which scans the second dial 11 of the marking carrier 5 and which is constructed in the same way as the scanning device 14 and comprises an opto-electrical sensor 20 and a light source 22 which is preferably designed as a light-emitting diode which are held by a carrier 21. The signals emitted by the opto-electrical sensor 20 when the markings 12 of the marking group β pass by are fed to the measuring and evaluation circuit 25 via a line 23.

Starting from the axis of rotation 1 can be placed through each of the two sensors 15 and 20 a dash-dotted radial beam 26 and 27, and it is basically the included by these two radial beams 26 and 27 with respect to the axis 1, which with the help of method according to the invention or the device according to the invention is measured. However, it is not necessary to define the zero point of this angle measurement in that the two radial beams 26 and 27 come to exactly coincide in the projection carried out in the direction of the axis 1. Rather, this zero point can be freely selected, so that the device, for example, directly converges or converges the angle between two bodies 2, 3 rotating in the direction of arrow F in the projection in the direction of axis 1. edges moving away from each other, the body 2, measures and displays directly.

As the input elements connected to the lines 18 and 23, the measuring and evaluation circuit 25 comprises two transmitters 28 and 29, which not only serve to shape the sensor signals but also serve to make ideal, ie, from the real markings 12 located on the marking carrier 5 to gain temporally practically punctiform marks, the time intervals of which can then be measured very precisely by the time measuring unit 30 connected downstream of the sensors 28 and 29.

Furthermore, the measurement includes. and evaluation circuit 25 an identification circuit 33 which recognizes which mark 12 of group a or β is currently passing sensor 15 or 20. This identification can be carried out with the help of the time intervals measured by the time measuring unit 30 for the marking pair immediately preceding the respective marking or the quotient of the time intervals of the immediately preceding marking pairs and / or by counting the number of the signals 28 or 29 coming signals occur. The identification circuit 33 supplies the respective markings individualizing information to a control circuit 35, which centrally controls the functional sequences in the entire measuring and evaluation circuit according to a predefinable program. Since this control circuit 35 must also receive information about the respective working state of the other circuit components in order to carry out its tasks properly, it is connected to them by transmission lines operating in both directions, which is symbolized in FIG. 1 by double lines provided with arrows at both ends.

1 essentially has three signal output lines 40, 41 and 42, of which the first two 40 and 41 each have a corresponding input of the identification circuit 33, a buffer 50, a comparison circuit 51 and a first computing and Memory circuit 55 are connected. On these lines 40 and 41, the time measuring unit 30 gives the time interval measured values ​​ΔT_µµ or ÄT_{ß "} from which it has determined for the immediately consecutive, identified markings 12 of the signals 28 and 29 assigned to the group a or β.

In the calibration sound, these time interval measurements are ΔT_αµ and ATß, initially measured for any selected revolution of the marking carrier 5 and stored in the buffer memory 50 while maintaining their assignment to the associated markings and marking groups a and β.

At the next revolution, the buffer 50 outputs these values ​​in the order on lines 43 and 44 to the comparison circuits 51 in which the associated new values ​​appear on lines 40 and 41. If the pairs of values ​​originating from the two successive revolutions coincide with one another within predetermined tolerance limits, which is a sign that the rotational speed of the marking carrier 5 is constant during this period, the comparison circuit on lines 46 and 45 sends signals to the first arithmetic and memory circuit 55 which cause them to use the new time interval measurement values ​​also supplied to them on lines 40 and 41 for the calculation of calibration values ​​to be stored in the long term and from the time interval measurement values ​​likewise supplied by the time measurement unit 30 during these two revolutions for one and the same marking on one of the sensors 15 and 20 generated signals and thus a 360 ° rotation of the marking carrier 5 to calculate the rotational speed determined as constant during the two calibration revolutions.

As soon as a complete set of calibration values ​​for the absolute angular distances of their markings is contained in the first arithmetic and storage circuit 55 for at least one of the two marking groups a or β, it can continuously determine the current rotational speed for each of these marking groups from each time interval measurement value newly generated by the time measuring unit 30 Calculate w (t) of the marking carrier 5 independently of whether or not the comparison circuit indicates a correspondence with the associated value originating from the previous revolution.

The instantaneous rotational speed w (t) thus obtained, which can be redetermined as many times as there are markings in the relevant marking group during one revolution of the marking carrier 5, is transmitted by the first arithmetic and storage circuit 55 via line 47 to the second arithmetic unit - And memory circuit 57 supplied.

This second arithmetic and storage circuit 57 also receives, via lines 49 and 48, the calibration values ​​relating to the angular distances within each marking group a and β from the first arithmetic and storage circuit 55, and via line 42 from the time measuring unit 30, time interval measured values ​​Δθ_µνfor the signals coming from the markings contained in the different groups a and β.

From these "mixed" time intervals Δθ_µν first calculates and stores the second arithmetic and storage circuit 57 on the basis of an angular position of the two bodies 2 and 3 or arbitrarily selected as zero position of the radial beams 26 and 27 supplied command signal using the calibration values ​​obtained via the lines 49 and 48 and the instantaneous angular velocity w (t) of the marking carrier 5 obtained on the line 47, a plurality of measuring angle distances ab related to this zero point of the angle measurement_µν, of identified markings µ and v, of which one belongs to one group a and the other to the other group β.

If this value table producing the angular, zero-point-related correlation between the two marking groups a and β is finished, then the computing and storage circuit 57 can use Δθ from each newly acquired time interval_{µ, ν}(t), the associated instantaneous angular velocity w (t) of the marking carrier 5 and the stored measuring angle distance ϕ_µν, calculate the current angular position ϕ (t) between the two bodies and feed it via line 39 to a display unit 58 and / or further processing if it has been checked whether the angular velocity w (t) of the marking carrier during the measurement of Δθ_{µ, ν}(t) was constant.

Since all the variables used for the calculation of ϕ (t) were obtained from time interval measurements, the angle measuring device determines the angle sought with the accuracy and the resolving power of the time measuring unit 30 contained therein.

It should also be pointed out here that the signal transmission between the blocks of the measuring and evaluation circuit 25 shown in FIG. 1 can take place not only in series but also in parallel or in a mixed form. In the latter two cases, the connections represented by simple lines then comprise several parallel transmission lines.

Before going into the structure and function of the time measuring unit 30 in more detail below, the interaction between the marking disks 10 and 11 with the associated sensors 15 and 20 and the structure and mode of operation of the are to be first of all based on FIGS Signals of these sensors forming sensors 28 and 29 are described. Since the basic structure of the marking disks 10 and 11 and the associated sensor and sensor arrangements 15, 29 and 20, 28 is identical, only the marking disk 10, its associated sensor 15 and the sensor 29 connected downstream of it should be considered for this purpose.

In Fig. 2, a two-mark section of a scale disc 10 is shown, in which, unlike in Fig. 1, the large area of ​​the disc is opaque, so that the measuring light beam emanating from the light source can only reach the sensor when a marking 12 runs past the sensor. According to Fig. 2nd essentially from a differential photodiode 62, the light-sensitive surfaces 64, 64 of which are separated from one another by a narrow web 63. As can be clearly seen from FIG. 2, it is not important that the marking strips 12 have a particularly precisely defined geometric shape, since from them with the aid of the differential photodiode 62 and that described in more detail below with reference to FIG. 3 An ideal brand is abstracted, which is shown in Fig. 2nd is symbolized in each case by the dash-dotted lines 60. As also shown in FIG. 2, when a differential photodiode 62 is used, the line 18 comprises two wires which are insulated from one another and which, according to FIG. 3, are each connected to one of the two inputs of a differential amplifier 66 and, on the other hand, are connected to the system ground via load resistors 67.

In a particularly preferred embodiment it is provided that in each of the lines leading from the photosensitive surfaces 64, 64 of the differential photodiode 62 to one of the inputs of the differential amplifier 66, an amplifier, not shown in FIG. 3, with a high amplification factor is switched on by the pre-amplify the light emitted surfaces of the photodiode before the difference is formed. This makes it possible to achieve that in Fig. 4th in the third line of the output signal of the differential amplifier 66 reproduced from above has a very steep and therefore precisely fixed and precisely detectable zero crossing.

The output of the differential amplifier 66 is connected via a capacitor 68 to the negative input of a comparator 70, which is connected to the system ground via a resistor 69, for DC decoupling. The positive input of the comparator 70 is connected via a resistor 71 to an adjustable stabilized voltage source, shown in simplified form as a potentiometer 72, which supplies the reference potential required to detect the zero crossing of the output signal output by the differential amplifier 66. The output of the comparator 70 is connected on the one hand to the input of the first of two inverters 74 serving as buffers and on the other hand is fed back via a capacitor 73 to generate a defined hysteresis to the positive input of the comparator 70. The inverters 74 can be, for example, TTL modules which serve to shape the signal output by the comparator 70 in such a way that it can be fed into the subsequent circuit units without further ado.

The mode of operation of the encoder shown in FIG. 3 will now be explained with reference to FIG. 4, in which the signals are reproduced over a time axis, which occur at points G to K 'of the encoder circuit 29 when a series of markings on the sensor 15 passes by.

Every time the measuring light beam is interrupted by a marker 12, or in the embodiment according to Fig. 2nd is let through, a half-wave occurs at points G and H, the signal at point G being offset in time from the signal at point H, since differential photodiode 62 is arranged in such a way that its separating web 63 is approximately in a radial direction The axis of rotation of the marking disk 10 extends, so that first the one and then the other of the two photosensitive surfaces 64, 64 of the diode 62 is swept by the marking.

From the half-waves occurring at the points G and H, the differential amplifier 66 generates the signal 1, which then has the steep zero crossing shown in FIG. 4 and thus forms an ideal mark if the circumferential width of the markings 12 is greater than the width of the separating web 63 and is smaller than the diameter of the differential photodiode 62.

The zero crossing of the signals 1 is recognized with the aid of the comparator 70, which is given a corresponding comparison level by the potential source 72. The square-wave signals K appear at the output of the comparator 70, the rising edges of which are very steep and have a clear temporal connection with the zero crossings of the signals I. The time delay s occurring between each such zero crossing and the associated rising pulse edge is shown in FIG. 4th very exaggerated. In fact, it is so small that any short-term fluctuations in this delay do not affect the accuracy of the measurement result. Since the same also applies to the signal delays occurring at the buffer inverters 74, the signal K 'occurring at the output of the second inverter 74 can be equated to the signal K at the output of the comparator. The rising edges of the rectangular pulses K 'thus clearly and at least for a short time represent ideal marks assigned to the optical lines of gravity 60 of the marking strips 12, the time intervals AT of which _a at a given angular velocity of the marking carrier, form an exact measure for the absolute values ​​of the angular distances of these optical lines of gravity. Conversely, if these absolute angular distances are known, the instantaneous angular velocity m (t) of the marking carrier 5 can be calculated very precisely from the measured values ​​Δ.

One for the highly precise measurement of the time intervals ΔT_αµ,AT_ß, and Δθ_µν preferred time measuring unit 30 is shown schematically in FIG. 5. It is based on the principle that, on the one hand, the oscillation periods of a free-running, quartz-controlled oscillator 78 are counted with the aid of a counter 79 in order to determine rough time measurements. Since this, as the sole measuring method, either requires a very high-frequency oscillating oscillator or would only lead to relatively imprecise measurement results, on the other hand, the temporal position of the signals, the time intervals of which are to be determined, measurements are provided within an oscillator period.

As can be seen from the above, time intervals must be measured both between signals that all occur at one of the sensors, ie belong to markings of the same group, and between signals, one of which is from one sensor 15 and the other comes from the other sensor 20. The time intervals of the first type are in the present context with Δθ_αµ or AT_ß, where the first index identifies the associated marker group and the second is a counting index, while the time intervals of the second type with Δθ_µν are symbolized, where µ stands for the corresponding, identified marking of the one group and v for the associated, identified marking of the other group.

To record the above-mentioned rough measurement values, it is sufficient to provide a counter which counts the oscillation periods of the quartz-controlled oscillator 78 (FIG. 5), from which the next count value is read out into the computer and memory 81 under the direction of the sequence controller 80 whenever one of the two Encoder 28, 29 emits a signal.

The computer and memory 81 is also connected to the transmitters 28 and 29 via the lines 111 and 121 and can therefore recognize whether the count value read belongs to a marking signal of the a or the β group or whether both transmitters 28, 29 are within the have given a signal before the previous oscillator period. As already mentioned, the latter case can easily occur since the two sensors 15 and 20 can repeatedly assume angular positions in which a marking from group a or β runs past them exactly or almost simultaneously.

The result of this is that a separate time / amplitude converter circuit 88 or, respectively, for determining the time measurement values ​​for each transmitter 28, 29 98 is provided with the associated additional circuits, since a time / amplitude converter circuit alone is unable to detect the occurrence of another signal during the measurement of the time interval of a signal originating, for example, from the transmitter 28 until the next rising oscillator pulse edge and also during a subsequent dead time Detect encoder 29 and determine its time interval until the next oscillator pulse edge.

The time measuring unit 30 therefore comprises two separate channels for the fine time measurements, each of which is connected downstream of one of the two sensors 15, 20 or transmitters 28, 29. Since the structure of these two fine time measuring channels is identical, a description of the channel downstream of the transmitter 28 is sufficient. In this description, the reference symbols for the corresponding circuit units of the second channel are added in parentheses when reference symbols are specified.

The time / amplitude converter circuit 88 (98) provided for acquiring the time measurement values ​​essentially comprises a capacitor which, after triggering by a START signal, is charged with the aid of a constant current source, so that after a short initial phase the voltage drop across it is also reduced the time increases linearly until the charging process is ended by a STOP signal. The amplitude of the signal present after the STOP signal at the output of the time / amplitude converter circuit thus represents a measure of the time elapsed between the START and the STOP signal.

5, the time / amplitude converter circuit 88 (98) is now used in such a way that it is started by the signals coming from the transmitter 28 (29) via the OR gate 84 (94) and by the next, the following rising edge of an oscillator pulse is stopped. This STOP signal is fed to it under control of the sequencer 80 via an AND gate 85 (95), which is used to mask out the oscillator pulses that are not required, an OR gate 86 (96) and a delay element 87 (97), the meaning of which will be explained in more detail below is explained.

The time / amplitude converter circuit 88 (98) therefore measures for each encoder signal with a high degree of accuracy the time extended by a constant value t predetermined by the delay element 87 (97), between the occurrence of the rising edge of this encoder signal and the next rising pulse edge of the free-running oscillator passes.

Since the counter 79 supplies the associated rough measurement value in digital form, it is expedient to also digitize the analog output signal amplitudes of the time / amplitude converter circuit 88 (98), which is done with the aid of the analog / digital converter 92 (102). The calculation of the time interval measurements AT of interest_«1" AT_ß, or Δθ_µν then performs the arithmetic and storage circuit 61.

The sequence control 80, which can be constructed, for example, as a Johnson counter, carries out, in addition to the control of the AND gates 85 and 95 already mentioned, a whole series of further functions, which in the following also relate to the description of the other components of the time measuring unit 30 are explained, insofar as they are essential.

Of the other components of the time measuring unit 30, the START-STOP controller 82 is first to be mentioned, which sends the START or STOP signals to the corresponding inputs of the time / amplitudes via the OR gates 84 (94) and 86 (96) Converter circuit 88 (98) can deliver. Basically, two different types of functions can be distinguished, depending on whether the time / amplitude converter circuit 88 (98) is verified by means of these internally generated START-STOP signals by comparison with the frequency standard formed by the quartz-controlled oscillator 78 or whether to eliminate Runtime differences and non-linearities a correction value is to be generated.

In the first case, the START-STOP controller 82 receives a command signal from the sequencer 80 via the input E1 (E3), whereupon it first emits a START signal via the line 112 (122), which is exactly the same as that via the line 120 supplied oscillator oscillation is synchronized, for example, coincides exactly with the falling edge of one of the square-wave pulses emitted by the quartz-controlled oscillator 78. For example, a half period of the oscillator oscillation later, ie on the next rising pulse edge, the controller 82 then delivers a STOP signal via line 113 (123), so that it becomes possible to use the very high value of the amplitude value of the output signal of the time / amplitude converter circuit 88 (98) generated in the meantime compare exact and constant oscillation of the quartz oscillator. This comparison can also be made with several half or full periods. Furthermore, it is possible to vary the number of oscillator periods used from calibration measurement to calibration measurement, so that a whole calibration scale results for the output amplitudes of the time / amplitude converter circuit 88 (98).

In the second case, the sequence controller 80 delivers a command signal to the input E2 (E4) of the START-STOP controller 82, which causes it to start and stop the lines 112 (122) and 113 (123) at the same time. Emit signal. Since the STOP signal arrives at the time / amplitude converter circuit 88 (98) delayed by the delay element 87 (97) by the time τ, this generates an output signal corresponding to this time period τ. Since the time t is chosen so that it is slightly longer than the start-up period in which the output signal of the time / amplitude converter circuit increases in a non-linear manner, the output signal obtained in this way represents a correction value which is suitable for the influence of this Eliminate non-linearity from a measurement or calibration value just obtained. This correction value also records transit time differences, which may exist between the signal path supplying the START signal to the time / amplitude converter circuit and the corresponding STOP signal path due to parasitic effects.

Each of these measured values ​​A was obtained in such a way that the time / amplitude converter circuit 88 (98) was in operation not only during the period Δt to be measured but also during the additional period τ. If this time period τ was also appended at the end of the respective measuring time At + t, the special choice of t means that each measuring or A calibration value A at the output of the time / amplitude converter circuit 88 (98) a substantially non-linear component a obtained during the first t time units and a component ΔA obtained subsequently and strictly proportional to the period of interest Δt that is actually of interest:<maths id="math0006" num=""><img file="EP0040359B1_D0006.tif" /></maths>the size of these two parts is initially unknown. Therefore, under the direction of the sequence control 80, the measured value A, which can be a time interval or a calibration measured value, is first stored in a first (third) sample-and-hold circuit 89 (99). Immediately thereafter, the time / amplitude converter circuit 88 (98) is reset to the initial state and the control unit 80 issues a command signal to the input E2 (E4) of the START-STOP control 82, which then simultaneously starts in the manner described above - and STOP signal on lines 112 (122) and 113 (123), whereby the time / amplitude converter circuit 88 (98) is caused, to generate exactly the essentially non-linear part a belonging to the above equation over the period t. This portion a is then transferred to a second (fourth) sample-and-hold circuit 90 (100) under the control of the sequence controller 80, whereupon a differential amplifier 91 (101) according to<maths id="math0007" num=""><img file="EP0040359B1_D0007.tif" /></maths>can calculate the value ΔA, which is strictly proportional to the period of interest actually of interest, and can forward it to the analog / digital converter 92 (102).

These processes are summarized again in FIGS. 6 and 7 in the form of signal diagrams.

6 shows two pulses emitted by the transmitter 29 over a time scale in the top line. Since the transmitter 29 is connected downstream of the sensor 15 scanning the marking group a (FIG. 1), the m-th and m + 1- th marking should be assigned to this group a.

In the two lines below, GEB 28 (N) and GEB 28 (M), the pulses n and n + 1 of the transmitter 28, which immediately follow these pulses from the transmitter 29, are connected downstream of the sensor 20 which scans the marking group β (FIG. 1), shown in two different time periods, namely once in the time period N in which the zero point of the angle measurement is determined and by determining the relevant measuring angle distances ϕ_µν the value table producing the angular connection between the two marking groups a and β is determined, and on the other hand in a period M in which, through appropriate measurements, an angular position deviating from the zero position _ϕ(M) of the two bodies 2 and 3 is to be determined.

The top line GEB 29 (N, M) therefore has a double function in FIG. 6 in connection with the two lines GEB 28 (N) and GEB 28 (M), namely the relative position of the two sensors 28 and 29 to represent originating impulses in two different time periods (and therefore of course also for different revolutions of the marking carrier 5) and in two different angular positions of the two bodies 2 and 3. This bracketing of the very different periods N (zero point determination) and M (measurement of the new angular position) by means of the one line GEB 29 (M, N) results in the prerequisite for purely technical reasons that the angular velocity of the marking carrier is the same in these two periods (o (N) - w (M)). It is expressly pointed out here that this condition does not have to be met in the general case, since in the present method, unlike the diagram of FIG. 6 not the time intervals measured in different time periods, but the relationships resulting from these time intervals by multiplication with the currently prevailing angular velocity w of the marking carrier 5 and thus independent of w according to equations (1).

Nevertheless, the somewhat restrictive representation for Fig. 6 was chosen because it can be used to show very clearly that when determining the angular position ϕ (M) for successive "mixed" marking pairs m, n and m + 1, n +1 very different time intervals Δθ_mn and Δθ_{m + 1, n + 1} can result because the distances between the markings m and m + on the one hand and n and n + 1 on the other are very different due to the fact that the marking carrier 5 is produced with little effort. Nevertheless, there are two identical measured values ​​die for the angular position ϕ (M) that is assumed to be unchangeable during the period M₁(M) and ϕ₂(M) because the stored measuring angle distances ϕ_mn and ϕ_{m + 1, n + 1}correctly reflect the existing irregularities in the division of the two groups a and β.

The fourth signal line OSZ 78 (M) of FIG. 6 shows the output pulses of the free-running quartz-controlled oscillator 78 which occurs in the period M and which oscillates with the period 2.

Below this, FIG. 6 shows the values ​​actually detected for determining the time intervals of interest, namely the time intervals Δt to be measured as fine time measurement values ​​with the aid of the time / amplitude converter circuits 88 and 98 (FIG. 5)₁, Δt₂, Δt₃ and At₄ between the encoder pulses m, n, m + 1 and n + 1 and the immediately next, rising oscillator pulse edge and the associated count values ​​z, determined with the aid of the counter 79 (FIG. 5)₁, e.g.₂, e.g.₃ and Z₄.

As can be seen directly from FIG. 6, the "mixed" time intervals Δθ that are initially of interest can be_µν Calculate from the directly measured values ​​according to the following equations.<maths id="math0008" num=""><img file="EP0040359B1_D0008.tif" /></maths><maths id="math0009" num=""><img file="EP0040359B1_D0009.tif" /></maths>

It also follows directly from FIG. 6 that these "mixed" time intervals with the angle measurement values ​​interess of interest₁(M) and ϕ₂(M) are linked by the following relationships:<maths id="math0010" num=""><img file="EP0040359B1_D0010.tif" /></maths><maths id="math0011" num=""><img file="EP0040359B1_D0011.tif" /></maths>It can be seen that the equations (1) result from these equations by simple transformation.

The time intervals ΔT also shown in FIG. 6_αmand ΔT_ßn satisfy the following relationships:<maths id="math0012" num=""><img file="EP0040359B1_D0012.tif" /></maths><maths id="math0013" num=""><img file="EP0040359B1_D0013.tif" /></maths>and can be used to determine the current angular velocity using the corresponding stored calibration values:<maths id="math0014" num=""><img file="EP0040359B1_D0014.tif" /></maths><maths id="math0015" num=""><img file="EP0040359B1_D0015.tif" /></maths>

It is clear that in principle it is sufficient to monitor the angular velocity of the marking carrier with the aid of only one marking group a or β.

In any case, FIG. 6 clearly shows that all the values ​​required for determining the instantaneous angular position &lt;p (M) from the fine time measurement values ​​Δt_rand the counts e.g._r (r - 1,2,3 ...) can be obtained.

Since, as already mentioned, it is not sufficient that the instantaneous angular velocity w (M) of the marking carrier 5 during the determination of the "mixed" time intervals Δθ_{µ, ν} is measured, but must actually be constant during these time periods, then for example for the "mixed" time interval Δθ_{m + 1, n + 1} by measuring the time intervals ΔT_αm and ΔT_{αm + 2} (the latter is no longer shown in FIG. 6), the angular velocity values ​​determined are used to check the constancy of this angular velocity.

If it appears that the angular velocity was constant in the relevant periods, this measured value is used to calculate the associated angular distance ϕ_r(M) used.

To increase the accuracy of the measurement, one can, for example, specify that 50 or 100 such values ​​are to be formed and the mean value should be calculated from them. Depending on whether the angular velocity of the marking carrier changes several times or not at all in the relevant period M, these 50 or 100 measured values ​​are then recorded somewhat more slowly or very quickly.

Fig. 6 also shows that the counter 79 counts the pulses of the oscillator 78 continuously (count values ​​z_r). Since the counting capacity of counter 79 is not unlimited, it must start counting again at the latest after reaching a maximum count value. With this reset, the connection to the previous count values ​​z 1, 2 can be lost, so that it is expedient to ensure that the counter 79 is only reset when no time intervals of interest are being measured. If the angular position of the two bodies 32 changes frequently, it may happen that there is no time period suitable for resetting the counter 79. In this case it will be expedient to change the time measuring unit 30 shown in FIG. 5 in such a way that a second counter is provided which counts the pulses of the oscillator 78. Each of these two counters would then be a marking group a or Assigned to β and would, for example, inevitably always be reset to zero when the zero mark of the group concerned passes the sensor. From the knowledge of the number of markings present in each marking group a and β, the computer and memory 81 could then always produce the correlation between the count values ​​of the two counters required to form the “mixed” time intervals.

Obtaining the fine time measurement values ​​Δt shown in FIG. 6_r will now be shown in connection with FIG. 7.

In Fig. 7 two signals are shown in the top line I, as z. B. arise in the transmitter 28 at point I when two markings, for example the n-th and the n + 1-th mark of the group β, pass by the sensor 20. The zero crossings of the two signals 1 represent the ideal marks abstracted from the real markings, their time interval ΔT_{ß, n} should be measured.

In the second line, FIG. 7 shows the pulses appearing at the output of the OR gate 84. First of all, these are the two pulses each assigned to the signals 1, which reach the gate 84 via the line 114 and are passed on by the latter. The rising edges of these pulses are delayed by s against the zero crossings of signals 1, but this delay - unlike shown - is extremely small and does not play a role even in the case of a very high measurement accuracy, since it over the short time ΔT_{ß, n} can be assumed to be constant.

To determine ΔT_{ß, n} it is therefore possible without errors to measure the time intervals between the rising edges of the pulses in question at the output of the gate 84.

Since these pulses fed to the gate 84 from the transmitter 28 also reach the counter 79 and the sequence control 80 via the line 111, the count value reached after the first of these two pulses can be read out from the counter 79. According to FIG. 7, in the third line of which the rectangular pulses of the oscillator 78 are shown from above, this is the count value z + 2.

In a corresponding manner, the counter value z + 6 is read out of the counter 79 after the second encoder pulse.

Thus one obtains ΔT for the time interval of interest_{ß, n} the rough measurement<maths id="math0016" num=""><img file="EP0040359B1_D0016.tif" /></maths>when Q is the period length of the oscillator oscillation.

Due to the fact assumed here that only a few oscillator pulses occur between the individual encoder pulses, this coarse measurement value is not suitable, the time interval AT of interest_{β, n} to represent with sufficient accuracy.

Therefore, at the same time as the rising edge of each output pulse of the gate 84 corresponding to an encoder signal, the time / amplitude converter circuit 88 is started, the output signal of which, as shown in the bottom line of FIG. 7, initially in a non-linear manner, but at the latest after the time t grows strictly proportionally with time.

The rising pulse edges of the oscillator 78 immediately following the rising edges of the pulses originating from the transmitter 28 at the output of the gate 84 are identified in FIG. 7 with the count values ​​z + 2 and z + 6 and of the former have the time-measured values time intervals of interest Δt₁ and Δt₃. The above-mentioned rising pulse edges of the oscillator 78 are passed via the gates 85 and 86 to the delay element 87 which, as shown in the second line from the bottom in FIG. 7, has a STOP command to the time / amplitudes with the predetermined delay time τ Converter circuit 88 passes on.

Thus, at the output of the time / amplitude converter circuit there are times Δt₁ + t and Δt₃+ t the amplitudes A, and A₃ are available, which are each input into the previously deleted first sample-and-hold circuit 89. At the same time, the time / amplitude converter circuit 88 is reset to its output level.

With the next falling oscillator pulse edge, the START-STOP control 82, activated by the sequence control 80 via its input E2, simultaneously generates a START and a STOP pulse.

The STAKT pulse appears at the output of gate 84 with virtually no delay, as shown in line 2 of FIG. 7, and causes the output signal of the time / amplitude converter circuit 88 to rise again in the bottom line of this figure.

The STOP pulse passes through delay circuit 87 and appears at its output with delay τ. After each measurement value acquisition, the time / amplitude converter circuit is put into operation again for the time τ, which leads to the generation of the correction values ​​a also shown in the bottom line of FIG. 7₁ and a₃ leads.

These correction values ​​are each input into the second sample-and-hold circuit 90 and, through the differential amplifier 91, of the measurement values ​​A currently contained in the first sample-and-hold circuit₁ or A₃ subtracted so that according to the equations<maths id="math0017" num=""><img file="EP0040359B1_D0017.tif" /></maths><maths id="math0018" num=""><img file="EP0040359B1_D0018.tif" /></maths>into the analog / digital converter 92 the time measurement values ​​Δt of interest₁ and At₃ strictly proportional amplitude values ​​can be entered.

From this, the computer and memory 81 can then determine the time measurement values ​​and according to the equation<maths id="math0019" num=""><img file="EP0040359B1_D0019.tif" /></maths>the desired time interval ΔT_ß,_n to calculate.

It should be pointed out here that the time intervals ΔT_ß,_n T_{ß, n + 1 '} .... add up between successive markings of a group without the measuring errors being summed up. It follows from the equation above<maths id="math0020" num=""><img file="EP0040359B1_D0020.tif" /></maths>It can be seen that with the exception of Δt₁ all fine time measurement values ​​At₃, At₅... fall out so that their mistakes cannot add up.

The measurement of the correction values ​​a₁, a₃ ... introductory internally generated START pulses do not necessarily have to be temporally correlated in the manner shown with the falling oscillator pulse edges which directly affect the recording of the measured values ​​A₁, A₃ follow rising rising pulse edges. It is only important that they are generated in good time after the previous measurement value acquisition that the correction value a_l, a₃ ... represents a reasonable measure of the essentially non-linear part contained in the associated measured value. In addition, it must be ensured that the correction value acquisition is completed before the next measurement or calibration value acquisition begins.

Furthermore, in FIG. 7 there is a period ΔT_ß,_n Calibration detection taking place for re-calibration of the time / amplitude converter circuit 88 is shown with the aid of the quartz oscillator 78.

For this purpose, the START-STOP control 82 triggered by the sequence controller 80 via the input E1 first generates a START pulse which is exactly synchronized with the output signal of the oscillator 78. In Fig. 7, this is the third pulse from the left in line G 84.

Just as precisely synchronized with the oscillator signal, the START-STOP controller 82 outputs a STOP signal via line 113, delayed by a predetermined number of oscillator half or full periods. In FIG. 7, for the sake of simplicity, it is assumed that the STOP pulse is generated exactly one quartz half period after the START pulse. It also passes through the delay element 87, so that the time / amplitude converter circuit is stopped after the time Q / 2 + c; a signal with the amplitude A is then at its output_E which, like a normal measured time value, is first buffered in the first sample-and-hold circuit 89. Thereupon, by controlling the input E2 of the START-STOP control 82 in the same way as described above for the time measurement values, a correction value a_E generated so that the differential amplifier 91 the calibration value ΔA that is actually of interest_E can generate, which is then also digitized and processed by the computer and memory 81.

Since there is generally a sufficiently large number of oscillator pulses between successive encoder pulses, it is ΔT within each time period_ß,_ν possible to re-calibrate the time / amplitude converter circuit 88 in the manner just described.

Corresponding functions also run for the second time / amplitude converter circuit 98, which is therefore also continuously calibrated with the aid of the quartz-controlled oscillator 78 and the measured values ​​of which are also partially corrected by eliminating the non-linear rise.

8 shows a somewhat different embodiment of a measuring arrangement according to the invention. The parts corresponding to the parts shown in FIG. 1 are identified by the same reference numerals, but with an apostrophe. The most important difference to the embodiment according to Fig. 1 consists in that here the marking carrier 5 'consists of a circular cylinder, on the outer lateral surface of which the markings 12' differing in their light reflectivity from the immediately surrounding areas of the marking carrier are arranged in two groups a 'and β'. These markings also consist of elongated strips which are essentially perpendicular to the direction of rotation R 'of the marking carrier 5', ie so here extend essentially parallel to the surface lines of the cylinder 5 '. Here, too, the two marking groups a 'and β' are spatially separated from one another and are each scanned by a scanning device 14 'or 19', but here in the incident light method.