DE19548385C2 - Method for determining the angular position of an axis of rotation of an object by a computer - Google Patents

Description

The invention relates to a method in which it is possible from two sinusoidal sensors supplied by two direction sensors Signals to an absolute angular position of an axis of rotation average with a value between 0 ° and 360 °.

The direction sensors are designed such that each as an output signal depending on the angular position a sinusoidal signal on the axis of rotation to be evaluated point.

The determination of an angular position on the axis of rotation is in many technical areas of great importance.

Examples of application areas are in the automotive industry rich, where it is necessary, for example, the position of the Crankshaft or the camshaft, the position continues of the gearbox or a contactless potentio to realize meters.

The determination of an angular position of an axis of rotation of a However, the subject is also in other technical fields of considerable importance, also in the area of work machine tools. Furthermore, this process is in all tech niche areas where an angular position applies sition of an axis of rotation of an object to measure.

DE 32 01 005 describes a device for error correction Position measuring systems known. This is used to measure the Relative position of two objects the division of a scale, the is connected to an object from a scanning unit geta  which is connected to the other object. In doing so Correction traces applied on a magnetic foil. Hall elements are also used in this device puts.

This facility means in addition to the ones necessary anyway Direction sensors a considerable additional effort due to additional correction tracks and the Hall elements.

Furthermore, in the known device performed measurement procedures that supplied by the direction sensor Sine-cosine signal converted into a digital signal. This occurs several thousand times per revolution.

This is in the event that you only have a period over 360 ° wants to measure, not usable.

The same applies to the known magnetic length or Angle measuring device, which is described in DE 32 14 794.

DE 40 29 828 A1 discloses a rotation angle detection unit knows, from two phase-shifted cosine signals to determine an angle the angle range from 0 ° to 360 ° is divided into four areas. The division takes place such that a standardization of the Co sinusoidal signals is performed. For the standardized Cosi a comparison of their amounts is carried out. On finally one of the two cosine signals is checked, whether the value of the respective cosine signal is greater than Zero. Depending on the two comparisons, an Winkelbe rich selected, which is used to determine the angle of rotation becomes.

One disadvantage of this method is that that normalization of the cosine signals at the beginning of the process rens is required.

The invention is therefore based on the problem of a method to determine the angular position of an axis of rotation of a Ge to specify the object by means of a computer which mentioned disadvantages of the known measuring devices ver avoids.

The problem is solved by the method according to claim 1 solved.

Two sinusoidal phases are shifted from each other signals from two offset by approximately 90 °, side by side ordered directional sensors evaluated.

This also results in the phase shift of the two sinusoidal signals measured by the direction sensors a shift of about 90 °.  

These two sinusoidal waveforms are within egg ner period from 0 to 360 ° further processed.

The entire interval from 0 ° to 360 ° is now at least divided into four areas, each one of the two sinusoidal waveforms as approximately linear is sought.

Depending on the amplitude values of the two sinusoidal ones Waveforms for an angular position to be determined The axis of rotation of the object is now a relative angle in one of the at least four areas.

This is done by adjusting the phase and adjusting the amount of a previously determined and stored reversal function of an all common sinusoidal signal. The relative angle becomes considering the selected area now in egg In the last step, the absolute angular position of the axis of rotation determined.

The significant advantage of the process can be seen in the fact that a one-time determination of the values of an inverse function only part of a general sinusoidal signal, namely an approximately linear section of the sinusoidal Signals corresponding to the selected ranges, is sufficient.

It is advantageous to have an upper bound as well as a lower one Barrier from the intersection of the two out of phase to determine sinusoidal signals. The respective area as an indication of the corresponding phase and amount adjustments solution is then derived from the simple Ab ask whether a first sinusoidal signal or a second sinusoidal signal over an upper bound or is below the lower bound.

The absolute angle is now determined starting from from the values of the reversal function determined only once  on. It is not necessary with every determination of the angular position tion again the reverse functions of those of the direction sensors sensors delivered sinusoidal signals each teln. This saves considerably the computing time required for the procedures performed by the computer as well as significant Storage space required, since it is only part of a reversal function of a general sinusoidal waveform ge must be saved.

Advantageous further developments of the method according to the invention result from the subclaims.

As a result, the method according to the invention is further combined times and thus further reduces the computing time requirement.

It is advantageous to have fixed, simple criteria for choosing the respective range depending on the current value of the most sinusoidal signal and the second sinusoidal Si gnals to use. By doing so, the Re further reduction in the time required to carry out the method graces.

It is also advantageous to reverse the function of the at least an approximately linear part of a general sine shaped signal by summation of tangential to the to determine the respective straight line equations men.

It is also advantageous to use Be calculation of individual values and the determined Store values in tabular form. By forming this "Look-up table" will significantly reduce the rake time required.

Are application-typical deviations from the direction sensors actually delivered sinusoidal signals is known  it is advantageous to do this when determining the reversal function to be taken into account and compensated for.

It is possible that the computer, with a single circulation the angular position from 0 ° -360 °, the criteria for bil the areas and the parameters of the inverse function itself measures, and thus calibrates itself.

In order to avoid negative values when determining the angular position, it is also advantageous to add an offset value to the inverse function, which will later be taken into account when calculating the absolute angle ρ _abs .

To increase the accuracy of the procedure is a tempera door compensation for the individual direction sensors is carried out advantageously.

To increase the accuracy of the results provided it advantageous to the first sinusoidal signal and / or second sinusoidal signal before performing the procedure reinforce rens.

An embodiment is shown in the figures and is explained in more detail below.

Show it

Fig. 1 is a sketch in which the functional principle of a Rich direction sensor, which is implemented as a GMR sensor, is shown;

Figure 2 shows a basic arrangement of a shaft with a permanent magnet mounted on the shaft and a crossed directional sensor pair, with which the angular position of the shaft is determined.

Fig. 3 is a sketch in over an angular range of an angle φ from 0 ° to 360 °, the amplitudes AW of the first sinusoidal signal and said second sinusoidal signal is Darge are;

Fig. 4 is a sketch in which the division of the entire angular range of 360 ° in this case four areas in which a part of the first sinusoidal signal or the second sinusoidal signal is seen as approximately linear;

Fig. 5 is a flowchart in which individual Verfahrensschrit te of the process are shown;

Fig. 6 is a block diagram in which the entire measuring arrangement, within the scope of which the method can be carried out, is provided, consisting of an analog part, a computer, and an output;

Fig. 7 is a diagram in which the principle that underlies the process, so a crossed direction sensor pair and a rotating permanent magnet are shown;

Contribute Figure 8, two bridge circuits tion to Temperaturkompensa of the entire arrangement.

Fig., A circuit arrangement is performed with a sensor signal processing 9;

Referring to Figs. 1 to 9, the invention is further tert erläu.

In Fig. 1 the principle of operation is shown of a direction sensor.

In this particular embodiment, a so-called called Giant Magneto-Resistive (GMR) sensor GMR gene.

However, it can be any direction sensor that is one of the respective current angular position of the axis of rotation of an object G. outputs a sinusoidal curve as a sensor signal, ver be applied.

A permanent magnet PM mounted on the object G, for example a rotatable shaft DW, is seen as the signal transmitter (see FIG. 2).

A magnetic field is generated by this permanent magnet PM testifies which couples contactlessly to the GMR sensor GMR.

The GMR sensor GMR has at least three layers, one soft magnetic layer WS with a magnetization M1, a metallic intermediate layer MZ, through which the soft magnetic cal layer WS and a hard magnet described below table layer HS are decoupled, as well as the hard magnetic layer HS with a fixed magnetization M2.

In general, a GMR sensor exhibits a change in resistance ΔR / R that is proportional to the cosine of the angular difference α between the field direction of the magnetization M1 soft magnetic layer WS and the fixed magnetization M2 of the hard magnetic layer HS.

With a rotation of the object G, for example the rotatable shaft DW by 360 ° results for the contra change in position, which is measured by means of a measuring current MS in the GMR Sensor GMR can be measured, a cosine signal course.  

Thus, a 180 ° field direction encoder can be used with a GMR sensor be made because you can only do it from the absolute contradiction cannot determine whether you are on the fall ling branch or on the rising branch of each ten cosine curve.

However, in order to determine an angular position that can move in the angular range from 0 ° to 360 °, it is therefore necessary to use two GMR sensors GMR1, GMR2, the two GMR sensors GMR1, GMR2 rotating by approximately 90 ° are arranged. They thus form a crossed pair of direction sensors GS (see FIG. 2).

This basic structure of a permanent magnet PM, whose magnetic field acts on the crossed sensor pair GS, is once again, detached from the rotatable shaft DW, shown in FIG. 7.

When changing the relative position of the permanent magnet PM with respect to the crossed sensor pair GS, both GMR- Sensors GMR1, GMR2 waveforms, each one cosi have a nusoidal or sinusoidal course.

When the angle of the permanent magnet PM changes from 0 ° to 360 °, i.e. one full revolution of the rotatable shaft DW speaking, a first direction sensor GMR1 delivers a first sinusoidal signal Sig1 and a second direction sensor GMR2 to the first sinusoidal signal Sig1 by about 90 ° phase-shifted second sinusoidal signal Sig2.

An example of the phase-shifted two sinusoidal signal waveforms Sig1, Sig2 is shown in FIG. 3, the respective amplitude of the sinusoidal signals Sig1, Sig2 being plotted over an angular range of 0 ° to 360 °, the amplitude values A being assumed such that the values of the sinusoidal signals are between 0 and 2.

This course, illustrated as an example, limits the general validity of possible sinusoidal signals in no white se one.

In this example shown in FIG. 3, the first sinusoidal signal Sig1 results in:
Sig1 = cos (ϕ + 45 °) + 1
and the second sinusoidal signal Sig2 results in:
Sig2 = cos (ϕ - 45 °) + 1.

A position angle bezeichnet denotes the respective win between the current position of the permanent magnet a predetermined rest position of the permanent magnet PM.

In Fig. 4 is now a subdivision of the sine curves shown in Fig. 3 Darge into four areas I, II, III, IV is shown, in which either the first sinusoidal signal Sig1 or the second sinusoidal signal Sig2 is regarded as approximately linear becomes.

In order to improve the accuracy, it can be done in a further be advantageous of the method according to the invention, the Entire interval from 0 ° to 360 ° in more than four areas to divide. This improves the accuracy of the linear sewing production improved.

It has proven to be advantageous to include the areas in this way share that the boundaries of the areas by the Intersection points of the first sinusoidal signal Sig1 with the second sinusoidal signal Sig2 are determined.

However, this is not a necessary assumption is only that in the areas a branch is one of the  two sinusoidal signals Sig1, Sig2 approximately as can be viewed linearly.

For one of the areas, an inverse function becomes one general sinusoidal signal determined. The investigation can be as precise as possible according to the possibilities ten, for example the resolution of the computer. A Possibility to determine the inverse function for each area lies in single tangential straight lines for individual points on the sinusoidal curve put together.

An alternative procedure is the calculation individual values of the inverse function in the respective area and storing the values in a table, such called "look-up" table.

The logical decisions like dividing the quadrants, can also be used with analog comparators and logic gates be performed.

In addition, the inverse functions of the line equations be formed analogously.

Furthermore, the look-up table of the reverse straight line could also can be stored in an EPROM that uses its address data an A / D converter. The data that the EPROM has on output A / D values can be programmed as required.

In this embodiment, the reverse radio is determined tion for the selected area in a memory for Execution of the procedure required computer, saved.

FIG. 4 also shows an intuitive, very simple choice of an upper barrier OG and a lower barrier UG, which will be described in more detail below.

In this embodiment, the upper bound results OG and the lower barrier UG also from the intersection th of the two phase-shifted sinusoidal signals Sig1 and Sig2.

However, the upper bound is possible without restriction OG and the lower barrier UG to choose differently.

A basic block diagram, in which the entire network arrangement with an analog part AT, the computer R and an output A, is described in FIG. 6.

Here, a first channel K1 and a are in the analog part AT second channel K2 for recording the first sine shaped signal Sig1 and the second sinusoidal signal Sig2 provided.

In the analog part AT individual means are provided, with de improvements in the results achieved by the process nisse can be achieved, for example by a temperature compensation or by amplifying the individual sine waves shaped signals Sig1, Sig2.

These training courses will be described in detail below wrote.

An analog / digital converter AD is provided in the computer R. hen. In the analog / digital converter AD this is done by the Ana log part AT prepared first sinusoidal signal Sig1 and the second sinusoidal signal Sig2 in digital sinusoidal Signals implemented.

A means for signal evaluation SA is also provided, in which the main steps of the invention Procedure.  

Furthermore, a means for signal output PWM is an output voltage in the form of a pulse width modulation voltage Darge poses. The pulse width modulated voltage becomes an output A, in which, if desired, smoothing the pulse width modulated voltage is carried out.

Fig. 5 shows a flow chart with individual method steps of the method. For better clarity, however, the process is already shown in this flowchart for the individual special cases of the exemplary embodiment.

For an angular position of the rotatable shaft DW to be determined, the amplitude value of the first sinusoidal signal Sig1 and the amplitude value of the second sinusoidal signal Sig2 are measured by the first direction sensor GMR1 and the second direction sensor GMR2 for the respective position. In a first step 501 it is checked whether the value of the second sinusoidal signal Sig2 is greater than the upper bound OG. If this is the case, the angle is thus in the first area I (see FIG. 4) and the amplitude value of the first sinusoidal signal Sig1 is used 502 for further determination of the angular position. In a further step SO ₃ , a relative angle ϕ _rel determined using the stored inverse function. The determination is carried out in the following manner from the amplitude value Ampl (Sig1) of the first sinusoidal signal Sig1:
ϕ _rel = arcsin (Ampl (Sig1) - 1) + 45 ° + offset

In a last step (step 504), an absolute angle ϕ _abs is now determined from the relative angle ϕ _rel . The following formula is used for this:
ϕ _abs = 90 ° - ϕ _rel + offset

The addition of an offset can be advantageous in order to avoid the occurrence of negative values during the determination of the angular position, that is to say the absolute angle ϕ _abs , but is not necessary for the method.

However, the value of the second signal Sig2 is not greater than the upper bound OG, it is now checked whether the value of the he most sinusoidal signal Sig1 is smaller than the lower one Barrier UG (step 511).

If this is the case, then in this special embodiment, the angle is in the second area II (see FIG. 4). The value of the second sinusoidal signal Sig2 is therefore used for further calculation (step 512). In accordance with the procedure described above, the relative angle ϕ _rel is again determined from the amplitude value Ampl (Sig2) of the second sinusoidal signal Sig2 using the inverse function (step 513):
ϕ _rel = arcsin (Ampl (Sig2) - 1) + 45 ° + offset

From the relative angle ϕ _rel , the absolute angle ϕ _abs is now determined for this case according to the following rule (step 514).
ϕ _abs = 180 ° - ϕ _rel + offset However, if the amplitude value Ampl (Sig1) of the first sinusoidal signal Sig1 is not less than the lower bound UG, it is checked whether the amplitude value Ampl (Sig2) of the second sinusoidal signal Sig2 is smaller as the lower barrier UG (step 521).

If this is the case, the angle is in the third region III (see FIG. 4) and the amplitude value Ampl (Sig1) of the first sinusoidal signal Sig1 is used to further determine the angular position (step 522).

Again, the relative angle ϕ _{rel is} determined using the stored inverse function, this time according to the following rule (step 523):
ϕ _rel = arcsin (Ampl (Sig1) - 1) + 45 ° + offset

The absolute angle is again determined from the relative angle ϕ _rel (step 524):
ϕ _abs = 180 ° + ϕ _rel + offset

However, if the value of the second sinusoidal signal Sig2 is not less than the lower bound UG, then it is further checked whether the value Ampl (Sig1) of the first sinusoidal signal Sig1 is greater than the upper bound OG (step 531). If this is the case, the angle is therefore in the fourth area IV (cf. FIG. 4) and the value of the second sinusoidal signal Sig2 is used for further determination of the angular position (step 532).

Here, too, the relative angle ϕ _{rel is determined} using the inverse function (step 533). In this case, this is done in the following way (step 534):
j _rel = arcsin (Ampl (Sig2) - 1) + 45 ° + offset

The absolute angle ϕ _{abs is} determined 534 from the relative angle 53 _rel. This is done according to the regulation:
ϕ _abs = 270 ° + ϕ _rel + offset

From this representation it can be seen that in each case to calculate the relative angle ϕ _rel depending on the values of the first sinusoidal signal Sig1 and the second sinusoidal signal Sig2, an amount adjustment and a phase adjustment for the reversed function once determined, which is even only for one Part of a general sinus-shaped signal had to be determined, takes place.

The absolute angle ϕ _abs is also only determined by taking into account the respective range in which the angle, as determined by the values of the first sinusoidal signal Sig1 and the second sinusoidal signal Sig2, is located.

The results of the procedure can be obtained from one of the following The further training described described can be improved.

In Fig. 8 this two circuits by about 90 ° twisted bridges are shown in the two direction sensors GMR1, are connected in the form of a full-bridge GMR2. Since a temperature compensation of the basic resistance of the direction sensors GMR1, GMR2 is achieved. Due to tolerances of the basic resistance of the individual direction sensors GMR1, GMR2, the output signal can be superimposed by an offset voltage, the temperature response of which, however, is negligible.

By connecting the two direction sensors GMR1 and GMR2 in the form of a full bridge is therefore a temperature comm compensation of the basic resistance of the direction sensors GMR1, GMR2 achieved.

In Fig. 9 is a circuit arrangement with which a processing of the sensor signals supplied is achieved. A voltage signal U _{A that can be} tapped off from the full bridges is in the range of some +/- 10 milli-volt in the exemplary embodiment presented and, in order to utilize the accuracy of the A / D converter AD located in the computer R as well as possible, be raised to a signal amplitude of 5 volts.

For this purpose, a first amplifier stage V1 and a second Amplifier stage V2 used.

In order not to burden the tapped signal U _A with a measuring current, an isolation amplifier is selected as the first amplifier stage V1, which ensures a first pre-amplification of the signal, for example, by a factor of 10.

In the second amplifier stage V2 is a signal amplified by the first amplifier stage V1 signal UV _A with a invertie in power operational amplifier circuit is amplified to a signal amplitude of 5 volts and a possibly provided offset adjusted so that the signal minimum is at 0 volts.

An output signal U _AA thus obtained is now fed to the analog / digital converter AD of the computer R. In the computer R, the method for evaluating the angle described above is carried out, and the result, that is to say the value of the angle, is output, for example with the aid of a timer, as a pulse-width-modulated voltage PWM.

To use the pulse width modulated output signal PWM with egg ver output signal from a conventional potentiometer To make it comparable, it can be advantageous in the output A still smooth the pulse width modulated signal PWM for example with the help of an R / C link.

It is also possible that the computer, at a one-time around the angular position from 0 ° - 360 °, the criteria en to form the areas and parameters of reverse radio tion measures itself, and thus calibrates itself.

The following procedure is possible, for example:
If the upper bound is determined by the upper intersection of the two signals Sig1 and Sig2, the value of the upper bound is determined by the computer itself.

The value of the upper bound is determined by the fact that applied magnetic field is rotated until the first Si Signal Sig1 and the second signal Sig2 have the same value point. This value will decrease as the value for the upper bound saved.

Likewise, the lower bound is automatically determined method.

If the absolute angle ρabs is in the first range or in the fourth range, the following steps should be provided in the event that angle ranges greater than 90 ° are permitted for the value range of the relative angle:
If one obtains absolute angle values ρ _abs in the first area which are smaller than 0 °, it must be taken into account that the angle area then "jumps" from the first area to the fourth area. For this purpose, the amount of the negative absolute angle ρabs in this case must be determined and in order to obtain the real absolute angle ρabs, the amount of the negative absolute angle ρabs must then be subtracted from 360 °. This also gives the correct absolute angle ρabs for this special case.

The same procedure is followed in the event that the absolute angle ρ _{abs is} greater than 359 °.

Claims (6)

1. A method for determining the angular position of a rotary axis of an object by means of a computer from a first sinusoidal signal and a second, phase-shifted sinusoidal signal and from a stored course of an inverse function of a sinusoidal signal,

• 1. in which a range of values from 0 ° to 360 ° is divided by an upper and a lower barrier into at least four areas, the upper barrier being a first intersection of the first sinusoidal signal with the second sinusformi signal and the lower barrier second intersection of the first sinusoidal signal with the second sinusoidal signal, and
• 2. in which the area is selected according to the following criteria:
  • 1. If the value of the second sinusoidal signal is above the upper limit, a first range is selected,
  • 2. if the value of the first sinusoidal signal is below the lower limit, a second range is selected,
  • 3. if the value of the second sinusoidal signal is below the lower limit, a third range is selected,
  • 4. if the value of the first sinusoidal signal is above the upper limit, a fourth range is selected,
• 3. a relative angle within the selected range is determined from the stored curve of the inverse function of the sinusoidal signal, taking into account the phase adjustment and the amount adjustment, and
• 4. in which, depending on the selected area, the absolute angular position is determined by adding an area-dependent angle.

2. The method according to claim 1, in which the stored course of the reverse function of the si nus-shaped signal by composing the course li approximating straight line equations is determined. 3. The method according to claim 1 or 2, in which the stored course of the reverse function of the si nus-shaped signal in tabular form by determining individual values of the inverse function is determined. 4. The method according to any one of claims 1 to 3, in which an offset value is added to the sinusoidal signal becomes. 5. The method according to any one of claims 1 to 4, in which temperature compensation is provided. 6. The method according to any one of claims 1 to 5, in which the first sinusoidal signal and / or the second si nut-shaped signal is amplified.