METHOD AND SYSTEM FOR COMPUTING THE PHASE SHIFT OR AMPLITUDE OF A THREE PHASE SYSTEM

Abstract

A method and system for computing the phase shift or the amplitude of an electromagnetic three-phase system. The method comprises the following steps of: detecting vector values corresponding to an electromagnetic quantity by three sensors, the three sensors delivering signals that are offset from each other substantially by 0, 120 and 240; computing changed vector values by logically adjusting one of the detected vector values to a phase of 0; and iteratively computing the phase shift of the three-phase system using the changed vector values.

Claims

1. A method for computing at least one of the phase shift or the amplitude of an electromagnetic three-phase system, comprising: detecting vector values corresponding to an electromagnetic quantity by three sensors, the three sensors delivering signals that are offset from each other substantially by 0, 120 and 240; computing with a processor changed vector values by logically adjusting one of the detected vector values to a phase of 0; and iteratively computing with said processor the phase shift of the three-phase system using the changed vector values.

2. The method according to claim 1, wherein the vector values are calculated by combining oppositely disposed sensors.

3. The method according to claim 1, wherein the iterative computation is carried out by said processor using a stored coefficient table.

4. The method according to claim 1, wherein the iterative computation is carried out by said processor using a Cordic method.

5. A system for computing an angle, comprising: at least three sensors for detecting an electromagnetic quantity, wherein the at least three sensors are arranged substantially at equal angles about a circle; a processor receiving values from the at least three sensors and computing one of the phase shift or the amplitude of an electromagnetic three-phase system, comprising: detecting vector values corresponding to an electromagnetic quantity by three sensors, the three sensors delivering signals that are offset from each other substantially by 0, 120 and 240; computing changed vector values by logically adjusting one of the detected vector values to a phase of 0; and iteratively computing the phase shift of the three-phase system using the changed vector values.

6. The system according to claim 5, wherein the sensors are Hall sensors.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0010] The invention will now be explained in more detail with reference to the following figures. It will be understood that the embodiments and aspects of the invention described in the figures are only examples and do not limit the protective scope of the claims in any way. The invention is defined by the claims and their equivalents. It will be understood that features of one aspect or embodiment of the invention can be combined with a feature of a different aspect or aspects of other embodiments of the invention. This invention becomes more obvious when reading the following detailed descriptions of some examples as part of the disclosure under consideration of the enclosed drawings. Referring now to the attached drawings which form a part of this disclosure. There are shown:

[0011] FIG. 1 is an overview of a rotatable permanent magnet and the sensor in an integrated circuit;

[0012] FIG. 2 is the values of a three-phase system;

[0013] FIG. 3 is the magnetic vector values; and

[0014] FIG. 4 is the sequence of the method.

DETAILED DESCRIPTION

[0015] The object of the present invention is fully described below using examples for the purpose of disclosure, without limiting the disclosure to the examples. The examples present different aspects of the present invention. To implement the present technical teaching, it is not required to implement all of these aspects combined. Rather, a person skilled in the art will select and combine those aspects that appear sensible and required for the corresponding application and implementation.

[0016] FIG. 1 shows an overview of an angle detection system 5 of the present description, which computes the rotational angle of a rotatable permanent magnet 10. The rotatable permanent magnet 10 in the represented embodiment is a four-pole magnet and is arranged above a sensor plate 20. The sensor plate 20 has six sensors for detecting the magnetic field of the permanent magnet 10, which are arranged at an angular distance of approximately 60 on a circular line 30 around a symmetry center 50. The sensors 25 are horizontal Hall effect sensors which, for example, can deliver signals that correspond to the value of the detected magnetic field.

[0017] The signals from the respectively oppositely disposed sensors 25, i.e. offset by 180 in each case, can be combined in order to deliver one single signal Su, Sv, Sw for each pair. This combination is represented in FIG. 1 and shows the normalized signals Su, Sv, Sw for the three field vectors U, V and W, which are combined from the pairs of sensors 25.

[0018] The angle detection system 5 also includes a processor 40 which receives the signals from the sensors 25 and processes these according to the following method.

[0019] FIG. 3 shows the respective three magnetic field vectors U, V and W. The present invention is based on the idea of turning back one of these vectors, for example the U vector, to an angle of 0, to be able to thus compute the angle. The rotational angle can be computed subsequently using the known CORDIC method. Accordingly, the present invention relates to the expansion of the CORDIC method to a three-phase system.

[0020] The method for computing the rotational angle is carried out step by step using the following rules. From the three measured sensor signals Su, Sv, Sw and a rotational angle , new signals Su, Sv, Sw and a new rotational angle are computed:

Su=SuA*(SvSw),(1)

Sv=SvA*(SwSu),(2)

Sw=SwA*(SuSv),(3)

=+R*arctan({square root over (3)}*2.sup.n)(4)

[0021] The parameter A is chosen as A=R*2.sup.n, wherein n is an integer. By choosing a power of two, the multiplication is reduced to a shift operation. The parameter R is either +1 or 1 and corresponds to the rotation direction, thus the sign of SwSv.

[0022] The method starts with the measured values Su, Sv, Sw, with =0 and n=1 and is iterated several times. Thereby the iteration requires only addition operations and can do without multiplication. For computing the rotational angle , for example for an angle resolution of 1 degree, a coefficient table can be used with the values of arctan({square root over (3)}*2.sup.n) for n=1, 0, . . . , 10.

TABLE-US-00001 Max. rotation Rotation Rotation * 3 arctan [] range 2 3.46410 73.89789 222.77401 1 1.73205 60.00000 148.87612 0.5 0.86603 40.89339 88.87612 0.25 0.43301 23.41322 47.98273 0.125 0.21651 12.21635 24.56951 0.0625 0.10825 6.17839 12.35316 0.03125 0.05413 3.09820 6.17477 0.015625 0.02706 1.55023 3.07656 0.007813 0.01353 0.77526 1.52633 0.003906 0.00677 0.38765 0.75107 0.001953 0.00338 0.19383 0.36342 0.000977 0.00169 0.09691 0.16960 0.000488 0.00085 0.04846 0.07268 0.000244 0.00042 0.02423 0.02423

[0023] The column max. rotation range serves for verifying the method. From the uppermost line, it can be seen that the rotation range can detect a maximum value of 222.

[0024] In a further embodiment, the signals Su, Sv, Sw are scaled with the value cos(p) in each iteration step, wherein p=arctan({square root over (3)}*2.sup.n). Thus, for example (from (1))

Su=cos(p)*(SuA*(SvSw))(5)

[0025] Thereby the amplitude scaling of the components is maintained and the value of Su at the end of the iteration corresponds to the amplitude of the original measured signal. The equation (5) requires three multiplications per iteration step, i.e. in 10 iteration steps a total of 30 multiplications are used. In order to minimize this computation effort, the equation can be computed at first without the scaling cos(p). The scaling is carried out once subsequently at the end of all iteration steps, since the product of all cos(p) is constant for a fixed number of iteration steps. Therefore, only three multiplications are carried out in this aspect. The value of the scaling can also be stored in the coefficient table.

[0026] The vector values can also be scaled while the iteration steps are carried out, so that the vector values do not become too large, so as to avoid overflows in the computer unit, for example. This scaling can be effected using a shift operation and has to be taken into account in the final scaling.

[0027] The method of the present description results from the properties of a three-phase system. It is known that the sum of the three field vectors U, V and W in a three-phase system must be zero. The following formula results therefrom:

{right arrow over (p.sub.u)}+{right arrow over (p.sub.v)}+{right arrow over (p.sub.w)}={right arrow over (0)}(6)

[0028] The sum of the respective component, for example on the x-axis, is therefore:

p.sub.u,x+p.sub.v,x+p.sub.w,x=0(7)

[0029] Each of the field vectors has the same length V:

|{right arrow over (p.sub.u)}||{right arrow over (p.sub.v)}|=|{right arrow over (p.sub.w)}|=V(8)

This yields:

|{right arrow over (p.sub.u)}||{right arrow over (p.sub.v)}|=|{right arrow over (p.sub.w)}|=3V(9)

[0030] The components on the respective x-axis and y-axis are:

p.sub.u,x.sup.2+p.sub.v,x.sup.2+p.sub.w,x.sup.2=3/2V.sup.2(10)

p.sub.u,y.sup.2+p.sub.v,y.sup.2+p.sub.w,y.sup.2=3/2V.sup.2(11)

|{right arrow over (p.sub.u)}|=p.sub.u,x.sup.2+p.sub.u,y.sup.2=V.sup.2(12)

[0031] Thus the component in the y-direction can be determined. From (12) the length of the vector is derived:

p.sub.u,y.sup.2=V.sup.2p.sub.v,x.sup.2(13)

[0032] The following formulas result therefrom:

[00001]$\begin{array}{cc}{p}_{u,x}^2=\frac{p_{u,x}^2+p_{v,x}^2+p_{w,x}^2}{1.5}-\frac{1.5}{1.5}.Math..Math.p_{u,x}^2& \left(14\right)\\ p_{u,x}^2=\frac{-0.5.Math..Math.p_{u,x}^2+p_{v,x}^2+p_{w,x}^2}{1.5}& \left(15\right)\end{array}$

[0033] Thus from (5):

[00002]$\begin{array}{cc}{p}_{u,x}^2=\frac{-0.5.Math.{\left(-p_{v,x}^2-p_{w,x}^2\right)}^2+p_{v,x}^2+p_{w,x}^2}{1.5}& \left(16\right)\end{array}$

[0034] (11) is multiplied by 2:

[00003]$\begin{array}{cc}{p}_{u,x}^2=\frac{-p_{v,x}^2-2.Math.p_{v,x}.Math.p_{w,x}-p_{w,x}^2+p_{v,x}^2+p_{w,x}^2}{3}& \left(17\right)\end{array}$

[0035] With the binomial formula:

[00004]$\begin{array}{cc}{p}_{u,y}^2=\frac{{\left(p_{v,x}-p_{w,x}\right)}^2}{3}& \left(18\right)\end{array}$

[0036] Like in (1) this yields:

[00005]$\begin{array}{cc}{p}_{u,y}=\frac{1}{\sqrt{3}}.Math.\frac{\left(p_{v,x}-p_{w,x}\right)}{}& \left(19\right)\end{array}$

[0037] From the Cordic method, a rotation matrix is used. The new values

[00006]$\left(\begin{array}{c}x\\ y\end{array}\right)$

after each iteration are derived from the old values

[00007]$\left(\begin{array}{c}{x}_{\mathrm{old}}\\ y_{\mathrm{old}}\end{array}\right)$

as follows:

[00008]$\begin{array}{cc}\left(\begin{array}{c}x\\ y\end{array}\right).Math.=\left(\begin{array}{cc}\cos .Math..Math.& -\sin .Math..Math.\\ \sin .Math..Math.& \cos .Math..Math.\end{array}\right).Math.\left(\begin{array}{c}{x}_{\mathrm{old}}\\ y_{\mathrm{old}}\end{array}\right)& \left(20\right)\end{array}$

[0038] Therefore:

[00009]$\begin{array}{cc}\left(\begin{array}{c}x\\ y\end{array}\right).Math.=\frac{1}{\cos .Math..Math.}.Math.\left(\begin{array}{cc}1& -\tan .Math..Math.\\ \tan .Math..Math.& 1\end{array}\right).Math.\left(\begin{array}{c}{x}_{\mathrm{old}}\\ y_{\mathrm{old}}\end{array}\right).Math.=& \left(21\right)\end{array}$

[0039] Let us now consider the computation of the X component for the U vector:

[00010]$\begin{array}{cc}x=\frac{1}{\cos .Math..Math.}.Math.\left(1-\tan .Math..Math.\right).Math.\left(\begin{array}{c}{p}_{u,x,\mathrm{old}}\\ p_{u,y,\mathrm{old}}\end{array}\right)& \left(22\right)\end{array}$

[0040] Thus from (19):

[00011]$\begin{array}{cc}{p}_{u,x}=\frac{1}{\cos .Math..Math.}.Math..Math.p_{u,x,\mathrm{old}}-\tan .Math..Math..Math.\frac{1}{\sqrt{3}}.Math.\left(p_{v,x,\mathrm{old}}-p_{w,x,\mathrm{old}}\right).Math.& \left(23\right)\end{array}$

[0041] Similar formulas yield the values for p.sub.v,x and p.sub.w,x. To simplify the Cordic method, is chosen such that

[00012]$\frac{1}{\sqrt{3}}.Math..Math.\tan .Math..Math.=2^n$

for the n-iteration. The root from using the root of 3 results from the modification of the Cordic method for the three-phase system.

[0042] Accordingly, an algorithm can be programmed with the following steps. This is shown in FIG. 4.

[0043] The starting point is the step 400. In this case, the angle is =0. The iteration in question is the first one (i=1) and n=1. In the step 410 the rotation direction is determined by the sign (sgn) of the following formula:

Rotation direction=sgn(p.sub.w,xp.sub.v,x)(24)

The values and p.sub.wx are computed in the step 420 as follows:

p.sub.u,x=p.sub.u,x,old2.sup.n(p.sub.v,x,oldp.sub.w,x,old)(25)

p.sub.v,x=p.sub.v,x,old2.sup.n(p.sub.w,x,oldp.sub.u,x,old)(26)

p.sub.w,x=p.sub.w,x,old2.sup.n(p.sub.u,x,oldp.sub.v,x,old)(27)

[0044] Accordingly, the angle has the following value after the first iteration:

Angle.sub.new=Angle.sub.old+Rotation direction.Math.arctan(2.sup.n.Math.{square root over (3)})(28)

[0045] In the next step 430 the value of the iteration is raised by 1, i.e. iteration i=i+1, and the value of the rotation is reduced by half, i.e. N=n1. As long as the value of the iteration lies below the maximum number of iterations (step 440), the computation in step 420 is repeated, wherein the value of the Angle.sub.old corresponds to the value Angle.sub.new computed in the previous iteration.

[0046] When the maximum number of iterations is reached, the value Angle.sub.new is taken.

[0047] The coefficient table with the arctan values (third column) is stored in the processor 40 and speeds up the computation. This coefficient table is implemented in the form of a look-up table, to which fast data access is possible.

[0048] As can be seen in this coefficient table, the change of the angle becomes increasingly small the greater the number of iterations there are. When the required accuracy has been reached, the computation is concluded.

[0049] This method can process the data from the three-phase system directly, since the algorithm can turn the vectors by over 220 (first iteration in the coefficient table). Due to the rotation range that includes over 180, the method is robust vis--vis erroneous decisions. By combining signals of oppositely disposed sensors 25, offsets are compensated as well.

[0050] The angle detection system 5 of the present invention is used, for example, in motors and inverters and in sensors, for example for accelerator pedals.

[0051] The foregoing description has been presented for purposes of illustration and description. It is not intended to be exhaustive or to limit the invention to the precise form disclosed, and modifications and variations are possible in light of the above teachings or may be acquired from practice of the invention. It is intended that the scope of the invention be defined by the claims appended hereto, and their equivalents. The entirety of each of the aforementioned documents is incorporated by reference herein.

LIST OF REFERENCE NUMBERS

[0052] 5 angle detection system [0053] 10 permanent magnet [0054] 20 sensor plate [0055] 25 sensors [0056] 30 circular line [0057] 40 processor [0058] 50 symmetry center