METHOD FOR DETECTING A ROTOR POSITION ANGLE IN AN ELECTRIC MACHINE AND CONTROL FOR IT, AS WELL AS A DRIVE SYSTEM EQUIPPED WITH IT

Abstract

Method (10) for detecting a rotor position angle (W) in an electric machine, wherein magnetoresistive sensors are interconnected in measuring bridge circuits comprising two full bridges (VBs; VBc) each with two half bridges (HB), wherein sensor signals (sinPos, sinNeg) for a sine component (sin) of the rotor position angle (W) are supplied by the half bridges (HB) of one full bridge (VBs) and sensor signals (cosPos, cosNeg) for a cosine component (cos) of the rotor position angle (W) are supplied by the half bridges (HB) of the other full bridge (VBc). are supplied, wherein for each full bridge (VBs; VBc) a difference is formed from the sensor signals (sinPos, ..., cosNeg) supplied by the half bridges (HB) in order to generate measurement signals (sinDiff; cosDiff) with amplitudes that are greater than the amplitudes of the sensor signals (sinPos, ... cosNeg), wherein in the event of an error (12) occurring in one of the sensor signals (cosNeg') for the associated full bridge (VBc) a substitute measurement signal (cosSkal) is calculated using at least one of the other non-errorful sensor signals (cosPos, sinPos, sin Neg) (13'), characterized by the fact that the substitute measurement signal (cosScal) is calculated (13') by applying a scaling factor (SFK) to the non-faulty sensor signal (cosNeg) from the associated full bridge (VBc), which results from the difference between the non-faulty sensor signals (sinPos, sinNeg) from the other full bridge (VBs).

Description

[0001] The invention relates to a method for detecting a rotor position angle in an electrical machine according to the preamble of claim 1, and a control system operating thereafter as well as a drive system equipped therewith according to the preamble of one of the independent claims

[0002] In the field of electrical machines and drives, especially those equipped with brushless DC motors, control is often achieved with the help of sensors that continuously detect the rotor position angle and use this to determine control parameters, particularly for commutation. Typically, magnetoresistive sensors are used for this purpose. These utilize one of the well-known ARM, GRM, or TRM effects to generate a sensor signal when the rotor changes position. This signal changes, in particular, depending on the current rotor angle and usually has a sinusoidal waveform. The individual sensors are often interconnected in measuring bridge circuits, creating two complete full bridges (usually Wheatstone circuits) each consisting of half bridges.The full bridges are arranged offset from one another in the direction of rotation of the rotor so that one full bridge supplies sensor signals for determining the sine component of the desired rotor position angle, while the other full bridge supplies sensor signals for determining the cosine component. In many applications, a difference is formed between the sensor signals in order to generate measurement signals with amplitudes that are greater than the amplitudes of the sensor signals. With the usual difference formation, the amplitudes are amplified by a factor of 2. Such a sensor technology is also described in the article "Measuring angle and linear position with ARM sensors" by Thomas Tzscheetzsch; the article was published in July 2015 in the specialist journal "Elektronikpraxis", issue no. 14 and can be downloaded from the internet at the following link (URL): . http: / / files.vogel.de / vogelonline / vogelonline / issues / ep / 2015 / 014.pdf

[0003] The method mentioned above is also used, among other things, in electric drives designed as electric power steering systems. The method forms the starting point for the present invention and is to be described here with reference to the attached Fig. 1 are described: The Fig. Figure 1 shows a schematic representation of two full bridges, each comprising four sensors and thus consisting of two half bridges HB, each with two sensors. One full bridge VBc supplies sensor signals for the cosine component cos of the desired rotor position angle W; the other full bridge VBs supplies sensor signals for the sine component sin. The time profiles of the positive sensor signals sinPos and cosPos as well as the negative sensor signals sinNeg and cosNeg are shown. The output signals of the sensors are therefore positive and negative sine or cosine waves, which are represented by the movement or rotation of the respective magnet attached to the rotor. In order to be able to perform a more precise angle calculation, a difference is formed between the sensor signals sinPos, sinNeg, cosPos and cosNeg. This is because by forming the difference, e.g.Using (differential) operational amplifiers, differential signals in the form of sine and cosine signals with higher amplitudes can be generated. The amplitudes of the differential signals are typically larger than the amplitudes of the sensor signals by a scaling factor of 2. The differential signals are in the . Fig. 1 are represented as sinDiff and cosDiff and form the sine and cosine components of the desired rotor position angle W.

[0004] One advantage of differential calculation is that, during analog-to-digital conversion on a µC or ASIC, ADC ranges can be utilized as fully as possible. However, differential calculation means that the values ​​of each half-bridge signal are no longer available to the system or controller.

[0005] The angle W can be calculated from the sine and cosine signals. The more accurate the angle, the more efficiently a motor system can be operated, where the rotor position angle forms a reliable input variable for the control or regulation of a drive. As the right-hand curve in Fig. 1 shows, a strictly linear relationship (ideally 1:1) between the calculated rotor position angle W and the actual physical angle on the rotor can be achieved.

[0006] Thus, a method for detecting a rotor position angle in an electrical machine is known from the prior art, in which magneto-resistive sensors are interconnected in measuring bridge circuits comprising two full bridges VBs and VBc, each with two half bridges HB, wherein the sensor signals sinPos, sinNeg for a sine component sin of the rotor position angle W are supplied by the half bridges of one full bridge VBs and the sensor signals cosPos, cosNeg for a cosine component cos of the rotor position angle W are supplied by the half bridges of the other full bridge VBc, and wherein for each full bridge VBs and VBc a difference is formed from the sensor signals supplied by the half bridges in order to generate measuring signals (sinDiff and cosDiff, respectively) with amplitudes that are greater than the amplitudes of the sensor signals sinPos, sinNeg, cosPos and cosNeg, respectively.

[0007] A particular problem arises when a differential signal fails, as this makes it impossible to accurately calculate the angle, e.g., for motor or drive control. This situation is particularly unpleasant with power steering systems and is generally only solved unsatisfactorily: Either the system shuts down immediately in the event of a problem, which results in no longer providing steering assistance and a reduction in the overall availability of steering assistance. Or, in the event of a problem, further system support must be activated using additional sensors, which, however, requires the expense of additional components.

[0008] Furthermore, from the generic document DE 11 2011 104 630 T5, a method for detecting a rotor position angle in an electrical machine is known, wherein in the event of a fault in one of the sensor signals for the associated full bridge, a substitute signal is calculated from the non-faulty sensor signal of the corresponding full bridge.

[0009] Based on this, the present invention addresses the problem of improving a method of the type mentioned above. A control system operating according to this method and a drive system equipped with it are also to be proposed, which should be particularly suitable for use in electric power steering systems.

[0010] The problem is solved by a method having the features of claim 1 as well as by a control system operating thereon and a drive system equipped therewith having the features of one of the independent claims.

[0011] Accordingly, if a fault occurs in one of the sensor signals (e.g., cosNeg'), a substitute measurement signal (cosSkal) is calculated for the corresponding full bridge (here: VBc) using at least one of the remaining non-faulty sensor signals (cosPos, sinPos, sinNeg). This means that if a fault occurs in a half-bridge, the usual differentiated measurement signal (e.g., cosDiff) is not calculated for the corresponding full bridge, but rather a substitute measurement signal (hereinafter referred to as cosSkal) using the remaining sensor signals supplied by the non-faulty half-bridges.

[0012] The equivalent measurement signal (e.g. cosSkal) is calculated by scaling the non-faulty sensor signal (cosPos), which is supplied by the non-faulty half-bridge of the corresponding full bridge, according to the fault-free subtracted measurement signal (here sinDiff), i.e. by applying a scaling factor (e.g. SFK=2) which results from the difference formation of the sensor signals (sinPos, sinNeg) of the other, non-faulty full bridge.

[0013] A controller for an electric machine, in particular for a brushless direct current (BLDC) motor, which operates according to the method is also proposed. Furthermore, a drive system equipped with the controller, in particular an electric motor-driven power steering system, is also proposed.

[0014] The invention is described in more detail below with reference to the accompanying drawings, which show the following schematic representations: Fig. 1 illustrates the starting point of the present invention based on the structure of a sensor with two full bridges and the usual signal processing of the sensor signals; Fig. 2a-b illustrate the problem case when a half-bridge fails or is faulty, so that a short circuit to +5 volts occurs and one of the supplied sensor signals (here sinPos') is faulty, which also makes the difference signal sinDiff" unusable; Fig. 2c-d illustrate the effects on the determined or calculated rotor angle W if no compensation measures are carried out; Fig. 3a-c illustrate another problem case and its effects, where there is a short circuit to +3 volts and one of the sensor signals (here sinNeg') is faulty; Fig. 4a-c illustrate how such a problem case (here short circuit from cosPos to +3 volts) can be solved using the method according to the invention by calculating a replacement measurement signal to compensate for the failure (here of cosPos); Fig. 5a-c illustrate how the method according to the invention can also be used to solve a problem case in which a sensor signal in each of the two full bridges is faulty or fails for further signal processing; Fig. 6 illustrates the performance of the method according to the invention using an additional radius diagnosis; Fig. 7 shows a schematic flow diagram for the method according to the invention; and Fig. 8 shows the schematic structure of a drive system in the form of a power steering system operated with a control system operating according to the method.

[0015] With reference to the Fig. 1, the starting point of the present invention is first described:

[0016] The sensor signals generated by the half-bridges HB can be described by the following sine or cosine functions: SinPos=ASinPos*Sin(F*a+φ)+OSinPos SinNeg=ASinNeg*Sin(F*a+φ+180°)+OSinNeg SinNeg=ASinNeg*−Sin(F*a+φ)+OSinNeg CosPos=ACosPos*Cos(F*a+φ)+OCosPos CosNeg=ACosNeg*Cos(F*a+180°)+OSinNeg CosNeg=ACosNeg*−Cos(F*a+φ)+OCosNeg

[0017] Formulas (I)-(III) refer to the lower full bridge VBs, which supplies sensor signals for the sine component sin; and formulas (IV)-(VI) refer to the upper full bridge VBc, which supplies sensor signals for the cosine component cos, whereby a difference is formed from the respective two sensor signals (e.g., cosPos and cosNeg) in order to obtain a difference signal (cosDiff) with a higher amplitude.

[0018] The formulas (I), (II) and (IV), (V) describe the respective signal as a mathematical sine or cosine function, whereby a 180° phase shift must be taken into account for the respective negative signal (sinNeg or cosNeg).

[0019] Formulas (III) and (VI) are alternative representations to formulas (II) and (V), respectively, in which the negated function (-sin or -cos) is used instead of the 180° phase shift.

[0020] The factors AsinPos , AsinNeg , A CosPos , A CosNeg are the respective amplitudes of the half-bridge signals. The factors O SinPos , O SinNeg , O CosPos , O CosNeg are the respective offsets of the half-bridge signals. In most cases, the offset values / voltage for typical sensors are exactly half the supply voltage of the respective half-bridge (offset = supply voltage / 2).

[0021] The factor F, which appears in all formulas, is the frequency of the sine / cosine waves during one revolution of a magnet. The factor F depends on the number of pole pairs and the magnetic measuring principle. For an AMR sensor, F=2, and for other measuring technologies, such as GMR and TMR, F=1. If the number of pole pairs of the magnet increases, "F" is doubled. For example, for a measuring system with an AMR sensor and a magnet with a pole pair number of 1, the factor is F=2*1=2. For a measuring system with a TMR sensor and a magnet with a pole pair number of 1, the factor is F=1*1=1.

[0022] The factor -α, which appears in all formulas, is the current position of the rotor during one revolution, i.e., the mechanical angle of the rotor in an electric motor, for example. The factor -φ is an additional phase shift of the signals. This is ideally zero degrees.

[0023] The difference signals can be described using formula (I) and formula (III) for the sine and using formula (IV) and formula (VI) for the cosine as follows, ie the derivation for the difference sine signal is as follows: Formulas for the difference sine wave as a single mathematical sine function: SinDiff=SinPos−SinNeg SinDiff=ASinPos*Sin(F*α+φ)+0SinPos−(ASinNeg*−Sin(F*α+φ) +0SinNeg) SinDiff=ASinPos*Sin(F*α+φ)+0SinPos+ASinNeg*Sin(F*α+φ) -0SinNeg SinDiff=Sin(F*α+φ)*(ASinPos+ASinNeg)+0SinPos−0SinNeg

[0024] With O SinPos = O SinNeg and A SinPos = A SinNeg , the result is a sine wave that corresponds to the phase of the positive sine half-bridge signal and also has no offset component and twice the amplitude of the half-bridge signal. Thus, the scaling factor SFK = 2. −math. formula for the resulting difference sine signal SinDiff=SFK*ASinPos*Sin(F*α+φ)=2*ASinPos*Sin(F*α+φ)

[0025] The same applies to the difference cosine signal: Formulas for the difference cosine wave as a single mathematical cosine function: CosDiff=CosPos−CosNeg CosDiff=ACosPos*Cos(F*α+φ)+0CosPos−(AcosNeg*−Cos(F*α+φ)+0CosNeg) CosDiff=ACosPos*Cos(F*α+φ)+0CosPos+AcosNeg*Cos(F*α+φ)−0CosNeg CosDiff=Cos(F*α+φ)*(AcosPos+AcosNeg) +0CosPos−0CosNeg

[0026] With O CosPos = O CosNeg and A CosPos = A CosNeg , a cosine wave results which corresponds to the phase of the positive cosine half-bridge signal and also has no offset component and twice the amplitude (SKF = 2) of the half-bridge signal. −math. formula for the resulting difference cosine signal CosDiff=SFK*AcosPos*Cos(F*α+φ), where SFK=2

[0027] The invention is based on the knowledge that formulas (VIII) and (X) can only be used to obtain “true” difference signals (see sinDiff and cosDiff in Fig. 1) can be formed if all half-bridges provide the required sine or cosine waveforms (negative as well as positive).

[0028] This problem occurs when a half-bridge malfunctions and, for example, delivers a half-bridge signal or sensor signal with a fixed value, i.e., it does not generate a sensor signal with a sine or cosine waveform. This can occur, for example, in the case of a short circuit to a fixed voltage.

[0029] For example, one of the lower half bridges HB is in Fig. 1 is defective and delivers an unusable (e.g. constant) sensor signal sinPos = const.

[0030] The Fig. 2a-d illustrate such a problem case, where, for example, the sensor signal sinPos is faulty and here shows a constant signal curve due to a short circuit to +5V. The faulty signal is in Fig. 2a is denoted by sinPos'. In Fig. 2b it becomes clear that the difference signal sinDiff' is also faulty and is not suitable for angle calculation.

[0031] The Fig. 2c shows the curve of the ideal target angle W* compared to the incorrectly calculated angle W. This would result in a significant linearity error dW with conventional difference formation and angle calculation, which in Fig. 2d is illustrated.

[0032] Another exemplary problem case is illustrated by the Fig. 3a-c. Here, a short circuit from sinNeg to +3V leads to significant errors in the difference signal sinDiff' ( Fig. 3a) and finally to angular deviations from the ideal case ( Fig. 3b) or linearity errors ( Fig. 3c).

[0033] The invention is based on the new approach that in the case of a voltage drop in one of the half bridges (e.g. one of the lower HB in Fig. 1) errors occurring, the usual subtraction of the sensor signals for calculating the difference signal (processed measurement signals) sinDiff or cosDiff is not carried out, but a substitute measurement signal (hereinafter also called scaled signal) is calculated using the remaining sensor signals supplied by the non-faulty half-bridges.

[0034] If, for example, sinPos is faulty and the processed measurement signal sinDiff=sinDiff' becomes unusable (see Fig. 2a-d), then a substitute measurement signal sinSkal is calculated. This can preferably be achieved by calculating the substitute measurement signal by scaling the still error-free sensor signal (i.e., sinNeg) of this full bridge (here, VBs) to the error-free difference signal (i.e., cosDiff) of the other full bridge (VBc). In particular, the error-free sensor signal sinNeg is subjected to a scaling factor SFK, which results from the difference between the other error-free sensor signals cosPos and cosNeg.

[0035] In other words, the above formulas (VII) and (IX) are reformulated. The procedure will now be explained in detail using examples, with reference to the attached figures:

[0036] In formula (VI), the negative sine half-bridge signal sinNeg is short-circuited with "X" (e.g. to +3 volts). This results in the following equation for the differential signal sinDiff to be calculated (the problem case is shown in Fig. 3a-c): Formulas for the fault case where sinNeg is short-circuited to any supply voltage. The resulting differential signal is similar to the positive half-bridge signal, only with a different offset and amplitude: SinDiff'=SinPosSinNeg SinDiff'=ASinPos*Sin(F*α+φ)+0SinPos−X

[0037] The differential sine signal sinDiff' is faulty, but almost identical to the sine wave of the positive sine half-bridge signal sinPos, with the only difference being that the offset is subtracted by the amount X and the amplitude is smaller by a factor of 2 than a differential signal without errors (sinDiff).

[0038] This also applies to a short circuit for the positive sine half-bridge signal or also for the cosine half-bridge signals: Formulas for the error case where cosNeg is short-circuited to an arbitrary supply voltage. The resulting differential signal is similar to the positive half-bridge signal, only with a different offset and amplitude: SinDiff=X−SinNeg SinDiff=X−(ASinNeg*−Sin(F*α+φ)+0SinNeg)=X+ASinNeg*Sin(F*α+φ)−0SinNeg SinDiff=ASinNeg*Sin(F*α+φ)−0SinNeg+X CosDiff=CosPos−CosNegCosDiff=ACosPos*Cos(F*α+φ)+0CosPos−X CosDiff=X−CosNeg CosDiff=X−(AcosNeg*−Cos(F*α+φ)+0CosNeg)=X+AcosNeg *Cos(F*α+φ)−0CosNeg CosDiff=AcosNeg*Cos(F*α+φ)−0CosNeg+X

[0039] To illustrate the effect of short circuits on half-bridge signals, we refer again to the Fig. 2 and Fig. 3 Reference is made to: An error e.g. short circuit to 5V of a half-bridge signal (here e.g. Sin Pos ) would have the following effect (see Fig. 2a-d): For Sin Neg , Cos Pos and Cos Neg The amplitude should be 0.25 V; the offset should be 2.5 V and the frequency should be 1.

[0040] By forming the difference between the sinusoidal signals, a differential sinusoidal signal sinDiff' with an offset component and a smaller amplitude would be created (Formula XII), where the offset = -2.5+5= 2.5

[0041] An angle calculated from this (from sinDiff' and cosDiff) would no longer have a linear progression (see also Fig. 2d).

[0042] Depending on the error characteristics (short circuit to different levels) of a half-bridge signal, the angle curve has a different “shape” or “linearity error behavior” (compare with Fig. 3a-c).

[0043] The Fig. 2c and Fig. The angle curves shown in 3b above are only intended to show that, depending on the half-bridge error, any incorrect angle errors can occur.

[0044] If such an incorrect angle signal is provided, it would lead to major problems, for example, when controlling a motor in a steering system (unintended actuator functions, stiff steering, etc.). Therefore, this signal must no longer be used. The following measures are typically taken to address such problems: • Switching off the system e.g. steering assistance no longer available • Further support through additional sensors.

[0045] The present invention prevents immediate system shutdown ("sudden loss of assistant") in the event of clearly detected short circuits of a half-bridge signal to fixed voltage levels (DC). The costs of additional sensors can also be avoided.

[0046] According to the invention, these disadvantages (sudden shutdown or costs due to additional sensors) are avoided by using signals with high amplitudes at the half-bridge level. This is preferably possible with TMR technology. Other technologies such as AMR or GMR are also conceivable. However, since the sensitivity (~amplitude) is lower here, additional amplification at the half-bridge level would be required. This would mean that each half-bridge signal would have to be amplified separately, e.g. using operational amplifiers. Amplifying the half-bridge signals has the disadvantage that each amplifier stage introduces different drifts due to, among other things, temperature, age and tolerances. This would impair the actual operation, which delivers the best performance with differential signals.

[0047] TMR technology makes it possible to generate a high signal amplitude even without amplifying the half-bridge signals. This allows for high performance in normal operation (difference generation is still partially necessary) and allows for emergency operation (half-bridge signals are included with the system).

[0048] In the event of a fault, emergency operation is possible: The angle is calculated from the still clearly present half-bridge signals (sine wave(s) and cosine wave(s)). The method according to the invention is also explained using the flow chart in Fig. 7 described: Method 10 essentially comprises the following steps: After the start S, normal operation (block 11) is initially performed. In step 12, the half-bridges and their sensor signals are continuously checked for any errors.

[0049] In the error-free case, the rotor angle calculation is carried out in step 13 using the usual difference calculation.

[0050] However, if a fault is clearly detected on a half-bridge in step 12, the program branches to block 13' to determine a substitute measurement signal. It is then still possible to calculate an angle from sine and cosine (step 14).

[0051] The calculation of the equivalent measurement signal (replacement for the respective difference signal) can be done as follows (for better understanding, the errors on the sine waves are considered here. This also applies analogously to cosine half-bridge errors): Two half-bridge signals of a full bridge (e.g. cosine) are present and only one half-bridge signal is available.

[0052] A difference is calculated from the two half-bridge signals of a full bridge (e.g., cosine), and the half-bridge signal of the other full bridge is scaled so that it corresponds to the original difference signal of the full bridge in terms of amplitude, phase, and offset. This still ensures sufficient signal performance / angle quality, since a difference is still calculated for a full-bridge signal (e.g., cosine).

[0053] For further explanation, two error constellations or calculation steps for the full sinusoidal bridge are considered: Case A): At the lower full bridge VBs (see Fig. 1) only the sensor signal Sin Pos available -> The other sensor signal Sin Neg is clearly discovered with an error and is therefore not used further.

[0054] The available positive sine half-wave sinPos should now be scaled to the difference cosine signal.

[0055] The scaled signal corresponds to the difference size of the full sine bridge. −Formula for scaling sinPos so that it corresponds to the difference sine signal: sinSkal=2*(SinPos−0SinPos)=2*ASinPos*Sin(F*α+φ)=SinDiff

[0056] Case B): At the lower full bridge VBs (see Fig. 1) only the sensor signal Sin Neg available -> The other sensor signal Sin Pos is clearly discovered with an error and is therefore not used further.

[0057] The available negative sine half-wave should now be scaled to the difference cosine signal.

[0058] The scaled signal corresponds to the difference size of the sine bridge. For the negative half-waves, it should be noted that these should be multiplied by a "minus" (Formula III) for Sin Neg and (Formula VI) for Cos Neg . −Formula for scaling sinNeg so that it corresponds to the difference sine signal: sinScal=−2*(SinNeg−0SinNeg)=−2*(ASinNeg*−Sin(F*α+φ)+0SinNeg−0SinNeg)sinSkal=−2*ASinNeg*Sin(F*α+φ)=sinDiff

[0059] The same applies to the cosine bridge. This requires an adjustment to the cosine notation.

[0060] For a better understanding, the example shown here is where the positive cosine half-bridge signal cosPos is short-circuited to +3V (see Fig. 4a). This case has been reliably identified and thus, with regard to the cosine full bridge VBc (see Fig. 1) only the negative cosine half-bridge signal cosNeg can be used. The calculation is performed using formula (XVI).

[0061] With the help of the invention, the failure of a half-bridge or sensor signal can be compensated at any time (see also Fig. 4a-c).

[0062] It is also possible, if another half-bridge signal of another full bridge can no longer be used, to calculate an angle: Depending on the error case, the formulas (XV) and (XVI) can be used here and an angle can be calculated exclusively using the (two) scaled values ​​sinSkal and cosSkal.

[0063] The angle is then calculated from the still available error-free half-bridge signals (e.g. sinNeg and cosPos - see the Fig. 5a-c).

[0064] Then with Vdd = supply voltage of e.g. =5V: cosSkal=SKL*cosPos−OcosNeg=2*cosPos−2.5Volt(→2.5V=Vdd / 2) sinSkal=−SKL*sinNeg−OsinNeg=−2*sinNeg−2.5Volt(→2.5V=Vdd / 2)

[0065] The examples above show how an angle can be calculated from half-bridge signals even if a half-bridge signal from a full bridge may no longer be used ( Fig. 4a-c) or if even one half-bridge signal per full bridge is faulty ( Fig. 5a-c).

[0066] It is also possible to create diagnoses that can even be used during normal operation (angles calculated from differential signals).

[0067] Not only the angle can be calculated from the differential signals, but also that of the half-bridges. Using a "voter principle" (comparator / decision maker) and together with other diagnostics, an angle error due to incorrect signals can be clearly detected (see step 12 in Fig. 7).

[0068] Furthermore, it is possible to perform a radius diagnosis for the half-bridge signals, as sine and cosine values ​​are available. The radius diagnosis can be carried out using the Fig. 6 and can be used in step 15 of the procedure (see Fig. 7) are carried out: If the sine sin is plotted against the cosine cos, then, for example, the rotation of a magnet, a signal circle results. This circle then has a radius calculated from the sine and cosine. The radius can be calculated, for example, using the following formula: Radius=√(sin2+cos2)

[0069] Ideally, the radius should move along an ideal curve NC, referred to here as the "NormCircle" and shown in dashed lines. Since the analog signals vary slightly due to the system's sensor and environmental conditions, the radius is not always on the "ideal line" NC. Therefore, tolerance limits (shown here as LL and HL) are provided in the system. LL represents a lower limit ("Low_Limit") and HL an upper limit ("High_Limit"). The radius should be within this tolerance range or "band."

[0070] Here, too, it is possible, in normal operation (difference signals are still present), to additionally calculate a radius using the scaled sine and cosine values ​​and to detect errors more quickly using a voter principle.

[0071] The procedure ( Fig. 7) is carried out by a control STR, which controls an electrical machine, here e.g. a brushless DC motor M, which is installed in a drive system, here e.g. in an electric motor-driven power steering system L (see Fig. 8). The invention enables reliable control, whereby even in the event of sensor signal failure, at least the emergency operation described above is possible.

Claims

[1] Method (10) for detecting a rotor position angle (W) in an electrical machine, wherein magneto-resistive sensors are interconnected in measuring bridge circuits comprising two full bridges (VBs; VBc) each with two half bridges (HB), wherein the half bridges (HB) of one full bridge (VBs) supply sensor signals (sinPos, sinNeg) for a sine component (sin) of the rotor position angle (W) and the half bridges (HB) of the other full bridge (VBc) supply sensor signals (cosPos, cosNeg) for a cosine component (cos) of the rotor position angle (W) are supplied, wherein for each full bridge (VBs; VBc) a difference is formed from the sensor signals (sinPos, ..., cosNeg) supplied by the half bridges (HB) in order to generate measurement signals (sinDiff; cosDiff) with amplitudes that are greater than the amplitudes of the sensor signals (sinPos, ... cosNeg), wherein in the event of an error (12) occurring in one of the sensor signals (cosNeg') for the associated full bridge (VBc), a substitute measurement signal (cosSkal) is calculated (13') using at least one of the remaining non-faulty sensor signals (cosPos, sinPos, sinNeg), characterized by , that the substitute measurement signal (cosSkal) is calculated (13') by applying a scaling factor (SFK) to the non-faulty sensor signal (cosNeg) from the associated full bridge (VBc), which results from the difference formation of the non-faulty sensor signals (sinPos, sinNeg) from the other full bridge (VBs). [2] Method (10) for detecting a rotor position angle (W) according to claim 1, characterized by that the substitute measurement signal (cosSkal) is calculated (13') by applying an offset value (OsinPos) to the non-faulty sensor signal (cosNeg) from the associated full bridge (VBc), which offset value results from the level of the non-faulty sensor signals (sinPos,sinNeg) from the other full bridge (VBs). [3] Method (10) for detecting a rotor position angle (W) according to claim 1, characterized by that the scaling factor (SFK) is calculated from a difference between the amplitudes of the non-faulty sensor signals (sinPos,sinNeg). [4] Method (10) for detecting a rotor position angle (W) according to one of the preceding claims, characterized bythat in the event that in each of the two full bridges (VBs; VBc) an error occurs in one of the sensor signals (sinPos'; cosNeg'), that for the sine component (sin) and the cosine component (cos) of the rotor position angle (W) a substitute measurement signal (sinSkal, cosSkal) is calculated using the remaining non-faulty sensor signals (sinNeg; cosPos). [5] Method (10) for detecting a rotor position angle (W) according to claim 4, characterized by that the remaining non-faulty sensor signals (sinNeg; cosPos) are subjected to a scaling factor (SFK*) which results from the difference formation between the amplitudes of these non-faulty sensor signals (sinNeg; cosPos). [6] Method (10) for detecting a rotor position angle (W) according to claim 4 or 5, characterized bythat the respective substitute measurement signal (sinSkal; cosSkal) is calculated by applying an offset value (OsinNeg) to the non-faulty sensor signals (sinNeg; cosPos) which results from the level of the non-faulty sensor signals (sinPos,sinNeg). [7] Method (10) for detecting a rotor position angle (W) according to one of the preceding claims, characterized by that the presence of at least one faulty sensor signal (sinPos', cosNeg') is detected by means of comparison and decision operations which are carried out on all supplied sensor signals (sinPos, ... cosNeg), in particular according to the voter principle (12). [8] Method (10) for detecting a rotor position angle (W) according to one of the preceding claims, characterized by that a radius diagnosis for the sine and cosine components is carried out using the supplied sensor signals (sinPos, ... cosNeg) (15). [9] Controller (STR) for an electrical machine, in particular a brushless DC motor (M), and for detecting a rotor position angle (W) of the electrical machine, wherein magneto-resistive sensors are interconnected in measuring bridge circuits comprising two full bridges (VBs; VBc) each with two half bridges (HB), wherein the half bridges (HB) of one full bridge (VBs) supply sensor signals (sinPos, sinNeg) for a sine component (sin) of the rotor position angle (W) and the half bridges (HB) of the other full bridge (VBc) supply sensor signals (cosPos, cosNeg) for a cosine component (cos) of the rotor position angle (W), wherein the controller (STR) has signal processing means which, for each full bridge (VBs; VBc), from the sensor signals (sinPos, ..., cosNeg) supplied by the half bridges (HB), Form a difference to generate measurement signals (sinDiff; cosDiff) with amplitudes that are larger than the amplitudes of the sensor signals (sinPos, ... cosNeg), where the controller (STR) calculates a substitute measurement signal (cosSkal) for the associated full bridge (VBc) using at least one of the remaining non-faulty sensor signals (cosPos, sinPos, sin Neg) in the event of an error occurring in one of the sensor signals (cosNeg'), characterized by , that the controller (STR) calculates the equivalent measurement signal (cosSkal) by applying a scaling factor (SFK) to the non-faulty sensor signal (cosNeg) from the associated full bridge (VBc), which results from the difference between the non-faulty sensor signals (sinPos, sinNeg) from the other full bridge (VBs). [10] Control (STR) for an electrical machine according to claim 9, characterized by that the control (STR), in particular the signal processing means, are designed to carry out the method (10) according to one of claims 2 to 8. [11] Drive system, in particular electromotive power steering system (L), which is equipped with a controller (STR) and an electric machine, in particular a brushless DC motor (M), wherein the controller (STR) controls the electric machine and detects a rotor position angle (W) of the electric machine, wherein magneto-resistive sensors are interconnected in measuring bridge circuits which comprise two full bridges (VBs; VBc) each with two half bridges (HB), wherein the half bridges (HB) of one full bridge (VBs) supply sensor signals (sinPos, sinNeg) for a sine component (sin) of the rotor position angle (W) and the half bridges (HB) of the other full bridge (VBc) supply sensor signals (cosPos, cosNeg) for a cosine component (cos) of the rotor position angle (W), wherein the controller (STR) has signal processing means which, for each full bridge (VBs; VBc), sensor signals (sinPos, ...) supplied by the half-bridges (HB), cosNeg) to generate measurement signals (sinDiff; cosDiff) with amplitudes that are larger than the amplitudes of the sensor signals (sinPos, ... cosNeg), where . the controller (STR) calculates a substitute measurement signal (cosSkal) for the associated full bridge (VBc) using at least one of the remaining non-faulty sensor signals (cosPos, sinPos, sin Neg) in the event of an error occurring in one of the sensor signals (cosNeg'), characterized by , that the controller (STR) calculates the equivalent measurement signal (cosSkal) by applying a scaling factor (SFK) to the non-faulty sensor signal (cosNeg) from the associated full bridge (VBc), which results from the difference between the non-faulty sensor signals (sinPos, sinNeg) from the other full bridge (VBs).

Images