Method and electronic device for the pulse-modulated actuation of a load

Abstract

Method and electronic device for the pulse-modulated actuation of a load in a vehicle, a period duration (T.sub.PM) of a frequency (f.sub.PM) of the pulse modulation being able to be divided into an integer number (N) of sections (T.sub.STEP), the duration of each of which corresponds to a multiple of a period duration (T.sub.OSC) of a clock signal, and the method having the steps of: calculating a frequency (f.sub.PM+1, f.sub.PM) or period duration (T.sub.PM+1, T.sub.PM) of a period of the pulse modulation on the basis of underlying frequency modulation, and determining the duration of a respective section (T.sub.STEP) of a period duration (T.sub.PM) of the pulse modulation using the calculated frequency (f.sub.PM+1, f.sub.PM) or period duration (T.sub.PM+1, T.sub.PM) of a period of the pulse modulation.

Claims

1. A method for the pulse-modulated actuation of a load in a vehicle, a period duration (T.sub.PM) or a frequency (f.sub.PM) of the pulse modulation being able to be divided into an integer number (N) of sections (T.sub.STEP), the duration of each of which corresponds to a multiple of a period duration (T.sub.OSC) of a clock signal, wherein the method comprises: calculating a frequency (f.sub.PM+1, f.sub.PM) or period duration (T.sub.PM+1, T.sub.PM) of a period of the pulse modulation on the basis of underlying frequency modulation; and determining the duration of a respective section (T.sub.STEP) of a period duration (T.sub.PM) of the pulse modulation using the calculated frequency (f.sub.PM+1, f.sub.PM) or period duration (T.sub.PM+1, T.sub.PM) of a period of the pulse modulation; wherein the following steps are carried out in order to determine the duration (T.sub.STEP) of the sections of a period of the pulse modulation, where R.sub.STEP represents a remainder value, f.sub.OSC represents a frequency of the clock signal, N represents the number of sections and f.sub.PM represents the frequency of the pulse modulation: 1. Initializing the remainder (R.sub.STEP) with a value equal to zero at the beginning of a period of the pulse modulation; 2. Performing an allocation: $R_{\mathrm{STEP}}=R_{\mathrm{STEP}}+\frac{f_{\mathrm{OSC}}}{N}-f_{\mathrm{PWM}}$ at the start of a section; 3. Performing an allocation: R.sub.STEP=R.sub.STEP−f.sub.PM after a period of the clock signal; and 4. Terminating the section and performing the second step if the remainder (R.sub.STEP) is less than the frequency of the period of the pulse modulation (f.sub.PM), and otherwise repeating the third step.

2. The method as claimed in claim 1, wherein the frequency (f.sub.PM+1, f.sub.PM) or period duration (T.sub.PM+1, T.sub.PM) of a period of the pulse modulation is calculated by means of a recursive calculation, a frequency (f.sub.PM−1) or period duration (T.sub.PM−1) of a preceding period of the pulse modulation being used.

3. The method as claimed in claim 1, wherein the frequency (f.sub.PM+1) or period duration (T.sub.PM+1) of a subsequent period of the pulse modulation is calculated during an ongoing period of the pulse modulation.

4. The method as claimed in claim 1, wherein the frequency (f.sub.PM+1) or period duration (T.sub.PM+1) of the subsequent period of the pulse modulation is calculated precisely once during an ongoing period of the pulse modulation.

5. The method as claimed in claim 1, wherein the underlying frequency modulation follows a predefined frequency change profile.

6. The method as claimed in claim 1, wherein an approximation that the frequencies (f.sub.PM+1, f.sub.PM) of successive periods of the pulse modulation or the period durations (T.sub.PM+1, T.sub.PM) of the pulse modulation are approximately the same is used to calculate the frequency (f.sub.PM+1, f.sub.PM) or period duration (T.sub.PM+1, T.sub.PM) of the period of the pulse modulation.

7. The method as claimed in claim 1, wherein the duration (T.sub.STEP) of the sections of a period of the pulse modulation is iteratively determined separately for each section using the frequency (f.sub.PM) or period duration (T.sub.PM) of an ongoing period of the pulse modulation.

8. The method as claimed in claim 7, wherein the duration (T.sub.STEP) of a respective section of the period duration (T.sub.PM) of the pulse modulation is also determined on the basis of a ratio value (V.sub.PM) of the duration of a section (T.sub.STEP) to the period duration (T.sub.OSC) of the clock signal.

9. An electronic device for the pulse-modulated actuation of a load in a vehicle, comprising: at least one circuit for generating a pulse-modulated signal, said circuit may be operated using a clock signal, a period duration (T.sub.PM) of the pulse-modulated signal being able to be divided into an integer number (N) of sections (T.sub.STEP), the duration of each of which corresponds to a multiple of a period duration (T.sub.OSC) of the clock signal; said at least one circuit configured to generate a pulse-modulated signal of different period durations (T.sub.PM) or frequencies (f.sub.PM+1, f.sub.PM), the pulse-modulated signal being able to have, within a period duration (T.sub.PM), a non-integer mean value of ratio values (V.sub.PM) of the duration of the sections (T.sub.STEP) of a period of the pulse modulation to the period duration (T.sub.OSC) of the clock signal; wherein the following steps are carried out by at least one circuit in order to determine the duration (T.sub.STEP) of the sections of a period of the pulse modulation, where R.sub.STEP represents a remainder value, f.sub.OSC represents a frequency of the clock signal, N represents the number of sections and f.sub.PM represents the frequency of the pulse modulation: 1. Initializing the remainder (R.sub.STEP) with a value equal to zero at the beginning of a period of the pulse modulation; 2. Performing an allocation: $R_{\mathrm{STEP}}=R_{\mathrm{STEP}}+\frac{f_{\mathrm{OSC}}}{N}-f_{\mathrm{PRM}}$ at the start of a section; 3. Performing an allocation: R.sub.STEP=R.sub.STEP−f.sub.PM after a period of the clock signal; and 4. Terminating the section and performing the second step if the remainder (R.sub.STEP) is less than the frequency of the period of the pulse modulation (f.sub.PM), and otherwise repeating the third step.

10. The electronic device as claimed in claim 9, wherein said at least one circuit is configured to generate sections of different time periods (T.sub.STEP) within a period duration (T.sub.PM) of the pulse-modulated signal.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) Further embodiments emerge from the following description of exemplary embodiments on the basis of figures, in which:

(2) FIG. 1 schematically shows exemplary frequency modulation for generating a pulse-modulated signal (“PM signal”) using an integer divisor V=2;

(3) FIG. 2 schematically shows exemplary frequency modulation for generating a PM signal using a mean divisor V=2.5 over a period T.sub.PWM of the pulse modulation (PM period);

(4) FIG. 3 schematically shows an exemplary frequency profile during frequency modulation;

(5) FIG. 4 schematically shows an exemplary frequency profile using a recursive calculation of the pulse modulation frequency (“PM frequency”);

(6) FIG. 5 schematically shows an example of a binary sequential calculation of the term

(7) $a.Math.\frac{b}{c};$
and

(8) FIG. 6 schematically shows exemplary frequency modulation for generating a PM signal with a mean divisor V.sub.PWM=2.6 over a period T.sub.PWM of the pulse modulation.

DETAILED DESCRIPTION

(9) FIG. 1 shows an example of a period of a conventional PWM actuation. Synchronous digital circuits are operated by means of a base clock with the period duration T.sub.OSC or by means of a clock signal derived therefrom. A PWM period of the duration T.sub.PWM is divided in this case into N sections of the duration T.sub.STEP, each section of the length T.sub.STEP corresponding to an integer number of clock pulses of the period duration T.sub.OSC of the base clock. The ratio T.sub.STEP/T.sub.OSC in this determined by divisor V.sub.PWM. The divisor is V=2 according to the example in FIG. 1.

(10) Depending on a mean current to be set for the purpose of actuating a load, the actuation signal is “1” for a particular number of sections within the PWM period and the actuation is “0” for the further course of the PWM period. This PWM signal can be used to switch a pump driver, for example, on and off.

(11) The section width T.sub.STEP within a PWM period is constant in this case. Therefore, particular discrete period durations T.sub.PWM result on the basis of the base clock with the step width T.sub.OSC:
T.sub.PWM=T.sub.STEP.Math.N=T.sub.OSC.Math.V.sub.PWM.Math.N
where the values of V.sub.PWM and N are integers.

(12) This constitutes a considerable restriction, in particular in mixed-signal circuits, since the oscillator frequency with the period duration T.sub.OSC cannot be stipulated in an arbitrary manner in these cases. If, for example, an oscillator frequency of 50 MHz and N=1024 are chosen, only the discrete frequencies: f.sub.PWM1=48.8 kHz f.sub.PWM2=24.4 kHz f.sub.PWM3=16.3 kHz f.sub.PWM4=12.2 kHz . . .
can be achieved.

(13) FIG. 2 shows an exemplary embodiment in which the time periods T.sub.STEP of the steps within a PM period may have different lengths, these still being formed in the integer ratio to the base clock of the period duration T.sub.OSC, or having a different number of base clocks of the period duration T.sub.OSC. Since pure pulse-width modulation is not necessarily present, the more general expression pulse modulation (“PM”) is used. In contrast to the embodiment according to FIG. 1, the time periods T.sub.STEP of the steps have different divisors V.sub.PM, with the result that the divisor V.sub.PM has a mean value of 2.5 over the period duration T.sub.PM of the PM. This enables comparatively finely tunable frequency modulation of the period durations T.sub.PM of the PM, as a result of which it is possible to distribute interference energy to a wide frequency band. Within a PM period, the maximum error caused by this procedure is therefore a base clock T.sub.OSC.

(14) FIG. 3 illustrates, by way of example, a frequency profile during frequency modulation. Starting from the center frequency f.sub.PM=1/T.sub.PM, the frequency during the ongoing operation of an underlying system or during the actuation of a load is varied in a triangular manner between the two extreme values f.sub.MIN and f.sub.MAX by means of a frequency-modulated voltage or a frequency-modulated current. In this case, the modulation frequency f.sub.FM=1/T.sub.FM determines the speed of the frequency change.

(15) It is explained below how, starting from the frequency profile according to FIG. 3, a non-integer value V.sub.PM can be calculated in order to obtain PM actuation corresponding to the example in FIG. 2.

(16) The following values are given for the calculation: f.sub.PM=1/T.sub.PM Desired center frequency of the PM actuation f.sub.MIN=1/T.sub.MIN Minimum PM actuation frequency f.sub.MAX=1/T.sub.MAX Maximum PM actuation frequency f.sub.FM=1/T.sub.FM Modulation frequency f.sub.OSC=1/T.sub.OSC Base clock of the digital circuit

(17) The frequency profile according to FIG. 3 can be generally represented as a function:

(18) $f\left(t\right)=\frac{f_{\mathrm{MA}X}-f_{MIN}}{t_1}.Math.t+f_{MIN}=\frac{2.Math.\left(f_{\mathrm{MA}X}-f_{MIN}\right)}{T_{FM}}.Math.t+f_{MIN}\mathrm{for}0\le t\le t_1,\mathrm{and}f\left(t\right)=\frac{f_{MIN}-f_{\mathrm{MA}X}}{t_2-t_1}.Math.\left(t-t_1\right)+f_{\mathrm{MA}X}=-\frac{2.Math.\left(f_{\mathrm{MA}X}-f_{MIN}\right)}{T_{PM}}.Math.\left(t-t_1\right)+f_{\mathrm{MA}X}\mathrm{for}{t}_1\le t\le t_2.$

(19) The calculation of the frequency for any desired times t is not absolutely necessary inside a digital circuit for PM actuation since, in particular, the distance between two successive rising signal edges is relevant as the period duration T.sub.PM. It therefore generally suffices to calculate the PM frequency with the period duration T.sub.PM once for each PM period. A recursive method which is explained on the basis of FIG. 4 may be used to calculate the PM frequency f.sub.PM+1 of the next PM period, in which case the following apply:

(20) $\mathrm{df}=f_{\mathrm{PM}+1}-f_{\mathrm{PM}}=\pm \frac{f_{\mathrm{MIN}}-f_{\mathrm{MAX}}}{T_{FM}/2}.Math.\mathrm{dt}=\pm \frac{2.Math.\left(f_{\mathrm{MIN}}-f_{\mathrm{MAX}}\right)}{T_{FM}}.Math.T_{\mathrm{PM}}{f}_{\mathrm{PM}-1}=\mathrm{df}+f_{\mathrm{PM}}=\pm \frac{2.Math.\left(f_{\mathrm{MIN}}-f_{\mathrm{MAX}}\right).Math.f_{\mathrm{FM}}}{f_{\mathrm{PM}}}+f_{\mathrm{PM}}{f}_{\mathrm{PM}+1}=\frac{k}{f_{\mathrm{PM}}}+f_{\mathrm{PM}}\mathrm{with}k=\pm 2-\left(f_{\mathrm{MAX}}-f_{\mathrm{MIN}}\right).Math.f_{FM}.$

(21) The sign ± in this case respectively distinguishes whether there is a rising or falling edge of the PM frequency. The step widths df and dT change with the PM frequency result from the PM period duration, which changes with the PM frequency.

(22) In order to determine the frequency f.sub.PM+1, a calculation of the following form is carried out:

(23) $a.Math.\frac{b}{c}\mathrm{with}a=\left(f_{\mathrm{MAX}}-f_{\mathrm{MIN}}\right),b=f_{\mathrm{FM}}\mathrm{and}c=f_{\mathrm{PM}}.$

(24) There are countless ways of calculating the term

(25) $a.Math.\frac{b}{c}.$

(26) In this case, the exemplary embodiment corresponding to FIG. 5 shows a possible resource-efficient procedure for the binary, sequential calculation thereof. According to this calculation, neither a multiplier nor a divider is required, which is advantageous inside digital circuits since adders, subtractors and shift operations can be implemented in a comparatively simple manner and multipliers and dividers can be implemented in a considerably more complicated manner.

(27) In the calculation example according to FIG. 5, the following bit widths were selected for the variables a, b, and c: n.sub.a=3, n.sub.b=7 and n.sub.c=4.

(28) The following method steps may be carried out: 1. A value Erg is initialized with 0, a step counter is likewise set to 0 and a maximum value (>=n.sub.b−n.sub.c) which specifies the computation accuracy is stipulated for the step counter. 2. If the first n.sub.c bits of the variable b are greater than or equal to the value of c, c is subtracted from the first n.sub.c bits of b. At the same time, a is added to the value Erg. If the first n.sub.c bits of b are less than c, this step is skipped and step 3 is carried out. 3. The step counter is incremented by 1. If the step counter has not reached the maximum value, b and Erg are shifted one bit to the left and the method jumps to step 2. If the counter has reached the maximum value, the computation operation is terminated.

(29) Erg now represents the result of the computation operation involving multiplication and division. This calculation is carried out in each PM period in order to determine the PM frequency f.sub.PM+1 of the next PM period. A division remainder R often occurs during this operation and is preferably taken into account in the next iteration step or when calculating the PM frequency of the PM period which follows the next PM period:

(30) $f_{\mathrm{PM}+1}=\frac{k}{f_{\mathrm{PM}}}+\frac{R}{f_{\mathrm{PM}-1}}+f_{\mathrm{PM}}\mathrm{with}k=\pm 2.Math.\left(f_{\mathrm{MAX}}-f_{\mathrm{MIN}}\right).Math.f_{\mathrm{FM}}.$

(31) In this case, f.sub.PM−1 represents the PM frequency of the preceding PM period, f.sub.PM represents the PM frequency of the current PM period and f.sub.PM+1 represents the PM frequency of the next PM period. However, this would require an additional division operation and a plurality of additional subtraction and shift operations. For further simplification when calculating the next PM frequency, it is assumed that two successive PM periods are very similar, with the result that: f.sub.PM−1≈f.sub.PM.

(32) The following is obtained:

(33) $\begin{array}{cc}{f}_{\mathrm{PM}+1}=\frac{R}{f_{\mathrm{PM}-1}}+\frac{k}{f_{\mathrm{PM}}}+f_{\mathrm{PM}}\approx \frac{R+k}{f_{\mathrm{PM}}}+f_{\mathrm{PM}}& \mathrm{Equation}1\end{array}$

(34) The divisor V.sub.PM+1 of the subsequent PM period can be determined from the frequency f.sub.PM+1 calculated using equation 1:

(35) 0$T_{\mathrm{STEP}}=V_{\mathrm{PM}+1}.Math.T_{\mathrm{osc}}{V}_{\mathrm{PM}+1}=\frac{T_{\mathrm{STEP}}}{T_{\mathrm{OSC}}}=\frac{T_{\mathrm{PM}}/N}{T_{\mathrm{OSC}}}=\frac{f_{\mathrm{OSC}}}{N.Math.f_{\mathrm{PM}+1}}$

(36) The value V.sub.PM+1 is generally not an integer. As already explained for FIG. 2, the step widths within a PM period have different lengths as a result. A division operation is also expediently avoided here for reasons of saving resources.

(37) In digital circuits, the number N of steps with the period duration T.sub.STEP within a PM period is preferably selected as a power of two, that is to say N=2.sup.M, where M is an integer. As a result, the division by N corresponds to a shift operation by M digits. The expression f.sub.OSC/N is expediently constant within PM actuation, thus making it possible to achieve a particularly resource-efficient calculation.

(38) The actual time period T.sub.STEP of the sections of a PM period or the number of base clock periods T.sub.OSC of the respective section is iteratively determined using the following algorithm: 1. A remainder R.sub.STEP is initialized with 0 at the beginning of a pulse modulation period. 2. At the start of a section of the length T.sub.STEP, the following allocation:

(39) $R_{\mathrm{STEP}}\overleftarrow{=}{R}_{\mathrm{STEP}}+\frac{f_{\mathrm{OSC}}}{N}-f_{\mathrm{PM}}$
is carried out, where

(40) $R_{\mathrm{STEP}}+\frac{f_{\mathrm{OCS}}}{N}-f_{\mathrm{PM}}=R_{\mathrm{STEP}}+f_{\mathrm{PM}}.Math.\left(V_{\mathrm{PM}}-1\right)$
and f.sub.PM and V.sub.PM represent the values f.sub.PM+1 and V.sub.PM+1 determined in the preceding PM period using equation 1. 3. After a base clock period T.sub.OSC, the following allocation

(41) $R_{\mathrm{STEP}}\overleftarrow{=}{R}_{\mathrm{STEP}}-f_{\mathrm{PM}}$
is carried out. 4. If the remainder R.sub.STEP is less than f.sub.PM, the step T.sub.STEP is finished and the method jumps to 2 in order to determine the length of the next step, otherwise 3 is repeated.

(42) The frequency f.sub.PM+1 of the next period of the pulse modulation can therefore be recursively determined using the parameters f.sub.OSC/N, f.sub.MIN, f.sub.MAX, f.sub.FM, and f.sub.PM, in which case f.sub.OSC/N, f.sub.MIN, f.sub.MAX, and f.sub.FM, in particular, can also remain constant over a multiplicity of periods of the pulse modulation.

(43) FIG. 6 shows an example of a calculation of the step width T.sub.STEP. The following input variables were used for this example: f.sub.OSC=6.656 MHz N=2.sup.8=256 f.sub.PM=10 kHz
These input variables result in divisors V.sub.PM of 2.6.

(44) A remainder R.sub.STEP is retained from step to step and makes it possible to achieve the desired value of V.sub.PM=2.6 on average.

(45) According to FIG. 6, and also for the embodiment according to FIG. 1, the PM actuation signal is oriented to the number of steps of the step width T.sub.STEP irrespective of the fact that this step width now varies slightly. Within the PM period T.sub.PM, the actuation signal is “1” for a particular number of steps of the step width T.sub.STEP and the actuation is “0” for the further course of the PM period.

(46) Estimation of the Maximum Total Error of Half a PM Period

(47) The simplification f.sub.PM−1≈f.sub.PM results in an error F which can occur during each iterative computation step:

(48) $F=.Math.\frac{R+k}{f_{\mathrm{PM}}}-\left(\frac{R}{f_{\mathrm{PM}-1}}+\frac{k}{f_{\mathrm{PM}}}\right).Math.=.Math.\frac{R}{f_{\mathrm{PM}}}-\frac{R}{f_{\mathrm{PM}-1}}.Math.=.Math.R.Math.\frac{f_{\mathrm{PM}-1}-f_{\mathrm{PM}}}{f_{\mathrm{PM}-1}.Math.f_{\mathrm{PM}}}.Math.$

(49) Within half a PM period, the errors F of a PM period are added to form a total error F.sub.G of half a PM period:

(50) $F_G=.Math.\underset{T/2}{.Math.}F.Math.=.Math.\underset{T/2}{.Math.}R.Math.\frac{f_{\mathrm{PM}-1}-f_{\mathrm{PM}}}{f_{\mathrm{PM}-1}.Math.f_{\mathrm{PM}}}.Math.$

(51) The maximum total error F.sub.MAX of half a PM period can be estimated by:

(52) $F_{\max }=.Math.\underset{T/2}{.Math.}F.Math.<f_{\max }.Math.\frac{f_{\mathrm{MAX}}-f_{\mathrm{MIN}}}{f_{\mathrm{MIN}}.Math.f_{\mathrm{MIN}}}$

(53) This estimation is composed of three considerations, in particular: Remainder R can never become greater than the current frequency f.sub.PM. If the remainder R is replaced with the maximum frequency f.sub.MAX, an estimation is carried out according to the above. The error becomes maximal, the smaller the denominator becomes, and the frequencies f.sub.PM−1 and f.sub.PM are therefore each replaced with f.sub.MIN. The iteration formula results in a sum of the enumerators in the form: f.sub.1−f.sub.2+f.sub.2−f.sub.3+f.sub.3−f.sub.4 . . . =f.sub.MAX−f.sub.MIN.

(54) If an estimation in % is required, it is also possible to divide by the minimum PM frequency f.sub.MIN to give the following:

(55) $F_{\mathrm{MAX}}\left(\%\right)<\frac{f_{\mathrm{MAX}}.Math.\left(f_{\mathrm{MAX}}-f_{\mathrm{MIN}}\right)}{f_{\mathrm{MIN}}.Math.f_{\mathrm{MIN}}.Math.f_{\mathrm{MIN}}}$

(56) For exemplary frequency modulations, the following results arise for the maximum total error of half a PM period:

(57) $\begin{array}{cc}f=10\mathrm{kHz}\pm 1\mathrm{kHz}{F}_{\mathrm{MAX}}<\frac{11.Math.2}{9.Math.9.Math.9.Math.1000}\approx 0.00301\%& \mathrm{Example}1\end{array}$$\begin{array}{cc}f=100\mathrm{Hz}\pm 10\mathrm{Hz}{F}_{\mathrm{MAX}}<\frac{110.Math.20}{90.Math.90.Math.90}\approx 0.301\%& \mathrm{Example}2\end{array}$

(58) In order to minimize the error, the extreme points f.sub.MIN and f.sub.MAX are preferably used as supporting points at which the recursive calculation is restarted by setting f.sub.MAX=f.sub.PM and f.sub.MIN=f.sub.PM in order to calculate f.sub.PM+1 if the PM frequency is exceeded or is the same.