Method for Automatic Frequency Adaptation of Filters During Operation in Closed Control Loops

Abstract

The present invention relates to a method for adjusting the resonance frequency of a loop filter in a delta-sigma modulator, e.g. in an angular rate sensor, to a predetermined frequency value, wherein the sigma-delta modulator comprises an input terminal, which is connected to the loop filter, a quantizer, which is connected to an output of the loop filter, and a feedback branch, which couples an output of the quantizer back to the input terminal. The method comprises the following steps: Optional rough adjustment of the resonance frequency of the filter by means of the regulating variable of a second oscillator, input of a filter input signal of the loop filter into a frequency adjustment circuit, determination of a noise spectrum of the filter input signal in a first frequency band and a second frequency band, wherein the first frequency band and the second frequency band are arranged symmetrically around the predetermined frequency, comparison of the noise spectra and creation of an adjustment signal that leads to a frequency adjustment when the noise spectra deviate from one another, and feedback of the adjustment signal of the frequency adjustment circuit to a control input of the loop filter for setting the filter frequency in response to the comparative result.

Claims

1. A method for adjusting the resonance frequency of a loop filter in a delta-sigma modulator to a predetermined frequency value, wherein the delta-sigma modulator comprises: an input terminal that is connected to the loop filter, a quantizer that is connected to an output of the loop filter, and a feedback branch that couples an output of the quantizer back to the input terminal, wherein the method comprises the following steps: inputting a filter input signal of the loop filter into a frequency adjustment circuit, determining a noise spectrum of the filter input signal in a first frequency band and a second frequency band, wherein the first frequency band and the second frequency band are arranged symmetrically around the predetermined frequency, comparing the noise spectra and generating an adjustment signal that causes a frequency adjustment when the noise spectra deviate from one another, and feeding back the adjustment signal of the frequency adjustment circuit to a control input of the loop filter for setting the filter frequency in response to the comparative result.

2. The method according to Claim 1, wherein the first and the second frequency band are each demodulated into a base band for comparing the noise spectra and wherein the overall noise power in the two frequency bands is compared.

3. The method according to Claim 1, further comprising the step of an initial adjustment of the resonance frequency of the loop filter through entry of the predetermined frequency value as a starting value.

4. A delta-sigma modulator with an input terminal that is connected to the loop filter, a quantizer that is connected to an output of the loop filter, and a feedback branch that couples an output of the quantizer back to the input terminal, wherein the delta-sigma modulator further comprises: a frequency adjustment circuit whose input is connected to an input of the loop filter for receiving a filter input signal of the loop filter and whose output is coupled back to a control input of the loop filter, wherein the frequency adjustment circuit has a first and a second demodulator branch that are operated to determine a noise spectrum of the filter input signal in a first frequency band and a second frequency band, wherein the first frequency band and the second frequency band are arranged symmetrically around the predetermined frequency, and wherein the frequency adjustment circuit further has a comparative unit that is operated to compare the noise spectra and to generate an adjustment signal, which causes a frequency adjustment when the noise spectra deviate from one another, at the output of the frequency adjustment circuit.

5. The delta-sigma modulator according to Claim 4, wherein the frequency adjustment circuit further comprises a pre-amplifier for amplifying the filter input signal that is arranged between the input of the frequency adjustment circuit and the demodulator branches.

6. The delta-sigma modulator according to Claim 4, wherein the first demodulator branch has a first multiplier that is operated to multiply the filter input signal with a first comparative frequency that is lower than the predetermined frequency value, wherein the second demodulator branch has a second multiplier that is operated to multiply the filter input signal with a second comparative frequency that is higher than the predetermined frequency value and wherein the first and the second comparative frequencies are arranged symmetrically around a predetermined frequency value.

7. The delta-sigma modulator according to Claim 4, wherein each of the demodulator branches has at least one filter element.

8. The delta-sigma modulator according to Claim 4, wherein each of the demodulator branches has a squaring device that is connected to an adding device to determine a difference of the signals that are applied to the outputs of the squaring devices and to output a difference value.

9. The delta-sigma modulator according to Claim 4, wherein each of the demodulator branches respectively has an absolute value element for determining an absolute value and the outputs of the absolute values are connected to an adding device to determine a difference of the signals that are applied to the outputs of the absolute value elements and to output a difference value.

10. The delta-sigma modulator according to Claim 8, further comprising a control unit that is impinged with the difference value and that creates the adjustment signal based on said difference value.

11. A circuit arrangement for reading out of a capacitive angular rate sensor with at least one primary mass and at least one secondary mass that is connected to the primary mass, wherein the primary mass is excited to a primary oscillation during operation and wherein the secondary mass is deflected from a resting position in a direction that is transversal to the primary oscillation when the angular rate sensor rotates around a sensitive axis that is transversal to the direction of the primary oscillation and to the direction of the deflection of the secondary mass, wherein the circuit arrangement comprises: a delta-sigma modulator with at least one control loop to perform a force feedback that resets the secondary mass into its resting position by means of applying a reset signal, wherein the reset signal forms a modulator output signal of the delta-sigma modulator, wherein the delta-sigma modulator is formed by a delta-sigma modulator according to claim 4 and wherein the predetermined frequency value is the frequency of the primary oscillation, a control unit for calculating and outputting an angular rate signal out of the modulator output signal.

12. The circuit arrangement according to Claim 11, wherein the frequency of the primary oscillation is controlled by means of a closed phase-locked control loop that has an oscillator, and wherein the phase-locked control loop is connected to a control input of the loop filter for receiving a control signal of the oscillator.

13. The circuit arrangement according to Claim 11, as far as dependent on Claim 6, wherein the phase-locked control loop can be operated further to output the first and the second comparative frequency to the frequency adjustment circuit.

14. The circuit arrangement according to Claim 11, further comprising a memory device for saving the adjustment signal.

15. A coriolis angular rate sensor with at least one primary mass and at least one secondary mass that is connected to the primary mass, wherein the primary mass is excited to perform a primary oscillation during operation and wherein the secondary mass is deflected in a direction that is transversal to the primary oscillation when the Coriolis angular rate sensor rotates around a sensitive axis, wherein the Coriolis angular rate sensor has a circuit arrangement according to Claim 11 for reading out an angular rate signal.

Description

[0045] The FIGS. show:

[0046] FIG. 1 a schematic diagram of an angular rate sensor based on the Coriolis effect;

[0047] FIG. 2 a schematic diagram of a secondary control loop based on the principle of delta-sigma modulation for the operation of an angular rate sensor;

[0048] FIG. 3 a graph of the spectral power density of the output of the secondary control loop in case of an incorrect adjustment between the primary resonance frequency and the filter frequency of the loop filter;

[0049] FIG. 4 a first known arrangement for readjustment of the filter frequency:

[0050] FIG. 5 a second known arrangement for readjustment of the filter frequency;

[0051] FIG. 6a a graph of the spectral power density of the output of the secondary control loop in case of a match between the primary resonance frequency and the filter frequency of the loop filter;

[0052] FIG. 6b a graph of the spectral power density of the output of the secondary control loop in case of incorrect adjustment between the primary resonance frequency and the filter frequency of the loop filter;

[0053] FIG. 7a a further known arrangement for readjustment of the filter frequency;

[0054] FIG. 7b a further known arrangement for readjustment of the filter frequency;

[0055] FIG. 8a a schematic diagram of a secondary control loop with input test signals;

[0056] FIG. 8b a schematic diagram of a circuit for evaluating the test signals that were input in the secondary control loop;

[0057] FIG. 9a a graph of the spectral power density of the input of the electric filter of the secondary control loop as well as the frequency bands evaluated according to the invention in case of a match between the primary resonance frequency and the filter frequency of the loop filter;

[0058] FIG. 9b a graph of the spectral power density of the input of the electric filter of the secondary control loop as well as the frequency bands according to the invention in case of incorrect adjustment between the primary resonance frequency and the filter frequency of the loop filter;

[0059] FIG. 10 a schematic diagram of an angular rate sensor based on the Coriolis effect with a frequency adjustment circuit according to an exemplary embodiment of the present invention.

[0060] An advantageous embodiment of the present invention will be described in the following with reference to the FIGS. 9 and 10. According to the present invention, the signal V.sub.fllt that was input in the loop filter is tapped and input in a frequency adjustment circuit 100 for adjusting the filter frequency f.sub.t of the loop filter H.sub.f(s) 110 to the resonance frequency f.sub.d of the primary oscillation. In the displayed embodiment, the system for frequency adjustment of the electric filter H.sub.f(s) comprises a component for rough initial alignment of the frequencies f.sub.t with f.sub.d and a component for frequency adjustment that works in the background during operation.

[0061] The component for rough alignment is in particular necessary to balance the fluctuations during production of the angular rate sensor and therefore the variation of the primary resonance frequency f.sub.d. For this purpose, the control signal of the oscillator of the phase-locked loop (PLL) that exists in the primary control loop is used. This oscillator is aligned by the components within the primary control loop with the primary resonance frequency of the angular rate sensor. Through an alignment of the control characteristic of the oscillator with the frequency control of the electric filter in the secondary control loop, a rough alignment of the frequencies f.sub.t and f.sub.d can therefore take place for example at the start of the operation.

[0062] To reach the required accuracy, the present invention suggests a frequency adjustment that runs in the background in addition. The principle of frequency adjustment according to the invention is thereby based on an evaluation of the noise formation of the closed control loop at the input of the electric filter V.sub.fllt in a differential band around the primary resonance frequency f.sub.d.

[0063] FIGS. 9a and 9b show the spectrum (power density spectrum, PSD) of the signal V.sub.fllt as a function of the frequency f. FIG. 9a thereby displays the frequency components for f.sub.d=f.sub.t schematically whereas FIG. 9b refers to the case in which f.sub.d<f.sub.t applies. According to the invention, the values for the power density spectrum of the signal V.sub.fllt are compared to one another respectively in a first frequency band 102 and in a second frequency band 104 that are arranged symmetrically around the resonance frequency f.sub.d. In the case of f.sub.d=f.sub.t, the curve 106 is symmetrical in relation to the resonance frequency f.sub.d of the primary oscillation and the difference of the values for the power density spectrum of the signal V.sub.fllt disappears. On the other hand, the curve 106 is asymmetric in relation to the resonance frequency f.sub.d of the primary oscillation in case of f.sub.d≠f.sub.t. FIG. 9b shows the case of f.sub.d<f.sub.t. For a too low filter frequency, the position of the frequency bands 102, 104 is shifted respectively in the other direction in relation to the curve 106.

[0064] If a difference signal is formed in this case, said signal will not be equal to zero and can be used to generate an adjustment signal for the control input of the loop filter. By means of this comparison, the position of the current filter frequency can consequently be estimated with reference to the resonance frequency of the primary oscillation.

[0065] As already explained with reference to FIG. 2 above, the signal V.sub.fllt as described by the equation (5) depends on the Coriolis force F.sub.c, the quantization noise n.sub.q as well as the filter functions H.sub.s(s) and H.sub.f(s) of the secondary mass and/or of the loop filter.

[0066] Contrary to the output signal Y of the ΔΣM, both the quantization noise n.sub.q as well as the angular rate signal F.sub.c are suppressed by the electric filter H.sub.f(s) in an advantageous way at the input of the electric filter V.sub.fllt.

[0067] The system is further based on the idea of demodulating the bands 102, 104 displayed in FIG. 9 individually into the base band to subsequently compare the overall noise and/or signal power in the two bands. This allows for an accurate determination of whether the current filter frequency is too high or too low.

[0068] Due to this differential approach in the frequency range, also an angular rate signal that is not completely suppressed and still existing does not cause any disruption of the setting process. Respectively one component emerges in band 102 and/or band 104 due to the two side bands of the amplitude-modulated angular rate signal. As the filter according to the above equation (5) suppresses the angular rate signal F.sub.c in the same way as the quantization noise the angular rate signal can consequently also be used for control.

[0069] FIG. 10 shows the fundamental structure of an angular rate sensor 108 with the automated frequency adjustment according to the invention of the loop filter 110 of the secondary control loop according to an advantageous embodiment.

[0070] For rough alignment of the frequency f.sub.t of the electric filter 110 [with the transfer function H.sub.f(s)] with the primary resonance frequency f.sub.d, a signal V.sub.tg, which is proportional to the control signal of the oscillator in the phase-locked loop (PLL) of the primary control loop, is generated. This signal is offset (for example through addition) with the control signal V.sub.t of the frequency adjustment.

[0071] To determine the current frequency of the electric filter based on the noise formation, the signal V.sub.fllt is at first pre-amplified with H.sub.v(s) and frequencies that are clearly above the primary resonance frequency f.sub.d are suppressed. In the next step, the signal is demodulated by means of two multipliers and the frequencies f.sub.a,b=f.sub.d±f.sub.1 in order to evaluate the resulting base band signals further.

[0072] The required frequencies f.sub.a,b can be generated accurately in a very simple way and show a negative (f.sub.a) and/or positive (f.sub.b) frequency shift towards the primary resonance frequency f.sub.d. Through multiplication, we consequently obtain signals with frequencies that were previously above and/or below f.sub.d by f.sub.1 and subsequently at 0 Hz. This enables a very easy removal of other signal components with successive low-pass filters. The frequencies f.sub.a,b can be applied to the multipliers either by means of sine-shaped voltages or, in order to reduce the complexity, by means of square-wave voltages.

[0073] In the following step, either the power and/or signal strength of both signals is measured through squaring or through formation of absolute values and subsequently the difference is formed. The difference variable is used as an error variable for a control device that advantageously has a l-portion. The control device can thereby either be operated continuously or reset periodically in connection with a plus/minus sign evaluation and a digital logic.

[0074] The value for the regulating variable V.sub.t obtained after the first adjustment can be saved in an advantageous way in order to obtain the right frequency f.sub.t faster after a subsequent start. For this purpose, the frequency adjustment circuit 100 can have an appropriate memory device.

[0075] The principle for initial alignment according to the invention uses the oscillator, which already exists in the primary control loop, and consequently requires only a minimal additional switching effort in contrast to the known arrangements. In addition, the accuracy to be achieved through the additional circuit for adjustment of the frequencies has to be only in the range of approximately 10% of the primary resonance frequency.

[0076] In contrast to the known methods, the principle for adjustment of the frequencies according to the invention also allows for the adjustment of strongly non-linear filters (such as Gm-C filters) during operation. In addition, there is no dependence on the sensor transfer function in contrast to the known methods.

[0077] Due to the closed control loop used, there is no requirement for the circuit to align specific components particularly well to the actual filter in case of the frequency adjustment according to the present invention. In addition, a significantly faster and more accurate frequency adjustment than in case of examining discreet signal components in the spectrum of the output Y of the secondary control loop is reached through the integration of a large noise signal band. As no additional signals whatsoever need to be input in the secondary control loop or as the secondary control loop has to be disconnected for a short time, no impairment of the functionality for signal readout can occur.

[0078] Due to the simple components that are to be implemented very efficiently in the analog domain, the estimated surface requirement of the suggested implementation is significantly lower than in the solutions according to the current state of the art.

[0079] In sum, the low surface requirement allows for the realization of a more cost-efficient system with reduced energy consumption through the insensible Gm-C filters and a higher accuracy than with known arrangements.

[0080] The scope of application of the present invention, however, is not limited to angular rate sensors. In addition, the invention is also applied for the operation of other sensors such as acceleration sensors. Furthermore, in particular the frequency adjustment during operation can also be applied in analog-to-digital converters with closed control loops (e.g. ΔΣM ADC). Besides, the concept can be applied for any type of filters in which a broadband excitation of the filter occurs and therefore in which the transfer function of the filter is depicted in an approximate way based on one of the signals in the application.