Rapid baseline recovery method for irregular frequency content large dynamic range unipolar data signal

Abstract

A technique dynamically adjusts a corner frequency of a high pass filter in response to a time domain amplitude value of an output signal to maintain accuracy while quickly recovering a baseline value of the signal. The system may be configured to filter a frequency domain value of the input signal based on the filtering characteristic. The filtering characteristic may be dynamically set in response to a time-domain value of the amplitude of the output signal. The filtering characteristic may include a corner frequency, and the system may include a high-pass filter configured to filter frequency components below the corner frequency. The system is configured to dynamically change the corner frequency from a first value to a second value in response to a time-domain value of the amplitude crossing a threshold. The system may dynamically vary the corner frequency during a time interval after the crossing.

Description

[0001] Government Interest Statement

[0002] This invention was made with support from the U.S. Government under Contract No. N00019-16-G-0021 / N00019-19-F-0019, which was awarded by the U.S. Government. The U.S. Government has certain rights in this invention. Technical Field

[0003] The present disclosure relates to a fast baseline recovery method for a single-pole data signal with irregular spectrum and large dynamic range. Background Art

[0004] The input signal processing data from many systems often includes unipolar signals with irregular shapes, irregular data spacing, a large dynamic range, and may also include a constant component (which may also be referred to as the continuous, direct current, or DC component of the signal). For signal processing, it is often necessary to filter out the constant portion of the signal, thereby setting a "zero" baseline from which the data (e.g., the unipolar signal) is measured. With this filtering, the engineer or signal processing technician can direct the processing and analysis efforts to the relevant high-frequency data content. Therefore, a simple, common signal processing technique is to filter the input signal using a high-pass filter (HPF) or other filter to set a baseline for high-frequency content data analysis. However, this filtering still presents many non-trivial problems when considering unipolar data signals. Summary of the invention

[0005] In one embodiment, a system is provided, including: a filter circuit, a corner frequency (corner-frequency) up-adjustment circuit, a corner frequency down-adjustment circuit, and a selector circuit. The filter circuit may have an input and an output, and may be configured to filter an input signal received from the input of the filter circuit based on a corner frequency, and send the filtered input signal to the output of the filter circuit. The corner frequency up-adjustment circuit may be configured to increase the corner frequency. The corner frequency down-adjustment circuit may be configured to reduce the corner frequency. The selector circuit may be coupled to the output of the filter circuit, and coupled to one or both of the corner frequency up-adjustment circuit and the corner frequency down-adjustment circuit. The selector circuit may be configured to select the corner frequency up-adjustment circuit or the corner frequency down-adjustment circuit based on the filtered input signal. In some cases, the filter circuit includes a high-pass filter, and filtering the input signal based on the corner frequency includes filtering out frequency components with frequencies lower than the corner frequency. In some cases, the filter circuit includes a low-pass filter, and filtering the input signal based on the corner frequency includes filtering out frequency components higher than the corner frequency. In some cases, the selector circuit includes: a threshold terminal or node, a switch, and a comparator. The switch may have a first terminal coupled to the output of the filter circuit, a second terminal coupled to one or both of the corner frequency up-adjustment circuit and the corner frequency down-adjustment circuit, and a control input. The comparator may have a first comparator input, a second comparator input, and a comparator output, the first comparator input being coupled to the output of the filter circuit, the second comparator input being coupled to a threshold terminal or node, and the comparator output being coupled to the control input of the switch. In some cases, selecting between the corner frequency up-adjustment circuit and the corner frequency down-adjustment circuit based on the filtered input signal includes switching between the corner frequency up-adjustment circuit and the corner frequency down-adjustment circuit in response to the filtered input signal crossing a threshold at the threshold terminal or node. Switching between the corner frequency up-adjustment circuit and the corner frequency down-adjustment circuit in response to the filtered input signal crossing the threshold may include switching between the corner frequency up-adjustment circuit and the corner frequency down-adjustment circuit by comparing the filtered input signal with a threshold voltage corresponding to the threshold by the comparator. In some cases, the selector circuit includes a multiplexer having: a first terminal coupled to a corner frequency up-adjustment circuit, the corner frequency up-adjustment circuit configured to perform a first bit shift; a second terminal coupled to a corner frequency down-adjustment circuit, the corner frequency down-adjustment circuit configured to perform a second bit shift; and a control input coupled to a comparison circuit, the comparison circuit configured to perform a comparison of a filtered input signal with a threshold. Selecting between the corner frequency up-adjustment circuit and the corner frequency down-adjustment circuit based on the filtered input signal may include: selecting, by the multiplexer, between the corner frequency up-adjustment circuit and the corner frequency down-adjustment circuit based on the control input.

[0006] In another embodiment, a system is provided that is configured to filter frequency domain values ​​of an input signal based on a filtering characteristic. The filtering characteristic may be dynamically set in response to a time domain value of an amplitude of an output signal. In some cases, the filtering characteristic includes a corner frequency, and the system includes a high pass filter configured to filter out frequency components having a frequency lower than the corner frequency. In some cases, the system is configured to dynamically change the corner frequency from a first value to a second value in response to a time domain value of the amplitude decreasing below a threshold, the second value being higher than the first value. In some cases, the system is configured to dynamically change the corner frequency from a first value to a second value in response to a time domain value of the amplitude increasing above a threshold, the second value being lower than the first value. In some cases, the system is configured to dynamically change the corner frequency within a time interval after the time domain value of the amplitude crosses the threshold. In some cases, the time interval is 3 clock ticks of a system clock or less. In some cases, the time interval is 1 clock tick of a system clock or less. In some cases, the time interval is less than 1 microsecond. In some cases, the time interval is short enough so that a signal baseline is preserved, thereby allowing single-pole data measurements consistent with application requirements. In some cases, the time interval is short enough so that the signal baseline is preserved, thereby allowing single-pole data measurement consistent with a predetermined accuracy threshold. In some cases, the system is a multi-channel system. The amplitude can be a corresponding amplitude among a plurality of amplitudes corresponding to a plurality of signals. The system can be configured to set a filtering characteristic in response to a corresponding value of any one of the plurality of amplitudes crossing a threshold. In some cases, the output signal is a digital output signal. Filtering the frequency domain value can include, for example, shifting the digital output signal. In some cases, the system includes a multiplexer configured to dynamically select a displacement value based on the amplitude of the output signal. In some cases, the output signal is an analog output signal. Setting the filtering characteristic can include setting a switch state (e.g., a transistor or another switch) based on the output of a comparator that compares the time domain value of the amplitude with a threshold voltage. Other types of filters can be used. For example, in some cases, the filtering characteristic includes a corner frequency, and the system includes a low-pass filter configured to filter out frequency components above the corner frequency.

[0007] In another embodiment, a method for filtering frequency domain values ​​is provided, comprising: dynamically setting a filter characteristic in response to a time domain value of an amplitude of an output signal; and filtering frequency domain values ​​of an input signal based on the dynamically set filter characteristic. In some cases, the input signal is a digital input signal received by a computing device. Dynamically setting the filter characteristic may include: dynamically setting the filter characteristic by the computing device. Filtering the frequency domain values ​​of the input signal may include: filtering the frequency domain values ​​of the input signal by the computing device.

[0008] In some cases, the computer instructions may be configured to perform a high pass filtering method with a variable filtering characteristic. The variable filtering characteristic may be dynamically adjusted in response to a time domain value of an amplitude of an output signal of the high pass filtering method. The variable filtering characteristic may include a variable corner frequency.

[0009] In another embodiment, a system having a first input channel and a second input channel is provided. The system can be configured to filter frequency domain values ​​of the first input signal and the second input signal based on a filtering characteristic. The filtering characteristic can be dynamically set in response to time domain values ​​of amplitudes of a first output signal and a second output signal of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 is a block diagram illustrating the architecture of a system configured to filter frequency domain values ​​of an input signal according to an embodiment of the present disclosure.

[0011] Figure 2A is a circuit diagram showing details of an analog high pass filter circuit.

[0012] Figure 2B is a circuit diagram showing details of a digital high-pass filter circuit.

[0013] Figure 3A is a circuit diagram showing details of a dynamic analog high pass filter circuit according to an embodiment of the present disclosure.

[0014] Figure 3B is a circuit diagram showing further details of a high pass filter circuit configured to dynamically set filtering characteristics in response to time domain values ​​of the amplitude of an output signal according to an embodiment of the present disclosure.

[0015] Figure 3C 1 is a circuit diagram showing control logic of a multi-channel high-pass filter circuit according to an embodiment of the present disclosure, the multi-channel high-pass filter circuit being configured to dynamically set filtering characteristics in response to time-domain values ​​of the amplitudes of multiple signals. The circuit can be configured to transmit control signals for threshold detection within a single system clock tick, three clock ticks, or another short time interval.

[0016] Figure 4A A signal filtered using a single corner frequency is shown graphically.

[0017] Figure 4B The slow baseline recovery of a saturated signal filtered with a single corner frequency is graphically shown.

[0018] Figure 5A More detail of a signal filtered using a single corner frequency is shown graphically.

[0019] Figure 5BMore details of the slow baseline recovery of a saturated signal filtered with a single corner frequency are shown graphically.

[0020] Fig. 6A A signal filtered with a dynamically set corner frequency in accordance with an embodiment of the present disclosure is graphically illustrated.

[0021] Figure 6B Graphically illustrated is the baseline restoration of a saturated signal filtered with a dynamically set corner frequency in accordance with an embodiment of the present disclosure.

[0022] Fig. 7A More details of a signal filtered with a dynamically set corner frequency in accordance with an embodiment of the present disclosure are graphically shown.

[0023] Figure 7B Further details of baseline restoration of a saturated signal filtered with a dynamically set corner frequency in accordance with an embodiment of the present disclosure are graphically shown.

[0024] Fig. 8A is a flowchart of a method for dynamically filtering frequency domain values ​​according to an embodiment of the present disclosure.

[0025] Figure 8B is a flowchart of a method for dynamically filtering frequency domain values ​​based on a time domain amplitude of an output signal according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0026] The disclosed techniques are used to dynamically adjust the corner frequency of a filter, such as a high pass filter (HPF), in response to the time domain data content of a signal. Thus, the disclosed techniques can be used to maintain data accuracy while quickly restoring the baseline value of a signal, even in the case of signals with a high dynamic range. Furthermore, the disclosed techniques can be used to avoid the need for additional filtering stages that typically introduce spurious low frequency artifacts.

[0027] In some embodiments, the disclosed technology can be implemented in a system that filters frequency domain values ​​of an input signal based on a filter characteristic. The filter characteristic can be dynamically set in response to a time domain value of an output signal amplitude. The filter characteristic can include a corner frequency, and the system can include, for example, a high pass filter configured to filter out frequency components below the corner frequency. The system is configured to dynamically change the corner frequency from a first value to a second value in response to the time domain value of the amplitude crossing a threshold.

[0028] General Overview

[0029] As mentioned above, incoming signal processing data from many systems often includes unipolar signals with irregular shapes, irregular data spacing, large dynamic ranges, and possibly constant or DC components. In signal processing, such as in DSP, it is often desirable to eliminate the constant portion of a signal, thereby creating a "zero" baseline for measuring the data. The engineer or signal processing technician can then focus processing and analysis efforts on the remaining high-frequency data content. Therefore, a common and effective signal processing technique is to filter the incoming signal with an HPF to set a "zero" baseline for high-frequency content data analysis. However, when the frequency content of the data is inconsistent, the correct selection of the cutoff frequency (also called the corner frequency or pole) of the HPF can be complicated. For example, in some cases, a corner frequency that is too high may quickly restore the baseline value of the signal, but may filter out some potentially important components of the signal of interest, resulting in inaccurate processing. In other cases, a corner frequency that is too low may slowly restore the baseline value, resulting in the possibility of loss of acquired data or loss of accuracy due to inappropriate baseline location.

[0030] Thus, according to some embodiments, techniques and systems are provided herein that can dynamically adjust the corner frequency of a HPF, low-pass filter, band-pass filter, or other filter in response to data content, maintaining data accuracy even while rapidly restoring the baseline value of the signal (including in the case of signals with high dynamic range). For example, the disclosed systems and methods can be used to address key data acquisition challenges by restoring the baseline significantly faster, thereby reducing the time it takes for the system to be ready for the next signal and maintaining or even improving signal quality. The system can also be used to avoid the necessity of other filtering stages that often introduce spurious low-frequency artifacts. For example, the system is useful when the data frequency content or occurrence rate changes or necessarily approaches the bandwidth of the HPF. Thus, the disclosed systems and methods can be widely used in the field of digital signal processing (DSP).

[0031] System Architecture

[0032] Figure 11 is a block diagram showing the architecture of a system 100 configured to filter frequency domain values ​​of an input signal according to an embodiment of the present disclosure. In this embodiment, the system 100 includes an input line 102, a filter circuit 104, a selector circuit 106, a corner frequency up-adjustment circuit 108, a corner frequency down-adjustment circuit 110, and an output line 112. The input line 102 can receive an input signal. For example, the input signal can include unipolar data and / or can have an irregular shape, an irregular data interval, a large data dynamic range, and / or a constant DC component, as described above. Therefore, the input signal 102 can be filtered by the filter circuit 104. The filter circuit 104 can include, for example, an HPF, which filters out signal components with frequencies below a cutoff frequency, which can also be referred to as a corner frequency or a pole. In some embodiments, the value of the cutoff frequency can be dynamically adjusted based on the time domain value of the output signal 112. In various embodiments, the filter circuit 104 can include another type of filter, such as a low-pass filter, a band-pass filter, or some other filter.

[0033] In some embodiments, the selector circuit 106 (which may include a unipolar data arrival detector) can select the value of the cutoff frequency based on the output signal 112. For example, in some cases, a cutoff frequency that is too high may cause inaccurate processing. In other cases, a cutoff frequency that is too low may slowly restore the baseline value of the signal. Therefore, the selector circuit 106 can adjust the level of the cutoff frequency in response to the time domain value of the output signal 112 to prevent these problems. For example, the selector circuit 106 can monitor the output to determine when the unipolar data arrives, so that the switching corner frequency becomes favorable. In this embodiment, the selector circuit 106 is located in a parallel or alternating path between the filter circuit 104 and the output 112.

[0034] In some implementations, the selector circuit 106 can select a parallel branch of the circuit 100 that includes the corner frequency up-adjustment circuit 108 or the corner frequency down-adjustment circuit 110. Specifically, the corner frequency up-adjustment circuit 108 can increase the value of the cutoff frequency of the HPF, while the corner frequency down-adjustment circuit 110 can decrease the value of the cutoff frequency, as disclosed herein.

[0035] In some embodiments, the selector circuit 106 selects a branch based on a comparison of the output signal 112 with a threshold value, as disclosed herein. Since the output 112 of the filter has a zero baseline, such a comparison with the threshold value is both meaningful and practical. In contrast, the input signal 102 may retain a DC component, so a meaningful comparison with the threshold value cannot be made.

[0036] In this embodiment, the system 100 provides two alternative options for the cutoff frequency, depending on the branch selected by the selector circuit 106. However, in some cases, the system may have any number of alternative values, such as three. In addition, the system may provide a disable control that the user may select to fix the corner frequency to a static value, as described below. Note that such a static value may have a different value than the values ​​in branches 108 and 110.

[0037] Finally, after the cutoff frequency corresponding to the branch 108 or 110 is selected by the selector circuit 106 , the filter circuit 104 filters the input signal and outputs the filtered signal to the output line 112 .

[0038] Circuit Details

[0039] Figure 2A is a circuit diagram showing details of analog high pass filter circuit 200. Analog HPF circuit 200 can filter out frequency domain components of a signal from input line 202 whose frequencies are below a cutoff frequency or corner frequency. In this embodiment, circuit 200 can reduce the voltage from input 202 across an RC combination of capacitor 204 having a capacitance C and resistor 206 having a resistance R. Output 208 is equivalent to the voltage across resistor 206.

[0040] The cut-off frequency or corner frequency is related to the time constant τ=1 / RC of the resistor and capacitor combination, that is, it is related to the charging time of capacitor 204. Specifically, when capacitor 204 is only partially charged, current can continue to flow in circuit 200 because the voltage drop V across capacitor 204 is C =Q / C is less in magnitude than the voltage from input 202. For frequencies much higher than the corner frequency (e.g., for frequencies greater than 170% of the corner frequency, 200% of the corner frequency, or 400% of the corner frequency), the voltage changes sign before capacitor 204 is fully charged, so V C The amplitude is always smaller than the input voltage.

[0041] Thus, the alternating current can flow continuously in response to a high frequency input signal (e.g., for frequencies greater than 170% of the corner frequency, 200% of the corner frequency, or 400% of the corner frequency). In contrast, for frequencies much lower than the corner frequency (e.g., for frequencies less than 90% of the corner frequency, 50% of the corner frequency, or 25% of the corner frequency), capacitor 204 has time to fully charge, and thus the output signal will decay the input signal after a period of τ. Thus, circuit 200 represents an analog implementation of a HPF, but digital implementations are also possible, such as Figure 2B As shown in the embodiment.

[0042] Figure 2B is a circuit diagram showing details of a digital HPF circuit 250. The digital HPF circuit 250 can filter out frequency domain components of a signal from an input line 252 whose frequencies are below a cutoff frequency or corner frequency. As described above, the HPF 250 can be used to remove DC and lower frequency components of an input signal for high frequency content data analysis, thereby creating a "zero" or corrected or adjusted baseline from which data is measured. In some embodiments, the disclosed systems and methods can use different types of filters to filter out other ranges of frequency components. For example, the system can use a low pass filter to filter out frequencies above a cutoff frequency, or can use a band pass filter to filter out frequencies outside of a frequency band or range. Although the HPF circuit 250 is an implementation of a HPF circuit with digital electronics, in some embodiments, the system can use a HPF or another filter with analog electronics or with both digital and analog electronics.

[0043] In some embodiments, HPF 250 can utilize spectral inversion, which involves subtracting the output of a low-pass filter to achieve high-pass filtering. That is, since the output of the low-pass filter is the low-frequency component of the original input signal, subtracting the low-pass output from the original input leaves only the high-frequency component. In some embodiments, HPF 250 can utilize an infinite impulse response (IIR) spectral inversion filter.

[0044] In some embodiments, HPF 250 may be an integer filter that can only process integer-valued signals. However, at each time step, HPF 250 may output a fraction of the previous output and the current input. Therefore, in order to maintain numerical accuracy, it may be necessary to amplify the input signal to a large integer and then scale down the output after filtering. Therefore, the HPF may set the corner frequency using a scaling factor K that is implemented by right-shifting the signal 264. Figure 2B , and may occur, for example, in a previous low pass filter stage. In some embodiments, the HPF 250 may use K=1 / 256, or an 8-bit right shift 264, as described below. Alternatively, in some embodiments, the HPF 250 may be implemented by floating point digital circuits, analog circuits, etc., rather than integer implementations, and scaling up and down may not be necessary.

[0045] In some embodiments, the corner frequency of the HPF 250 may be related to the scale factor K and the clock period dt of the system. For example, the corner frequency may be approximately equal to K / ((1-K)2πdt). For example, for an 8-bit right shift 264 and a clock period dt=10ns, the corner frequency may be approximately 62kHz. The tolerance of the corner frequency may vary from embodiment to embodiment. In some example cases, the tolerance is symmetrical, such as ±10%, ±6%, ±4%, or ±2%. In other cases, the tolerance may be asymmetrical, such as -6% / +4%.

[0046] In some embodiments, the HPF circuit 250 may include a bypass multiplexer 253 that may select a circuit branch based on an HPF bypass signal. In this embodiment, if the HPF bypass signal selects branch A, the HPF function of the circuit 250 may be bypassed. If the HPF bypass signal selects branch B, the HPF function of the circuit 250 may be activated, i.e., the circuit 250 may perform high pass filtering, as described in this embodiment.

[0047] In this embodiment, if the HPF bypass signal selects branch B, the input signal 252 may be combined with the input signal from the adder 254 (also referred to herein as an adder). Figure 2B The signals of the lower branch of the circuit 250 are added. Next, along the upper branch of the circuit 250, the input signal 252 can pass through the register 256 and then output to the output line 258 of the HPF circuit 250. In some embodiments, the register herein can refer to a 1 / Z block, where Z is the impedance. Note that in some embodiments, the register 256 can be a 24-bit register. Alternatively, any other register can be used.

[0048] In some embodiments, the circuit 250 may further include one or more additional registers, which may cause a delay in the final output, such as a delay of one clock cycle.

[0049] As can be seen from the lower branch, inverter 259 can be used to perform subtraction in adder 254. For example, inverter 259 can invert its input signal by taking the two's complement of the signal. Continuing along the lower branch of circuit 250, the output of adder 260 and register 262 can form a signal to be inverted by inverter 259.

[0050] Feedback path 264 of circuit 250 may carry the shifted output of register 256. For example, feedback path 264 may apply a shift, such as a right shift of 8 bits and carry the most significant bit (MSb), i.e., reintroducing the MSb from the left. In some embodiments, the shift may be a right shift of the two's complement number of the signal. Note that the shift 264 may occur between register 256 and adder 260. In some embodiments, the MSb carry is needed because the value being shifted (HPF output) may be negative. Therefore, the system may use the MSb carry, which may correspond to a right shifted two's complement value. This shift may correspond to dividing the signal amplitude by 2 8 =256, and produces the HPF cutoff frequency. For example, for a 100MHz system clock rate (i.e., 10ns clock period), divide by 2 8 This corresponds to a HPF cutoff frequency of approximately 62kHz.

[0051] Thus, HPF circuit 250 may filter out frequencies below the cutoff frequency from input signal 252 and output the filtered signal to HPF output line 258. HPF 250 may be used for signal processing, such as DSP, to remove a constant or DC portion of a signal, thereby shifting the baseline of the signal to zero (e.g., by removing a DC offset, if present).

[0052] As described above, the optimal selection of the HPF corner frequency can be complicated when the frequency content of the data is inconsistent. For example, in some cases, a high corner frequency may filter out potentially important components of the signal, while a low corner frequency may slowly restore the baseline value of the signal. The disclosed systems and methods can address these challenges. For example, the following Figure 3A and Figure 3B The dynamic HPF circuit of an embodiment can dynamically adjust the corner frequency in response to the data content. In addition to the dynamic HPF, in some embodiments, a dynamic low-pass filter, a dynamic band-pass filter or other dynamic filters are also possible.

[0053] Figure 3A is a circuit diagram showing details of a dynamic analog high-pass filter circuit 300 according to an embodiment of the present disclosure. Figure 2A Similar to the circuit 200 of the embodiment of the present invention, the circuit 300 reduces the voltage from the input 302 across the RC combination of a capacitor 304 and a resistor 306. The circuit 300 also includes additional components in parallel with the resistor 306, such as a switch 308 (e.g., a field effect transistor (FET)), a second resistor 310, a comparator 312 with hysteresis, and a threshold voltage signal 314.

[0054] Comparator 312 may compare output voltage 316 (e.g., a voltage measured or sensed from an output terminal or node) to a threshold voltage 314 (e.g., present at a threshold reference terminal or node). In some embodiments, comparator 312 may compare the voltage across resistor 306 (which may be equal to output voltage 316) to threshold voltage 314. If the signal is less than threshold voltage 314, comparator 312 may open switch 308 so that resistor 310 is connected in parallel with resistor 306, thereby reducing the net resistance in series with capacitor 304. This in turn will reduce the time constant τ=1 / RC and increase the cutoff frequency because capacitor 304 can charge faster with more current flowing through circuit 300. Conversely, if the signal is greater than threshold voltage 314, comparator 312 may close switch 308, thereby increasing the net resistance and reducing the cutoff frequency.

[0055] In one embodiment, the switch 308 may be a FET. However, in some other embodiments, the circuit 300 may use another type of switch 308 instead of a FET. For example, any switch 308 that can respond to changes in the amplitude of the output signal quickly enough by switching may be used. For example, an analog switch, a relay, a vacuum switch, or other switch may be used as the switch 308. In some embodiments, a sufficiently fast response may refer to switching the corner frequency within a sufficiently short time interval to maintain the baseline of the signal, thereby ensuring accurate measurement. In some embodiments, the required level of accuracy may depend on the performance requirements of a particular application. Therefore, in some cases, a trade-off or optimization may be made between the selection of the corner frequency and the switching time interval. For example, depending on the performance requirements of the application, the time interval may be 3 clock ticks of the system clock or less, 1 clock tick of the system clock or less, 1 microsecond or less, or may be any other time period or interval.

[0056] As described above, in various embodiments, the disclosed systems and methods can be designed so that the time intervals are short enough because the signal baseline is preserved, thereby allowing single-pole data measurements consistent with application requirements and / or predetermined accuracy thresholds, which can be based on such application requirements. In such embodiments where the performance and / or accuracy thresholds are looser, longer time intervals can be selected, thereby enabling the use of lower corner frequencies. For example, for some precision guidance and sensing applications, very high performance and / or accuracy may be required, such that the time interval is 1 clock tick of the system clock or less. In another embodiment, the disclosed systems and methods can be implemented within a mechanical damping system, so the time interval can be 100 milliseconds or 10 milliseconds. In yet another embodiment, the disclosed systems and methods can be used to filter RF data, so the time interval can be on the order of 100 nanoseconds or 10 nanoseconds.

[0057] As shown, comparator 312 may include hysteresis to reduce the chance of cutoff frequency oscillation due to noise fluctuations. In an embodiment, the hysteresis in comparator 312 may be very small, for example, the hysteresis may be only sufficient to prevent oscillation under zero input noise conditions.

[0058] Since the comparator 312 of the circuit 300 can dynamically adjust the cutoff frequency according to the input signal, the dynamic analog HPF circuit 300 can filter the input signal while still maintaining the accuracy of the signal above the cutoff frequency and quickly restore the baseline value of the signal. Specifically, when the input signal is greater than the threshold voltage 314, the circuit 300 can set a lower cutoff frequency, thereby retaining most of the frequency components of the input signal, thereby possibly retaining the shape and quantitative value of the signal. Conversely, when the input signal is lower than the threshold voltage 314, the dynamic HPF circuit 300 can set a higher cutoff frequency, thereby filtering out artifacts of lower frequencies (longer periods), thereby achieving faster baseline recovery.

[0059] Figure 3B is a circuit diagram showing further details of HPF circuit 330 according to an embodiment of the present disclosure, configured to dynamically set filtering characteristics in response to time domain values ​​of the amplitude of an input signal. For example, HPF circuit 330 may provide additional details of system 100 to filter frequency domain values ​​of an input signal, such as Figure 1 shown.

[0060] In some embodiments, the dynamic HPF 330 may utilize a spectrum inversion, and / or an IIR spectrum inversion filter, as described above. Figure 2B In some embodiments, the dynamic HPF 330 may be an integer filter, so it may be necessary to amplify the input signal and then reduce the output after filtering, such as Figure 2B Thus, the dynamic HPF 330 can set the corner frequency using a scaling factor K achieved by right shifting the signal 346. Figure 3B , and may occur, for example, in a previous low-pass filter stage. In some implementations, the HPF 330 may use K=1 / 256, or an 8-bit right shift 346, as described below. Alternatively, in some embodiments, the HPF 330 may be implemented by floating-point digital circuits, analog circuits, software, etc., rather than integer implementations, so that scaling up and down may not be necessary.

[0061] In some embodiments, the corner frequency of HPF 330 may be related to the scale factor K and the system clock period dt. For example, the corner frequency may be approximately equal to K / ((1-K)2πdt). For example, for an 8-bit right shift 346 and a clock period dt=10ns, the corner frequency may be approximately 62kHz.

[0062] In some embodiments, HPF circuit 330 may include a bypass multiplexer 333 that may select a circuit branch based on an HPF bypass signal. In this embodiment, if the HPF bypass signal selects branch A, the dynamic HPF function of circuit 330 may be bypassed. If the HPF bypass signal selects branch B, the dynamic HPF function of circuit 330 may be activated, i.e., circuit 330 may perform dynamic high pass filtering as disclosed herein.

[0063] In this embodiment, if bypass multiplexer 333 selects branch B, HPF circuit 330 may receive an input signal from HPF input line 332. The input signal may be summed with the signal from the lower branch of the circuit in adder 334 (also referred to as an adder), similar to the above. Figure 2B Next, in the upper branch of circuit 330, the input signal may pass through register 336 and then be output to the output line 338 of the HPF. In some embodiments, the register here may refer to a 1 / Z block, where Z is impedance. In some embodiments, the register may be clocked, while other blocks may not be clocked. Note that in some embodiments, register 336 may be a 24-bit register. Alternatively, any other register may be used.

[0064] In some embodiments, circuit 330 may further include one or more additional registers, which may cause a delay in the final output, such as a delay of one clock cycle.

[0065] With the above Figure 2B Similar to the embodiment of FIG. 3 , along the lower branch of circuit 330 , inverter 339 may be used to perform subtraction in adder 334 , such as by two's complement. Next in the lower branch of circuit 330 , the output of adder 340 and register 342 may form a signal to be inverted in inverter 339 .

[0066] In some embodiments, the lower branch of circuit 330 may include a multiplexer 344 that selects between circuit branches 346 and 348 based on a control signal A_CTRL 349 that may be set by a user. Branch 346 may carry the shifted output of register 336, while branch 348 may carry the selection of multiplexer 350. Specifically, if A_CTRL 349 is set to a first value that disables dynamic corner frequency selection, such as 0 or a Boolean false value, the signal of branch 346 may propagate through circuit 330. As described above, Figure 2B, branch 346 may apply a shift, such as an 8-bit right shift with an MSb carry, which may correspond to a cutoff frequency, such as approximately 62 kHz. Note that the shift 346 may occur between register 336 and multiplexer 344. Thus, when A_CTRL 349 is set to a first value, thereby disabling dynamic corner frequency selection, circuit 330 may behave as Figure 2B In contrast, if A_CTRL 349 is set to a second value, such as 1 or a Boolean true value, to enable dynamic corner frequency selection, the signal may continue through branch 348 of circuit 330 to multiplexer 350 .

[0067] Next, in some embodiments, the lower branch may include a multiplexer 350 that selects between circuit branches 352 and 354 and sends the selection to multiplexer 344 in branch 348. Specifically, multiplexer 350 may correspond to the above Figure 1 The selector circuit 106 in the embodiment, the branch 352 may correspond to the corner frequency up-adjusting circuit 108 , and the branch 354 may correspond to the corner frequency down-adjusting circuit 110 .

[0068] Branches 352 and 354 may carry the output of register 336 with two different shift values. Note that in some cases, there may be multiple branches corresponding to multiple corner frequencies. Multiplexer 350 may select a branch corresponding to the time domain value of the output signal amplitude. For example, multiplexer 350 may select a branch based on comparing the amplitude to a switching threshold, which may be a threshold that enables the system to determine that unipolar data has been reached.

[0069] In some embodiments, multiplexer 350 selects a branch based on a comparison of the amplitude of the output signal with a threshold, such as Figure 3C As shown in the embodiment of . Since the output of the filter has a zero baseline, this comparison with the threshold is both meaningful and practical. In contrast, the input signal may retain a DC component and thus cannot be meaningfully compared with the threshold.

[0070] In the following Figure 4A and Figure 4B In the embodiment of Figure 4A and Figure 4B In some embodiments, the switching threshold value may be a value in units of , such as 9,000. In other embodiments, the switching threshold value may be between 20 and 2,000. Alternatively, or in addition, the threshold value may be dynamic, such as the system may dynamically determine the threshold value based on many other factors. In some embodiments, the same switching threshold value may be used elsewhere in the disclosed system to be consistent with other parts of the system.

[0071] The select line of the multiplexer 350 may carry a signal D_CTRL 356 that may perform this comparison. For example, D_CTRL 356 may carry a Boolean signal, such as the result of comparing the amplitude to a switching threshold. In various embodiments, the switching threshold may be a fixed value, or may be adjusted by the system in response to other criteria, for example, depending on the specific details of the application. In some embodiments, the comparison may be made in absolute value, such as the absolute value of the signal amplitude may be compared to the switching threshold.

[0072] If multiplexer 350 selects branch 352, the shifted signal of branch 352 can be propagated through the circuit, for example, branch 352 can apply a 5-bit right shift with an MSb carry. This can correspond to setting a higher value for the HPF cutoff frequency, such as 500kHz or 600kHz. Therefore, the baseline value of the signal will be restored more quickly. If the amplitude of output signal 338 is below the switching threshold, multiplexer 350 can select branch 352.

[0073] If multiplexer 350 selects branch 354, the shifted signal of branch 354 can be propagated through the circuit, for example, branch 354 can apply a 12-bit right shift with an MSb carry. This can correspond to setting a lower value for the HPF cutoff frequency, such as 200 Hz or 4 kHz. Therefore, the accuracy of the signal will be better maintained. If the amplitude of output signal 338 is greater than the switching threshold, multiplexer 350 can select branch 354.

[0074] The filtered signal can then be passed to the HPF output line 338. Specifically, the signal can be filtered so that the input signal components (i.e., frequency domain values) with frequencies below the selected cutoff frequency are attenuated. Since the circuit 330 can dynamically select the cutoff frequency in response to the input signal, the circuit 330 can filter the input signal while still maintaining the accuracy of the signal above the cutoff frequency and quickly restore the baseline value of the signal. Specifically, when the output signal amplitude 338 is greater than the switching threshold, the system can set a low cutoff frequency so that most of the frequency components of the input signal are retained, so the shape and quantitative value of the signal may be retained. Conversely, when the output signal amplitude is less than the threshold, the system can set a high cutoff frequency so that lower frequency (longer period) artifacts are filtered out, resulting in faster baseline recovery.

[0075] Figure 3C3 is a circuit diagram illustrating control logic 360 of a multi-channel high pass filter circuit according to an embodiment of the present disclosure, which is configured to dynamically set filtering characteristics in response to time domain values ​​of the amplitudes of multiple signals 362. In this embodiment, a single switching threshold signal 364 can be compared with all channels 362 by a comparator. Alternatively, or in addition, a multiplexer can be used to compare the switching threshold signal 364 with the channels 362. For example, comparator 366 can select between channel 1 HPF output 368 and switching threshold signal 364 by comparing their values. Similarly, comparator 370 can select between channel 2 HPF output 372 and switching threshold signal 364, comparator 374 can select between channel 3 HPF output 376 and switching threshold signal 364, and comparator 378 can select between channel N HPF output 380 and switching threshold signal 364. In some embodiments, these comparisons can be made in absolute values, for example, in the case of a bipolar input signal, the absolute amplitude of the corresponding channel amplitude can be compared with the switching threshold signal 364.

[0076] The output of the multiplexer can then be passed to an OR gate 386, which can determine whether the signal from any channel is greater than the switching threshold. The output of the OR gate 386 can be passed to a D_CTRL 388, which can be used to select the corner frequency, as described above. Figure 3B . Thus, if the amplitude of the signal from any channel exceeds the switching threshold, the system can adjust the corner frequency of the HPF, for example by lowering it. In some embodiments, all channels 362 of the multi-channel system 360 can adjust the corner frequency simultaneously. In some embodiments, the circuit 360 is configured to transmit a control signal 388 for threshold detection within a short time interval after any amplitude exceeds the threshold (e.g., within a single system clock cycle, three clock cycles, 25ns, 100ns, or 1μs after the threshold is crossed).

[0077] As described above, in various embodiments, the disclosed systems and methods can be designed so that the time intervals are short enough because the signal baseline is preserved, thereby allowing single-pole data measurements consistent with application requirements and / or predetermined accuracy thresholds, which can be based on such application requirements. In such embodiments where the performance and / or accuracy thresholds are more relaxed, longer time intervals can be selected, thereby enabling the use of lower corner frequencies. For example, for some precision guidance and sensing applications, very high performance and / or accuracy may be required, such that the time interval is 1 clock tick of the system clock or less. In another embodiment, the disclosed systems and methods can be implemented within a mechanical damping system, so the time interval can be 100 milliseconds or 10 milliseconds. In yet another embodiment, the disclosed systems and methods can be used to filter RF data, so the time interval can be on the order of 100 nanoseconds or 10 nanoseconds.

[0078] Specifically, circuit 360 may transmit control signal 388 for threshold detection at a sufficiently short time interval so that the corner frequency can be switched quickly enough to maintain the baseline and ensure accurate measurement. In some embodiments, the required level of accuracy may depend on the performance requirements of a particular application. Therefore, in some cases, a trade-off or optimization may be made between corner frequency selection and switching time interval.

[0079] In addition to the above Figure 3A-3C In addition to the analog and digital embodiments of the disclosed dynamic filtering systems and methods, in some embodiments, the disclosed dynamic filtering systems and methods can also be implemented in mechanical systems. For example, the disclosed systems and methods can be implemented in a shock absorption system, such as in a shock absorber on an automobile or other vehicle. In one embodiment, the disclosed dynamic filtering systems and methods can provide a fast response when a rough road shock absorption response occurs, and can provide a slow response when the road is flat. In another embodiment, the disclosed dynamic filtering systems and methods can be used in the suspension system of an automobile. For example, if a car or other vehicle drives over a pothole or bump, the suspension can react quickly enough to avoid the disturbance being felt in the car.

[0080] Alternatively or additionally, the disclosed dynamic filter may also be implemented in various other systems, such as by floating point digital circuits, analog circuits, and / or software. For example, a software HPF may be configured with variable filtering characteristics, such as a variable corner frequency, which may be dynamically adjusted in response to a time domain value of the amplitude of an output signal of the software HPF. In one such embodiment, such a software HPF may sample its input and / or output signals at a high rate (e.g., having a Nyquist frequency of 10 MHz, 50 MHz, 100 MHz, 200 MHz, or higher). The software HPF may dynamically adjust the filtering characteristics or corner frequency during a brief period when the amplitude of the output signal exceeds or falls below a threshold, such as within 1 sampling period of such crossing, within 3 sampling periods of such crossing, within 10 sampling periods, etc.

[0081] Comparison results

[0082] Figure 4A A signal 400 filtered with a single corner frequency is graphically illustrated. Specifically, the signal 400 may be filtered with a HPF having a single static corner frequency. As shown, the measured height of the peak 402 is 7,462 units, while the ideal height of the peak 402 is 7,520 units. Thus, the measured height of the peak 402 is approximately 99% of its correct height. Thus, while the peak 402 is relatively tall in this embodiment, it may not be so tall as to saturate the HPF. Thus, the filtered signal 400 may be referred to as the nominal output of the HPF. The peak 402 maintains a relatively accurate shape compared to the input signal.

[0083] As shown, the baseline 406 remains depressed after the peak 402. This results in reduced accuracy for the smaller peak 404. For example, as shown, the height of the peak 404 is 342 units above the horizontal axis, but the baseline 408 near the peak 404 is still slightly below the horizontal axis. In this embodiment, the ideal height of the peak 404 is 492 units. Therefore, due to the depression of the baseline 408, the apparent height of the peak 404 may be somewhat erroneously reduced to approximately 70% of its correct value, resulting in slight data corruption of the peak 404. Nevertheless, the signal baseline does recover modestly after the peak 402. Therefore, 70% of the correct height of the smaller peak 404 relative to the baseline can be measured.

[0084] Figure 4B The slow baseline recovery of a saturated signal 450 filtered with a single corner frequency is graphically shown. Figure 4A As shown in the embodiment of FIG. 4 , signal 450 can be filtered with a HPF having a single static corner frequency. As shown, the height of peak 452 is 11,436 units, while the ideal height of peak 452 is 11,484 units. Therefore, in this embodiment, peak 452 is higher than Figure 4A402, so that the HPF may become saturated. As shown, due to the HPF saturation, the shape of the peak 452 may be inaccurate compared to the input signal. For example, due to the modification of the frequency content of the data signal by the HPF, the height of the peak 452 may be reduced. In addition, the top of the peak 452 is tilted, while the corresponding peak of the input signal may be smoother and concave downward, approximately symmetrical about its apex. This inaccurate tilted shape of the peak 452 may be a defect of the saturated filtered signal 450.

[0085] 44. In addition, the base of the smaller peak 454 is at a negative amplitude value, while the apex of the peak 454 is close to zero. Therefore, the peak 454 is difficult to measure and may be missed, for example, by an automatic peak locator, while the ideal height of the peak 454 is 500 units. This difficulty may be caused by the signal baseline 456 not recovering quickly enough after the larger peak 452. The baseline 456 remains below zero for most of the display domain of the signal 450, resulting in an incorrect baseline for the peak 454. This may be another defect of the filtered signal 450. The disclosed systems and methods can address such issues by dynamically selecting the corner frequency to minimize saturation, as disclosed herein.

[0086] Figure 5A Further details of the signal 400 filtered with a single corner frequency are shown graphically. In this embodiment, the signal 400 is from the above Figure 4A Detailed view of signal 400, wherein the vertical axis is expanded and the horizontal axis is contracted. Peak 504 shown in this example corresponds to Figure 4A Peak 404. As mentioned above Figure 4A As in the embodiment of FIG. 4 , the output from the HPF may be in a nominal state. As shown, the signal 400 may still have a slow recovery 502 of the baseline value. However, as described above Figure 4A As in the embodiment of FIG. 5 , the slowness of recovery 502 is not as severe as in the saturation case. For example, in this embodiment of the nominal signal, the total baseline recovery time (before the value of the baseline is dominated by noise) is shown to be approximately 800 (arbitrary units), while in the following Figure 5B In the embodiment of FIG. 5 , the total baseline recovery time of the saturated signal is shown to be approximately 2200.

[0087] Figure 5B The details of the slow baseline recovery of a saturated signal filtered with a single corner frequency are further illustrated graphically. In this example, signal 450 is from the above Figure 4B Detailed view of signal 450, where the vertical axis is expanded and the horizontal axis is contracted. Peak 556 shown in this example corresponds to Figure 4B Peak 454. As mentioned above Figure 4B As shown in the embodiment of FIG. 5 , the output from the HPF may be saturated. Therefore, the baseline value recovery 552 of the signal 450 is even better than that described above. Figure 5A For example, in this example, the overall baseline recovery time for the saturated signal is shown to be approximately 2200 (arbitrary units), while Figure 5A The nominal signal in the embodiment is then about 800.

[0088] Furthermore, the residual data artifact 554 dominates the baseline recovery process 552. Specifically, even though the slowly recovered baseline 552 remains below zero before the artifact occurs, the data artifact 554 subsequently falsely pushes the signal 450 above zero. Thus, in reality, the baseline does not fully recover to zero until 2700 hours.

[0089] Fig. 6A 600 is graphically illustrated using a dynamically set corner frequency filtered signal according to an embodiment of the present disclosure. The signal 600 may be based on Figure 4A The same input data as signal 400 in the embodiment is filtered using dynamic corner frequencies instead of a single corner frequency. Specifically, peak 602 is shown to have a height of 7,480, similar to Figure 4A In this embodiment, peak 602 has a precise shape, and smaller peak 604 can be measured with a correct height of approximately 459 units, as shown, because the baseline has been restored. Specifically, baseline 606 near smaller peak 604 is very close to the horizontal axis, as shown. Therefore, the disclosed system and method has improved the accuracy of peak measurement from Figure 4A The example rises from about 70% of the correct height in the embodiment to about 93% of the correct height in this embodiment.

[0090] also, Fig. 6A A control signal 608 is shown, which may correspond to switching a multiplexer (eg, Figure 3B For example, the control signal 608 may correspond to Figure 3B D_CTRL 356 of an embodiment, or Figure 3C 604. The control signal 608 is a control signal that is used to control the multiplexer 600. In this embodiment, the threshold has been set to 300 units, that is, when the time domain amplitude of the signal 600 exceeds 300 units, the control signal 608 instructs the multiplexer to switch to a low corner frequency. Therefore, as shown, the control signal 608 can instruct the multiplexer to switch to a low corner frequency during times 610 and 612 corresponding to peaks 602 and 604, respectively. At other times, the control signal 608 can instruct to remain at a high corner frequency. By switching the multiplexer, the control signal 608 can quickly restore the baseline while maintaining good signal quality for peaks 602 and 604.

[0091] Figure 6B6 shows a graphical representation of baseline recovery of a saturated signal 650 filtered with a dynamically set corner frequency according to an embodiment of the present disclosure. The signal 650 may be based on Figure 4B The same input data as signal 450 in the embodiment is filtered using dynamic corner frequencies instead of a single corner frequency. Specifically, in this embodiment, the height of peak 652 is 11,463, similar to the above Figure 4B Thus, in this embodiment, peak 652 is quite high, so that the HPF may become saturated, as shown in FIG. Figure 4B However, using the disclosed dynamic filtering system and method, peak 652 maintains a more accurate shape. Specifically, the top of peak 652 is substantially flat, Figure 4B The inclined peak 452 is in contrast.

[0092] Furthermore, since the baseline 656 has been restored, the correct height of the smaller peak 654 can be measured, as shown, to be approximately 488 units. Specifically, the measured height of the smaller peak 654 is 98% of its ideal height of 500 units, while Figure 4B In the case of a single frequency HPF, the height of the small peak 454 is too low to be detected. Note that due to noise, the measured height of a peak may be lower or even higher than its ideal height, but in this example, the measured height is still lower than the ideal height. In addition, as shown, the baseline 656 returns to zero faster than Figure 4B The baseline 456 in the embodiment is much faster. Note that the time it takes for the baseline to go below zero is significantly reduced.

[0093] Therefore, the baseline 658 of the peak 654 corresponds to a value close to zero, so the height of the peak 654 can be measured correctly, which is consistent with Figure 4B The opposite is true for peak 454 in . By significantly improving baseline recovery time, the disclosed systems and methods reduce the lag time required to recover from saturation peak 652 before the system is ready to measure subsequent signal 654 .

[0094] also, Figure 6B A control signal 660 is shown, which may correspond to switching a multiplexer (eg, Figure 3B 650). In this embodiment, the threshold for switching to a low corner frequency has been set to the time domain amplitude of signal 650, which is greater than 300 units. Thus, as shown, control signal 660 may instruct the multiplexer to switch to a low corner frequency during times 662 and 664 corresponding to peaks 652 and 654, respectively. At other times, control signal 660 may instruct to remain at a high corner frequency. Thus, control signal 660 may quickly restore the baseline while maintaining good signal quality for peaks 652 and 654.

[0095] Fig. 7A 600 is graphically shown in further detail using a dynamically set corner frequency filtered signal according to an embodiment of the present disclosure. In this embodiment, the signal 600 is from the above Fig. 6A Detailed view of signal 600, where the vertical axis is expanded and the horizontal axis is contracted. Figure 5A As in the embodiment of FIG. 6 , the signal 600 output from the HPF in this embodiment may be in a nominal state. However, using the disclosed system and method to dynamically set the corner frequency, the baseline in this embodiment may be greater than Figure 5A 702. As shown, the baseline 702 of the signal 600 does not lag below zero for an appreciable amount of time. Specifically, in this example, the overall baseline recovery time (until the value of the baseline is dominated by noise) is shown to be approximately 50 (arbitrary units) of the nominal peak, while the above Figure 5A The nominal signal in the embodiment is approximately 800.

[0096] Figure 7B 650 is graphically shown, in accordance with an embodiment of the present disclosure, for baseline recovery of a saturated signal 650 filtered with a dynamically set corner frequency. In this embodiment, signal 650 is from the above Figure 6B Detailed view of signal 650, where the vertical axis is expanded and the horizontal axis is contracted. Figure 5B Compared to the artifact 554 in the embodiment of the present invention, the amplitude of the residual data artifact 754 is significantly reduced. Specifically, when the output signal amplitude is greater than the switching threshold, the system can set a low cutoff frequency for filtering. Such a low cutoff frequency can cause most of the frequency components of the input signal to be retained, so it is likely that the shape and quantitative value of the signal will be retained. For example, the peak of the false residual data artifact 754 is approximately half the amplitude of the artifact 554. In addition, the time span of the residual data artifact 754 is five times shorter than that of the artifact 554.

[0097] In this embodiment, the baseline 752 of the filtered signal also recovers much faster than in the case of a single corner frequency. When the output signal amplitude is less than the switching threshold, the system can set a high cutoff frequency for filtering. Such a high cutoff frequency may cause low-frequency components of the input signal to be filtered out, so low-frequency artifacts that take a long time to correct are likely to be missing from the filtered signal 650. As shown in this embodiment, the total time for baseline recovery 752 is about 700 (arbitrary units) after the saturation peak, while the above Figure 5B The total time of the saturation signal in the embodiment is approximately 2200.

[0098] method

[0099] Fig. 8A800 is a flow chart of a method 800 for dynamically filtering frequency domain values ​​according to an embodiment of the present disclosure. In various embodiments, the method 800 may utilize a HPF circuit (e.g., Figure 3A , Figure 3B and Figure 3C The dynamic HPF circuit in the embodiment of the present invention is used to perform.

[0100] like Fig. 8A As shown, the dynamic filtering frequency domain value first sets 802 the filtering characteristics in response to the time domain value of the output signal amplitude. For example, the filtering characteristics may include a cutoff frequency or corner frequency for filtering the frequency domain value, such as using an HPF, a low pass filter, a band pass filter, or other filter. In some embodiments, the system can adjust the filtering characteristics within a time interval (e.g., within 25ns, 100ns, or 1μs) after the signal amplitude exceeds a predefined threshold. The threshold can be a fixed value, or can be adjusted by the system based on additional criteria. In some cases, the time interval can be a few clock ticks, such as 3 or 4 clock ticks. This can correspond to an almost immediate time interval, such as less than 40 nanoseconds, or less than 50 nanoseconds. In another embodiment, the time interval is less than 1 microsecond.

[0101] Next, the dynamic filtering frequency domain values ​​continue to filter 804 the frequency domain values ​​of the input signal amplitude based on the filtering characteristic. For example, filtering 804 frequency domain values ​​can include using an HPF, a low pass filter, a band pass filter, or other filters with a cut-off frequency or a corner frequency corresponding to the filtering characteristic. In the case of an HPF, filtering 804 frequency domain values ​​can include filtering out frequency components of frequencies below the filtering characteristic, such as a cut-off frequency or a corner frequency. Alternatively, in the case of a low pass filter, filtering 804 frequency domain values ​​can include filtering out frequencies above the filtering characteristic, such as a cut-off frequency or a corner frequency. In the case of a band pass filter, filtering 804 frequency domain values ​​can include filtering out frequencies outside a frequency band or range. For example, the filtering characteristic can include a lower cut-off frequency of a frequency band or range, a higher cut-off frequency of a frequency band or range, or both. Then, the band pass filter can filter out 804 frequencies below a low cut-off frequency and / or above a high cut-off frequency.

[0102] Figure 8B 8 is a flow chart illustrating a method 850 for dynamically filtering frequency domain values ​​based on the time domain amplitude of an output signal according to an embodiment of the present disclosure. In various embodiments, the method 850 may be performed using a HPF circuit, such as the one described above. Figure 3A , Figure 3B and Figure 3C In some embodiments, the method 850 shows a single cycle of evaluation of the output signal and can be repeated at a high frequency, for example, the method can be repeated with each clock tick of the system.

[0103] like Figure 8B As shown, dynamically filtering frequency domain values ​​based on time domain amplitude begins by determining 852 whether the time domain value of the output signal amplitude is above a threshold. For example, the multiplexer can select a branch based on comparing the amplitude to a switching threshold, as described above. Figure 3B For example, the selection line of the multiplexer can be based on a Boolean comparison signal D_CTRL, which can compare the amplitude with the switching threshold. In the case of a multi-channel signal, the comparison signal D_CTRL can be the result of comparing all signals with the threshold, as described above. Figure 3C In some embodiments, the comparison may be performed by one or more comparators, such as Figure 3C In some embodiments, the disclosed system is configured to provide threshold detection control within a short interval after a threshold is crossed, for example, within a single clock tick, three clock ticks, 25ns, 100ns, or 1 μs of the crossing.

[0104] In some cases, the system may compare the signal amplitude to multiple thresholds, such as multiple values ​​for setting the HPF corner frequency.

[0105] Next, if the time domain value of the signal amplitude is lower than the threshold, the frequency domain value is continuously dynamically filtered, and the corner frequency is set 854 to a high value, such as 500kHz or 600kHz. Specifically, this situation may correspond to the above Fig. 6A . By determining that the signal amplitude is below the threshold, the system can determine that the signal is not saturated. Therefore, setting the HPF corner frequency 854 to a high value results in filtering out many of the low frequency components of the signal. However, since the signal below the threshold is not sufficient to saturate the HPF, the baseline value is still expected to be restored after filtering.

[0106] If the time domain value of the output signal amplitude is higher than the threshold, the frequency domain value is continuously dynamically filtered and the corner frequency is set to a low value 856, such as 200 Hz or 4 kHz. Specifically, this situation may correspond to the above Figure 6B . By determining that the signal amplitude is above a threshold, the system can determine that the signal is saturated. Therefore, setting the HPF corner frequency to a low value of 856 results in retaining most of the frequency components of the signal, including many low frequency components. Therefore, even if the initial signal peak saturates the HPF, the baseline value can be quickly restored after filtering due to the low corner frequency. In addition, by setting a low value of 856 for the corner frequency, the system can improve the accuracy of the signal shape compared to a single corner HPF.

[0107] The system can set the corner frequency by selecting the branch of the circuit. For example, the system can be as above Figure 3BThe system may select the appropriate circuit branch in response to the time domain signal amplitude, and then the displacement or other circuit corresponding to the selected branch may set the cutoff frequency. Note that the selection of the cutoff frequency may change dynamically as the signal amplitude changes in the time domain. For example, in some embodiments, the system may adjust the cutoff frequency within a time interval after each crossing or falling below a threshold value of the signal amplitude, for example, within 25ns, 100ns, or 1μs after each crossing. In some cases, the system is configured to dynamically change the corner frequency within a time interval after the time domain value of the amplitude exceeds the threshold value. In some cases, the time interval may be a few clock ticks, such as 3 or 4 clock ticks. This may correspond to an almost immediate time interval, such as less than 40 nanoseconds, or less than 50 nanoseconds. In another example, the time interval is less than 1 microsecond.

[0108] In embodiments where the system uses multiple thresholds, the system can select among multiple circuit branches to set the appropriate cutoff frequency. For example, if there are N cutoff frequencies resulting in (N+1) amplitude ranges, the system can select among (N+1) circuit branches corresponding to the (N+1) cutoff frequencies.

[0109] Next, the system can filter out 858 frequency components below the corner frequency. For example, the system can use an HPF to filter out low frequency components. In some embodiments, the system can use another type of filter to filter out frequency components in other ranges. For example, the system can use a low pass filter to filter out frequencies above the cutoff frequency, or can use a band pass filter to filter out frequencies outside of a frequency band or range.

[0110] The description of the disclosed embodiments is presented for the purpose of illustration and description. It is not intended to be exhaustive or to limit the disclosure to the precise form disclosed. Many modifications and variations may be made in light of the disclosure. The scope of the disclosure is not limited by this detailed description, but by the appended claims.

Claims

1. A system comprising: A filter circuit having an input and an output and configured as filtering an input signal received from an input of the filter circuit based on a corner frequency, and passing the filtered input signal to an output of the filter circuit; a corner frequency up-adjustment circuit configured to increase the corner frequency; a corner frequency down-adjustment circuit configured to reduce the corner frequency; and A selector circuit is coupled to the output of the filter circuit and to one or both of the corner frequency up-adjustment circuit and the corner frequency down-adjustment circuit, the selector circuit being configured to select the corner frequency up-adjustment circuit or the corner frequency down-adjustment circuit based on the filtered input signal.

2. The system of claim 1, wherein: The filter circuit includes a high pass filter, and filtering the input signal based on the corner frequency includes: filtering out frequency components below the corner frequency; or The filter circuit includes a low pass filter, and filtering the input signal based on the corner frequency includes filtering out frequency components above the corner frequency.

3. The system according to claim 1, wherein: The selector circuit comprises: Threshold terminal or node; switch, which has a first terminal coupled to the output of the filter circuit, a second terminal coupled to one or both of the corner frequency up-adjustment circuit and the corner frequency down-adjustment circuit, and control inputs; and A comparator having a first comparator input coupled to the output of the filter circuit, a second comparator input coupled to the threshold terminal or node, and a comparator output coupled to the control input of the switch.

4. The system of claim 3, wherein: Selecting between the corner frequency up-adjustment circuit and the corner frequency down-adjustment circuit based on the filtered input signal includes: switching between the corner frequency up-adjustment circuit and the corner frequency down-adjustment circuit in response to the filtered input signal crossing a threshold at the threshold terminal or node; and Switching between the corner frequency up-adjustment circuit and the corner frequency down-adjustment circuit in response to the filtered input signal crossing the threshold includes switching between the corner frequency up-adjustment circuit and the corner frequency down-adjustment circuit by comparing the filtered input signal with a threshold voltage corresponding to the threshold by the comparator.

5. The system of claim 1, wherein: The selector circuit includes a multiplexer having: a first terminal coupled to the corner frequency up-adjustment circuit, the corner frequency up-adjustment circuit being configured to perform a first shift; a second terminal coupled to the corner frequency down-adjustment circuit, the corner frequency down-adjustment circuit being configured to perform a second shift; and a control input coupled to the comparison circuit, the comparison circuit configured to perform a comparison of the filtered input signal with a threshold value; and Selecting between the corner frequency up-adjustment circuit and the corner frequency down-adjustment circuit based on the filtered input signal includes selecting, by the multiplexer, between the corner frequency up-adjustment circuit and the corner frequency down-adjustment circuit based on the control input.

6. A system configured to filter frequency domain values ​​of an input signal based on a filtering characteristic, wherein: The filter characteristic is dynamically set in response to a time domain value of the amplitude of the output signal.

7. The system according to claim 6, wherein: The filtering characteristic includes a corner frequency, and the system includes a high pass filter configured to filter out frequency components having frequencies lower than the corner frequency.

8. The system of claim 7, wherein: The system is configured to dynamically change the corner frequency from a first value to a second value, the second value being higher than the first value, in response to the time-domain value of the amplitude falling below a threshold; or The system is configured to dynamically change the corner frequency from a first value to a second value, which is lower than the first value, in response to the time-domain value of the amplitude rising above a threshold.

9. The system according to claim 7, wherein: The system is configured to dynamically change the corner frequency within a time interval after the time domain value of the amplitude crosses a threshold.

10. The system of claim 9, wherein: This time interval is less than 1 microsecond; The interval is 3 ticks of the system clock or less; or The time interval is short enough so that the signal baseline is preserved, thereby allowing unipolar data measurement consistent with a predetermined accuracy threshold.

11. The system of claim 6, wherein: The system is a multi-channel system; The amplitude is a corresponding amplitude among a plurality of amplitudes corresponding to a plurality of signals; and The system is configured to set the filter characteristic in response to a corresponding value of any one of the plurality of amplitudes crossing a threshold value.

12. The system of claim 6, wherein: The output signal is a digital output signal; and Filtering the frequency domain value comprises, for example, shifting the digital output signal.

13. The system of claim 12, comprising a multiplexer configured to dynamically select the shift value based on the amplitude of the output signal.

14. The system of claim 6, wherein: The output signal is an analog output signal; and Setting the filtering characteristic includes setting a switch state based on an output of a comparator that compares the time domain value of the amplitude to a threshold voltage.

15. The system according to claim 6, wherein: The filtering characteristic includes a corner frequency, and the system includes a low pass filter configured to filter out frequency components above the corner frequency.

16. A method for filtering frequency domain values, comprising: dynamically setting a filter characteristic in response to a time domain value of the amplitude of the output signal; and Filters the frequency domain values ​​of the input signal based on dynamically set filter characteristics.

17. The method according to claim 16, wherein: The filter characteristic includes a corner frequency, and filtering the frequency domain value includes filtering out frequency components having a frequency lower than the corner frequency via a high pass filter.

18. The method of claim 17, wherein: Dynamically setting the filter characteristic includes: dynamically changing the corner frequency from a first value to a second value, the second value being higher than the first value, in response to the time domain value of the amplitude falling below a threshold; or Dynamically setting the filter characteristic includes dynamically changing the corner frequency from a first value to a second value, which is lower than the first value, in response to the time-domain value of the amplitude rising above a threshold.

19. The method according to claim 16, further comprising: receiving a plurality of signals, wherein the amplitude is a corresponding amplitude of a plurality of amplitudes corresponding to the plurality of signals; and The filter characteristic is set in response to a corresponding value of any one of the plurality of amplitudes crossing a threshold value.

20. The method of claim 16, wherein: The input signal is a digital input signal received by a computing device; Dynamically setting the filtering characteristic includes: dynamically setting the filtering characteristic by the computing device; Filtering the frequency domain value of the input signal includes: filtering the frequency domain value of the input signal by the computing device.

Images