An ADPLL fractional spurious suppression device and optimization method based on a spurious notch filter

By introducing a spurious notch filter into the digital loop of a fractional digital phase-locked loop (PLL), and using an infinite impulse response filter to selectively suppress fractional spurious signals, the problem of effectively suppressing periodic errors in existing technologies is solved, achieving a highly efficient spurious suppression effect while reducing the impact on PLL performance.

CN123074484APending Publication Date: 2026-11-10BEIJING UNIV OF TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611174719.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-04
Publication Date
2026-11-10

AI Technical Summary

Technical Problem

Existing fractional digital phase-locked loops (PLLs) have shortcomings in suppressing fractional spurious signals. They mainly rely on digital time converters and their calibration circuits, which makes it difficult to effectively suppress periodic residual errors that have entered the PLL signal path. Furthermore, traditional methods may affect the PLL bandwidth, locking speed, and power consumption.

Method used

By introducing a spurious notch filter into the digital loop signal path of a fractional digital phase-locked loop, the periodic error components that have entered the loop are selectively suppressed. An infinite impulse response filter is used to form a notch at the target frequency, thereby achieving targeted suppression of fractional spurious signals.

Benefits of technology

It effectively suppresses fractional spurious emissions, reduces the impact on the overall performance of the phase-locked loop, improves spectral purity and adjacent channel interference performance, and reduces circuit design complexity and power consumption.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123074484A_ABST
    Figure CN123074484A_ABST
Patent Text Reader

Abstract

This invention discloses an ADPLL fractional spurious suppression device and optimization method based on a spurious notch filter, belonging to the field of mixed-signal integrated circuit design in electronic communications. This invention incorporates a spurious notch filter in the digital loop signal path of a phase-locked loop (PLL). The spurious notch filter determines the digital angular frequency corresponding to the target spurious frequency based on frequency parameters such as the reference frequency and the fractional division ratio, and generates filter coefficients based on this digital angular frequency. This causes the notch filter to form a narrow-band deep notch at the target spurious frequency, selectively attenuating the periodic error component while allowing other non-target frequency signals to pass normally. This invention extends fractional spurious suppression from the digital time converter and its calibration circuit to the digital loop filtering stage, effectively reducing fractional spurious signals on both sides of the output carrier while avoiding significant impacts on PLL bandwidth, locking speed, and phase noise. It is suitable for high-performance fractional digital PLL frequency synthesizers.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention designs a spurious notch filter for ADPLL fractional spurious optimization, belonging to mixed-signal integrated circuit design in the field of electronic communication. Background Technology

[0002] Wireless communication technology is a crucial foundation of modern information society, and its continuous evolution drives the rapid development of mobile communications, the Internet of Things (IoT), intelligent transportation, satellite communications, and high-precision positioning. With the widespread application of fifth-generation (5G) mobile communication technology and the deepening research into sixth-generation (6G) mobile communication technology, wireless communication systems are evolving towards higher data rates, lower transmission latency, higher spectrum utilization, and lower bit error rates. To meet the demands of multi-band, multi-standard, and broadband communication, radio frequency transceivers need to provide stable, low-noise, and high-frequency-resolution local oscillator signals over a wide frequency range. Therefore, higher requirements are placed on the output spectral purity, frequency switching speed, power consumption, and integration of frequency synthesizers.

[0003] Phase-locked loops (PLLs) are key modules in RF transceivers for generating local oscillator frequencies and synthesizing frequencies. The phase noise and spurious performance of their output signals directly affect the error vector amplitude, adjacent channel leakage, receiver sensitivity, and anti-blocking capability of the communication system. Compared to integer-division PLLs, fractional-division PLLs can achieve smaller frequency steps and higher frequency configuration flexibility while maintaining the same reference frequency. Therefore, they are widely used in multi-band wireless communication systems, high-speed data interfaces, and high-precision clock generation circuits. Especially with the development of digital integrated circuit technology, fractional-digital PLLs employing digital loop filters, digitally controlled oscillators, and digital calibration modules offer advantages such as high programmability, good process portability, and ease of integration with digital auxiliary calibration, making them an important implementation form for highly integrated frequency synthesizers.

[0004] However, when fractional digital phase-locked loops (PLLs) achieve non-integer frequency division ratios, they typically require a digital modulator to modulate the frequency division control word or phase compensation control word. The quantization error generated by the digital modulator has a certain periodicity. When this quantization error passes through the nonlinear module or gain mismatch module in the PLL, it easily transforms into a periodic phase error, forming discrete fractional spurious noise on both sides of the output carrier. Unlike continuously distributed random phase noise, fractional spurious noise is usually concentrated at one or more specific frequency positions. Even if its amplitude is small, it may fall into adjacent channels or the receiving bandwidth, thereby reducing the spectral purity of the output signal and deteriorating the system's error vector amplitude and adjacent channel interference performance. Therefore, fractional spurious noise is one of the important indicators for evaluating the performance of fractional digital PLLs, and its effective suppression is a key issue in the design of high-performance frequency synthesizers.

[0005] In existing fractional digital phase-locked loops (PLLs), a digital time converter (DTC) is typically used to compensate for the fractional quantization error generated by the digital modulator in the time or phase domain. Ideally, the controllable delay generated by the DTC can offset the fractional phase error in the frequency division feedback signal, thereby reducing the periodic phase disturbance at the phase detector input. However, practical DTCs suffer from gain errors, integral nonlinearity, differential nonlinearity, delay unit mismatch, and variations in power supply voltage, temperature, and manufacturing processes, making it difficult for the actual delay generated to perfectly match the fractional phase error to be compensated. The residual compensation error may still exhibit significant periodicity and be transmitted to the oscillator control terminal through the PLL, ultimately forming fractional spurious signals in the output spectrum.

[0006] To mitigate the impact of the aforementioned non-ideal factors, existing technologies have proposed various improvement schemes, including digital time converter gain calibration, nonlinear calibration, digital predistortion, multi-channel digital time converters, differential digital time converters, and pseudo-differential digital time converters. These schemes primarily reduce periodic residual errors by improving the linearity, matching accuracy, or error estimation accuracy of the digital time converter, thereby reducing these errors at the location where fractional phase errors occur or are compensated. While these methods can improve fractional spurious performance to some extent, their suppression effectiveness generally still depends on the resolution, dynamic range, linearity, and calibration accuracy of the digital time converter itself. As the output frequency of phase-locked loops increases and spurious performance becomes increasingly stringent, digital time converters need to achieve smaller time resolutions and higher linearity, leading to increased circuit design difficulty, calibration complexity, chip area, and power consumption.

[0007] Furthermore, when the structure of a digital time converter lacks room for further improvement, relying solely on increasing calibration accuracy to suppress fractional spurious emissions often requires significant hardware and power consumption, and residual fractional spurious emissions may still remain. Therefore, most existing fractional spurious emission suppression methods focus on the digital time converter and its calibration circuit, primarily compensating for periodic phase errors at the error source. However, for periodic residual errors that have already entered the phase-locked loop (PLL) signal path, there is a lack of effective means to selectively suppress them from the perspective of the PLL's closed-loop transfer characteristics.

[0008] If a conventional low-pass method is used to reduce the loop gain at the fractional spurious frequency, it is usually necessary to reduce the PLL bandwidth. However, this will simultaneously reduce the loop's tracking capability and locking speed, and may also alter the PLL's in-band noise performance. Fractional spurs are characterized by their relatively concentrated frequency locations, narrow spectral width, and frequency dependence on the fractional-division control parameters. Therefore, it is necessary to selectively suppress the discrete frequency components corresponding to the fractional spurs without significantly altering the overall PLL bandwidth and non-target frequency response.

[0009] Therefore, there is an urgent need for a fractional spurious suppression circuit and method for fractional digital phase-locked loops (PLLs). This method can overcome the limitations of traditional solutions that mainly rely on the structural optimization and error calibration of digital time converters. It can suppress fractional spurious signals and their harmonic components in a targeted manner from the perspective of PLL loop signal transmission and frequency response, and minimize the impact of the spurious suppression process on PLL bandwidth, locking speed, phase noise, and power consumption. Summary of the Invention

[0010] To address the problem of periodic control word changes caused by fractional quantization errors in fractional digital phase-locked loops, resulting in fractional spurious signals, this invention proposes a spurious notch suppression method and apparatus for fractional digital phase-locked loops.

[0011] During the operation of a fractional digital phase-locked loop (PLL), the quantization error generated by fractional modulation causes the control word of the digitally controlled oscillator (DCO) to exhibit periodic changes. These periodic changes periodically modulate the output phase or frequency of the DCO, resulting in fractional spurious signals on both sides of the output carrier. Existing technologies typically reduce these errors by addressing the digital time converter and its calibration circuit. However, this invention focuses on the signal transmission process within the PLL loop, selectively suppressing the periodic error components that have already entered the loop.

[0012] Periodic signals can be decomposed into one or more sinusoidal components of different frequencies. For any target frequency component, by generating a signal with an amplitude corresponding to that frequency component but opposite in phase, and then superimposing the two signals, the frequency components can be canceled out or significantly reduced. Based on the above principle, this invention introduces a spurious notch filter into the digital loop signal path of a fractional phase-locked loop, thereby suppressing the periodic components in the loop signal corresponding to fractional spurious signals, while allowing other non-target frequency components to pass through essentially normally.

[0013] The spurious notch filter determines the frequency location of the fractional spurious based on the reference frequency of the fractional digital phase-locked loop, the fractional division ratio, or other parameters related to the fractional modulation state, and forms a notch at the corresponding frequency. When the fractional division ratio or the operating frequency changes, the spurious notch filter can adjust the notch position accordingly to maintain a correspondence with the changed fractional spur frequency.

[0014] The spurious notch filter can generate a canceling component with the opposite phase to the target periodic component through delay, phase adjustment, addition / subtraction operations, feedback operations, or other digital signal processing methods. Specifically, it can be implemented using a digital notch filter, a finite impulse response filter, an infinite impulse response filter, or other digital processing structures capable of attenuating at the target frequency. This invention uses an infinite impulse response filter to implement a single notch unit, which can generate a deep notch using zeros and then compress the stopband width using poles near the zeros. Therefore, only a lower order is needed to form a very narrow and deep attenuation at the target spurious frequency, while minimizing changes to the loop signal at other frequencies.

[0015] Specifically, the technical solution designed in this invention is a fractional spurious suppression device based on a fractional digital phase-locked loop (PLL). The device is disposed in the digital loop signal path of the fractional digital PLL and is used to selectively attenuate the periodic signal components in the digital loop signal path that correspond to the fractional spurious frequency.

[0016] The device includes:

[0017] A coefficient generator is configured to receive frequency parameters related to the operating state of the fractional digital phase-locked loop, determine the digital angular frequency corresponding to the target spurious frequency based on the frequency parameters, and generate filter coefficients based on the digital angular frequency.

[0018] A digital notch filter, coupled to the coefficient generator, is configured to receive the filter coefficients and form a notch at the target spurious frequency based on the filter coefficients to filter the signal in the digital loop signal path, wherein the amplitude response of the digital notch filter at the target spurious frequency is lower than the amplitude response at other non-target frequencies.

[0019] Furthermore, the digital notch filter is an infinite impulse response filter, whose transfer function has a pair of conjugate zeros at the angle corresponding to the target spurious frequency on the unit circle in the z-plane, and a pair of conjugate poles at the same angle with a radius of r, where 0 <r<1。

[0020] Furthermore, the transfer function of the infinite impulse response filter is:

[0021]

[0022] Wherein, ω0 is the digital angular frequency corresponding to the target spurious frequency, and r is the radius of the conjugate pole.

[0023] Furthermore, the time-domain difference equation corresponding to the infinite impulse response filter is:

[0024]

[0025] Where x[n] is the input signal at the current sampling time, y[n] is the output signal at the current sampling time, x[n-1] and x[n-2] are the input signals at the previous and previous two sampling times, respectively, and y[n-1] and y[n-2] are the output signals at the previous and previous two sampling times, respectively.

[0026] Furthermore, the coefficient generator includes a CORDIC calculation module configured to perform iterative calculations on the digital angular frequency to obtain the cosine value of the digital angular frequency; the filter coefficients include the cosine value; the coefficient generator also includes a coefficient register configured to latch the calculated filter coefficients; when the frequency parameter changes, the coefficient generator recalculates and updates the filter coefficients in the coefficient register; when the frequency parameter does not change, the CORDIC calculation module stops working or turns off the clock.

[0027] Furthermore, the digital notch filter is implemented using the Verilog hardware description language and includes an input delay register, an output feedback delay register, a multiplier, an adder, and a coefficient register; the frequency parameters include the reference frequency of the fractional digital phase-locked loop and the fractional part of the fractional division ratio; the target spurious frequency is determined based on the reference frequency and the fractional part of the fractional division ratio.

[0028] Furthermore, the target spurious frequency includes the fundamental fractional spurious frequency and at least one harmonic frequency; the digital notch filter is a cascaded plurality of filters, each stage of the digital notch filter receiving filter coefficients corresponding to different harmonic orders, so as to form notches at the fundamental fractional spurious frequency and each of the harmonic frequencies respectively.

[0029] A fractional spurious suppression method based on a fractional digital phase-locked loop using a spurious notch filter includes:

[0030] Obtain the frequency parameters related to the operating state of the fractional digital phase-locked loop;

[0031] Determine the digital angular frequency corresponding to the target spurious frequency based on the frequency parameters;

[0032] Filter coefficients are generated based on the digital angular frequency;

[0033] The filter coefficients are provided to a digital notch filter in the digital loop signal path of the fractional digital phase-locked loop, so that the digital notch filter forms a notch at the target spurious frequency and filters the signal in the digital loop signal path. The amplitude response of the digital notch filter at the target spurious frequency is lower than the amplitude response at other non-target frequencies.

[0034] The digital notch filter is an infinite impulse response filter; the step of generating filter coefficients based on the digital angular frequency includes:

[0035] Calculate the cosine value of the digital angular frequency;

[0036] The transfer function coefficients of the infinite impulse response filter are generated based on the cosine value and the preset pole radius parameter.

[0037] The frequency parameters include the reference frequency of the fractional digital phase-locked loop and the fractional part of the fractional division ratio; the step of determining the digital angular frequency corresponding to the target spurious frequency based on the frequency parameters includes:

[0038] The target spurious frequency is determined based on the reference frequency and the fractional part of the fractional frequency division ratio.

[0039] The digital angular frequency is determined based on the target spurious frequency and the sampling frequency or update frequency of the digital notch filter.

[0040] When the frequency parameters change, the acquisition, determination, and generation steps are re-executed to update the filter coefficients;

[0041] When the frequency parameters do not change, the current filter coefficients remain unchanged;

[0042] The target spurious frequency includes the fundamental fractional spurious frequency and at least one harmonic frequency; the method includes:

[0043] Cascade multiple of the aforementioned digital notch filters;

[0044] Each stage of the digital notch filter generates filter coefficients corresponding to different harmonic orders, so that each stage of the digital notch filter forms a notch at the fundamental fractional spurious frequency and each of the harmonic frequencies.

[0045] Through the above technical solution, the present invention extends the suppression of fractional spurious signals from the digital time converter itself to the phase-locked loop digital loop. By selectively canceling specific frequency components of periodic errors, fractional spurious signals are suppressed, while reducing the impact on other frequency signals in the loop and the overall loop performance. Attached Figure Description

[0046] Figure 1 This is a system block diagram of the infinite impulse response filter in this invention.

[0047] Figure 2 This is the Bode plot of the z-domain transfer function of the spurious notch filter in this invention.

[0048] Figure 3 This is a simulation result diagram of spurious optimization after applying the spurious notch filter to ADPLL in this invention. Detailed Implementation

[0049] This invention comprises two parts: a coefficient generator and an infinite impulse response filter.

[0050] The first part is the coefficient generator. The location of fractional spurious signals in the phase-locked loop is related to the reference frequency and the fractional division ratio. Therefore, the fractional spurious frequency to be suppressed can be determined based on the reference frequency and the fractional part of the fractional division ratio. Let the target spurious frequency to be suppressed be f. spur The sampling frequency or update frequency of the digital notch filter is f. s Then the digital angular frequency ω0 corresponding to the target spurious frequency is:

[0051]

[0052] Since the notch position of the digital notch filter is determined by the digital angular frequency ω0, the coefficient generator first calculates the target digital angular frequency ω0 based on the reference frequency, the fractional division ratio, and the sampling frequency of the digital filter, and then calculates the cosine value cos(ω0) corresponding to this digital angular frequency. In one embodiment, the coefficient generator uses the CORDIC algorithm to iteratively calculate the digital angular frequency ω0 to obtain the corresponding cos(ω0), and outputs the calculation result to the coefficient input port of the digital notch filter. After the coefficient calculation is completed, the calculation result is latched in the coefficient register, and the CORDIC calculation module can stop working or turn off the clock; when the reference frequency or the fractional division ratio changes, the new filter coefficients are recalculated, so that the notch position can be adjusted according to the fractional spurious position.

[0053] The second part is the digital IIR filter. To create a notch at the target digital angular frequency ω0, a pair of conjugate zeros are placed on the unit circle in the z-plane at the position corresponding to the angular frequency ω0, so that the amplitude response of the digital filter at this frequency is zero or close to zero. Simultaneously, a pair of conjugate poles are placed at the same angle and with a radius of r, where:

[0054]

[0055] The conjugate poles are located inside the unit circle to reduce the stopband width of the notch filter. The closer r is to 1, the narrower the stopband of the notch filter, thus reducing the impact on the loop signal at adjacent frequencies while suppressing the target fractional spurious signals.

[0056] Based on the locations of the zeros and poles mentioned above, the transfer function of the digital spurious notch filter is:

[0057]

[0058] The corresponding time-domain difference equation can be obtained from the transfer function:

[0059]

[0060] Where x[n] is the input signal of the digital spurious notch filter at the nth sampling time, and y[n] is the output signal of the digital spurious notch filter at the nth sampling time; x[n-1] and x[n-2] are the input signals at the previous sampling time and the previous two sampling times, respectively; y[n-1] and y[n-2] are the output signals at the previous sampling time and the previous two sampling times, respectively. Let the coefficients a0 = -2cos(ω0), b0 = 2rcos(ω0), b1 = -r 2 The time-domain difference equation is reduced to the following:

[0061]

[0062] Based on the aforementioned difference equation, a digital spurious notch filter can be constructed from an input delay register, an output feedback delay register, a multiplier, an adder, and a coefficient register, and implemented digitally using the Verilog hardware description language. The digital spurious notch filter is configured with a coefficient input port for receiving cos(ω0); when the pole radius r is fixed, the remaining filter coefficients can be directly generated based on cos(ω0) and the fixed r. By changing the input cos(ω0), the angles of the zeros and poles in the z-plane can be changed, thereby allowing the digital spurious notch filter to generate corresponding narrowband notches at different target frequency locations.

[0063] When it is necessary to suppress fractional spurious components of higher harmonics, multiple of the above-mentioned spurious notch filters can be cascaded, and different cos(ω0) coefficients can be input for each stage, so that each stage forms notches at the fundamental frequency fractional spurious components and their different harmonic positions, thereby meeting the application scenarios with more stringent fractional spurious requirements.

Claims

1. An ADPLL fractional spurious suppression device based on a spurious notch filter, characterized in that, The device is installed in the digital loop signal path of the fractional digital phase-locked loop and is used to selectively attenuate the periodic signal components in the digital loop signal path that correspond to the fractional spurious frequency. The device includes: A coefficient generator is configured to receive frequency parameters related to the operating state of the fractional digital phase-locked loop, determine the digital angular frequency corresponding to the target spurious frequency based on the frequency parameters, and generate filter coefficients based on the digital angular frequency. A digital notch filter, coupled to the coefficient generator, is configured to receive the filter coefficients and form a notch at the target spurious frequency based on the filter coefficients to filter the signal in the digital loop signal path, wherein the amplitude response of the digital notch filter at the target spurious frequency is lower than the amplitude response at other non-target frequencies.

2. The ADPLL fractional spurious suppression device based on a spurious notch filter according to claim 1, characterized in that, The digital notch filter is an infinite impulse response filter. Its transfer function has a pair of conjugate zeros at the angle corresponding to the target spurious frequency on the unit circle in the z-plane, and a pair of conjugate poles at the same angle with a radius of r. <r<1。 3. The ADPLL fractional spurious suppression device based on a spurious notch filter according to claim 2, characterized in that, The transfer function of the infinite impulse response filter is: ; Wherein, ω0 is the digital angular frequency corresponding to the target spurious frequency, and r is the radius of the conjugate pole.

4. The ADPLL fractional spurious suppression device based on a spurious notch filter according to claim 3, characterized in that, The time-domain difference equation corresponding to the infinite impulse response filter is: ; Where x[n] is the input signal at the current sampling time, y[n] is the output signal at the current sampling time, x[n-1] and x[n-2] are the input signals at the previous and previous two sampling times, respectively, and y[n-1] and y[n-2] are the output signals at the previous and previous two sampling times, respectively.

5. The ADPLL fractional spurious suppression device based on a spurious notch filter according to claim 1, characterized in that, The coefficient generator includes a CORDIC calculation module configured to perform iterative calculations on the digital angular frequency to obtain the cosine value of the digital angular frequency; the filter coefficients include the cosine value; the coefficient generator also includes a coefficient register configured to latch the calculated filter coefficients. When the frequency parameter changes, the coefficient generator recalculates and updates the filter coefficients in the coefficient register; When the frequency parameters do not change, the CORDIC calculation module stops working or turns off the clock.

6. The ADPLL fractional spurious suppression device based on a spurious notch filter according to claim 1, characterized in that, The digital notch filter is implemented using the Verilog hardware description language and includes an input delay register, an output feedback delay register, a multiplier, an adder, and a coefficient register. The frequency parameters include the reference frequency of the fractional digital phase-locked loop and the fractional part of the fractional division ratio. The target spurious frequency is determined based on the reference frequency and the fractional part of the fractional division ratio.

7. The ADPLL fractional spurious suppression device based on a spurious notch filter according to claim 6, characterized in that, The target spurious frequency includes the fundamental fractional spurious frequency and at least one harmonic frequency; the digital notch filter is a cascaded plurality of filters, each stage of the digital notch filter receiving filter coefficients corresponding to different harmonic orders, so as to form notches at the fundamental fractional spurious frequency and each of the harmonic frequencies respectively.

8. An optimization method for ADPLL fractional spurious suppression based on a spurious notch filter, characterized in that, include: Obtain the frequency parameters related to the operating state of the fractional digital phase-locked loop; Determine the digital angular frequency corresponding to the target spurious frequency based on the frequency parameters; Filter coefficients are generated based on the digital angular frequency; The filter coefficients are provided to a digital notch filter in the digital loop signal path of the fractional digital phase-locked loop, so that the digital notch filter forms a notch at the target spurious frequency and filters the signal in the digital loop signal path. The amplitude response of the digital notch filter at the target spurious frequency is lower than the amplitude response at other non-target frequencies.

9. The ADPLL fractional spurious suppression optimization method based on a spurious notch filter according to claim 8, characterized in that, The digital notch filter is an infinite impulse response filter; the step of generating filter coefficients based on the digital angular frequency includes: Calculate the cosine value of the digital angular frequency; The transfer function coefficients of the infinite impulse response filter are generated based on the cosine value and the preset pole radius parameter. The frequency parameters include the reference frequency of the fractional digital phase-locked loop and the fractional part of the fractional division ratio; the step of determining the digital angular frequency corresponding to the target spurious frequency based on the frequency parameters includes: The target spurious frequency is determined based on the reference frequency and the fractional part of the fractional frequency division ratio. The digital angular frequency is determined based on the target spurious frequency and the sampling frequency or update frequency of the digital notch filter.

10. The ADPLL fractional spurious suppression optimization method based on a spurious notch filter according to claim 8, characterized in that, When the frequency parameters change, the acquisition, determination, and generation steps are re-executed to update the filter coefficients; When the frequency parameters do not change, the current filter coefficients remain unchanged; The target spurious frequency includes the fundamental fractional spurious frequency and at least one harmonic frequency; the method includes: Cascade multiple of the aforementioned digital notch filters; Each stage of the digital notch filter generates filter coefficients corresponding to different harmonic orders, so that each stage of the digital notch filter forms a notch at the fundamental fractional spurious frequency and each of the harmonic frequencies.