Analog Filter Bandwidth Tuning With Two-Tone DFT Calibration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing analog filter tuning methods in wireless devices face performance degradation due to bandwidth variations caused by manufacturing and temperature changes, with limited calibration accuracy and long calibration times, and lack dynamic control of the quality factor (Q) for improved tracking performance.
Innovation Solution
A digital tuning system that uses a two-tone calibration signal and discrete Fourier transform (DFT) for precise control of the analog filter's bandwidth, dynamically adjusting the quality factor (Q) and minimizing hardware requirements, allowing for concurrent measurement and faster calibration.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a tracking loop is used to vary R/C filter parameters to compensate for manufacturing variations, then bandwidth tracking is achieved, but calibration time becomes long (approximately 10-msec) and calibration accuracy is limited due to pseudorandom signal variance
Solution Approach 1:
The patent uses a periodic two-tone calibration signal instead of a pseudorandom signal. The two-tone signal consists of sinusoidal tones at frequencies f1 and f2, which are applied periodically to the filter. This periodic action enables faster convergence of the tracking loop because the signal structure is predictable and allows for more efficient correlation-based measurement, reducing calibration time from 10-msec to approximately 1-msec while maintaining high accuracy.
Solution Approach 2:
The patent changes the signal type parameter from pseudorandom to periodic two-tone sinusoidal. This parameter change fundamentally alters the measurement approach, allowing the use of correlation techniques that are much faster than the averaging required for pseudorandom signals. The specific frequencies of the two tones are chosen to be within the filter bandwidth, and their ratio is optimized to maximize the slope of the magnitude response for accurate bandwidth measurement.
2Measurement precision
If in-band tone and band-edge tone are used to tune the filter with separate non-concurrent measurements, then reference measurement is achieved, but additional time is required and correction accuracy is lost due to low slope of magnitude response
Solution Approach 1:
The patent merges the in-band reference tone and the band-edge measurement tone into a single two-tone calibration signal that is applied concurrently to the filter. Both tones are present simultaneously, allowing parallel measurement of the reference level and the filtered response. This eliminates the sequential measurement approach and achieves both reference establishment and bandwidth measurement at the same time, reducing calibration time while improving accuracy through the steep slope of the filter response at the band-edge tone frequency.
3Measurement precision
If dedicated analog circuitry is used to measure R/C time constant, then measurement capability is achieved, but die area and current drain increase due to analog circuit complexity
Solution Approach 1:
The patent replaces the dedicated analog measurement circuitry with a digital signal processing approach. Instead of using analog comparators and timers to measure the R/C time constant, the system uses a digital processor to generate the two-tone calibration signal and to measure the filter response. The bandwidth and time constant are determined through digital correlation and spectral analysis of the filtered calibration signal, eliminating the need for complex analog measurement circuits and reducing die area and power consumption.
4Reliability
If known bandwidth tracking methods are used, then basic tracking is achieved, but BER sensitivity is degraded by 0.5-dB and EVM is degraded from 2% to 8.4% due to large tracking errors of up to 12.5%
Solution Approach 1:
The patent implements a closed-loop feedback system where the measured bandwidth from the two-tone calibration signal is continuously fed back to adjust the filter parameters. The digital processor calculates the actual bandwidth from the ratio of the two tone amplitudes after filtering, compares it to the target bandwidth, and generates correction signals to adjust the R/C time constants of the filter stages. This feedback mechanism eliminates the large tracking errors (12.5%) of known methods, achieving BER sensitivity degradation of less than 0.1-dB and EVM degradation of less than 1%, thereby significantly improving reliability.
Data Source
AI summary
A digital tuning system (250) for changing a cutoff frequency of an analog filter (132) includes digital synthesizers (292 and 294) for producing a two-tone calibration signal (196) applied to an input of the filter after a quality factor of the filter is increased. The filter includes at least one R/C circuit with two resistors (304 and 306) for changing the quality factor and arrays (308 and 310) of capacitors for changing the cutoff frequency. The amplitude of the magnitude responses (409 and 411) of the filter to each tone (405 and 407) is measured by a two discrete Fourier transform single-frequency bin power detection circuits (253 and 254) while the filter is sequenced through a plurality of capacitance settings. An optimal capacitance for the R/C circuit is selected by comparing, to a pre-selected value, a difference between the responses of the filter to each tone, for each capacitance setting.