Digital Predistortion Capture Selection Across Power Ranges
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional digital predistortion (DPD) algorithms for power amplifiers, particularly those using GaN transistors, face challenges in accurately modeling nonlinear distortions due to charge trapping effects, leading to impaired signal quality and increased error vector magnitude (EVM) in communication systems.
Innovation Solution
Implementing power-specific capture selection in DPD methods by establishing boundaries of power ranges in feedback signals, allowing for more representative modeling of signal distortions across the full range of signal magnitudes, including both high and low-power regions, to improve the accuracy of predistortion coefficients.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional DPD algorithms use only high-power captures for modeling, then the model fits high-power regions well, but the model accuracy deteriorates in low-power regions due to charge trapping effects
Solution Approach 1:
The patent segments the power amplifier output power range into multiple power ranges (e.g., first power range, second power range, third power range). For each power range, separate capture sets are identified and used to determine power range-specific predistortion coefficients. This segmentation allows the model to adapt to different operating conditions, particularly addressing charge trapping effects in low-power regions while maintaining accuracy in high-power regions.
Solution Approach 2:
The patent applies local quality by making the predistortion model power-range-dependent. Instead of using a single global model, different predistortion coefficients are determined for different power ranges based on captures from corresponding power ranges. This local optimization ensures that the model accurately represents the power amplifier behavior in each specific operating region, particularly improving low-power region accuracy where charge trapping effects are prominent.
2Device complexity
If DPD model uses captures from limited power ranges, then the model complexity is reduced, but the signal quality and EVM performance deteriorate due to incomplete representation of full power range distortions
Solution Approach 1:
The patent implements a dynamic model adaptation mechanism where the system determines multiple power ranges based on the actual operating conditions and dynamically selects appropriate capture sets for each power range. The predistortion coefficients are updated based on the current power range, making the model adaptive to varying signal conditions. This dynamic approach improves signal quality and EVM performance by ensuring the model always uses relevant captures for the current operating point, while avoiding the complexity of a single monolithic model.
3Ease of manufacture
If DPD uses traditional single-model approach, then the implementation is simpler, but the ability to reduce nonlinear distortions across full power range is insufficient
Solution Approach 1:
The patent segments the predistortion modeling into multiple power range-specific models. Each power range has its own capture set and predistortion coefficients determined from captures within that power range. This segmentation approach systematically addresses nonlinear distortions across the full power range by treating each range separately, ensuring that charge trapping effects in low-power regions and other nonlinearities in high-power regions are all properly compensated.
Solution Approach 2:
The patent employs feedback mechanisms where the system monitors the output power range and uses this information to select the appropriate predistortion coefficients. The feedback loop ensures that the correct power range-specific model is applied based on current operating conditions, enabling effective reduction of nonlinear distortions across the full power range while maintaining implementation feasibility through structured organization of multiple models.
Data Source
AI summary
Digital predistortion methods with power-specific capture selection are disclosed. An example method includes receiving a feedback signal indicative of a power amplifier output and establishing boundaries of multiple ranges of powers in the received signal by analyzing signal statistics in windowed intervals of multiple trial captures. At least one range established in this manner may include the highest value, and at least one other range may include the lowest value of the maximum powers determined for the trial captures. The method further includes updating a power amplifier model based on one or more captures of the feedback signal in each of the K ranges, and using the model to apply digital predistortion to an input signal. By specifically targeting regions of lower power and combining these with high-power captures, the model can be made more representative of the signal as a whole, and signal quality may be improved.