Cross-Spectrum Phase Correction Using GPS Time-Stamped Signals
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for computing cross-spectra from data acquisition units with independent sampling clocks result in phase mismatch issues, leading to inaccurate results, especially in large-scale measurements requiring synchronized clocks across long distances.
Innovation Solution
The system corrects phase mismatch between FFT spectra by using GPS time stamps and established measurement parameters to determine the relative starting points of different data acquisition units, allowing for phase adjustment during cross-spectrum calculations, enabling accurate cross-spectral analysis without the need for physical synchronization cables.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If distributed data acquisition units with independent sampling clocks are used to expand measurement coverage, then measurement area and system scalability are improved, but phase mismatch between channels occurs leading to inaccurate cross-spectrum results
Solution Approach 1:
The system performs preliminary time stamping of data samples using GPS synchronization before cross-spectrum analysis. Each data acquisition unit records the absolute time of each sample, enabling post-processing alignment of channels without requiring physical clock synchronization during data collection
Solution Approach 2:
The invention replaces the mechanical/electrical synchronization system (physical synchronization cables connecting ADC clocks) with a time-stamping system based on GPS time signals and digital record-keeping of sample timestamps, eliminating the need for physical clock distribution infrastructure
2Measurement precision
If physical synchronization cables are used to synchronize ADC sampling clocks, then cross-spectrum measurement accuracy is improved, but system cost and installation complexity increase significantly
Solution Approach 1:
The invention extracts the synchronization function from the physical cable infrastructure and implements it through independent time-stamping at each data acquisition unit using GPS signals, eliminating the need for physical synchronization cables while maintaining measurement accuracy
Solution Approach 2:
GPS time signals serve as an intermediary reference that each data acquisition unit independently uses to timestamp its samples, replacing the direct physical connection (synchronization cable) between master and slave ADC clocks with an indirect time-reference mediation
3Measurement precision
If GPS time stamping and phase correction methods are implemented, then cross-spectrum accuracy is improved without physical sync cables, but computational complexity increases
Solution Approach 1:
The system changes the approach from synchronizing clock frequencies to recording and correcting time offsets. Instead of maintaining identical sampling rates through physical synchronization, the system records the actual sampling time of each channel and applies phase corrections based on measured time differences, transforming a frequency-synchronization problem into a time-offset correction problem
Data Source
AI summary
For cross-channel spectral analysis of measurement data from multiple recording units with independent sampling clocks, a processing method corrects phase mismatch between the data received over the different channels. Blocks of sampled measurement data are buffered in a hardware logic circuit and timestamps are associated with successive blocks through a hardware interrupt to a GPS receiver of each recording unit. For each first channel data block, the block's starting point, a closest point in time in a data block of the second channel, and the starting point of that second channel data block are determined, using GPS timestamps associated with those data blocks, nominal sampling rate and block size. Phase correction based on the time offset between starting points of the pairs of data blocks and the interval between starting points of successive blocks is applied in the frequency domain after a time-to-frequency domain transformation. Multiple frames of phase-corrected spectra may then be averaged. Only a subset of samples in each data block need be used based upon a specified overlap ratio.