Machine Learning Optimizes FFE Taps for High-Speed Signal Testing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current test and measurement systems, particularly for high-speed signal testing, take significant time to perform complex measurements like Transmitter and Dispersion Eye Closure Quaternary (TDECQ) due to the time-consuming process of optimizing 5-tap feed-forward equalizer (FFE) taps, which hampers production efficiency and increases costs in manufacturing.
Innovation Solution
Employing machine learning techniques, specifically a deep learning network, to optimize FFE taps by converting waveforms into tensor arrays and training the network to predict optimized tap values, thereby reducing the measurement time to less than a second per device under test.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional methods are used to optimize FFE taps for TDECQ measurement, then measurement accuracy is maintained, but measurement time increases significantly
Solution Approach 1:
The patent pre-calculates and stores optimal FFE tap values for various signal conditions in a lookup table during system initialization. When performing TDECQ measurements, the system directly retrieves pre-computed tap values based on measured signal characteristics, eliminating the need for time-consuming real-time optimization calculations while maintaining measurement accuracy.
Solution Approach 2:
The patent creates simplified models or lookup tables that copy the essential characteristics of complex FFE optimization results. Instead of performing full optimization calculations during measurement, the system uses pre-generated tap value sets that replicate the optimal performance, significantly reducing computation time while preserving measurement precision.
2Measurement precision
If FFE tap optimization is performed for each device under test, then measurement accuracy is improved, but production throughput decreases
Solution Approach 1:
The system performs FFE tap optimization in advance and stores the results in lookup tables. During production testing, devices under test benefit from these pre-computed optimal tap values without requiring individual optimization cycles, thereby maintaining measurement accuracy while dramatically increasing production throughput.
Solution Approach 2:
The patent creates universal lookup tables containing FFE tap values that can be applied across multiple device types and test conditions. A single pre-computed set of tap values serves multiple devices and measurement scenarios, eliminating the need for device-specific optimization and enabling high-volume production testing.
3Manufacturing precision
If traditional optimization algorithms are used, then tap values are accurately determined, but test duration increases to seconds per device
Solution Approach 1:
The patent computes optimal FFE tap values during system initialization or offline preparation and stores them in lookup tables. During actual production testing, the system retrieves these pre-determined tap values instantaneously, reducing test duration from seconds to milliseconds while preserving the accuracy achieved through traditional optimization algorithms.
Solution Approach 2:
The patent replaces the mechanical computation process of real-time optimization algorithms with an information-based retrieval system. Instead of executing complex mathematical iterations during testing, the system uses pre-computed data structures and lookup operations, substituting computational mechanics with efficient data access that maintains accuracy while reducing time.
Data Source
AI summary
A test and measurement instrument has an input configured to receive a signal from a device under test, a memory, a user interface to allow the user to input settings for the test and measurement instrument, and one or more processors, the one or more processors configured to execute code that causes the one or more processors to: acquire a waveform representing the signal received from the device under test; generate one or more tensor arrays based on the waveform; apply machine learning to the one or more tensor arrays to produce equalizer tap values; and apply equalization to the waveform using the equalizer tap values to produce an equalized waveform; and perform a measurement on the equalized waveform to produce a value related to a performance requirement for the device under test. A method of testing a device under test includes acquiring a waveform representing a signal received from the device under test, generating one or more tensor arrays based on the waveform, applying machine learning to the one or more tensor arrays to produce equalizer tap values, applying the equalizer taps values to the waveform to produce an equalized waveform, performing a measurement on the equalized waveform to produce a value related to a performance requirement for the device under test.