Information Extraction Using Power Spectrum Imaging

Christiansen, Joel Timothy, Cole, Edward I. · 2026

Read Information Extraction Using Power Spectrum Imaging by Christiansen, Joel Timothy, Cole, Edward I. (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/doe-osti-3943084-information-extraction-using-power-spect.txt.

Introduction

Darlene M. Udoni, Joel T. Christiansen, and Edward I. Cole of Sandia National Laboratories report a test method that reads the health of electronic parts from their power noise. The report, SAND2026-26549R, covers power spectrum imaging. The printed byline leads with Udoni, a name the catalog record leaves out.

The team drove test diodes with square waves from 5 kilohertz to 10 megahertz at just under one volt, then recorded how each part's power spectrum changed. Stacking scans builds a two-dimensional image, with frequency along one axis and part variation along the other. Small shifts in the pattern can flag a weak or altered part without opening it. Failure analysts get a way to pull useful facts from parts that still sit on the board, a help when hunting faults in complex gear.

Catalog record

CREATORChristiansen, Joel Timothy, Cole, Edward I.
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE24 KB
LICENSEPublic Domain
RIGHTSPUBLIC DOMAIN
SOURCEDOE OSTI

Source: https://www.osti.gov/biblio/3943084

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

SAND2026-26549R SANDIA REPORT SAND20XX-XXXX Printed Click to enter a date Information Extraction Using Power Spectrum Imaging Darlene M. Udoni, Joel T. Christiansen, Edward I. Cole Prepared by Sandia National Laboratories Albuquerque, New Mexico 87185 and Livermore, California 94550 CONTENTS Abstract ............................................................................................................................................................... 3 Acknowledgements............................................................................................................................................ 4 Executive Summary ........................................................................................................................................... 7 Acronyms and Terms ........................................................................................................................................ 8 1. Introduction ................................................................................................................................................. 9 2. Setup and approach .................................................................................................................................. 10 3. Experiments and tests .............................................................................................................................. 11 4. Results ......................................................................................................................................................... 17 Conclusion ........................................................................................................................................................ 18 References ......................................................................................................................................................... 19 Distribution....................................................................................................................................................... 20 LIST OF FIGURES Figure 1. Block Diagram .................................................................................................................................10 Figure 2. Zener Diode Fixture .......................................................................................................................11 Figure 3a. Single Line Spectra ........................................................................................................................12 Figure 3b. PSI of Discrete Device.................................................................................................................12 Figure 4a. PSI Indicating Fonset ...................................................................................................................13 Figure 4b. PSI Collected on Microcontrollers .............................................................................................13 Figure 5a. Raw PSI Collected on U.S. Gov ASIC.......................................................................................14 Figure 5b. Difference PSI Collected on U.S.Gov ASIC ............................................................................14 Figure 5c. Difference PSI Collected Showing Line Scan Ray 9 and 10 ...................................................15 Figure 6ab. Line Scan Plot Showing all 48 U.S Gov ASICs ......................................................................15 Figure 7a. Ray 9 Scan of U.S. Gov ASICs ...................................................................................................16 LIST OF TABLES Table 1. Six Conditions .............................................................................. Error! Bookmark not defined. 5 This page left blank 6 EXECUTIVE SUMMARY Power Spectrum Analysis (PSA) detects differences between components by measuring the frequency response of power fluctuations in biased devices subjected to dynamic stimuli [1]. PSA signatures associated with individual devices under test are found to be sensitive to subtle changes between devices. This method has been effective in detecting subtle differences in electrical components, yielding critical insights for applications in supply chain management, defect detection, failure analysis, and aging studies [2]. Current implementation of PSA relies heavily on the experience of Subject Matter Experts (SMEs) to determine the stimulus conditions to collect data. Different analysis methods have been developed to extract information from PSA signatures [3]. The key to successful PSA implementation is to select and optimize the dynamic stimuli for the spectral results obtained from the device under test. To readily acquire and analyze a wider range of input conditions, a new method has been developed to generate an image of multiple spectra, Power Spectrum Imaging (PSI). PSI displays a range of stimulus frequencies instead of a specific frequency. The developed software generates a PSI with user defined pixel density from line scan spectra. Parameters are customizable enabling applications for specific devices allowing for the selection of an input voltage, frequency, resolution bandwidth, step size and frequency span. PSI is then used for visualizing and analyzing multiple device frequency responses in a single image. The resulting PSI data has some similarities to spectrograms used in acoustics [4]. 7 ACRONYMS AND TERMS Acronym/Term Definition ASIC Application-Specific Integrated Circuit PSA/PSI Power Spectrum Analysis/Power Spectrum Imaging SME Subject Matter Expert DUT Device Under Test USG United States Government 8 1. INTRODUCTION 1.1 Power Spectrum Analysis (PSA) Basics Power Spectrum Analysis (PSA) is a technique that detects differences between electronic components by measuring the frequency response of power fluctuations in biased devices subjected to dynamic stimuli. PSA has proven effective in identifying subtle variations in electrical components, providing critical insights for applications such as supply chain management, counterfeit detection, defect identification, failure analysis, and aging studies. 1.2 Motivation and Purpose Current PSA implementations rely heavily on the expertise of Subject Matter Experts (SMEs) to select stimulus conditions and interpret results. The success of PSA depends on optimizing dynamic stimuli and data collection patterns that influence the device under test (DUT). To address these challenges, we developed Power Spectrum Imaging (PSI), a method that generates images from multiple PSA spectra, enabling comprehensive visualization and comparison of stimulus conditions. 9 2. SETUP AND APPROACH 2.1 Distinguishing PSI from PSA PSI extends traditional PSA by allowing customizable parameters including input voltage, frequency, resolution bandwidth, step size, and frequency span. Unlike PSA, which typically uses a single stimulus frequency, PSI employs a range of input frequencies to generate a greyscale image from line scan spectra. This approach provides an enhanced data set for analysis and visualization. 2.2 Experimental Setup A block diagram of the experimental setup is shown in Figure 1. Key equipment includes: • Signal Generator: Keysight N5183B MXG Microwave Analog Signal Generator • Spectrum Analyzer: Rohde & Schwarz FSW Signal and Spectrum Analyzer • Biasing Circuit: Custom programmable power supply for device biasing • Data Acquisition: National Instruments PXIe-5162 Digitizer Figure 1. Block Diagram of Setup 2.3 Components Tested • Discrete Zener diodes • Microcontrollers of various memories • U.S. Government ASIC Devices 10 3. EXPERIMENTS AND TESTS 3.1 Discrete Components: Zener Diodes Initial proof-of-concept experiments were centered on developing the best parameter settings. Typically, when performing PSA, the SME will bias the sample in forward bias through power and ground configuration. But since Zener Diodes specialize in reverse bias, the data was collected in reverse bias with an applied square wave. A total of 42 Zener Diodes were selected from two different manufacturers-32 from manufacturer A and 10 from manufacturer B. A total of six conditions were carefully selected by the SME which comprised of a frequency range of low to high (1kHz to 100kHz) values. These frequencies were selected to encompass the typical operating range of most devices. The frequency input and span varied while keeping other parameters constant for direct comparison. The parameters that remain constant are the number of peaks and the resolution bandwidth (RBW). Table 1. Diodes were placed into a fixture (Figure 2) and data was collected using existing software, PSA Ultimate (LabView based) as well as newly developed LabControl (C++ based). The new software is fully customizable so the user can adjust input voltages, frequencies, RBW, step size and frequency span. Conditio Signal Signal Analyzer Span # of Peaks RBW n Generator (resolution Input band width) C1 1kHz 1kHz-200kHz 200 500Hz C2 5kHz 5kHz-1MHz 200 500Hz C3 10kHz 10kHz-2MHz 200 500Hz C4 20kHz 20kHz-4MHz 200 500Hz C5 50kHz 50kHz-10MHz 200 500Hz C6 100kHz 100kHz-20MHz 200 500Hz Table 2. Six Conditions Figure 2. Zener Diode Fixture 11 Our experiments were performed on both Zener Diodes and U.S. Government ASIC devices. But during the first few months of conducting experiments, it was determined this approach was not the best way to go about collecting data. Our results indicated no single stimulus was universally optimal. This motivated the use of multiple stimuli across a frequency range to collect a 2D grayscale image. Spectral data in a line scan in the x-axis and the equipment variation in the y-axis. Again, utilizing discrete Zener diodes and a range of stimulus frequencies, PSA spectra (normalized to a standard) were acquired at multiple frequencies spanning a range of stimulus frequency values (5kHz to 10MHz) and an input voltage of 0.95V with a square wave applied between VDD and VSS. The individual spectra were then displayed as intensity lines in an image. The resulting images produced both the differences in normalized PSA spectra between devices but also how those differences change with applied frequency. Figure 3a is an example of a single normalized line spectra collected on discrete devices and Figure 3b is the PSI taken. Note the bright to dark phase shifts along the ‘rays’ (lines from upper right to lower left corner in the image). These were not observed in earlier

Read the complete text