Agilent Technologies, Inc.

Keysight Technologies, Inc.

Hewlett-packard Company

Steven D Draving

Allen Montijo

David Lee Gines

John C Kerley

Brent Anthony Holcombe

David Eskeldson

Kenneth Rush

Marlin E Viss

Dennis J Weller

Thomas J Zamborelli

Christopher P Duff

Ralph Urban

Min Jie Chong

A method and system measure a characteristic of a signal under test (SUT) using a signal measurement device. The method includes receiving and digitizing the first and second copies of the SUT through first and second input channels to obtain first and second digitized waveforms; repeatedly determining measurement values of the SUT characteristic in the first and second digitized waveforms to obtain first and second measurement values, which are paired in measurement value pairs; multiplying the first and second measurement values in each of the measurement value pairs to obtain measurement products; determining an average value of the measurement products to obtain an MSV of the measured SUT characteristic; and determine a square root of the MSV to obtain an RMS value of the measured SUT characteristic. The RMS value substantially omits variations not in the SUT, which are introduced by only one of the first and second input channels.

Measuring error in signal under test (SUT) using multiple channel measurement device

A method and system are provided for reducing noise in a time domain waveform of a signal under test (SUT). The method includes performing cross-correlation of multiple first complex signals and multiple second complex signals, respectively, from the SUT to provide multiple cross-correlated signals, respectively, where the cross-correlated signals have amplitude components and no phase components from the SUT, and where the first and second complex signals include uncorrelated noise, respectively. The method further includes determining an average of the cross-correlated signals to provide an average cross-correlated signal with reduced uncorrelated noise; obtaining a representative phase component from one of the first complex signals or the second complex signals; and combining the representative phase component with the average cross-correlated signal to provide a representative complex signal corresponding to the SUT with reduced uncorrelated noise.

System and method for reducing error in time domain waveform of a signal under test (SUT)

A method and system measure a characteristic of a signal under test (SUT) using a signal measurement device. The method includes receiving the SUT through first and second input channels; digitizing first and second copies of the SUT to obtain first and second digitized waveforms; repeatedly determining first and second measurement trends to obtain measurement trend pairs; cross-correlating the first and second measurement trends in each measurement trend pair to obtain cross-correlation vectors; extracting zero-displacement values from the cross-correlation vectors, respectively; summing the zero-displacement values to obtain a sum of measurement products for the measurement trend pairs; divide the sum of zero-displacement values by a total number of measurement products to obtain an average value of the measurement products, corresponding to MSV of the measured SUT characteristic; and determining a square root of the average value of the MSV to obtain an RMS value of the measured SUT characteristic.

A method measures a characteristic of a SUT using a signal measurement device having multiple input channels. The method includes digitizing first and second copies of the SUT in first and second input channels to obtain first and second digitized waveforms; repeatedly determining measurement values of the SUT characteristic in the first and second digitized waveforms to obtain first and second measurement values, respectively, each second measurement value being paired with a first measurement value to obtain measurement value pairs; multiplying the first and second measurement values in each of the measurement value pairs to obtain measurement products; determining a mean-squared value (MSV) of the SUT characteristic measurement; and determining a square root of the MSV to obtain a root-mean-squared (RMS) value of the measured SUT characteristic, which substantially omits variations not in the SUT, which are introduced by only one of the first or second input channel.

A measurement instrument and associated method: receive at a first input an aggressor signal (e.g., a supply voltage) from a device under test (DUT); capture samples of the aggressor signal; receive at a second input a victim signal from the DUT, wherein the received victim signal includes jitter induced thereon from the aggressor signal; capture samples of the received victim signal; apply the captured samples of the aggressor signal and the captured samples of the received victim signal to a predetermined model, which represents the jitter induced onto the received victim signal from the aggressor signal, to produce a system of equations; ascertain a plurality of unknown parameters in the predetermined model from the system of equations; and apply the predetermined model with the ascertained parameters to the captured samples of the aggressor signal to estimate the jitter induced onto the received victim signal from the aggressor signal.

System and method of analyzing crosstalk induced jitter

Various illustrative embodiments pertain to a signal quality evaluation system having a decision feedback equalizer (DFE) and a signal quality evaluator. The DFE receives an input signal containing symbols that represent digital data and uses the symbols to generate multiple detection thresholds. Each detection threshold is one of several detection thresholds that can be generated by the DFE by processing one or more symbols present in the input signal prior to a current clock cycle of a clock that is recovered from the input signal. The signal quality evaluator uses the detection thresholds provided by the DFE to detect transitions in the input signal. The signal quality evaluator may execute jitter measurements and/or time interval error (TIE) measurements by evaluating the transitions in the input signal.

Jitter analysis systems and methods

A measurement instrument and associated method: receive at a first input an aggressor signal (e.g., a supply voltage) from a device under test (DUT); capture samples of the received supply voltage; receive at a second input a victim signal from the DUT, wherein the received victim signal includes a crosstalk voltage induced thereon from the supply voltage; capturing samples of the received victim signal; applying the captured samples of the received supply voltage and the received victim signal to a predetermined model which represents crosstalk voltage induced onto the received victim signal from the supply voltage waveform, to produce a system of equations, wherein the model includes a plurality of unknown parameters; ascertain the plurality of unknown parameters in the predetermined model from the system of equations; and apply the predetermined model with the ascertained parameters to the captured samples of the received supply voltage to estimate the crosstalk voltage.

System and method of analyzing crosstalk and intersymbol interference for serial data signals

A measurement instrument and associated method: receive at a measurement instrument at least one victim signal from a device under test (DUT), the victim signal including crosstalk interference from one or more aggressor signals which are not received by the measurement instrument; extract from the victim signal an ideal data pattern for the received victim signal, where the ideal data pattern does not include intersymbol interference (ISI), a noise component, or crosstalk interference to the victim signal; ascertain from the received victim signal and the ideal data pattern the ISI for the victim signal; produce a difference signal as a difference between: (1) the received victim signal; and (2) a sum of the ideal data pattern and the ISI; and ascertain from the difference signal a sum of the noise component and the crosstalk interference from one or more aggressor signals which are not received by the measurement instrument.

System and method of analyzing crosstalk without measuring aggressor signal

A system for filtering a data signal includes an input configured to receive the data signal through a transmission medium and a filter configured to remove distortion from the received data signal using equalization coefficients. The system further includes a processing unit configured to determine dynamically the equalization coefficients of the filter without using a predetermined training pattern in the received data signal.

Method and apparatus for determining equalization coefficients

Equalized Acquisition Record are prepared from Original Acquisition Records reflecting Total Jitter and from an existing description of DDJ. Removal of timing DDJ alters the locations of edges associated with data events. Voltage DDJ adjusts the asserted voltage in the central portion of a Unit Interval. One technique for equalizing timing jitter variably interpolates along the existing Original Acquisition Record to discover plausible new voltage values to assign to existing sample locations along the time axis. Another technique construes the desired amount of correction for each data event as an impulse that is applied to a Finite Impulse Response Filter whose output is a Voltage Correction Waveform having a smoothed voltage excursion. Time variant voltage values output from the Finite Impulse Response Filter are collected into a Voltage Correction Waveform Record having entry times found in the Original Acquisition Record. An entry by entry addition of these two Records produces the Equalized Acquisition Record.

Method for equalizing a digital signal through removal of data dependent jitter

Separation and analysis of measured Total Jitter (TJ) begins with a suitably long arbitrary digital test pattern, from which an Acquisition Record is made. A Time Interval Error (TIE) or Voltage Level Error (VLE) Record is made of the Acquisition Record. A Template defines a collection of associated bit value or transitions that are nearby or otherwise related to a bit location of interest, and has associated therewith a collection of Descriptors and their respective Metrics. Each Descriptor identifies one of the various different patterns of bit value or transitions that fit the Template. The TIE/VLE Record is examined, and a parameter is measured for each instance of each Descriptor for the Template. The collection of measured parameters for each particular Descriptor are combined (e.g., averaging) to produce the Metric for that Descriptor. A Look-Up Table (LUT) addressed by the different possible Descriptors is loaded with the associated discovered Metric, which is a plausible value for Data Dependent Jitter (DDJ) at that bit. DDJ separates from TJ because DDJ is correlated with the Descriptors, while Periodic Jitter (PJ) and Random Jitter (RJ) can be expected to average to near zero over a sufficient number of instances of a given Descriptor. Identified instances of DDJ are individually removed from corresponding locations of TJ found for the entire waveform (the original TIE/VLE Record) to leave an Adjusted TIE/VLE Record that is PJ convolved with RJ.

Method of finding data dependent timing and voltage jitter for different bits in an arbitrary digital signal in accordance with selected surrounding bits

Measurement of jitter in a system uses a digital test sequence including many repetitions of a test pattern. An Acquisition Record is made of the entire test sequence. A complete Time Interval Error (TIE) Record is made of the Acquisition Record. The complete TIE Record is separated into a collection of Component TIE Records, one for each transition in the test pattern, and that collectively contain all the different instances in the test sequence of that transition in the test pattern. An FFT is performed on each component TIE Record, and the component FFTs are processed to obtain timing jitter data for the digital signal.

Method and apparatus using re-sampled tie records to characterize jitter in a digital signal

Discovery of RJ assumes an Adjusted TIE Record for TJ from which DDJ has been removed. What remains is PJ+RJ, from whose Fourier Transform PJ is 'synthetically de-convolved' to leave just RJ: Calculate the Power Density Spectrum of PJ+RJ, and determine a threshold that indicates a PJ component. Identify in the PDS the largest frequency component that exceeds the threshold, otherwise there is no significant PJ and PJ+RJ can be taken as RJ. If a frequency component exceeds the threshold, take the largest and calculate what the convolution of it with the FT of the Transition Pattern would be if this circumstance were to occur in isolation, and then remove it from PJ+RJ. Repeat with continued iterations, until there are no further PJ components.

Finding random jitter in an arbitrary non-repeating data signal

Discovery of DDJ within measured Total Jitter (TJ) begins with a suitably long digital Test Pattern, from which an Acquisition Record is made. A Time Interval Error/Voltage Error Record is made of the Acquisition Record. A Template defines a collection of associated bit value or transitions that are nearby or otherwise related to a bit location of interest, and through the applied test pattern produces a sequence of Data Symbols. The TIE/VLE Record is examined, and a parameter is measured for each Data Symbol as it occurs in the Test Pattern. A regression technique may be use to find coefficients for a DDJ Calculator whose inputs are the Data Symbols and whose output is respective values of DDJ. Subsequent separation of DDJ from TJ is possible because DDJ is correlated with the Data Symbols, while Periodic Jitter (PJ) and Random Jitter (RJ) can be expected to average to near zero over a sufficient number of instances of a given Data Symbol.

Finding data dependent jitter with a DDJ calculator configured by regression

A method and apparatus for analyzing jitter in a signal. The signal is buffered to form a sequence of signal vectors and a time interval error function is calculated for each the signal vector. A time interval error spectrum is then calculated by transforming the time interval error function using a discrete Fourier transform. An average time interval error spectrum and an average power spectral density function are then calculated as averages of the amplitude of the time interval error spectrum and its amplitude squared, respectively. Parameters of the signal jitter are estimated from the average time interval error spectrum and the average power spectral density function.

Growing community of inventors

Colorado Springs, CO, United States of America

Steven D Draving