Digital acquisition instruments like oscilloscopes and digitizers are incorporating increasingly large acquisition memories. Acquisition memories with lengths in the gigasample (GS) range are commonly available. The advantage of long-acquisition memories is that they can capture longer-time records. They also support higher sampling rates at any given record duration, providing better time resolution.
The downside of these long records is the time required to analyze them. Most users couple the instrument to a host computer and transfer the data records to the host computer for post-acquisition analysis.
The longer the record, the longer the transfer time and the slower the testing. Many instruments have added tools to allow internal analysis within the instrument, allowing only the results of the analysis to be transferred instead of all the raw data. This can save a great deal of time during testing. This article will investigate the use of several of those analysis tools.
Case study: The startup of an SMPS
Testing the startup of a switched-mode power supply (SMPS) provides an example of a relatively long acquisition. Figure 1 shows an example of a 10-ms acquisition covering the startup of an SMPS.
Figure 1 A 10-ms acquisition showing the startup of an SMPS, including the drain-to-source voltage (channel 1-yellow), drain current (channel 2-red), and gate drive voltage (channel 3-blue) of the FET switch. Source: Art Pini
This record, sampled at 250 megasamples per second (MS/s), has 2.5 million samples per channel. That is a lot of data to render on a screen with a 1920 x 1080 pixel resolution. The oscilloscope acquires and stores all the data, but when more than 1920 samples are being displayed, it compacts the displayed data. Rather than just sparse the signal records, which might cause the loss of significant data points, it detects the significant peaks and valleys and includes those values on the display. This enables users to find significant events within the compacted displays.
Basic measurements
There are three acquired waveforms. The drain-to-source voltage (VDS), drain current (ID), and gate-to-source voltage (VGS) of the primary FET switch. The test will look at the variation in these signals as the power supply controller powers up the supply. Some basic measurements of the signal amplitudes are made and displayed. The peak-to-peak amplitudes of VDS and ID, as well as the amplitude of VGS, are shown as parameters P1, P2, and P3, respectively.
The frequency of the VGS signal and the number of edges contained in the acquisition appear as parameters P4 and P5. Amplitude measurements are taken once per acquisition. Time measurements such as frequency, period, width, and duty cycle are made once per waveform cycle. So, the frequency measurement includes all 1163 cycles acquired. This is an example of “all instance measurements.” This feature ensures that every cycle in the signal is captured in the measurement.
Zoom in on the details
All the acquired data is stored in the instrument’s acquisition memory and can be expanded using zoom traces to see the details, as shown in Figure 2.
Figure 2 Zoom traces provide horizontally or vertically expanded views of the acquired traces, allowing detailed study of the elements of each acquired waveform. Source: Art Pini
In the figure, the zoom traces of the acquired waveforms are horizontally expanded and displayed at 5 ms per division, with a horizontal expansion of 200:1. The zoom traces are taken from the area of the acquired waveform highlighted with higher intensity. The SMPS uses pulse width modulation (PWM) to control its output power.
The zoom traces show the variations in the amplitude and duty cycle of the waveforms just after the gap at 456 ms in the acquired waveforms. The zoom traces are locked together to keep the displayed waveform time synchronous. They can be scrolled horizontally or vertically to show the details in any part of the source waveforms.
Finding desired events in long records
The question of locating areas of interest in these long acquisitions has several answers. Histograms of measured parameters can display the range of values and the number of measurements made by the instrument. A measurement track displays any measurement value versus time. The track can be aligned with the source waveform to show where in the acquisition that value occurs. Some instruments offer scanning functions to map where, in the long record, specific values of a measured parameter occur. These features are extremely useful in analyzing long records.
Histograms
A histogram plots the number of measured values occurring in a small range of measured values (known as a bin) against the nominal measured value. It counts the number of measurements in each bin. Figure 3 shows a histogram measuring the duty cycle of the VGS waveform.
Figure 3 The histogram of the duty cycle measurement of the VGS waveform shows the distribution of measured values with a mean value of 28.4%, a maximum value of 38.9%, and a minimum value of 0.3%. Source: Art Pini
The histogram shows the range of values encountered in a measurement. This example shows that the most commonly occurring value of the duty cycle is 31.6%. This is read from the X@peak parameter (P4). The range of duty cycle values is from 0.3 to 38.9%. This data is based on 1164 measurements shown in the total population measurement (totp – P8). How is the location of the maximum value of the duty cycle found? A measurement track matches measurements to a specific cycle in the acquired waveform.
Measurement track
A measurement or parameter track is a waveform comprised of a series of measured parameter values plotted against time at the same sample rate as the source waveform on which the measurement was made. It is time synchronous with the source waveform.
Figure 4 is an example of a measurement track based on the duty cycle at level (duty@lv) measurement of the VGS waveform.
Figure 4 The trace F1 is the track of duty@lv parameter (P1) values over the entire acquisition. It is time synchronous with the trace of channel 3. Source: Art Pini
The track function, located beneath the source waveform, illustrates how the duty cycle of the gate drive signal changes over time during SMPS startup. After a brief gap, it rises steadily until it reaches a plateau, then drops to a relatively stable value.
The parameter maximum (max-P2) reads the maximum value of the duty cycle as 38.59%. The parameter horizontal location of the maximum (x@max-P3) locates the maximum at 3.12 ms after the trigger (zero time). The parameter markers (blue dashed lines) mark these values on the track display. The center of the zoom traces can be set to 3.12 ms, and the zoom traces are expanded about that point to show the specific cycle of each waveform with the maximum duty cycle.
The VGS voltage appears in zoom trace Z3. The duty cycle at level is read for that specific cycle of the VGS signal in parameter P4, confirming that it is the cycle with the maximum duty cycle value. The track function helps locate specific waveform events within the long record without manually scrolling through the whole waveform to find them.
Tracks can show a variety of characteristics, such as peaks, valleys, periodicity, or rate of change. Periodicity in a track of frequency or phase provides information about frequency or phase modulation, respectively. In this example, the track has a nearly linear slope as the controller adjusts the duty cycle. The rate of change is of interest and can be easily measured, as shown in Figure 5.
Figure 5 Using the relative horizontal cursors to determine the rate of change of the duty cycle of the VGS waveform over the linearly changing portion of the track. Source: Art Pini
The relative horizontal cursors read the slope of the waveform between the cursor lines. This value is displayed and highlighted by an orange box in the waveform annotation box for the math trace F1 as 7.03 k% per second (7% per millisecond).
WaveScan—automatic scan and search of long waveforms
The oscilloscope used in this example has a scan and search engine called WaveScan that can locate unusual events in a single capture or scan for a specific measurement event in multiple acquisitions over a long period. WaveScan has over twenty search modes for analog or digital channel acquisition events. Figure 6 shows an example of a search using WaveScan to find all instances of a duty cycle measurement greater than 38%.
Figure 6 Using WaveScan, an automatic search tool, to search the VGS waveform for duty cycle values greater than 38%. Source: Art Pini
The WaveScan setup dialog box shows the search criteria set up to find duty cycle values greater than 38%. WaveScan can search based on measurements, waveform edges, non-monotonic edges, and serial data patterns. A numeric search, such as those on measurements, can be based on values greater, less, than, within a range, outside of a range, or for the rarest events.
In the example, the search is based on measuring the duty cycle at level with values greater than 38%. The search results are marked with red lines on the source trace and appear in the table in the upper left corner. Each event matching the search criteria is listed in the table in the order of occurrence. The maximum duty cycle value of 38.859%, located previously, is item 3. The table entries are hyperlinked to the Zoom trace, and if one is selected, it will center that event in the Zoom trace. In the example, event six is selected. The zoom trace Z1 has been centered on the sixth cycle with a duty cycle greater than 38%, highlighting its location at 3.2295 ms.
Post-acquisition analysis tools
Modern instruments offer longer acquisition times and include a host of tools to aid in analyzing the data generated. Features such as compaction, zoom, histograms, track, and WaveScan enable various analyses and measurements. The tools also augment the measurements by annotating them numerically or graphically on the display. These features enable local analysis, which accelerates testing and reduces the amount of data that needs to be transferred to external computers.
Arthur Pini is a technical support specialist and electrical engineer with over 50 years of experience in electronics test and measurement.
Related Content
- Why deep memory matters
- Understanding the importance of acquisition memory
- Oscilloscope special acquisition modes
- FFTs and oscilloscopes: A practical guide
- Basic oscilloscope operation
The post Analysis of large data acquisitions appeared first on EDN.
