How to Use Measurement Statistics to Set Up Triggers

Figure 1.  Histogram of the different pulse widths occurring
in a pulse-width modulated rectangular pulse train.

Triggering is an essential element in all modern digital oscilloscopes.  The trigger synchronizes the oscilloscope’s data acquisition with a user-specified event on the signal, be that an edge, threshold crossing or a specific signal characteristic. Teledyne LeCroy Smart Triggers can trigger oscilloscope acquisitions based on properties such as a period, width, low signal amplitude, slew rate or signal loss. These trigger types are ideal for capturing transient events like glitches, but they require knowing at least a range of possible values for the trigger to detect.

Intermittent transient events and glitches are among the most frustrating problems to detect and solve. This is especially true if you have no idea about the nature of the transient. However, you can use the oscilloscope’s measurement tools to help locate these bothersome transients, then use that information to set up your trigger to capture them when they occur. Here’s how.

Consider the pulse train shown in Figure 1. This is a pulse-width modulated signal with a nominal frequency of 2.57 kHz acquired using an Edge trigger and measured using several automatic measurement parameters.

There is obvious pulse width modulation around -1.5 ms and 1.5 ms, but the question is if there are any glitches—excessively narrow pulses—occurring in this signal.  The oscilloscope has a Glitch Smart Trigger, but using it requires knowing the glitch width you want to capture. How do you find that, when you haven't found the glitch?

The answer is to use the measurement statistics. If you acquire the waveform a number of times using an Edge trigger and measure the pulse width with the measurement statistics on, you get the mean, minimum and maximum pulse width measurements, along with the total number of measurements on which those statistics were calculated. Note the Num 4.425e+3 (4,225) for the P3 width(C2) parameter—that is how many pulse widths were measured and calculated into those statistics. 

The statistical information is amplified by the use of the histicon (iconic histogram) beneath the P3 width parameter. On many oscilloscopes, clicking the histicon causes a full histogram to be displayed in another grid.  

The histogram grid has a completely different horizontal scale than the pulse grid, in micro-seconds instead of milliseconds, because it reflects the scale of the pulse widths measured by the P3 parameter, not the time of the acquisition. The bulk of the histogram shows a range of widths running from about 150 𝛍s to a maximum of about 240 𝛍s. That is the normal width modulation range, where the majority of pulses fall. However, it also shows an outlier pulse width occurring near 0 𝛍s. The P3 statistics indicate a minimum value of 166.5 ns, or 0.1665 𝛍s, which would plot around 0 on that grid.

The vertical scaling of the histogram represents the number of measurements within a small range of measured values called a “bin”.  The bin containing the 0 𝛍s glitch width shows a count of about 16 measurements, out of the total 4,225 width measurements calculated. So, the glitch occurs about 0.000038% of the time, or about once every 263 measurements. Therefore, acquiring more than that number of pulses should assure capture of at least one glitch. That tells you how long your acquisition needs to be.

Figure 2: Glitch Smart Trigger trigger setup based on the
measured width values of the waveform.

Once we know the range of glitch values from the width statistics, we can set up the Glitch Smart Trigger as shown in Figure 2.

In this case, since the width of the glitch is known from the measurement, we set the upper limit to a value a bit above the glitch width of 166 ns. A value of 200 ns was chosen. The lower limit is much less critical because the histogram shows only the single glitch width, but a value of 150 ns was used based on the half-bin width of the histogram. 

Once we’ve triggered the oscilloscope using this Glitch trigger, we can verify the trigger setup and investigate the glitch with some additional measurement-based tools, as shown in Figure 3.

Figure 3: Waveform captured with Glitch Smart Trigger
shows glitch at center trigger time. 
The Track plot also confirms glitch location.

The Glitch trigger successfully triggered an acquisition every time the 166 ns width glitch was found. In Figure 3, 25 cycles of the source waveform captured are shown in the upper-left grid, with the glitch occurring near the trigger time right at the center. A zoom of the source in the upper-right grid shows a detail of the glitch. The triangular trigger icon at the bottom of the grid marks the trigger point, which is coincident with the end of the glitch.  

The histogram of the P3 width measurement is shown in the lower-right grid. Note that the height of the bin corresponding to the 0.166 𝛍s glitch shows a count of 30 occurrences. This is equal to the number of trigger occurrences, verified by the number of P1 peak-to-peak amplitude measurements, which are made one per acquisition. So, every acquisition includes a glitch—as it should since that is our triggering event—we are not missing any. Our trigger setup is working.

The fourth trace shown in the lower-left grid is the track of the P3 width measurement.  This is a time-synchronous plot of each width measurement, plotting width versus time.  Note the sharp drop in width at the trigger point corresponding to the location of the glitch.  This trace is time synchronous with the acquired waveform immediately above it, confirming the glitch location within the source trace.

There you have a simple way to determine how to setup a Glitch Smart Trigger based only on measurements of the signal itself.

Watch Steve Murphy's other advice for Getting Your Trigger to Do What You Want in our on-demand webinar.

See also "Take a Coffee Break and Learn...":

How to Layer Measurement Tools