Troubleshoot intermittent signals
Using any of several tools and techniques on a digital oscilloscope, you can locate an elusive waveform, set up a trigger, and capture the event.
By Mike Lauterbach, LeCroy -- Test & Measurement World, 11/1/2008 2:00:00 AM
Engineers often spend considerable time locating and isolating intermittent signals. You may find that a design works most of the time, but an occasional unwanted condition causes the product to fail. You never want a case where you think you’ve solved a problem only to see it arise after many units ship to customers.
Most oscilloscopes have advanced tools that help you locate intermittent problems. You can use them when necessary, but you should start with the easiest method to track down a short pulse, slow rise time, or other unwanted signal characteristic.
The value of persistence
 Figure 1. A color-graded persistence mode lets you see the frequency of a waveform’s occurrence. A signal normally returns to its baseline after the pulse (red), but a secondary pulse occasionally occurs (blue and violet). |
The fastest and easiest way to spot a rare signal shape is through an oscilloscope’s persistence mode. Many instruments use both intensity-gradient and color-gradient persistence. In intensity-gradient persistence mode, an oscilloscope will display a normal signal shape using bright pixels, and it will display rare (intermittent) occurrences with dim pixels. With color-gradient persistence, the most common signal shapes will appear in red. Less frequent events will appear in yellow and green, while rare events will appear in blue and violet. The waveform in Figure 1 normally returns to its baseline following a pulse (red), but on rare occasions, an unwanted secondary, lower-amplitude pulse (blue and violet) occurs a short time later.
Once you find the abnormal signal shape, you should set up a trigger to capture it the next time it occurs. Triggers let you see the abnormal shape more clearly, and they let you check the operation of other devices at the same moment the failure occurs. If the failure occurs once per second, for example, then you should trigger the oscilloscope only when the failure occurs. You won’t waste time looking at normal waveforms or clutter the screen with unnecessary information.
Use your knowledge of the intermittent signal shape to trigger when it occurs. Some of the most useful triggers for capturing intermittents are pulse width (sometimes called “glitch” to denote a very short pulse width), rise time (to spot edges that are too fast or too slow), runt (a piece of the signal with an amplitude that is smaller than normal), and dropout (when a signal that should have steady activity stops having transitions).
Move to the next step
Sometimes, an oscilloscope’s persistence mode displays signal shapes all over the screen, but none that let you see the failure condition. If you can define the “normal” signal shape, then you may be able to set the oscilloscope to trigger only when a different condition occurs. You don’t need to know the failure condition, just how to define a correctly shaped signal.
Suppose you are looking at a clock that shows excessive jitter at a certain frequency. You can tell the oscilloscope to trigger only when the time between two positive or two negative clock edges is outside a specified range. For example, a 100-MHz clock has a 10-ns interval, you might set the oscilloscope to trigger if a clock period is greater than 10.5 ns or less than 9.5 ns, but not trigger on clock signals in between those two values.
 Figure 2. By setting a trigger that activates when a waveform characteristic is outside a range, you can ignore acceptable conditions and capture only error conditions, such as the “staircase.” |
You can set up triggers based on pulse widths, intervals, duty cycles, rise times, fall times, and various other parameters, depending on your oscilloscope. Figure 2 shows an example where the “staircase” in the waveform causes an unacceptable width between two rising or falling edges.
When you can’t find an intermittent shape simply by viewing a waveform, you can get some clues about the error condition by using an oscilloscope’s measurement statistics. With statistics, you don’t need to know the failure condition of your circuit to make a guess at some parameters that could be affected by intermittent operation.
 Figure 3. Triggering on pulse width lets you find narrow pulses, which you can then analyze by changing an oscilloscope’s time-base setting. |
Suppose you’ve captured many triggered waveforms using a simple edge trigger and you use the oscilloscope to measure signal amplitude, rise time, fall time, pulse width, or other parameters. You can set the oscilloscope to display statistics denoting the maximum, minimum, and average values for each parameter. If you’re hunting intermittent waveforms, you should look for minimum or maximum values that are unexpectedly far away from the average.
The table shows the statistics calculated by an oscilloscope on a fairly slow signal with a mean pulse width of 195.773 μs and rise times and fall times of around 40 ns. If you look at the data, you’ll see a minimum pulse width of 164 ns—about three orders of magnitude shorter than normal. Because of that time difference, you won’t see the short pulse while looking at the typical waveform characteristics—the pulse will be compressed due to the oscilloscope’s time-base setting. Based on the statistics, you can set the oscilloscope to trigger on pulses of much shorter width (the time between a rising edge and the next falling edge) than normal and capture each occurrence of the error condition (Figure 3).
You can also use a histogram to get a clue about an error condition. Parameter statistics typically show the most recent value only, plus the average, minimum, maximum, and standard deviation. A histogram will display all the measured values of the parameter. Histograms let you count the rate of bad events, and you can put a cursor on the histogram to read the value of the offending signal shape. Thus, you can set up a trigger for that type of event.
 Figure 4. A histogram of parameter 1 (P1) shows the distribution of amplitudes. Rare events appear as long tails or small, separate peaks. |
Figure 4 shows an aberrant signal that appears in the tails of a histogram of a signal’s amplitude. The long tails indicate that occasionally one amplitude is considerably higher or lower than most others.
Sometimes, you need an oscilloscope to find an extremely rare shape in a captured waveform. Consider the upper trace in Figure 5, where a particular rise time occurs just once on the captured waveform. You can set the oscilloscope to search for signal characteristics that fall inside or outside a certain range. The undesired edge on the signal in the figure may occur because the signal got caught between pull-up and pull-down states, causing the edge to pause in its upward deflection.
 Figure 5. An oscilloscope can find an anomaly in a waveform’s rise time (top, indicated by red arrow) and let you zoom in on it (bottom). |
You can also use an oscilloscope to monitor a live signal, scan the signal for unusual signal shapes, and stop on an occurrence and save the screen image. Just set the oscilloscope to the sort of signal characteristic you would like to capture. The oscilloscope will monitor the live signal until it finds an occurrence. The oscilloscope can then stop with this waveform on the screen or archive it for later viewing and analysis and continue to look for more occurrences.
Figure 6 shows an example where the oscilloscope watches live data for non-monotonic edges (in either direction). When one occurs, the oscilloscope captures the waveform in the top trace, and a zoom shows the details in the bottom trace.
 Figure 6. Oscilloscopes let you monitor live waveforms (top trace) and zoom in on an intermittent failure such as a short glitch in the middle of the rising edge (bottom trace). The middle trace shows an overlay of several acquisitions in which a non-monotonic edge was found. |
Intermittent signal failures have long been a nemesis for design and test engineers. Today’s digital oscilloscopes offer a variety of tools that help engineers locate and troubleshoot the sources of these problems. Some of the tools are quite simple, requiring only a single button push. When the simple methods fail, you can use more advanced tools. Yet, even the advanced tools are fairly simple to set up and often give more insight into circuit behavior.
Table 1
. Statistics calculated for one slow signal reveal
an abnormally short pulse width (shown in blue).
| Measurement |
Width |
Rise time |
Fall time |
| Value |
165.180 µs |
41.092 ns | 38.327 ns |
| Mean |
195.773 µs |
42.441 ns | 38.491 ns |
| Min |
164.561 ns |
39.292 ns | 35.384 ns |
| Max |
239.538 µs |
118.305 ns | 40.761 ns |
| Standard deviation |
27.989 µs |
3.2923 ns | 854.04 ps |
| For Further Reading |
| Rowe, Martin, “DSO Displays: Almost as Good as Analog,” Test & Measurement World, February 2000. |
No related content found.
- 0 rated items found.
Datasheets.com Electronic Parts & Inventory Search
185 million searchable parts
- Part Number
- Description
- Inventory
- Products
- Manufacturers
View All Webcasts
