Divide and conquer signal anomalies
An oscilloscope’s segmented acquisition memory lets you isolate signal anomalies and increase screen update rate.
By Art Pini, LeCroy -- Test & Measurement World, 10/1/2009 2:00:00 AM
| Read more articles from our October 2009 issue. |
Real-time oscilloscopes digitize signals and store them in memory before processing and displaying waveforms. Knowing how to most effectively allocate that memory can help you get to the root cause of signal problems more quickly.
Under normal operation, an oscilloscope treats its entire acquisition memory as a single block, and it alternates between acquiring and displaying data. With segmented memory, the oscilloscope divides its acquisition memory into equal portions. Segmented memory reduces dead time between acquisitions because the oscilloscope does not update the screen until the entire acquisition memory is full. Depending on the oscilloscope model, dead time between acquisitions can be as little as 800 ns as opposed to tens of milliseconds in normal mode.
Segmented memory also lets you acquire only a waveform’s parts of interest. In waveforms that have significant idle time, you can trigger an acquisition to coincide with a characteristic of interest. Thus, the memory won’t fill with unwanted samples. Because the oscilloscope creates shorter records than when it uses normal memory, the effective acquisition sample rate increases, resulting in improved time resolution in the measurement.
Another advantage of segmented memory is that it time stamps each trigger event, showing the time of trigger, time between triggers, and time since the first trigger. If you trigger on an anomaly such as a runt pulse in a waveform, the trigger times will tell you the approximate rate at which the anomaly occurs—a potent diagnostic tool.
How segmented memory works
 |
Figure 1 shows how an oscilloscope divides its acquisition memory into segments, each storing a unique acquisition. You can set up a trigger on a characteristic such as pulse width, and each time the trigger conditions are satisfied, the oscilloscope will capture and store data into a new segment. In this example, the oscilloscope acquired 10 segments from a waveform.
Figure 2 shows an application of how segmented memory can help you analyze an anomaly. The waveform, captured using normal memory mode, consists of multiple pulses about 200 ns wide separated in time by about 1.5 µs. The waveform contains intermittent low-level pulses that you need to isolate and analyze. Because the pulses are about 200 ns wide, you can use a time-base setting of 100 ns/div when using segmented memory. That provides 1 µs across 10 divisions, which is enough to let you see the wanted and unwanted pulses.
 |
The number of samples per segment depends on the oscilloscope’s memory capacity and the number of segments you specify. For a given size of acquisition memory, the greater the number of segments, the fewer samples in each segment.
Figure 3 shows a setup for capturing the anomalies in the waveform from Figure 1. In this example, I selected time/division and the number of segments. The oscilloscope automatically allocated acquisition memory. This particular oscilloscope can store 400 ksamples of waveform data. Because I selected 100 ns/division and 80 segments, the oscilloscope will allocate 5 ksamples/segment (400 ksamples/80 segments). At a sample rate of 5 Gsamples/s, 5 ksamples spans 1 µs. That lets you see the wanted and unwanted pulses.
 |
Figure 4 shows the time stamps for acquisitions triggered by the presence of a runt pulse after the main pulse. Note that the time stamps, which include the absolute trigger times as well as time since the first trigger and time between triggers, indicate that these pulses occur over a period of 20 µs.
The 20-µs period provides insight into the nature of the condition, namely that these anomalous pulses occur at a 50-kHz rate. That rate could be the power-supply switching frequency or the display update rate of the unit under test. Knowing the rate of occurrence of the anomaly lets you link it to other system operations.
 |
You can also zoom in on a memory segment to view a waveform anomaly in greater detail. Turning on a zoom display automatically zooms to a single segment. In Figure 4, the zoom trace Z2 shows the 49th segment as an isolated trace, as shown by the number 49 in the zoom trace (Z2) annotation box. By using the horizontal zoom control, you can scroll through the entire sequence acquisition one segment at a time.
If you turn on display persistence, the oscilloscope automatically overlays all the segments on the display. That lets you see a waveform history using either intensity or color-mapped persistence.
 |
Using persistence is but one way to view data in segmented memory. In addition to the adjacent display mode shown in Figures 2, 3, and 4, there are other display types: overlay, waterfall, perspective, and mosaic. Figure 5 shows these.
The overlay display overlaps all the segments and allows the user to compare them. Waterfall displaces each segment trace by a small vertical offset, while the perspective display shifts them with both a vertical and horizontal offset. These two views let you see changes in the data over time. The final display mode is mosaic. Here, up to 80 traces are displayed individually. This, like a slide sorter, lets you search for particular segment features.
 |
You can also apply mathematical processing to data in segmented memory. Figure 6 shows the source sequence mode trace (top trace), the zoom of the 45th segment, the average of all 80 segments, and the envelope (minimum and maximum values) of all the segments. The oscilloscope is intelligent enough to automatically handle data as a series of waveforms. If you choose average or envelope, the oscilloscope will add all the segments before computing those parameters.
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
