The case for update rate as a "banner specification"
The current pantheon of banner specifications (the ones that oscilloscope vendors typically use at the top of their advertising) includes bandwidth, sample rate, memory depth, and price. However, update rate is equally important because it characterizes the oscilloscope's ability to capture both intermittent and repetitive events. It doesn't matter how much data the scope acquires—or how fast—if it spends disproportionably more time displaying data than acquiring it. And the speed of the display system doesn't matter if the oscilloscope's triggering circuits are slow to "re-arm" for the next acquisition.

Update rate is significant because it can impact debugging methodology. Consider the glitch mentioned above. If an engineer knows the glitch exists, it's easy to isolate it with a pulse-width trigger. However, it's the glitch that isn't suspected that causes the biggest problems. When a user gets a new circuit board, most "browse" from pin to pin. An oscilloscope with a fast update rate increases chances of finding the glitch during casual inspection—increasing the user's confidence in the board. If one isn't confident in the scope's display capability, the user will have to rely on its triggering system to search for each potential problem on each pin.

Another useful application of a fast update rate is manufacturing test. Many tests require multiple acquisitions on the same test point to increase measurement confidence. Large samples lead to better characterization of margins. Mask tests are an excellent example. A slow update rate forces a trade-off between lower test throughput and lower measurement confidence.

Even though a typical LCD or CRT updates at just 60Hz, one can still benefit from update rates on the order of hundreds of thousands of waveforms per second. All data is displayed, but each acquisition is overlaid using "persistence" algorithms which use color or intensity gradation to show frequency of occurrence. It's like looking at a histogram of the traces from above.

Characterizing update rate
Update rate is a dynamic characteristic. It is relevant to repetitive acquisitions, not "single-shot" measurements. And, it can vary with the scope's timebase settings, modes of operation, and its architecture.

Update rate is constrained by the "dead time" between acquisitions. The oscilloscope is blind to any events that occur during the dead time (Figure 1). There are several sources of dead time. The most important is the time it takes to display data in the acquisition memory.

At a fundamental level, some scopes simply have a faster data connection between acquisition and display than others. Architecture also matters. Some designs empty all data to the display before refilling it. Another approach is to "ping-pong" data from different acquisition memory banks to the display. A third technique is to queue data to the display. Other factors include the amount of memory processed and delay inherent in the triggering system.

Many oscilloscopes have special modes that can accelerate the update rate. They accomplish this by reducing the memory depth or bypassing most triggering circuitry. As a rule, these special modes require performance trade-offs (such as sample rate reductions or inability to execute even simple triggers) and should be used carefully.

If a scope has an external trigger and offers a frequency counter measurement, such as the Agilent DSO5054A, this is a simple experiment to perform. Use a 50Ω BNC cable to connect the external trigger to Channel 1 of the oscilloscope. Set the scope to auto trigger. The frequency counter measurement will count the number of triggers per second, which is a close approximation of the maximum update rate. If the scope does not offer an external trigger, the user can substitute a high-frequency source as the input.

Duty cycle
The concepts of update rate and dead time can be easier to visualize when placed in the context of duty cycle. For an oscilloscope, duty cycle is the percentage of time the scope is acquiring data. The more time spent acquiring data, the better chances are of seeing intermittent events.

If the oscilloscope's update rate is known, it is easy to calculate the update rate. We start with a more traditional definition of duty cycle:

Since an oscilloscope is a time-domain instrument that displays known periods of time, we can simplify this calculation.

For example, an oscilloscope is set at 2μs/division, and has a 4 GSample/s sample rate. The measured update rate is 19,300 waveforms per second.

Therefore, at this setting the oscilloscope is acquiring data 38.6% of the time.

Implications for your debugging methodology
Figure 2 shows how duty cycle changes with different oscilloscopes at different t/div settings. You'll notice two trends:

2. Duty cycle curves of real-world oscilloscope models.

1. Duty cycle can vary as much as 2.5 orders of magnitude between different oscilloscopes. If you have a 10-MHz microprocessor that makes a bad write to memory every 1 million cycles, the 30% duty cycle scope will display this error about 3 times a second. The 0.3% duty cycle scope will display it every 33 seconds. Implication: If the scope has a slower update rate, the user should rely upon triggers to find these intermittent problems. Casual browsing will not be sufficient.
2. Duty cycle increases as time base (sweep) slows. The amount of data in each acquisition increases faster than the update rate decreases. As scopes enter "roll mode" the duty cycle becomes effectively 100%. Implication: Slower sweep speeds increase your ability to view intermittent anomalies with all scope models.

These techniques assume rare or infrequent events. If looking for "one-time" events, hardware triggers remain the best option. However, for basic system characterization and debugging, a fast update rate allows greater insight into a system's behavior.

While this article focuses on update rate, but it's the interplay between update rate, sample rate and memory depth that determines the effectiveness of an oscilloscope and how confident the user will be in its measurements (Figure 3). When selecting an oscilloscope, users should be mindful of these three specifications, and how they fit the measurements that will be made.

About the Author
Phil Stearns is the Value Oscilloscopes Product Manager, Digital Validation Division, Electronic Measurements Group, Agilent Technologies.

Phil holds a BS in Electrical Engineering from the University of Cincinnati, and an MBA from Case Western Reserve University. Phil joined Hewlett Packard/ Agilent in 1996 and has held various positions marketing oscilloscopes and logic analyzers. Phil has also worked outside of Agilent in R&D, marketing, and consulting roles.

Phil is currently a product manager overseeing the value oscilloscope product lines the Design Validation Division based in Colorado Springs, Colorado. Outside of work, Phil tries to reverse the entropy caused by his three sons and coaches a youth track and cross country team.

Related Links
Choosing an Oscilloscope with the Right Bandwidth

Improve analog/mixed-signal simulator analysis using real-world, hardware-generated data