Why deep memory matters

The role of deep memory in oscilloscope performance

When it comes to oscilloscopes, bandwidth, sample rate, and memory depth are consistently cited as the three most important specifications. Memory depth determines the amount of waveform data that can be captured and stored for analysis. A larger memory depth allows for longer time durations to be captured and preserves more waveform details. This is especially beneficial when analyzing complex or intermittent signals, capturing rare events, or performing in-depth analysis and troubleshooting.

Acquisition memory and memory depth

Acquisition memory, also referred to as memory depth, refers to the number of samples an oscilloscope can store with each acquisition. It is determined by multiplying the sample rate (MSa/s) by the time captured. For example, an acquisition memory of 1 Mpts means that an oscilloscope acquires one million samples in a single acquisition on one channel. Even if multiple channels acquire the same amount of memory simultaneously, the acquisition memory is still 1 Mpts.

Oscilloscopes typically come with a predetermined amount of base acquisition memory as part of the standard purchase. In the past, manufacturers produced different versions of hardware with specific memory capacities. However, in the early 2000s, manufacturers adopted a more cost-effective approach by creating a single hardware platform with the deepest available memory, which could be enabled through software licensing.

This approach allowed users to start with a lower-priced oscilloscope that had less memory and then license additional memory as their needs evolved. It is important to refer to a datasheet to determine whether a specified memory value is the base memory included with the instrument or the maximum value associated with an additional paid option.

The benefits of increased memory are evident, but there’s a catch: more memory means increased processing requirements, which can result in slower overall operation.

Memory depth is not a static value

When purchasing any instrument, you may discover that a few attractive specifications are mutually exclusive. This also applies to oscilloscope manufacturers and their acquisition memory depth specifications. A promoted memory depth may not be available under certain scope settings due to tradeoffs within the oscilloscope architecture.

One common tradeoff, for instance, involves the number of channels. Oscilloscopes have a fixed amount of acquisition memory shared across channels, reference waveforms, and other functions. The maximum memory specification is generally applicable when half of the analog channels are active but decreases by a factor of two for each active channel that shares the same processing and storage path. For example, the memory depth may be 4 Mpts when only channel 1 is active but drops to 2 Mpts per channel when channel 2 is turned on.

Deep memory

The definition of deep memory varies among oscilloscope vendors and has evolved over time. Early digital oscilloscopes had memory measured in hundreds of points, which increased to Kpts in the 1990s. Modern oscilloscopes typically offer memory depths in the tens to hundreds of Mpts. Vendors may claim their instruments have deep memory. This is particularly true for older oscilloscope models that were considered deep when they were new but have low memory depth compared to current competitors who might provide up to 100 times more memory depth.

Deep memory benefits various types of embedded hardware testing, and it is especially useful when capturing long time intervals. Deep memory allows you to isolate an observable problem and trace it back to its source. It also helps in solving complex issues related to electromagnetic interference (EMI) and crosstalk.

Serial buses like I2C, SPI, RS-232, CAN or LIN, which are commonly used for digital designs, can be analyzed more effectively with deep memory. While protocol triggers assist in troubleshooting, visibility across multiple bursts or packets of data often requires capturing a larger time span. However, there are tradeoffs between reducing the sample rate to capture more time (which risks under-sampling the bus) and using segmented memory (which limits analysis capabilities and inter-segment visibility).

In applications that require further analysis, it is crucial to capture as much information as possible and analyze it afterwards. Oscilloscope tools and analysis applications, as well as offloading captured data to MATLAB or Python scripts, can facilitate in-depth analysis and insights.

Memory, sample rate, and bandwidth

Memory depth, sample rate, and bandwidth are closely related specifications. Having more memory allows users to capture a specific amount of time or extend the capture time at a given sample rate:

Memory = (Sample rate) * (Time captured).

With an increased memory depth, users can retain the needed sample rate or even use a faster sample rate for acquiring a combination of slow and fast signals. However, there is a limit to the memory depth when capturing more time by adjusting the timebase to a slower setting. Beyond this limit, the instrument must reduce its sample rate, potentially leading to undersampled signals and aliasing. This leads to invalid measurement results, and users may not realize that the sample rate is insufficient for the rated bandwidth of their oscilloscope.

Oscilloscopes do not provide notifications when the sample rate is inadequate for the rated bandwidth, making it challenging to identify undersampling or aliasing issues. Oscilloscopes with limited sample rate exacerbate this challenge.

More memory allows the user to capture extended time intervals while preserving an adequate sample rate for fast signals. When a user aims to capture a longer time interval, scopes with limited memory will inevitably reach their maximum memory capacity. Consequently, the instrument makes a tradeoff, decreasing the sample rate to accommodate the specified time period. This means that a shallow acquisition memory can result in a sample rate that is too low to correctly capture a signal (see Figure 1).

Figure 1 More memory means the instrument can capture more time without reducing sample rate. Source: Rohde & Schwarz

Most oscilloscopes come with a default memory limit to prevent performance issues when deep memory is enabled. For instance, one vendor may have a default limit of 10 Kpts while another may have a limit of 10 Mpts, even though both scopes offer higher maximum memory capacities.

Users need to manually adjust the scope settings to change the default limit and utilize more memory. This typically involves accessing a manual acquisition setting dialog where they can control parameters like sample rate.

Some oscilloscopes may not allow independent control over sample rate, timebase, and memory depth, leading to a frustrating user experience with limited workarounds. It is generally advisable to choose oscilloscopes that allow the user to independently control all three settings and offer the advantage of capturing off-screen acquisitions.

Segmented memory

Most oscilloscopes offer segmented memory mode, either as a standard feature or as an optional upgrade. This mode is useful for capturing signals with inactive periods, such as serial buses or RF chirps. In segmented memory mode, the oscilloscope saves memory space by capturing only the active parts of the signal. This enables single-shot captures across a longer time duration compared to continuous acquisition.

Segmented mode does not compensate for shallow memory. The amount of memory available per segment equals maximum memory divided by the number of segments. As shown in Figure 2, oscilloscopes with greater memory depth offer more powerful segmented memory, allowing for more segments, longer time per segment or higher overall sample rates.

Figure 2 Segmented memory saves only a capture window around the trigger event for more efficient memory utilization. More memory means more segments, more sample rate, or more time with each segment. Source: Rohde & Schwarz

It is important to note that segmented memory mode has some tradeoffs. It is a single-shot acquisition and does not work well in RUN repetitive mode. In addition, viewing the measurement results involves moving through multiple acquisition screens, and analysis across segments often has limitations.

Serial bus decode and memory

Oscilloscopes equipped with serial bus triggering and decode applications are useful for debugging and testing. However, the correlation between deep memory and the number of captured packets can be difficult to determine. Each protocol requires a certain sample rate for correct decoding, and deep memory with protocol decode can strain the oscilloscope’s processing resources, making it sluggish.

Next generation oscilloscopes, such as the R&S MXO 4, solve this problem with smart architectures. For example, an oscilloscope could have a duplicate sample path for protocol decode, enabling dual-path protocol analysis (see Figure 3). This technique allows a high number of packets to be decoded correctly, even with a slower analog sample rate. It also stores protocol packets as packets, which results in a much smaller packet information database, meaning that the instrument will have a faster update rate and better responsiveness.

Figure 3 R&S MXO 4 Series oscilloscopes have a dual-path protocol analysis. A separate packet decode memory means a more responsive instrument with a deterministic number of maximum packets that can be captured. Source: Rohde & Schwarz

The vital role of memory depth

Memory depth plays a vital role in oscilloscope performance, enabling the capture of longer time intervals and preserving waveform details. It enhances hardware testing and analysis by facilitating the decoding of serial bus protocols and providing a comprehensive view of signals. However, there are tradeoffs to consider, such as increased processing requirements and potential tradeoffs with the number of channels. Therefore, a careful comparison of manufacturer specifications is essential to make an informed decision when selecting the right oscilloscope for your needs.

Joel Woodward is an oscilloscope product planner at Rohde & Schwarz.

<!–
googletag.cmd.push(function() { googletag.display(‘div-gpt-ad-native’); });
–>

The post Why deep memory matters appeared first on EDN.

Become a member

Become a member

Become a subscriber

Become a sponsor