The capabilities of mixed signal scopes

The complexity of today's embedded systems designs is growing as engineers increase the use of advanced, high-speed micro-components in circuits to deliver smarter products to users in the computer, communications, defence and semiconductor industries.

Current embedded applications require higher computational power to control the rising number of functions and rich feature sets of the new digital designs, which in turn drives the need for higher clock and internal communication bus speeds.

The oscilloscope remains the test and measurement tool of choice for many embedded systems design engineers. These engineers frequently face situations where they need to simultaneously acquire and measure a combination of signal types and speeds including slow analog, fast digital and baseband RF signals.

However, design engineers encounter numerous problems when working with scopes that have not been optimised to tackle the mixed signals found in embedded designs.

Many engineers who rely on typical two- and four-channel scopes when verifying or debugging mixed analog and digital signals find their scopes to be inadequate when it comes to channel count, memory depth and triggering.

The interactions that occur in typical embedded system designs - isolating specific events in complex systems such as SDRAM and PCI buses as well as tracking a real-world signal such as sound, temperature or pressure through the digital control process - require more viewing capability than current oscilloscopes can provide.

Mixed-signal oscilloscope vs oscilloscope + logic analyser

The ability to test mixed-signal designs can be limited by the number of channels available on a conventional oscilloscope.

Although incorporating a logic analyser in the test process may make up for the limitations associated with the conventional oscilloscope, other limitations such as display update rate and trigger jitter must also be considered. Traditionally, engineers working with mixed-signal content have used oscilloscopes for debugging and characterising analog signals or making signal integrity measurements.

When many digital signals need to be viewed simultaneously, a logic analyser may be chosen. The logic analyser shows multiple channels of logic digital signals simultaneously so that an engineer can obtain a complete and accurate picture of what is really going on in a circuit.

However, a logic analyser shows only an idealised binary representation of the waveform, so the designer cannot see the analog characteristics of signals.

This is a reasonable compromise when the analyser is used to check functional operations that do not require assessment of the waveform details.

The Agilent Infiniium 54830D series mixed-signal oscilloscope (MSO) integrates the oscilloscope with a logic-timing analyser to enable engineers to perform all needed tests on their mixed-signal designs - including real-time waveform monitoring - while giving users familiar, oscilloscope-like controls for both the oscilloscope and logic channels. Both scope and logic channel signals are controlled and displayed on the same instrument over the same time span. The MSO helps engineers narrow-in on their design problems, reducing or eliminating the need for a logic analyser.

MegaZoom deep memory

The oscilloscopes within the Infiniium 54830 series incorporate MegaZoom deep-memory, which addresses the frustrations of current deep-memory users by delivering fast responsiveness and eliminating manual set-ups.

The 54833A and 54833D models all have MegaZoom.

Deep memory is required to find details buried in complex signals, to discover anomalies in the absence of good triggering events, and to correlate high-speed digital control signals with slower analog signals.

R&D engineers in communications, industrial design and automotive applications work with a mixture of analog and digital signals found in embedded systems designs and need deep-memory scopes.

According to Agilent studies, engineers have typically had to 'make do' with the traditional deep-memory technology of existing oscilloscopes.

They resort to edge triggering on the signal, and then autoscaling the signals to fit onto the display. This necessitates increasing the time displayed on the scope to make the entire signal viewable.

Finally, they stop any acquisition activities to zoom in and make measurements or search the data for problems.

Customer frustrations

Often, the first frustration that users experience with the traditional deep-memory oscilloscope occurs as the user zooms in on the data and discovers that the waveform has been under-sampled, resulting in either a biased or incorrect rendering of the signal.

This happens when the default memory on the scope is insufficient and the resulting sample rate used to capture the data has been too slow.

The user must then manually open a menu, decide how much memory is required for an adequate sample rate, set the memory depth and then reacquire the waveform. Reacquiring the waveform often is not possible as it may occur only once.

If and when the waveform is reacquired, users can zoom in on the waveform to search through the data only to discover a second frustration: the display response time slows dramatically.

This typically happens as the user makes control changes on the front panel, making it difficult to search through the captured data.

Understandably, this poor response time has often discouraged the use of deep memory on first-generation deep-memory oscilloscopes.

User frustrations can be attributed to how traditional deep-memory oscilloscopes work. These instruments always set a pre-determined, fixed-memory depth when the scope is autoscaled.

The amount of memory never changes and is preset by the manufacturer to an amount that generally provides the user with a fast display update rate, but doesn't maximise the scope's sample rate.

When the user wants to view a longer time interval, the memory depth available does not change, so as the time interval is increased, the resolution - or sample rate - decreases.

With traditional deep-memory oscilloscopes, users must manually increase the memory depth each time they work with complex signals because the default memory is usually insufficient to support the maximum sample rate required to accurately view and measure the signals.

Also, the resolution is sometimes insufficient to capture fast events, which results in missed events.

A second problem with the traditional deep-memory architecture is the responsiveness of the display with control changes.

Manufacturers set the memory depth to a fixed amount that is usually much less than the maximum available because the display responsiveness would be unacceptably slow when using the deepest memory.

When capturing a complex waveform with the deepest memory selected, these scopes can take up to eight seconds to respond to a control change. This makes zooming into, and searching through, complex waveforms an extremely tedious task.

Consequently, deep memory - a key but expensive feature - is relegated to a special mode and used only when absolutely needed.

MegaZoom offers a solution but it's useful to first consider the traditional deep-memory scope design.

With the older architecture, a single processor (scope CPU) is responsible for many tasks that are often performed sequentially. In most cases, the processor has the primary responsibility of processing the samples from the front-end A/D to their storage in the acquisition memory, drawing the scope waveforms on the display and monitoring front-panel configuration changes.

The entire waveform record is then sent to the CPU, creating a serious bottleneck. This slows scope operation and misses important waveform anomalies due to a slow display update rate and a lot of dead time.

With the MegaZoom, a custom acquisition and display system is connected to every input channel, and is responsible for fast analog-like display updates.

This enables the capture of up to 8 Mpts of data from the A/D at 2 GSa/s, or it can be interleaved to double the sample rate to 4 GSa/s, capturing 16 Mpts of data.

The technology contains a data processing engine for quickly reading the data from acquisition memory and for processing it for display and measurement.

This greatly reduces the quantity of data transferred to the CPU and also reduces any post-processing tasks. The net result is that MegaZoom substantially increases the waveform update rate and front-panel responsiveness of these mid-range 54830 series scopes.

Instant response

The deep-memory architecture in the 54830 series oscilloscopes provides instant response to control changes, even with the deepest memories, while always automatically setting the deep memory for maximum resolution.

Instant response to control changes means that the display updates as quickly with 16 Mpts memory depths as it does with 50 Kpts memory depths.

Users can quickly zoom into and then search through long and complex signals without delays. Deep memory is always on during scope operation, and can be used with no delays.

With automatic deep memory, the maximum resolution also is always available. Users need not stop what they are doing to calculate and manually set the memory.

When they select autoscale, MegaZoom begins to automatically calculate and set the memory depth. It then recalculates and resets the memory required for the maximum sample rate each time an engineer changes the horizontal scale (time/div) control.

When users need to view greater chunks of time, MegaZoom automatically adds memory to keep the maximum sample rate and resolution available.

The technology enables engineers to capture a waveform spanning a long time interval with the best resolution possible.

Therefore, when the user zooms into a signal, the true waveform, accurate measurements, and the capture and display of fast events are visible without using special modes or settings.

The 54833A oscilloscope comes standard with 500 kpts (1 Mpts single channel) of MegaZoom on each channel. The 54833D MSO comes standard with 2 Mpts of deep memory.

Memory options of 4 and 8 Mpts on each channel allow users to extend the time window that can be captured by up to 4 ms, at the maximum resolution of 4 GSa/s.

The 54833D Infiniium oscilloscopes are designed for R&D engineers in a broad range of industries, including those in computer and communications, who need to capture multiple channels of signals and who require deep memory to observe long serial data streams with high resolution, or to find details buried in complex waveforms.

The 54833D mixed-signal oscilloscope is suitable for mixed-signal design engineers using 16 or 32 bit embedded processors with one to two nanosecond signal edge rates.

When engineers decide to buy an oscilloscope, they usually try to match their unique measurement requirements. Infiniium offers a broad range of options that include:

• High-performance active probes that provide 1 GHz system bandwidth when used in conjunction with the 54833A/D 1 GHz models and feature small probe tips for tight places while providing a high dynamic range;
• Universal Serial Bus 2.0 test option, which enables pre-compliance testing of low/full speed USB devices within the scope, and runs certified USB scripts with embedded MatLab;
• Communication mask testing options that take the frustration out of communications testing to ensure that designs conform to industry standards;
• Voice control option that allows hands-free operation;
• Wedge adapters and logic analyser adapters that allow engineers to probe difficult-to-analyse digital signals;
• Differential probing from 200 MHz to 1 GHz;
• A time-correlation fixture for de-skewing, cross-triggering and importing scope waveforms and markers onto the 16700 analyser display.

This latest series of mid-range instruments features a PC-based architecture with the Pentium III 1 GHz processor and 256 MB CPU RAM. This non-proprietary architecture supports standard PC ports and devices and offers remote connectivity, file-sharing capabilities and peripheral support.

Remote operation of test equipment has long been problematic, often requiring dedicated telephone lines and PCs running custom software and remote software packages.

The 54830 series has what is claimed to be the first web-enabled deep-memory oscilloscope. This eliminates the need for special software, as well as any external computer connections to the oscilloscope.

Users can run the Windows XP Pro graphical user interface to assign an IP address to the scope, and then connect the standard 10Base-T/100Base-T interface to their LAN.

In addition to these web-enabled capabilities, the scope can use the Internet with 'email-on-trigger', an action that sends an email with an attached screen shot to the user when a trigger occurs.

An engineer who receives an email from an Infiniium scope can use any web browser from any location, to view and control the scope that's on the laboratory benchtop.