Rising data rates, hardware upgrades drive military test requirements
Higher data rates in military embedded systems can be both a boon and a burden: difficult to achieve and even more difficult to diagnose and debug. However, the capabilities of modern test equipment and associated software make it easier to deal with signal-integrity issues, difficult-to-probe hardware, and complex serial-data analysis.
In every original equipment manufacturer (OEM) market, design requirements for embedded systems call for lower materials costs, faster design and debugging cycles, and quick upgradeability to current technology where possible. Military development programs for platforms such as unmanned aerial vehicles (UAVs) are no exception to this trend, and have been among the widest adopters of open commercial off-the-shelf (COTS) architectures in an effort to drive down costs and design-cycle times.
However, today’s COTS platforms are delivering much higher data rates. Additionally, wireless streaming of high-definition data calls for high-speed processing, as do all types of next-generation military electronics for radar, electronic warfare, and communications.
The increasing data rates provided by the latest generation of open COTS architectures translate into complex signal-integrity issues from the chip level all the way to the system level. In general, military embedded systems are seeing hardware upgrades in the form of faster system processors and backplanes. Taking their cues from network infrastructure suppliers such as Cisco Systems and Juniper, key military contractors such as BAE Systems, Lockheed Martin, and Raytheon have adopted next-generation chipsets with data rates exceeding 10 Gbps.
One methodology for sorting out signal-integrity issues is to employ network analysis. Besides being very costly, traditional vector network analyzers (VNAs) are also notoriously difficult to use. A newer generation of signal-integrity network analyzers sidesteps these issues by connecting directly to the device under test and to PC-based software through a single USB connection.
These instruments, one example being Teledyne LeCroy’s SPARQ signal-integrity network analyzers, quickly measure multiport S-parameters (see Figure 1). Instruments such as the SPARQ analyzers quickly characterize multiport devices for development of measurement-based simulation models, design validation, compliance test, high-performance time-domain reflectometry, and PC board test.
What are S-parameters and why are they important in signal-integrity analysis? When we apply a signal to a network’s input, some fraction of that signal propagates through the network, while some amount reflects – or scatters – back through the port it entered. How much of our signal was reflected, and what caused the reflection?
The answer lies in the application of a mathematical construct called a scattering matrix (or S-matrix), which quantifies how energy propagates through a multiport network. The S-matrix is what allows an accurate description of the properties of incredibly complicated networks as simple “black boxes.” The S-matrix for an N-port network contains N2 coefficients (S-parameters), each one representing a possible input-output path. By “multiport network,” we could be referring to, for example, a cable or a microstrip line. How would that interface affect a 5-Gbps USB 3.0 signal?
S-parameters are a “frequency-domain” description of the electrical behavior of a network. They are complex numbers expressed in terms of both magnitude and phase; this occurs because both the magnitude and phase of the input signal are changed by the network’s losses, reflections, and propagation time.
A time-domain perspective on the performance of a multiport network requires time-domain reflectometry (TDR). Here, one sends a fast rising step edge into the device under test (DUT) and the reflected signal is measured. In addition, we can look at the signal that is transmitted through the DUT; this signal is the time-domain transmitted signal (TDT). The SPARQ instruments are able to perform both types of measurements. TDR measurements are more sensitive to the instantaneous impedance profile of the interconnect, while frequency-domain measurements are more sensitive to the total input impedance looking into the DUT.
A major difference between a signal-integrity network analyzer such as the SPARQ and traditional VNAs is in the calibration process. VNA calibration is a complex task involving multiple connections that can produce misleading results due to operator error. Calibration standards are built into the SPARQ analyzers, eliminating the need for costly external calibration modules. Calibration is accomplished quickly and without user intervention.
Often, OEMs build PC boards intended for military applications a little differently than those for commercial applications. Military-grade products must withstand larger temperature swings and severe vibration levels. The construction techniques used on such boards certainly improve reliability, but in some cases, the result restricts access to desired probing points.
What can test engineers do when they simply cannot physically attach a probe where needed for a measurement? Increasingly, the answer lies in a technique known as virtual probing. Virtual probing is a powerful signal-processing tool that enables the user to measure a signal anywhere within a system and then project a response at any other desired point. This feature works by using S-parameter files of the various components in the system to derive a filter that relates the desired measured signal to the acquired waveform.
For example, measurements can be made where the cleanest signal is available, usually at the transmitter, and the corrupted signal at the far end of the channel (at the end of the backplane) can be simulated, thus eliminating probe and instrument noise from the measurement. The derived filter takes into account all of the interactions among the elements of the system and transmitter signal, including differential to common-mode conversion and near-end and far-end crosstalk. Virtual probing also de-embeds probe and fixture responses from measurements, thereby improving the accuracy of signal-integrity measurements.
Many military embedded systems employ multilane signaling, making serial-link analysis a very complex affair. At today’s data rates, engineers must have access to various forms of signal analysis: eye diagrams, jitter, noise, and crosstalk. Anomalies in any of these parameters can trash a serial stream at the receiver. All of these analyses are difficult and time-consuming to perform without the proper toolset.
Fortunately, modern signal-analysis tools, when paired with a capable signal analyzer or oscilloscope, are able to demystify these analyses and shed a spotlight on signal defects on up to four lanes at once. An example of such tools is found in Teledyne LeCroy’s SDAIII-CompleteLinQ software, which provides a complete set of serial data analysis tools to form eye diagrams and decompose jitter into its component parts (see Figure 2).
Jitter is the term applied to signals that do not arrive at their destination at the proper time (whether early or late). The result of jitter is bit errors due to incorrect latching of registers. While virtually impossible to design a high-speed serial channel that is 100 percent free of bit errors, such a toolset can help determine the source of jitter, noise, and crosstalk.
Teledyne LeCroy 845-425-2000 www.teledynelecroy.com