Jim Mayrand, Mike Resso, and Dima Smolyansky
Agilent Technologies and TDA Systems
TechOnline

Moving data at 40-Gb/s over copper is no longer a theoretical exercise; it has become a reality. For physical-layer structures—the passive linear components, backplanes, IC packages, cables and connectors—are no longer the bottlenecks in high-speed digital systems. Recent advances in 4-port instrumentation now enable designers to accurately model 40-Gb/s differential structures. Furthermore, this latest in instrumentation development enables users to calibrate the 4-port time domain reflectometry equipment in just 40 minutes. This task would take up to four hours previously, if done at all.

Here is what has brought about the requirement of next generation test and measurement tools: Implementations of PCI Express and InfiniBand are reaching data rates in the neighborhood of 4-Gb/s. But the even newer standards, such as XAUI, OC-192, 10G Ethernet, and OC-768 aim even higher, to 40 Gb/s—and beyond. This upward spiraling trend of digital-data transmission speed creates signal-integrity challenges for physical-layer-device designers in their unending endeavor to keep up.

Figure 1:  Though many methods exist today for model extraction, measurement-based model extraction is a relatively new process yielding insight into high-frequency effects.

As bus and clock speeds have pushed past the gigabit-per-second mark, digital data no longer looks like textbook binary waveforms. It begins to exhibit analog behavior. This means that unwanted phenomena that can hinder data flow in signaling structures—such as reflections from discontinuities, dispersive loss, crosstalk, EMI radiation and susceptibility—must be successfully confronted and conquered.

So it is no surprise that differential-circuit topology is proliferating in design laboratories. The benefits of differential signaling include lower voltage swings, immunity from power supply noise, a reduced dependency on an RF groundplane and improved EMI performance.

But since any differential structure is inherently symmetric across any transverse plane, the challenge in realizing the benefits of such structures relates directly to just how symmetric it really is. Simple impedance and delay measurements of the copper-transmission lines etched on a backplane are no longer sufficient to ensure accurate analysis of gigabit interconnections. Now designers must push design rules to the limit. This requires concurrent time- and frequency-domain analysis.

In this article we introduce methods to achieve proper characterization of differential structures using a Time-Domain-Reflectometer (TDR) oscilloscope and a Vector Network Analyzer (VNA). Measurement accuracy and error correction techniques will be discussed for both time-domain and frequency-domain instrumentation. It will be demonstrated that accurate 4-port frequency dependent models can closely simulate performance of a differential structure—at data rates as high as 40 Gb/s.

Perfectly-designed devices would exhibit beneficial characteristics and would generate no unintended in-phase signals (common-mode signals. Any radiated external signal incident upon this ideal differential transmission line is considered a common mode signal and is rejected by the device. This is the familiar 'common mode rejection' which is the major benefit realized in a differential topology and is usually quantified as the Common Mode Rejection Ratio (CMRR) expressed in decibels.

Non-ideal differential transmission lines, however, do not exhibit these benefits. A differential transmission line with even a small amount of asymmetry will produce a common signal that propagates through the device. The radiated common signals are usually generated by adjacent RF circuitry or by the harmonics of digital clocks. But if properly-designed, differential devices will also reject noise circulating in the electrical ground plane, because the noise is common to both input terminals of the differential pair.

Asymmetry can be caused by any physical feature that is in one line of the differential pair but not the other, including solder pads, jags, bends, and digs. This mode conversion is a source of EMI radiation.

Mode-conversion analysis provides the designer with that insight so that EMI problems can be resolved earlier in the design stage.

To understand the test system laboratory configuration used in this design case study, refer to the flowchart in the frontispiece. Measurement-based, model extraction can be accomplished using a variety of methods.

The goal is to achieve an accurate model that can be simulated in either the time domain or the frequency domain, as depicted at the right side of the illustration. Most digital designers will focus on time-domain models and that will be our focus in this article. Either a topological model or behavioral model can be developed.

• The topological model is based on the physical structure of the device and can be very complex for a lengthy device exhibiting multiple impedance discontinuities. Though requiring multiple iterations, it is easily accomplished with today's standard PC computational power.
  
  
• The behavioral model is a "black box" approach and describes how the device behaves toward a particular stimulus. One type of behavioral model comprises Scattering (S) Parameters.

Both time-domain test equipment such as a Time Domain Reflectometer (TDR) and frequency domain test equipment such as a Vector Network Analyzer (VNA) can be used to measure prototype devices.

In general, the TDR is easier to use whereas the VNA is more accurate. In this experiment, measurements were made with a VNA using the Agilent N1930A Physical Layer Test System (PLTS) software to control the VNA via GPIB. This allowed for use of the automated calibration wizard and simplified this typically rigorous and error-prone process. The resultant, 4-port S-parameter data was exported to the TDA Systems IConnect MeasureXtractor Model Extraction Tool that, in turn, created an accurate time-domain Hspice model.

The TDA Systems IConnect MeasureXtractor was used at the model extraction tool. It was chosen because it is simple and easy to use. This extraction tool imports the impedance profile or 4-port S-parameters after the user performs the measurement with either a TDR or a VNA. The resultant model can be directly linked to a simulator, such as Hspice, Spectraquest and Smart Spice, subsequent to using a laptop to perform multiple iterations of model refinement. (Likewise in the case of a frequency-domain simulation S parameters could be fed to a simulator such as the Agilent Technologies Advanced Design System (ADS).

The convenience of comparing measured results with simulated results very quickly makes this approach an efficient way to check accuracy of the models. The role of the various tools employed in 4-port, differential signal modeling are listed and described in Table 1.

With a full 4-port measurement system, this stimulus/response test is performed on the reflected response and the transmitted response in both single-ended and differential modes. The TDR instrument accomplishes this task with a fast step with little overshoot, in concert with a wideband receiver to measure step response. Whereas the VNA employs a precise sine wave and sweeps a frequency range as the synchronized narrow-band receiver contained within the TDR detects the response.

It is this narrow-band receiver which makes possible the low-noise and high-dynamic range of the VNA. Whether the data-acquisition hardware is time-domain based or frequency-domain based, mixed-mode data is also compiled in a 4-port measurement system. The mixed-mode data refers to two specific test conditions—one being a differential stimulus and common response and the other being a common stimulus and differential response. This analysis discloses any undesirable anomalies due to asymmetry within the differential structure.

In Figure 2 is an overview of the test and measurement methodology used in the characterization of a differential structure. The first step in the process is to understand the sixteen S-parameters, step 1, and what information can be extracted from the sometimes overwhelming amounts of data. These S-parameters are discussed in detail in the section entitled 'Understanding 4-port Mixed Mode Analysis' of Utilizing TDR and VNA Data to Develop 4-port Frequency Dependent Models for Simulation.

Next, in step 2, the actual measurement is made to obtain these sixteen S-parameters. This can be accomplished with a variety of frequency domain instruments such as a VNA—or a time-domain instrument— such as a TDR.

Then, in step 3, the amount of loss in the differential structure is determined by observing the input differential insertion loss (SDD21). This will provide essential information with regard to the bandwidth of the device under test. Carefully analyzing mode conversion, step 4, is also invaluable.

A design can be optimized by determining the magnitude of mode conversion as a percentage of input signal and then locating just where in the physical structure an anomaly may exist that is causing the mode conversion.

Sometimes it is valuable to view reciprocity, as in step 5. By viewing the forward transmission and reverse transmission data, any masking effects caused by the TDR can be removed, thereby clarifying directional behavior.

A comparison of the two measurement-based modeling techniques has been presented that utilize 4-port, S-parameters. These modeling techniques were applied to a novel high-speed, differential interconnect, as described in detail in the reference.

The commonly-used Touchstone format of 4-port, S-parameters (.s4p) resulting from the 40-GHz VNA measurements was directly exported from Agilent's PLTS software and imported into the TDA Systems IConnect model extraction tool. After refining and optimizing the model with multiple iterations within IConnect, a simulated 40-Gb/s eye diagram was obtained. This 40-Gb/s eye diagram was then compared to the 40-Gb/s eye diagram derived from the internal PLTS eye-diagram generating algorithms. The correlation of these two simulated eye diagrams with each other was very good.

In conclusion, it is now possible to use one measurement system for both time and frequency domain information that will quickly identify design flaws that would otherwise degrade performance.