Serial Data Link Analysis Visualizer (SDLA Visualizer) Datasheet

Features & Benefits

• Complete Measurement Circuit De-embed Environment
  • Remove reflections, cross-coupling, and loss caused by fixtures, cables, probes allowing visibility of the true circuit behavior
• Simulation Circuit Embed Tools
  • Embed user defined channel models to simulate the signal at the end of the link without needing physical channels
• Receiver Equalization Flexibility
  • Open closed eyes caused by channel effects using receiver equalization techniques, such as Continuous Time Linear Equalizer (CTLE), Decision Feedback ( DFE), or Feed Forward (FFE) Equalization
  • Model silicon-specific receiver equalization using IBIS-AMI models to observe on-chip behavior
• Advanced Analysis and Modeling Capabilities
  • Flexible modeling including cascading of 4-port S-parameters (differential and single-ended), 2-port S-parameters, RLC Models, loss-less transmission lines, probes, or FIR filters
  • "What-If" analysis to evaluate the optimal transmitter settings using flexible insertion/removal of transmitter equalization parameters
  • Twelve test points allow simulated probing where physical probing is not practical
  • View the common mode, differential, or individual inputs of the signal
  • Simultaneously view and measure time-aligned waveforms at multiple test points using integrated DPOJET Jitter and Eye Analysis (opt. DJA) support
  • Comprehensive frequency and time domain plots, enable quick verification of S-parameters and test point transfer functions

Applications

• De-embedding the effects of cables, fixtures, and probes for silicon validation
• Characterization of backplane and embedded system performance using IBIS-AMI CTLE, DFE, or FFE receiver equalization
• Link budgeting and What-if Analysis with emulation of a range of channels
• Evaluate Transmitter Equalization (De-emphasis/Pre-emphasis)
• Reflection removal caused by impedance mismatch for high speed serial buses and DDR
• Characterize the performance for multi-gigabit standards such as SATA 6 Gb/s, USB 3.0, SAS 12 Gb/s, PCIe Gen 3, OIF CEI, XFP, IEEE 802.3 Ethernet

Serial Data Link Visualizer Overview

Acceleration of signaling speeds and shrinking geometries create a number of challenges for next generation multi-gigabit designs and test methodologies. Designs are evolving to address these challenges with advanced equalization techniques at the transmitter and receiver. Smaller form factors make signal access more difficult resulting in non-ideal probing points. This can lead to loss and reflections on the acquired signal due to impedance discontinuities that would not be present at the ideal measurement location.

The advanced techniques employed by the designs call for advanced measurement solutions. The challenge begins with the signal acquisition: capturing a signal through cables, probes and fixtures distort the signal shape; SDLA Visualizer allows you to de-embed the effects (reflections, insertion loss, and cross coupling) of the measurement circuit (cables, probes, and fixtures) from the waveform while taking into account the transmitter output and receiver input impedance. De-embedding these effects improves the accuracy of measurements and can make the difference between passing or failing a test.

After de-embedding the measurement circuit, what-if scenarios can be run by embedding a simulation circuit. This could consist of a simple ideal 50 ohm receiver load for transmitter characterization or could be a simulated channel that represents a backplane.

In many situations, after embedding the simulation circuit the effects of the channel loss result in a closed eye. Many analysis applications provide reference receiver equalizers; however, a limitation with this approach is that the algorithms typically do not match what is happening in real silicon. IBIS-AMI meets this need by allowing silicon vendors to model the actual receiver implementation in a way that does not expose intellectual property. SDLA Visualizer’s equalization features can emulate receiver equalization by applying an IBIS-AMI model, CTLE, FFE, or DFE equalization.

In other cases, the signal under test must be accessed with a high impedance probe, where the probe point is located at a position which results in capturing reflections due to impedance discontinuities that would not be present at the test point. These reflections can make it difficult or impossible to measure the system under test. A method to allow simulated probing while removing reflections is needed. Using SDLA, the reflections and any loading caused by the probe can be removed by modeling the system under test and the measurement circuit. This can be done by using S-parameters of the system or by using RLC and transmission line models.

Another challenge results from the transmitter signal shape. The signal is no longer a simple NRZ square-wave pattern. Designers alleviate high-frequency loss in the media with transmitter equalization features; that is, with pre-emphasis or de-emphasis of the transmitter waveform. Correspondingly, transmitter signals today need to be evaluated for this transmitter equalization; the SDLA software allows for complete transmitter equalization insertion or removal.

De-Embedding the Measurement Circuit

SDLA provides the ability to completely remove the insertion loss, reflections, and cross coupling effects caused by the measurement circuit while taking into account the characteristics of the system under test. The transmitter impedance is modeled by either a nominal impedance or two 1-port or one 2-port S-parameter model. The De-Embed block within SDLA allows for the definition of up to 8 block elements that describe the measurement circuit, for example fixtures, cables, and oscilloscope.

Each block can be described as one of the following:

• 4-port S-Parameter
• Two 2-port S-Parameters
• Transfer function
• FIR filter
• RLC model
• Lossless transmission line model
• High impedance probe

Within the De-Embed block there are two test points which allow for visibility within the cascaded network while viewing the signal with the loading of the measurement circuit.

When using high impedance probes, a 3-port probe model is necessary to accurately represent the interactions of the probe with the transmitter, receiver, and the oscilloscope. SDLA ships with 3-port probe models for many of the P75xx and P73xx series probes and input impedance models of the DPO/DSA/MSO70K series ocillsocopes.

SDLA De-Embed block with High Impedance Probe De-Embed enabled

Bandwidth Control

When de-embedding, it is critical to pay close attention to the signal to noise ratio of the signal at higher frequencies. While SDLA can automatically determine the bandwidth limit for the de-embed filter, it is often necessary to fine tune the filter. Some solutions are limited to a simple bandwidth limit control; however, in some cases this may not be sufficient. For this reason, SDLA Visualizer provides the ability to control the filter rolloff by adjusting the stopband in frequency and in dB. The examples above show how by using the fine tuning capability of SDLA, unwanted artifacts due to a simplified de-embed implementation are not present in the resulting signal.

SDLA Bandwidth and Stop Band Control

The eye diagrams above show the results of using the default stopband settings verus fine tuning. Notice that the preshoot on the signal has been reduced, eliminating the upper mask violation.

Mask hit violations on a PCIe 5 Gb/s signal without bandwidth fine tuning


After fine tuning the filter stopband, the mask hits have been eliminated

Embedding the Simulation Circuit

Once the effects of the measurement circuit have been removed, the simulation circuit can be defined in order to observe the signal without the effects of the measurement circuit. This is easily done in SDLA by defining the transmitter output impedance, which can be defined as a nominal value, two 1-port or one 2-port S-parameter model. The Embed block can then be configured to represent a simple load or a series of cascaded elements such as transmission lines and connectors. Like the De-Embed block, the Embed block allows for the definition of up to 8 cascaded block elements.

Each block can be described as one of the following:

• 4-port S-Parameter
• Two 2-port S-Parameters
• Transfer function
• FIR filter
• RLC model
• Lossless transmission line model
• High impedance probe

Applying Equalization

Whether using a simulated circuit or capturing the signal at the far end of the link it is often necessary to apply receiver equalization to open the eye before measurements can be performed. In most cases the lack of physical access makes it impossible to verify the receiver circuit behavior and monitor the effects of clock recovery and equalization. SDLA Visualizer provides a rich set of capabilities for RX equalization, including IBIS-AMI support, CTLE, DFE and FFE. Using the built in support for IBIS-AMI provides the advantage of using silicon specific equalization algorithms that provide the greatest level of correlation between what the receiver actually sees and the simulated waveform. When IBIS-AMI models are not available, SDLA Visualizer can be used to perform CTLE, DFE and FFE.

SDLA Visualizer IBIS AMI Receiver Equalizer Configuration

Complete Link Visibility

Within SDLA, twelve test points are provided with four moveable test points in the De-Embed and Embed paths, allowing the maximum flexibility to simulate probing the signal at any point within the system. Test Points in the De-Embed path show the signal with all the loading that is included in the measurement circuit, allowing visibility of the signal with the effects of loading from cables and probes. To visualize the signal without the effects of the measurement circuit, any of the test points along the simulation circuit can be observed. When computing the test points, SDLA automatically removes the delay so the embedded or de-embedded waveforms are time aligned. Alternatively, the delay can be preserved by selecting the “keep delay” option in the Test Point Menu as shown above.

When using SDLA Dual Input mode, each leg of the signal is processed independently through the network. In Dual Input mode the common mode, differential, or individual inputs of the signal can be viewed simultaneously. The effects of reflections, transmission lines, and crosstalk are taken into account by using all the S-parameter terms throughout the system to build the test point transfer functions.

Unlike other solutions which provide visibility of a single test point, SDLA Visualizer allows up to four test points plus the data and clock waveforms after DFE to be visible at one time. Together with DPOJET, jitter and timing properties of the signal at multiple test points can be measured simultaneously enabling quick comparisions and validation of the data.

SDLA Visualizer Test Point Menu with Individual Inputs, Differential and Common Mode Views enabled for a single test point

Measurements and Jitter Analysis

Signals processed with SDLA might require multi-UI measurements, jitter breakdown, jitter bathtub analysis, or other evaluation. DPOJET (Option DJA) is recommended for these tasks. SDLA and DPOJET work together to analyze jitter, generate eye diagrams of equalized waveforms and a number of other views and measurements. Using the Analyze feature of SDLA, DPOJET is automatically configured to plot an eye diagram and TIE measurement for all enabled test points.

Apply and Analyze menu for DPOJET configuration

Plot of Elements of the System

In many situations, the information about channels, fixtures, or cables, represented by an S-parameter touchstone file or other information about the system is not completely clear. A supplier might have, for example, supplied you with a touchstone file without properly labeling its ports. To resolve the port assignment and in general to present a view of network response, SDLA offers plots for each block in the De-Embed and Embed path. Both time and frequency domain plots of all 16 S-parameters elements are provided. Overlay plots of the 16 S-parameters allow for quick correlation between different elements for example to compare s21 and s12 to verify equivalence.

Plot types include:

• Magnitude vs. Frequency
• Phase vs. Frequency
• Impedance as a function of frequency
• Smith Chart
• Step Response
• Impulse Response

Plots are available for each block in the De-embed and Embed blocks along with Magnitude, Impulse Response, Step Response, and Phase plots for each test point.

Smith Chart allows analysis of the reflection coefficients plotted as impedance for each block in the Embed and De-embed paths


Full S-Parameter plot

Practical Applications of SDLA Visualizer

Cable De-Embedding

Many next generation serial standards require the de-embedding of cables and fixtures that are used to access the device under test. With SDLA Visualizer, the effects of the cables and fixtures can be easily removed by modeling using the De-Embed block. For differential systems a cable pair is often used to attach to the device under test. In many cases each side of the cable will be modeled as a 2-port S-Parameter file. SDLA Visualizer allows you to easily specify each side of the cable as separate 2-port S-parameter files. When using SDLA dual input mode, each test point in SDLA can then be represented as the common mode, differential, or individual inputs of the signal allowing for the maximum flexibility in viewing the test point. Processing and viewing the individual inputs of the signal allows differences in each input to be observed and analyzed quickly.

SDLA Visualizer Dual Input mode showing individual inputs of the raw waveform and after de-embedding to remove reflections

Reflection Removal

As form factors become smaller and smaller, the ability to probe the device under test at the desired test points becomes challenging. The resulting signal at the probe point often has reflections that would not be present at the ideal test point. These artifacts make it difficult or even impossible to measure the signal. With SDLA Visualizer, reflections can be removed and the desired test point can be simulated. In the example above, SDLA Visualizer has been used to remove reflections from a DDR signal by approximating the receiver impedance, the transmission line delay between the probe point and the receiver, and by specifying the package model of the receiver. Once the reflections have been removed, the signal can be measured using DDR Memory Bus and Electrical Validation Software (DDRA).

DDR acquired signal shown with reflections and after reflection removal with SDLA Visualizer

Channel Simulation

An important part of today’s evaluation of serial data links is the interaction between the measured waveform and the complex behavior of the interconnect channel. It is no longer possible to assume that if the transmitter output meets the eye diagram mask, a design will work against all channels up to a given loss. Instead, advanced link test methods need to acquire the true transmitter waveform, then test against several corner-case channels.

In some cases this test can be performed with real physical channels: by connecting the transmitter under test to the suite of normative test channels, one channel at a time. However, this methodology is still prone to errors. Certainly each physical device has tolerances against the standard’s ideal values; furthermore the physical channels may vary making a virtual channel more stable. Since physical corner-case channels are typically not conveniently available, virtual channels based on the network description file, for example S-parameter touchstone files can be used. Now the measurement on your transmitter under test simply involves acquisition of the transmitter signal. Then each channel can be embedded in software rather than using a separate physical connection for each channel model.

SDLA Visualizer supports channel simulation by using the Embed block to model the simulation circuit. Once the measurement circuit has been removed from the signal path, the channel model can be quickly entered into SDLA to characterize what-if scenarios. For those cases where the simulated channel causes a closed eye, SDLA Visualizer receiver equalization can be used to open the eye.

Receiver Equalization

Transmitter equalization is one technique used to compensate for loss and dispersion in the interconnect; another technique is the equalization at the receiver. Receiver equalization in most modern NRZ systems falls under either CTLE (Continuous Time Linear Equalization); FFE (Feed Forward Equalization, also known as LFE (Linear Feedback Equalization); or DFE (Decision Feedback Equalization). A receiver equipped with equalization is capable of decoding signals which, when viewed as an eye diagram, produces a closed eye. How can measurements be performed on a closed eye?

The equalization tools in the SDLA can open even a completely closed eye, with your own equalizer tap values; or, at a push of a button, optimized equalization tap values are found for you, even on random data. The speed of recalculation and the ease of use allow you to easily modify system parameters, such as the number and weight of taps, or the amount of pre- or de-emphasis; you can optimize the design, develop what-if scenarios, or confirm your simulations over a range of conditions.

PCI Express 3.0, for example, requires a specific equalization scheme which optimizes a CTLE and one tap DFE. SDLA Visualizer has built in support to find the optimal CTLE setting and DFE tap value to maximize the eye area of the signal as required in the PCI Express 3.0 specification. All the intermediate results are available, which can be used to verify the results against the expected values.

SDLA Visualizer with probe de-embed enabled showing all the block element options

Network Measurement Tools

Whenever S-parameters of the network are needed, SDLA will accept data from either VNA or a TDR-base S-parameter measurement system. We recommend Tektronix TDR hardware and Tektronix IConnect® software for high-quality network measurements. Among the advantages of using Tek TDR and IConnect is the preservation of the DC values in the Touchstone matrix, which is typically lost with other measurement methods. SDLA accepts network description based on other measurement methods, such as VNA data or field-solver calculation; the DC measurement results might need to be extrapolated. SDLA will extrapolate the data to DC in these cases.

Additional information

For additional information, we recommend the following resources: literature on the DPO/DSA/MSO70000 Series oscilloscopes, their probes and software applications; literature on Tektronix TDR/TDT and the IConnect network measurement software; and literature on SDLA package 80SJNB for sampling oscilloscopes.