Improved Measurements Overcome High-Speed Interconnect Challenges October 2013 Bob Buxton 1 EuMW Seminars 2013

Agenda Background S-parameter measurements - frequency range considerations Eye Diagrams and low frequency measurement data issues De-embedding Correlation – measurements and simulation Setting emphasis levels Superposition vs. true mode stimulus for active device measurement Resources Questions 2 EuMW Seminars 2013

Background – challenge for SI Engineers Compliance with higher data rate standards Cost/performance trade-offs Locating Defects Measurement – simulation correlation Dealing with test fixtures 3 EuMW Seminars 2013

8B/10B Encoder Serializer Equalizer Driver Clock/PLL 10 B/8B Decoder De- Serializer Equalizer DataRecovery CR/PLL Parallel Data Parallel Data Serial Data Jitter and Noise Channel Issues Challenges for SI Engineers 4 EuMW Seminars 2013

Types of Channel 5 EuMW Seminars 2013

Tx Rx Channel Issues - Loss 6 EuMW Seminars 2013

Channel artifacts (vias, impedance changes, ground plane issues etc.) Channel Issues - Structures 7 EuMW Seminars 2013

Backplane Transmission Measurement 8 EuMW Seminars 2013

Tx Rx Tx Rx TxRx FEXT NEXT Channel Issues – Crosstalk 9 EuMW Seminars 2013

Harmonic Content of 28 Gbps NRZ clock signal Attenuating harmonics distorts signal Ideally measure to 5 th harmonic Importance of Maximum Frequency Range 10 EuMW Seminars 2013

High Frequency VNA impact – resolution 11 TWO MISMATCHES SEPARATED By 2 mm (AIR) spanresolution 1)40 GHz3.75 mm 2)50 GHz3.0 mm 3)60 GHz 2.5 mm EuMW Seminars 2013

High Frequency VNA impact - causality Lack of causality means output appears to occur prior to stimulus Can cause unstable simulations Higher frequency data improves causality 12 Non-Causal Results EuMW Seminars 2013

Need for Low Frequency Data – 2 Reasons 1.DC Extrapolation 2.Sampling in the frequency domain and aliasing* – Max unambiguous time domain result: T max = 1/(2f s ) – Need to consider multiple reflections 13 * A Reverse Nyquist Approach to Understanding the Importance of Low Frequency Information in Scattering Matrices Daniel Dvorscak and Michael Tsuk, ANSYSInc, DesignCon 2013 EuMW Seminars 2013

VNA performance and DC Extrapolation Series of measurements made on same 40 inch line VNA 1 – 40 MHz to 40 GHz – Couplers for entire band VNA 2 – 4 MHz to 40 GHz – Hybrid of bridges and couplers 14 EuMW Seminars 2013

15 VNA performance and DC Extrapolation DC extrapolation depends on quality of S-parameter measurement Left trace from VNA limited low frequency performance –Coupler based –92 dB DR at 40 MHz Right trace from VNA with better low frequency performance –Bridges at LF –115 dB DR at 4 MHz EuMW Seminars 2013

Impact on step response MHz to 40 GHz data VNA 14 MHz to 40 GHz data VNA 2 Slope due to poor DC extrapolation Aliasing due to 40 MHz sampling Good low start frequency data improves DC extrapolation Low start frequency sets frequency domain sampling for low pass step response and hence alias free range EuMW Seminars 2013

Stability at low frequency also critical 17 Low frequency data issues (drift/instability…) can have an out-sized effect on transformed data because of its criticality to large-distance-scale structure. EuMW Seminars 2013

18 Eye diagrams – 10 Gbit 10 Gbit – data from VNA 1 – 40 MHz to 40 GHz 10 Gbit – Data from VNA 2 – 4 MHz to 40 GHz EuMW Seminars 2013

Importance of Low Frequency Range VectorStar displays a flat 25 ohm section as expected Measured Data range from to ohms (Composite picture of multiple screen captures showing measurements made on Beatty Standard) 19 Low noise on low frequency data give rock-steady results from sweep to sweep EuMW Seminars 2013

VectorStar Architecture: Two VNAs in One! < 2.5 GHz Low Band > 2.5 GHz High Band Port 1 Port 2 a2a2 a1a1 b1b1 b1b1 a1a1 b2b2 a2a2 b2b2 MS4640A Block Diagram (Fully Loaded Configuration) optional Bias 1Bias 2 20 EuMW Seminars 2013

Unique Hybrid VNA Architecture Two VNAs in parallel: Almost the only way to get 6 decades of coverage (from kHz to GHz frequencies) – Each receiver technology (sampler or mixer) used in its best range – Each coupling technology (coupler or bridge) used in its best range – Both share a common IF path and fully synthesized source 21

MethodStandards complexity Fundamental accuracy Sensitivity to standards Media preferences Type A (adapter removal) High High (refl.)Need good reflect and thru stds Type B (Bauer-Penfield) MediumHighHigh (refl.) Only need reflect standards, not great for coupled lines Type C (inner-outer) High Medium (refl.) More redundant than A so less sensitive but need good stds still Type D (2-port lines) Med Low for low-loss or mismatched fixtures Medium (line def’n.) Only need decent lines; match relegated to lower dependence; can handle coupled lines Type E (4 port inner-outer) High Medium (refl.) Somewhat redundant (like C) but need decent standards. Best for uncoupled multiport fixtures Type F (4-port uncoupled) Med Low for low-loss or mismatched fixtures Medium (line def’n.) Only need decent lines; match relegated to lower dependence; can handle coupled lines Type G (4-port coupled) Med Low for low-loss or mismatched fixtures Medium (line def’n.) Only need decent lines; match relegated to lower dependence; can handle coupled lines well Backplanes Best Accuracy Requires good repeatability De-embedding Methods available within VectorStar 22 EuMW Seminars 2013

Correlation - Measurement and Simulation Use Channel Modeling Platform Use time domain equipment to measure Eye and compare with simulated Eye 23 EuMW Seminars 2013

Use of Channel Modeling Platform A set of known structures Feed data into models Make measurements Make comparison 24 EuMW Seminars 2013

Correlation – Measurement and Simulation 25 Tx Rx Simulated Using measured S- parameter data Measured using Oscilloscope Input waveform with no emphasis Simulated Using measured S- parameter data Measured using Oscilloscope Input waveform with emphasis EuMW Seminars 2013

Challenge: Difficult to find the ideal emphasis settings from the many possibilities Problem: Searching for ideal settings while verifying the output waveform – takes an extremely long time – hard to explain why those settings are ideal. Solution: Use VNA-captured S-parameter data to apply inverse DUT characteristics to input waveform Setting Ideal Emphasis 26 EuMW Seminars 2013

Time and Frequency Domain – a merged solution 27 DUT (27inch PCB) Tap Settings EuMW Seminars 2013

MS4640B now with: Option 031 Dual Source Architecture Option 043 DifferentialView TM 28 EuMW Seminars 2013

VectorStar PROVIDES Broadest frequency span: 70 kHz to 70/110 GHz Best time domain analysis capability Widest range of calibration & de-embedding methods Upgradeable in frequency range, port count & option additions Choice of TMS or Superposition 29 Good S-parameter Data Poor S-parameter Data EuMW Seminars 2013

with DifferentialView and Dual Source True Mode Stimulus capability Differential, common and mixed mode S-parameters Adjust differential phase & amplitude Instant view of results during parameter change + _ + + _ _ + + __ Differential ModeCommon ModeMixed Mode 30 EuMW Seminars 2013

Port 1Port 2Port 3Port 4 Time-coherent in phase and amplitude Port 1Port 2Port 3Port 4 DUT Superposition and True Mode Stimulus 31 EuMW Seminars 2013

Applicability of the two methods Device to be measured:Superposition True Mode Stimulus Passive Balanced / Differential DUT Transmission LinesXX PCBXX Lumped ComponentsXX Passive FiltersXX Unshielded and Shielded Twisted Pair, Quad CablesXX Connectors / InterfacesXX Linear Active Balanced / Differential DUT Linear Amplifiers, Differential AmplifiersXX Linear Active FiltersXX Input / Output Match ADC / DACXX Non Linear Active Balanced / Differential DUT Devices in Compression / Saturation X Log Amplifiers X 32 EuMW Seminars 2013

Trade Offs SuperpositionTrue Mode Stimulus Type of VNASingle source VNADual source required Method of obtaining Differential and Mixed Mode Parameters Calculated Measured directly Type of DUTs Passive and active linear Necessary only for non-linear Available Frequency Range70 kHz to 110 GHz 70 kHz GHz (with VectorStar) Calibration Complexity Typical 4-port Typical 4-port plus calibration of dual sources Average Calibration Time T (Time depends on number of points, IF BW, skill of operator) Approx. 2T Calibration stability considerations Normal measurement calibration intervals Calibrate more frequently due to stability issues if VNA does not feature advanced correction algorithms Overall Solution Cost$$$ VectorStar MS4640B has a second source option and DifferentialView TM for true mode stimulus measurement Only recommended when device is non-linear 33 EuMW Seminars 2013

To Use or Not Use True Mode Stimulus? 34 Amplifier in non-linear region Difference between superposition and TMS apparent Amplifier in linear region Difference between superposition and TMS not discernable EuMW Seminars 2013

Broadest frequency span: 70 kHz to 70/110 GHz Best time domain analysis capability 4-port test set upgrades 2 port VectorStar to 4- port performance. Widest range of calibration & de-embedding methods Choice of TMS or Superposition Good S-parameter Data Poor S-parameter Data DifferentialView TM for Signal Integrity Measurements 35 EuMW Seminars 2013

DifferentialView TM for Broadband and mmWave Measurements Compact size and high performance make the Anritsu mmWave modules ideal for broadband differential analysis 36 EuMW Seminars 2013

DifferentialView TM True Mode Stimulus Interface DifferentialView offers easy configuration for differential and mixed mode measurements for thorough analysis 37 EuMW Seminars 2013

DifferentialView TM TMS Setup and Editing DifferentialView menu provides real time display of measurement parameters Immediately observe DUT performance changes with changes in setup Example of modifying phase sweep parameters while observing effects 38 EuMW Seminars 2013

Option 031 Dual Source eliminates the need for a transfer switch Provides up to 7 dB of additional power at 70 GHz MS4640B series improves noise floor specification as much as 9 dB Combined, results in improved dynamic range performance up to 16 dB at 70 GHz! + + _ _ Dual Source VectorStar a1a1 b1b1 a2a2 b2b2 xNxN xNxN VectorStar TM Dual Source Option 39 EuMW Seminars 2013

+ + _ _ Phase Offset at Test Port Non Phase Offset at DUT Input Dual Source VectorStar a1a1 b1b1 a2a2 b2b2 xNxN xNxN No TMS correction applied. Characterizing differential devices with a offset is the goal of component designers Nonlinear devices are sensitive to source mismatch Source mismatch will shift stimulus signals to non-ideal offset Without proper offset correction performance of device will vary Anritsu white paper discusses this issue Measurements of Nonlinear Differential Devices 40 EuMW Seminars 2013

+ + _ _ Phase Offset at Test Port Phase Offset at DUT Input Applying TMS correction within DifferentialView corrects offset shift Monitoring the applied signals (a3/a1) will provide an indication on the success of correction Opt 043 DifferentialView TM : Measure mismatch and apply correction during measurement Dual Source VectorStar a1a1 b1b1 a3a3 b3b3 xNxN xNxN Measurements of Nonlinear Differential Devices 41 EuMW Seminars 2013

Sweep to sweep phase variations from differential while driving nonlinear DUT at -12 dBm. 10 MHz to 50 GHz. Measured result showing < 1 degree sweep to sweep stability DifferentialView TM Phase Stability 42 EuMW Seminars 2013

Resource Materials MS4640B Technical Data Sheet Signal Integrity Application Brief White Papers – Overcoming High Speed Interconnect Challenges – Signal Integrity: Frequency Range Matters – High Data rates Require New De-embedding Techniques – Superposition vs. True balanced: What’s Required for Your Signal Integrity Application – True Mode Stimulus and Stability Application Note: Ideal Pre-Emphasis Constant Setting EuMW Seminars 2013

Solutions available to 20, 40, 50, 70 and 110 GHz 70 kHz start frequency and low frequency dynamic range of VectorStar TM critical to signal integrity application and to ensuring measurement-simulation correspondence MS4640B now offers DifferentialView TM and 2 nd internal source option for true mode stimulus measurements – Easy to use graphical set-up – Real-time view of measurement while making parameter changes – Advanced true mode stimulus corrections hold phase relationship at DUT Upgradeable in frequency, port count and options – E.g. can add second source and DifferentialView later Summary 44 EuMW Seminars 2013