Jitter Generation and Measurement for Test of Multigigabit

Jitter Generation and Measurement for Test of Multi-gigabit Serial Interconnects Sassan Tabatabaei, Freddy Ben-Zeev and Michael Lee

Purpose ü RJ, DDJ, and PJ injection for receiver test § Use delay line modulation § Programmable DDJ ü Jitter measurement methodology with continuous TIA (CTIA) § Markerless measurement

Outline ü Jitter injection issues ü Programmable jitter injection ü Continuous time interval analyzer (CTIA) ü Markerless and fast RJ/DDJ measurement methodologies

Jitter Injection Typical setup (e. g. , XAUI)

Jitter Injection ü Limited PJ frequency range ü Large RJ may cause spurious edges ü DDJ programmability very limited § Requires filter tuning § Typically not portable across applications with different bit rates ITC 2004 Jitter Models & Measurement for High-Speed Interconnects

Jitter Injection Programmable setup

DDJ Injection

Traditional Time Interval Analyzer (TIA) Operation Principles ITC 2004 Jitter Models & Measurement for High-Speed Interconnects

Continuous Time Interval Analyzer (CTIA) Operation Principles ITC 2004 Jitter Models & Measurement for High-Speed Interconnects

Jitter Measurement Definitions ü Random jitter (RJ) § Uncorrelated to Known sources § Gaussian distribution ü Data-dependent jitter (DDJ) § Inter-symbol interference § Duty-cycle distortion § Reflections

Jitter Measurement Definitions Time Interval Error (TIE) function: Ideal signal Real signal TIE(1) ITC 2004 Jitter Models & Measurement for High-Speed Interconnects TIE(2) TIE(3) TIE(4)

RJ Measurement ITC 2004 Jitter Models & Measurement for High-Speed Interconnects

Scaled FFT (ps) TIE sequence (ps) RJ Measurement 100 TIE 50 0 -50 -100 0 0. 01 0. 02 0. 03 0. 04 0. 05 0. 06 2 10 0. 07 0. 08 0. 09 Time (s) TIE FFT 0 10 -2 10 -4 10 0 0. 1 0. 2 0. 3 0. 4 0. 5 0. 6 0. 7 0. 8 0. 9 1. 0 Normalized frequency ITC 2004 Jitter Models & Measurement for High-Speed Interconnects

RJ Measurement ü Frequency domain deconvolution ü Insensitive to PJ and DDJ ü Predictable Repeatability § 3% for 8192 samples ü No external marker required

DDJ Measurement ITC 2004 Jitter Models & Measurement for High-Speed Interconnects

DDJ Measurement Bin population 120 Histogram for each pattern edge Edge 5 100 80 Edge 1 Edge 3 Edge 4 Edge 2 60 40 20 0 -40 0 40 TIE (ps) ITC 2004 Jitter Models & Measurement for High-Speed Interconnects 80 120

DDJ Measurement ü Time domain deconvolution § Per-edge time shift measurement ü Insensitive to RJ and PJ (averaging) ü No external marker required

DDJ Injection Example: Sinusoidal DDJ Edge shift (ps) 15 CTIA (GT 4000) Scope (86100 C) 10 5 0 -5 -10 0 20 ITC 2004 Jitter Models & Measurement for High-Speed Interconnects 40 60 80 Transition bit number 100 120

Measurement Results (RJ) RJ (ps) Accuracy CTIA TDS 7404 Linear fit 10 8 6 4 3 20 22 24 26 28 30 32 DRJ (ps) 0. 5 0 -0. 5 20 Noise generator attenuation (db)

Measurement Results (RJ) Immunity to DDJ 2. 5 Gbps, PRBS 7 pattern 5. 8 RJ (ps) 5. 7 5. 6 5. 5 5. 4 20 25 30 Injected DDJ (ps) ITC 2004 Jitter Models & Measurement for High-Speed Interconnects 35 40

Measurement Results (DDJ) Measured peak-to-peak DDJ (ps) Accuracy 50 45 40 PRBS 7 35 30 K 28. 5 25 CTIA ET-DSO 20 15 10 0 20 40 60 80 100 120 140 AWG amplitude (m. V) 160

Measurement Results (DDJ) 35 34 33 32 2 4 6 8 10 Injected RMS RJ (ps) 12 Measured peak-to-peak DDJ (ps) Immunity to RJ and PJ 34 33 32 31 30 0 50 Injected PJ (ps) 100

Conclusions ü Delay line modulation with locked AWG provide flexible and programmable jitter injection method. § No need for application-specific DDJ filters ü CTIA: markerless jitter separation § Virtual marker § Accurate § Fast: • No marker generation time • Compact data volume