抖动测量-深入浅出.pdf

《抖动测量-深入浅出.pdf》由会员分享，可在线阅读，更多相关《抖动测量-深入浅出.pdf（60页珍藏版）》请在三一文库上搜索。

1、t edge n t edge n+1 t edge n t ref n 抖动基本概念介绍以及测量方法概述 泰克科技(中国)有限公司 October 20002Jitter measurement 内容 1、抖动测试和分析基础 2、选择合适的抖动测试工具和方法 3、影响抖动测试结果的因素 4、使用示波器测试抖动的演示 5、问题讨论 October 20003Jitter measurement 抖动测试和分析基础 “边沿距理想位置的偏差” 抖动可能由以下原因引起 热噪声 注入的噪声（如EMI等） 电路不稳定 传输损耗 抖动测试基础术语 Period jitter Cycle-to-cycle

2、jitter Time Interval Error Clock jitter Data jitter Clock recovery Unit Interval BER October 20004Jitter measurement Period Jitter Period Jitter is the measurement of a signals period over a number of cycles Mean (the average of the period measurements) Std. Dev. (the RMS of the period measurements)

3、 Pk-Pk (the difference between the minimum and maximum period) What about jitter frequency or cycle-to-cycle requirements? t period October 20005Jitter measurement Cycle-to-Cycle Jitter Cycle to Cycle Jitter is the measurement of a signals change in period between adjacent cycles Mean (the average o

4、f the change in period measurements) Std. Dev. (the RMS of the change in period measurements) Pk-Pk (the difference between the minimum and maximum change) This is a period differential measurement! t period n t period n+1 t period n t period n+1 October 20006Jitter measurement Positive Cycle-to-Cyc

5、le Jitter t period n t period n+1 t period n t period n+1 October 20007Jitter measurement Time Interval Error TIE Jitter is the measurement of a signals timing error relative to a known or recovered clock ITU says TEITU says TE Mean (the average of the timing error) Std. Dev. (the RMS of the timing

6、error) Pk-Pk (the difference between the minimum and maximum error) This measurement requires a reference clock!This measurement requires a reference clock! t edge n t edge n+1 t edge n t ref n October 20008Jitter measurement Time Interval Error October 20009Jitter measurement Jitter Measurements P2

7、P3P4P1 Period Jitter = 18.3ps StdDv(0.990/1.010/0.980/1.020) 40ps p-p Cy-Cy Jitter = 36.1ps StdDv(0.020/-0.030/0.040) 70ps p-p TIE= 9.6ps StdDv(-0.010/0.000/-0.020/0.000) 20ps p-p 0.990ns1.010ns0.980ns1.020ns 0.0ns0.990ns2.000ns2.980ns4.000ns 0.020ns-0.030ns0.040ns -0.010ns0.000ns-0.020ns P Cy-Cy TI

8、E0.000ns * StdDv = MS ExcelStdDevA RMS of large populations October 200010Jitter measurement UI and BER Unit Interval The nominal period of one transmitted bit Bit Error Rate/Ratio Method to describe expected or measured data stream error rate or ratio of good bits to bad bits Generally specified to

9、 be lower than 1.0E-12 October 200011Jitter measurement Back to Time Interval Error TIE IDEAL SAMPLE POINT BECOMES THE REFERENCE FOR TIE Another view TIE = StdDev of all edges October 200012Jitter measurement 抖动测试和分析基础 有关抖动的新术语 Composite jitter Total jitter Random jitter Deterministic jitter Periodi

10、c jitter Duty cycle distortion Inter-symbol interference (DDJ) October 200013Jitter measurement Composite Jitter Historical Eye-Closure Measurement Jitter value including all Rj+Dj components Expressed as 1 sigma RMS or Pk-Pk Unbounded, result depends on measurement interval 11k hits = 124ps Pk-Pk57

11、2M hits = 173ps P-P = 22.7ps = 23.4ps October 200014Jitter measurement Random Jitter Jitter of a random nature has Gaussian distribution Histogram (estimate) pdf (mathematical model) Peak-to-Peak = unbounded! 1-sigma or RMS7-sigma October 200015Jitter measurement Deterministic Jitter Deterministic j

12、itter has non-Gaussian distribution and is bounded. Histogram = pdf (close enough) Peak-to-Peak NEVER A DUAL DIRAC HISTOGRAM! October 200016Jitter measurement Periodic Jitter TIE vs. time is a repetitive waveform Equivalent to Frequency Modulation (FM) Peak-to-Peak October 200017Jitter measurement D

13、uty Cycle Distortion Asymmetrical rise-time vs. fall-time Non-optimal choice of decision threshold 0.0v -0.1v Well, ALMOST never a dual-Dirac histogram! October 200018Jitter measurement 8不对称的上升边沿速率与下降边沿速率不对称的上升边沿速率与下降边沿速率 8不适当的判断门限选择不适当的判断门限选择 DCD 占空比失真 October 200019Jitter measurement Inter-Symbol

14、Interference ISI or DDj or PDj used interchangeably How a pattern effects subsequent bits Due to transmission line effects, reflections, etc. October 200020Jitter measurement 8ISI又称为又称为DDj数据相关抖动或数据相关抖动或PDj码型相关抖动码型相关抖动 8因为有限的带宽限制因为有限的带宽限制 8驱动器驱动器 Driver 8对比器对比器Comparator 8PCB线路与电缆的衰减与损耗 对经常切换的 线路与电缆的

15、衰减与损耗 对经常切换的“1,0,1,0,” 的高频信号，衰减比连续的的高频信号，衰减比连续的“1,1,1,1,0,0,0,0,”的低频信号 要来得厉害。所以长的连续不变码到达更高的电平，在跳变时需要更多的时间才能到 达门限电平，导致信号抖动。因为这个抖动的幅度与码型相关，所以又称码型相关抖 动。 的低频信号 要来得厉害。所以长的连续不变码到达更高的电平，在跳变时需要更多的时间才能到 达门限电平，导致信号抖动。因为这个抖动的幅度与码型相关，所以又称码型相关抖 动。 8因为阻抗不匹配导致信号发射。被发射的信号叠加在原由的信号导致幅度增加而最终 使转换电平所耗费的时间更多，从而产生抖动。 因为阻抗

16、不匹配导致信号发射。被发射的信号叠加在原由的信号导致幅度增加而最终 使转换电平所耗费的时间更多，从而产生抖动。 ISI 码间干扰 DDJ不一样的电平不一样的电平 October 200021Jitter measurement Total Jitter BER Random components Deterministic components Peak-to-Peak 1-sigma or RMS pdf: Tj = Dj Rj (convolution) option b: Pk-Pk: Tj = (N*Rj) + Dj , where N is desired sigma DJ RJL

17、rms RJR rms October 200022Jitter measurement 抖动分类的意义 理解抖动产生的原因就能准确的 定位到抖动 Random Jitter 使用低抖动的器件 减少器件个数 Deterministic Jitter 改善传绕和耦合 增加互连的带宽 改善时钟分配 改善电源、参考分割 October 200023Jitter measurement 内容 1、抖动测试和分析基础 2、选择合适的抖动测试工具和方法 3、影响抖动测试结果的因素 4、使用示波器测试抖动的演示 5、问题讨论 October 200024Jitter measurement 选择合适的抖动测

18、量工具和方法 实时示波器，如泰克DSA/DPO70604、DSA/DPO72004等 取样示波器，如泰克DSA8200+相应模块 频谱分析仪，如泰克RSA3408B 时间间隔分析仪 误码仪 October 200025Jitter measurement 可以测试哪些抖动？ ?A Real-TimeDSO (Digital Storage Oscilloscope) recovers the whole waveform? can measure anything. Currently it can characterize and analyze Serial Data Jitter, Cl

19、ock sources, PLLs, etc. Limitations are Frequency (or Bit Rate), Resolution (of spectra, of minute jitter, of multi-level modulation) ?A Real-Time Spectrum Analyzer excels with Complex modulations for Mobiles; Clocks, PLLs and their dynamic performance; etc. Limitations include Span (sub-100 MHz), a

20、nd BW (below 10 GHz), signals with great modulation spectrum ?Equivalent-timeSampling Oscilloscope /Comm. Signal Analyzer (CSA) Best BW for Serial Data. Currently the only with Noise analysis and BER Eye. Limitations include no Real-time capture repetitive patterns only, some jitter spectra aliased.

21、 Time Interval ErrorCycle-to-Cycle Period October 200026Jitter measurement ET-RT Basics: RT architecture - trigger on data and sample in real-time A realtime scope does not require a separate signal to trigger: the signal under test can act as the trigger for initiating fast real-time sampling to ac

22、quire a waveform. Data time voltage Minimum time between real-time sampled points is determined by the fastest sample rate the realtime scope is capable of. 40GSamples/sec results in 25ps between sampled points.* One trigger can initiate the real-time sampling of the entire record length *Maximum re

23、solution is higher through interpolation October 200027Jitter measurement ET-RT Basics: ET one-sample-per-trigger, non repetitive pattern ok for mask test When a clock signal is used to trigger the equivalent-time 8200 scope the sampled DATA signals generally create EYE PATTERNS (between clock trigg

24、ers the sampled DATA could be either a logical 1 or 0) Data voltage time Precision variable delay A clock trigger can be user-supplied or recovered from the data to trigger the equivalent time sampler Clock Eye patterns are the common result of clock-triggering in Equivalent time sampling: vectors a

25、re not drawn since adjacent samples can jump from logical 1 to 0 frequently Mask test October 200028Jitter measurement DPO示波器分析抖动 October 200029Jitter measurement DSOs exercising their RT muscle ?Transfer function of a PLL: the oscilloscope simultaneously captures the Output and the Input of the PLL

26、. The two are divided to yield the transfer function. Time Interval ErrorCycle-to-Cycle Period October 200030Jitter measurement 取样示波器分析抖动 October 200031Jitter measurement ET-RT Basics: ET - one-sample-per-trigger, repetitive pattern required for pulse streams An equivalent-time 8000 series sampling

27、oscilloscope requires a trigger signal: this is generally a user-supplied clock, a recovered clock, or a pattern sync signal synchronous to the signal. Data voltage time Pattern Sync A trigger is required for EACH sampled point in the high bandwidth equivalent-time 8000 scope Minimum time between eq

28、uivalent-time sampled points on high bandwidth 8000 scope can be adjusted to fractions of a picosecond Precision variable delay October 200032Jitter measurement The Jitter can be Analyzed Time Interval Error Jitter analysis used in the field of Serial Data transmission Separating jitter into its com

29、ponents provides increased precision and insight into root cause of BER performance Constituent Components of Jitter = Unbounded = Bounded Total Jitter (TJ) Duty-Cycle Jitter (DCD) Data Dependent Jitter (DDJ) Periodic Jitter (PJ) Deterministic Jitter (DJ) Random Jitter (RJ) October 200033Jitter meas

30、urement Real World: Jitter dominated signal impairment Sometimes the BER of a communications link is limited by just Jitter Jitter separation leads to insight into root cause BER extrapolations and bathtub curves can be accurately calculated Noise dominated signal impairment Sometimes it is noise th

31、at dominates the BER Jitter separation provides very little insight into root cause of BER performance Jitter & Noise signal impairment Often it is limited by both Jitter and Noise Jitter separation provides only a limited answer October 200034Jitter measurement 80SJNB Advanced Jitter, Noise and BER

32、 Analysis More Complete ? More Accurate BER Analysis = Unbounded = Bounded Noise Separation Random Noise (RN) Deterministic Noise (DN) Periodic Noise (PN) Data Dependent Noise (DDN) Total Noise (TN) Jitter Separation Total Jitter (TJ) Duty-Cycle Jitter (DCD) Data Dependent Jitter (DDJ) Periodic Jitt

33、er (PJ) Deterministic Jitter (DJ) Random Jitter (RJ) Bit Error Ratio (BER) 80SJNB the FIRST and ONLY oscilloscope solution to TOTAL BER analysis! NEW More Accurate Eye contour and BER Extrapolation October 200035Jitter measurement RTSA Real-Time Spectrum Analyzer Time Interval ErrorCycle-to-Cycle Pe

34、riod Trigger, Capture, Analyze Frequency October 200036Jitter measurement RTSA: Signal Source Analysis Measurement Overview 相噪相噪 功率噪声谱密度 相噪积分 相噪稳定时间相噪稳定时间 测量调频信号的相噪 随机抖动随机抖动 随机抖动随时间的变化 随机抖动稳定时间 最大周期性抖动最大周期性抖动 最坏情况下的随机抖动 噪声图噪声图 相噪随时间的变化 杂散杂散 Period October 200037Jitter measurement RTSA Signal Source

35、Analysis Measurement Suite Random Jitter Calculate Random Jitter (RJ) RJ settling time is measured as RJ vs time 3.2 msec at threshold 2.5 psec. Area of integrated RMS noise Calculated from integrated RMS noise Period October 200038Jitter measurement 内容 1、抖动测试和分析基础 2、选择合适的抖动测试工具和方法 3、影响抖动测试结果的因素 4、使

36、用示波器测试抖动的演示 5、问题讨论 October 200039Jitter measurement 影响抖动测试精度的因素有哪些？ 硬件平台指标和性能 带宽 采样率 高速采集内存大小 增量时间精度（DTA） 抖动噪声基底 触发抖动 软件测试和分析工具 您的测试对象 100M和1M的时钟都能测试么？ 合理的操作和设置 操作问题和问题讨论部分 October 200040Jitter measurement 0垂直比例设置: 所有情况下尝试将垂直幅度调节至6格摆幅或以上，这会减小垂直噪声的 影响。 0水平比例设置: 越多的采样数目越好。TDSJIT3/RT-EYE在分离Rj/Dj时, 需要单次

37、信号采 集最少100次重复的码型，唯一的需要考虑的是，当采样数目更多时，所需要采集的时 间与记录长度都更多。举例：1个2.5Gbps 27-1 PRBS伪随机重复码型，重复一次需要 127 X 0.4ns = 50.8ns， 100次重复所需的时间约为5us。 0采样率: 需要在边沿上采集最小2个点，3点效果会更好。这意味测量电路100ps 的上升 时间, 50ps/pt是最起码的采样间隔。一个200ps的边沿，最小要求100ps/pt的采样间隔, 不过50ps/pt将更好。 0泰克DPOJET对于垂直比例与水平比例设有Autoset功能, 基于信号的特性自动获取最佳 之设置值 0其他: 使用

38、Sin(x)/x正弦内插模式, 这会大大提升测量精度 0例子：使用上述的例子，使用50ps/pt的采样间隔(即20GS/s采样率)，记录5us, 需要 100K的记录长度。 深入进行精确的抖动测量 October 200041Jitter measurement 抖动噪声基底和触发抖动（时间单位） 触发抖动 示波器多次触发，多次采集时触发位置的时间偏差 使用实时示波器采集和分析抖动，测试结果不受触发抖动影响 抖动噪声基底 该指标综合所以影响抖动测试结果的硬件因素 反应示波器能够测试的最小抖动 实时示波器DSA70000的抖动噪声基底为400fs October 200042Jitter mea

39、surement 噪声 vs. 抖动噪声基底 JNF指标反应了噪声，孔径不确定度，时基稳定度等错误对抖动测试的影响 示波器进行抖动测试的最低极限 如何测试JNF 捕获一个非常稳定的时钟信号 测试每个边沿的TIE TIE的标准偏差即为JNF October 200043Jitter measurement 内存设置为2M时测试结果 October 200044Jitter measurement 内存设置为2M时结果细节 October 200045Jitter measurement 内存设置为2M时直方图 October 200046Jitter measurement 内存设置为10M时-

40、测试结果 October 200047Jitter measurement 内存设置为10M时结果细节 October 200048Jitter measurement 内存设置为10M时直方图 October 200049Jitter measurement 内存设置为25M时测试结果 October 200050Jitter measurement 内存设置为25M时结果细节 October 200051Jitter measurement 内存设置为25M时直方图 October 200052Jitter measurement 内存设置为100M时-测试结果 October 20005

41、3Jitter measurement 内存设置为100M时结果细节 October 200054Jitter measurement 内存设置为100M时直方图 October 200055Jitter measurement 4与误码仪的相关性：与误码仪的相关性： 4世面有不少号称世面有不少号称Rj/Dj分离的工具，但不见得所测试的结果互相关连。业内进行分离的工具，但不见得所测试的结果互相关连。业内进行Rj/Dj分离 的方法至今都是各厂家的专利， 分离 的方法至今都是各厂家的专利，并未有一套公认的标准与方法学。并未有一套公认的标准与方法学。 4在误码测试方面，误码仪是被公认的可信的仪器之一

42、，在误码测试方面，误码仪是被公认的可信的仪器之一，所以所以TDSJIT3所测试出来的所测试出来的BER 是否与误码仪所测试出来的相一致，是有力的证据证明其分离是否与误码仪所测试出来的相一致，是有力的证据证明其分离Rj/Dj与估算与估算BER的方法是 正确无误的。 的方法是 正确无误的。 4TDSJIT3/RT-Eye的的BER测试结果与误码仪的测试结果与误码仪的BER测试结果在测试结果在2%的差别范围之内。的差别范围之内。 DPOJET 结果的相关性 October 200056Jitter measurement In Closing Measurement and Jitter Analy

43、sis packages available on todays oscilloscopes are very effective tools for characterization, debug and analysis, and conformance testing todays current and emerging serial data standards. Thank you for attending! Application notes, additional information at October 200057Jitter measurement Total Ji

44、tter BER PJ PJ DCD DCD ISI/DDJ ISI/DDJ ?DJ is NOT a linear SUM ?DJ != DCD + PJ + DDJ October 200058Jitter measurement Rj/Djfrom Tj BER Assume bi-modal distribution (dual-Dirac) Slope is 1/Rj Intercept is Dj October 200059Jitter measurement Rj/Djfrom Tj BER Assume bi-modal distribution (dual-Dirac), measure Tj at two BER Fit curve to points, slope is Rj, Intercept is Dj Measured Tj 10-7 Measured Tj 10-4 DjxRj Estimated Tj 10-12 x7.4 x10.4 x14.1 October 200060Jitter measurement 动手实验 使用Tektronix实时数字荧光示波器DSA72004DPOJET分 析软件进行抖动测试