AND9075 - Understanding Data Eye Diagram …

1 Semiconductor Components Industries, LLC, 2015 June, 2015 Rev. 11 Publication Order Number: AND9075 /DAND9075/DUnderstanding data EyeDiagram Methodology forAnalyzing high SpeedDigital SignalsIntroductionThe data eye Diagram is a methodology to represent andanalyze a high speed digital signal. The eye Diagram allowskey parameters of the electrical quality of the signal to bequickly visualized and determined. The data eye Diagram isconstructed from a digital waveform by folding the parts ofthe waveform corresponding to each individual bit intoa single graph with signal amplitude on the vertical axis andtime on horizontal axis. By repeating this construction overmany samples of the waveform, the resultant graph willrepresent the average statistics of the signal and willresemble an eye. The eye opening corresponds to one bitperiod and is typically called the Unit Interval (UI) width ofthe eye Diagram . An ideal digital waveform with sharp riseand fall times and constant amplitude will have an eyediagram as shown in Figure 1 1.

2 Ideal high speed Digital Signal with Eye DiagramObviously, this ideal eye Diagram offers little additionalinformation beyond the time domain waveform world high speed digital signals suffer significantimpairments including attenuation, noise, crosstalk, data eye Diagram for a typical high speed digital signalis shown in Figure 2 below. Notice how the Diagram moreresembles the shape of an 2. Typical high speed Digital Signal with Eye DiagramData Eye Diagram FundamentalsHigh speed Digital SignalingDigital signaling is the transmission of baseband data overa cabled connection. This data is usually modulated or codedaccording to the telecommunication protocol set asa standard for the intended interface. The type of basebandcoding (commonly called Line Coding) chosen fora standard best optimizes performance, given the electricalcharacteristics of the data and the transport medium. Legacyhigh- speed digital standards including USB usea form of non-return to zero (NRZ) for the data codingwhere a high (positive) pulse represents a logic one anda low (negative) pulse a logic zero.

3 By controlling the dataformat ( , bit stuffing, etc) to make the number of ones andzeros equal, NRZ waveforms can be DC balanced and limitthe DC content in the signal. This allows the signal to becapacitively (or AC) coupled and also provides commonmode voltages or DC power to be combined with the signalon the same coding using return-to-zero (RZ) signaling isusually not used on high - speed digital interfaces because ithas excessive bandwidth requirements due to its inherentDC imbalance which in turn adds significant DC controlling DC content and bandwidth, dataformatting can also enable clock recovery for the high speedsignal. For reliable clock recovery at the receiver, themaximum number of consecutive ones or zeros needs to bebounded to a reasonable number since the waveform clockis recovered by observing transitions in the receivedsequence. A common method to bound the number of onesand zeros in a waveform is to use 8b/10b encoding.

4 8b/10bis a line code that maps 8-bit symbols to 10-bit symbols toprovide DC-balance for the waveform and also provideenough state changes to allow reasonable clock line coding, the signal is put through a physicalchannel. The line-coded signal can either be directly put ona transmission line or can be level shifted and pulse shapedto reduce its bandwidth, improve EMI/RFI performance ormatch system interfacing requirements. Low VoltageDifferential Signaling (LVDS) is a commonly used interfacestandard for high speed digital signals. By providinga relatively small signal amplitude and tight electric andmagnetic field coupling between the two differential lines,LVDS significantly reduces the amount of radiatedelectromagnetic noise and power lost to Diagram BasicsAs stated previously, the data eye Diagram isa representation of a high speed digital signal that allows keyparameters of the electrical quality of a signal to be quicklyvisualized and determined.

5 The requirements for high speeddata signals mentioned in the previous section are some ofthe key metrics that can be measured using eye diagrams are used to characterize a high speed signalsource or transmitter (receiver testing usually requires biterror rate testing). A typical eye Diagram test setup is shownin Figure 3 3. Eye Diagram Test SetupPulse GeneratorOscilloscopeDUTDataClockThe Pulse Generator is required to generate a random bitpattern since the eye Diagram is a statistical average of manythousands or even millions of samples of a pattern required for a particular data standard is definedby the protocol and is usually a pseudorandom bit sequenceof hundreds or thousands of bits. The high speed samplingoscilloscope has a typical bandwidth of 10 to 25 GHz inorder to capture the full characteristics of the the persistence mode of the oscilloscope,the superposition of millions of time-domain waveformscan be representative eye Diagram is shown below along withsome of the typical measurements that can be performed onthe 4.

6 Typical Eye Diagram MeasurementsAll of the measurement results are the statistical averageof the samples of the waveform at the point measurements are defined as follows:One LevelThe one level in an eye pattern is the mean value of a logicone. The actual computed value of the one level comes fromthe histogram mean value of all the data samples capturedinside the middle 20% (40 to 60% points) of the eye LevelThe zero level in an eye pattern is the mean value of a logiczero. The zero level is computed from the same 40 to 60%region of the baseline area during the eye period as the AmplitudeEye amplitude is the difference between the one and zerolevels. The data receiver logic circuits will determineswhether a received data bit is a 0 or 1, based on the HeightEye height is a measure of the vertical opening of an eyediagram. An ideal eye opening measurement would be equalto the eye amplitude measurement. For a real eye diagrammeasurement, noise on the eye will cause the eye to a result, the eye height measurement determines the eyeclosure due to noise.

7 The signal to noise ratio of the highspeed data signal is also directly indicated by the amount ofeye Crossing PercentageThe crossing level is the mean value of a thin verticalhistogram window centered on the crossing point of the eyediagram. The eye crossing percentage is then calculatedusing the following equation:Eye Crossing % = 100 * [(crossing level zero level)/(onelevel zero level)]Eye crossing percentage gives an indication of duty cycledistortion or pulse symmetry problems in the high speeddata signal. Figure 5 below shows an example ofa waveform with bad pulse symmetry (shown left) with theresultant eye Diagram with eye crossing of 75% (shownright). Eye crossing percentage is valuable for measuringamplitude distortions caused by differences in the one- andzero-level durations. It also reveals pulse symmetryproblems for diagnosis. When the eye crossing symmetryvalue deviates from the perfect 50% point, the eye closes andthus the electrical quality of the signal is 5.

8 Waveform and resultant eye Diagram with 75% eye crossing percentageBit PeriodThe bit period is a measure of the horizontal opening of aneye Diagram at the crossing points of the eye and is usuallymeasured in picoseconds for a high speed digital signal ( ,200 ps is used for a 5 Gbps signal). The data rate is theinverse of bit period (1/bit period). The bit period iscommonly called the Unit Interval (UI) when describing aneye Diagram . The advantage of using UI instead of actualtime on the horizontal axis is that it is normalized and eyediagrams with different data rates can be easily WidthEye width is a measure of the horizontal opening of an eyediagram. It is calculated by measuring the differencebetween the statistical mean of the crossing points of the TimeRise time is a measure of the mean transition time of thedata on the upward slope of an eye measurement is typically made at the 20 and 80 percentor 10 and 90% levels of the TimeFall time is a measure of the mean transition time of thedata on the downward slope of an eye measurement is typically made at the 20 and 80 percentor 10 and 90 percent levels of the is the time deviation from the ideal timing ofa data -bit event and is perhaps one of the most importantcharacteristics of a high speed digital data signal.

9 Tocompute jitter, the time deviations of the transitions of therising and falling edges of an eye Diagram at the crossingpoint are measured. Fluctuations can be random and/ordeterministic. The time histogram of the deviations isanalyzed to determine the amount of jitter. The p-p jitter isdefined as the full width of the histogram, meaning all datapoints present. RMS jitter is defined as the standarddeviation of the histogram. The units for a jittermeasurement on a high speed digital signal are normally Signals and Differential Eye PatternMeasurementsDifferential signals offer superior noise immunity andoverall improved signal integrity, which is highly desirablein the transmission and distribution of high - speed for conducting eye pattern measurementsinvolve using trace math operations to view these signalsboth separately and as a pair. On a modern high speedoscilloscope, eye and pulse pattern measurements can bemeasured separately (Ch1 and Ch2) and with trace math(Ch1 Ch2, Ch1 + Ch2).

10 By examining and overlayingmeasurements of the signal both separately and combined,differential and common mode effects including skewcaused by common mode imbalance and noise can Pattern Diagnostics and Mask ComplianceThe quality of a high speed digital signal can be quicklydetermined by using a compliance mask overlay on the eyediagram display. A typical mask includes both time andamplitude limits. An eye Diagram with compliance masks isshown in Figure 6 6. Eye Diagram with Compliance MaskIn the above Diagram , the gray block areas represent keep-out areas. In order to pass the mask compliance test,the transmitter output must not have ANY samples fallwithin the keep-out mask regions are defined as follows: Top region: The maximum expected value. Voltagesexceeding this value will fail. Middle Region: The size and shape of the eye diagramas defined by the interface standard for the high speeddigital signal. Bottom Region: The minimum expected below this value will Capacitance Products for high SpeedSignal QualityIt is vital for high speed interfaces to maintain impedancematching throughout the signal paths.