Eye Chart Summary Report

Directory

1. Formation of eye images ......................................................................................................................... ............. 2

1.1 Traditional method of eye graph generation ...................................................................................................... 2

1.2 Real-time eye graph generation method .......................................................................................................... 3

1.3 Comparison of two methods  ................................................................................................................... 4

2. Description of the structure and parameters of the eye diagram  .................................................................................................................... 4

Structure of the 2.1 eye chart ......................................................................................................................... .. 4

2.2 Main parameters of the eye chart  ....................................................................................................................... 5

2.2.1 Extinction ratio ......................................................................................................................... ... 5

2.2.2 Intersection point ......................................................................................................................... ... 5

2.2.3 Q factor ......................................................................................................................... ..... 6

2.2.4 signal rise time, fall time ...... ..... ..... ................... ..... ..... ..... ..... .............. ........... 6

2.2.5 Peak-to-peak jitter and RMS jitter ....... ..... ..... .............. ....... ..... .................. ......... 6

2.2.6 Signal to Noise ratio ......................................................................................................................... ... 6

3. The relationship between eye chart and system performance  ................................................................................................................. 7

4. The relationship between eye image and BER  ......................................................................................................................... 7

4. How to get an open eye view  ..................................................................................................................... 8

5. Relevant knowledge of impedance matching  ..................................................................................................................... 9

5.1 Series Terminal Matching ......................................................................................................................... .. 9

5.2 Parallel Terminal Matching ......................................................................................................................... 10

6. Eye diagram Frequently Asked questions analysis  ....................................................................................................................... 10

7. Summary ......................................................................................................................... ....................... 17

1. Formation of eye images

The eye graph is a series of digital signals accumulated on the oscilloscope , the shape of the image is similar to the eye, it is called eye diagram.

When using the afterglow oscilloscope to observe the transmitted data signal, the time signal of the measured system is used to control the scanning of the oscilloscope by external triggering or external synchronization, because the scanning period is exactly the integer times of the measured signal period. Therefore, the observation on the oscilloscope screen is a stable pattern formed by multiple random symbol waveforms. This figure looks like the eye, called the digital signal of the eye.

The general signal that the oscilloscope measures is a bit or a period of time waveform, more reflects the details of the information. The eye graph reflects the overall characteristics of all digital signals transmitted over the link. Such as:

1.1 Traditional method of eye graph generation

The CLK of the sampling oscilloscope is typically a user-supplied clock, a recovery clock, or a code synchronization signal synchronized with the data signal itself.

Figure: The principle of eye diagram formation of sampling oscilloscope

1.2 Real-time eye graph generation method

The real-time oscilloscope completes all data sampling with a single trigger, without the need for additional sync signals and trigger signals. The clock is usually restored by the software PLL method.

Figure: Real-time oscilloscope eye diagram formation principle

Another type of:

Figure: Real-time oscilloscope eye diagram formation principle

1.3 Comparison of two methods

1. Traditional methods measure 100 to 1000 times times slower than real-time eye-graph production methods.

2. The traditional method of eye graph generation is not high in real-time eye graph generation.

3. The traditional method introduces the trigger circuit noise and the clock recovery circuit noise.

4. Traditional Methods Eye Chart analysis function is limited.

2. Description of the structure and parameters of the eye diagram

Structure of the 2.1 eye chart

Figure: Structural framework of Eye diagram

Note: When the binary signal is transmitted, the eye image has only one "eye", and when the ternary code is transmitted, two "eyes" are displayed. As shown in the following:

2.2 Main parameters of the eye chart

The main parameters of eye diagram are as follows: extinction ratio, intersection point, Q factor, rise time of signal, descent time, peak-to-peak jitter, RMS jitter, signal-to-noise ratio , etc.

2.2.1 Extinction ratio

The extinction ratio is defined as the value of the "1" level over the "0" level in the eye graph . According to different rates, different transmission distances, different laser types, the extinction ratio requirements are not the same. General for FP/DFB Direct-tuned lasers require extinction ratio of not less than 8.2dB. The extinction ratio of the EML electro-absorption lasers is not less than 10dB. There is no maximum value for extinction ratio in ITU-T, but this does not mean that extinction ratio can be infinitely large, extinction ratio is too high, will lead to the laser chirp coefficient is too large, resulting in channel cost exceeded, not conducive to long-distance transmission.

It is generally recommended that the actual extinction ratio and the minimum extinction ratio are large 0.5~1.5db. This is not an absolute value, the reason given such a value is afraid of extinction ratio is too high, the signal deterioration after transmission is too strong, resulting in error generation or channel cost exceeded.

2.2.2 Intersection point

The intersection is where two eyes intersect, and its proportions reflect the size of the signal's duty. As the transmission process, the pulse width of the optical signal will widen, resulting in the intersection of the receiving side relative to the transmitting side

Move. In order to facilitate long-distance transmission, to ensure that the receiver side of the cross-point ratio of about 50%, so that the sensitivity of the receiver side is the best, we generally recommend in the sending side of the intersection of the position slightly lower, the general transmission side of the cross-point ratio of the proposed control in the 40%~45%.

2.2.3 Q factor

The Q factor can be measured by the oscilloscope, which comprehensively reflects the quality of the eye graph. The higher the Q factor the better, the better the quality of the eye image. Q-factor is generally affected by noise, optical power, whether the electrical signal from the beginning to the end of the impedance matching (about impedance matching, after the detailed description) and other factors. Generally speaking, the line of the "1" level in the eye chart is thinner and smoother, the higher the Q factor. In the case of no light attenuation, the Q-factor of the sending side light eye graph should not be less than 12, the higher the better. The Q-Factor received should not be less than 6, the higher the better.

Rise time and fall time of 2.2.4 Signal

The rise time and falling time of the signal reflect the rising and falling speed of the signal, and generally refer to the time of the 20%~80% change of the whole signal amplitude. It is generally required that the rise and fall time should not be greater than 40% of the signal period. If the 9.95G signal is required to rise, the descent time is not greater than 40ps.

2.2.5 Peak-to-peak jitter and RMS Jitter

Peak-to-peak jitter and RMS Jitter can qualitatively reflect the jitter size of the signal, as a comparative reference, these two measured values are smaller and better. Quantitatively measured output jitter or to a special test jitter of the instrument, such as Agilint 37718, Acterna ant20-se in the measurement jitter when the instrument generally need to preheat more than 30 minutes to ensure that the measured value is relatively accurate.

2.2.6 Signal to Noise ratio

Snr can also qualitatively reflect the quality of the signal, as a comparative reference. The greater the value of this measurement, the better, generally in the transmitter side of the measured value is more than 30dB. Quantitative measurement requires the use of a spectrum analyzer.

3. The relationship between eye chart and system performance

Through the research and experiment can draw the overall system performance situation from the eye diagram, here we combine the eye chart frame to give the following conclusion.

Figure: The structure of the eye diagram

1. The best sampling time should be "eyes" open the most time;

2. The slope of the bevel of the eye image determines the sensitivity of the system to the sampling timing error, the larger the slope, the more sensitive to the timing error;

3. The vertical height of the shadow area of the eye image indicates the distortion range of the signal;

4. The horizontal axis of the center of the eye map corresponds to the ruling threshold level;

5. Over 0 points of distortion is the length of the shadow on the horizontal axis, some of the receiver's timing standards are determined by the average position of the decision threshold point, so the greater the distortion of 0 points, the more unfavorable to the timing standard extraction.

6. The half of the interval between the upper and lower shadow areas at the sampling time is the noise tolerance, and the noise instantaneous value is more than it can error judgment.

4. The relationship between eye image and BER

After research and analysis, we can get the relationship between eye graph and BER as shown below:

Using the eye image analysis jitter to get the bath tub curve (Bath Tub), the error rate estimation, the results as shown:

Figure: Eye image of the bath tub curve and error rate analysis diagram

4. How to get an open eye view

1) PCB Trace length

Short lines are not always available, but short-run lines mean low losses.

2) Line width

Wide walking line can reduce skin effect.

3) Reduce the dielectric constant of the plate

The dielectric loss (dielectric Loss) is reduced, but costs are increased.

4) Signal pre-emphasis and equalization processing

The loss of the high-frequency component caused by the signal hopping on the compensating line is treated by the pre-weighting of the jump-deflection (pre-emphasis).

5) Other receiver side equalization processing.

5. Relevant knowledge of impedance matching

Impedance matching is the specific mating relationship between the load impedance and the internal resistance of the power supply or the impedance of the transmission line wave. Impedance matching refers to the need for the load impedance to be equal to the characteristic impedance of the transmission line during energy transmission, where the transmission does not produce reflections, indicating that all energy is absorbed by the load. Conversely, there is energy loss in the transmission. In high-speed PCB design, whether the impedance match or not is related to the quality of the signal.

There are two common ways of impedance matching

5.1 Series Terminal Matching

Under the condition that the signal source end impedance is lower than the characteristic impedance of the transmission line, a resistor R is strung between the source and transmission line of the signal, which matches the output impedance of the source end with the characteristic impedance of the transmission line, and suppresses the reflection of the signal reflected from the load side.

Tandem matching is the most common method of terminal matching. It has the advantage of low power consumption, no additional DC load to the drive, and no additional impedance between the signal and ground, and only one resistor element is required.

Matching resistor selection principle: the sum of the matched resistor value and the output impedance of the actuator equals the characteristic impedance of the transmission line. Common CMOS and TTL drivers whose output impedance varies depending on the level of the signal. Therefore, for TTL or CMOS circuit, it is impossible to have a very correct matching resistor, only compromise. The chain topology of the signal network is not suitable for the use of series terminal matching, all the load must be connected to the end of the transmission line.

Common applications: General CMOS, TTL circuit impedance matching. The USB signal also samples this method to do impedance matching.

5.2 Parallel Terminal Matching

In the case of low impedance of the signal source, the input impedance of the load end is matched with the characteristic impedance of the transmission line by increasing the shunt resistor, which is to eliminate the load-end reflection. The implementation form is divided into two types: single and double resistance.

Matching resistor selection principle: In the case of high input impedance of the chip, the shunt resistance value of the load end must be similar or equal to the characteristic impedance of the transmission line in the form of a single resistor, and the value of each shunt resistor is twice times the characteristic impedance of the transmission line in the form of double resistance.

The advantages of parallel terminal matching are simple, the obvious disadvantage is that it will bring DC power consumption: the DC power consumption of the single resistor is closely related to the duty ratio of the signal, and the dual resistance mode has DC power consumption regardless of whether the signal is high or low, but the current is less than one half of the single resistance mode.

Common applications: high-speed signal application more.

6. Eye diagram Frequently Asked questions analysis

Let's take a look at some typical good eye graphs and some problematic eye graphs and analyze where these eye problems are.

The following is a better 622M of the light eye diagram, we can see more symmetrical, eyeliner is very thin, extinction ratio moderate, Q factor is high, reached 24.

The following is a 622 of the eye chart without the STM-4 filter, you can see the eye image of the liner is thinner, especially the rise and Fall Edge, "1" Level a little fluctuation, this is because the low-pass filter, the signal of high frequency

Harmonics are not filtered out, and each harmonic component is superimposed to form a corrugated square wave. We see that even if the "1" level is uneven, the Q factor still reaches 21.7.

The following is a better 2.5G, the eye is symmetrical, more thin eyeliner, "0", "1" level are relatively smooth, extinction ratio moderate, Q factor is higher.

The following is a better 10G eye diagram, symmetrical eye diagram, the eye image is relatively thin, especially "0", "1" level, rising, falling edge slightly thicker, visible signal jitter is large, extinction ratio is moderate, Q factor is higher. The intersection point is slightly higher, in actual debugging, you can turn the intersection down a little bit.

Overall, the higher the rate, the worse the quality of the eye image. This is mainly caused by two aspects: the first is jitter, the higher the rate, jitter is more difficult to control; the second is noise, because the test process is generally to add a corresponding low-pass filter, the 10G signal of the low-pass filter bandwidth of about 8ghz,622m signal low-pass filter bandwidth of about 500MHz, from 500mhz~ 8GHz This frequency range of noise, by the filter of the 622M signal is filtered out, but not by the 10G signal filter, so from the eye view, 10G signal noise greater.

Problematic eye-image analysis

The following is a problematic 622M eye graph, this eye graph problem is more, we come to analyze each:

First of all, the eye chart has a very obvious two rising, falling edge (commonly known as double eyelid); "0" level "1" level is not flat, the signal has overshoot, undershoot, low extinction ratio is only 4.1dB; The reason for these phenomena is that the impedance mismatch of the signal causes the signal to overshoot, undershoot and multipath. This eye chart also illustrates another problem, that the eye chart is able to cover the template is just the basic requirement of the eye diagram, not the only one. Let's see, the side of this eye chart is still a certain amount of margin from the template.

Let's look at the following 622M eye diagram, the problem is that the noise is very large, it is estimated that the signal filtering is not handled well.

The following is a 2.5G eye chart, the overall quality is good, the problem is that the eye image is a bit crooked, asymmetrical, this with the laser modulation characteristics have a certain relationship.

The following 2.5G eye chart, the problem is that the jitter is large, note compared with the previous eye graph, its rise, falling edge is thicker, pay special attention to compare its peak-to-peak jitter, RMS jitter, are larger than.

The following 2.5G eye image is worse, the eye image twist to twist, rise, fall are very slow, signal quality is not good, Q factor is only 6.4. The extinction ratio is also very low, only 6.6dB. This can be caused by a problem with the drive, the laser itself, or a very mismatched impedance.

One of the following is a 2.5G eye chart, you can clearly see the rising edge of the eye is ringing, there may be two reasons: the first is the signal line above impedance mismatch, the second is the edging oscillation caused by the direct-modulated laser ring.

The following is a 10G eye chart, there are two problems with eye diagram, the first extinction ratio is too low, only 10dB. Eye graph "1" level is very coarse, uneven. Possible causes are: signal mismatch caused.

The following 10G eye chart does not have its measurement data, but the rise and fall edge of the eye graph can be seen from the rough, the signal jitter is relatively large.

The following is a 10G eye chart, the problem with this eye graph is that the noise is relatively large, where do you see it? Please note that the eye image of the rise, Fall, "1" level are relatively coarse, the whole eye map scattered more, very dirty.

The above three eye graphs we analyzed three kinds of cases which caused the bad eye graph: Impedance mismatch, jitter, noise. How do these three kinds of situations look from the eye chart?

If the "1" level line of the eye graph is coarse and uneven (above), then the impedance mismatch is the cause. Solve the problem from the guarantee from the beginning to the endpoint impedance matching. If the rising and falling edge of the eye graph is thicker (in the middle), then it is caused by jitter, the problem should be reduced from the signal jitter, such as: improve the input clock quality, reasonable design phase-locked loop, especially the low-pass filter part started. If the eye image is more coarse (all), then is the noise caused by, in general, the power supply noise, the ground loop is not smooth or the signal around a large source of interference caused, to solve the problem is to start from these aspects.

For the eye chart, can not be measured by a ruler, the higher the rate, the more difficult to ensure the quality of the eye image; the farther the target transmission distance, the better the quality of the eye image is, and the optical module with the data and clock input is more than the image quality of the optical module with data input, especially in the jitter aspect;

7. Summary

In this study, the eye diagram is only a summary of the level of the report, but a more elementary understanding of the depth of the theory and application of the eye diagram needs further research and experiments.

Eye Chart Summary Report