Statistical Method of near-end crosstalk report

Near-end crosstalk NEXT)

Most users carefully compare the numbers when viewing a technical data table or sales data, and make purchase decisions based on the cable performance data displayed on paper. However, in addition to the obvious performance data, the attenuation crosstalk ratio ACR is gradually becoming a new standard for measuring the cable performance of unshielded twisted pair wires.

ACR consists of two parts: attenuation and crosstalk. Attenuation is a parameter that can be calculated using an approximate equation. It remains unchanged between the line pairs of the entire cable. Its curves can be accurately expressed by several data points. Crosstalk cannot be accurately expressed using a curve. The currently adopted EIA/TIA TSB-36 specification is an equation that represents an approximate value of crosstalk that increases with frequency. However, there may be several dB differences in the crosstalk performance of different linewise pairs.

Many manufacturers have recently published standards on cable crosstalk performance. However, some of these tests may not accurately represent the performance of the cable. Next we will briefly describe the crosstalk problem and introduce a measurement method. At the same time, we will also recommend a method to accurately report the measurement results, it can more accurately report the crosstalk performance of the product.

Crosstalk

Crosstalk refers to the transfer of energy from one line to another. This transfer brings noise interference to the network system and shortens the effective signal transmission distance. Industry-standard measurements of this type of energy transfer are called near-end crosstalk NEXT ). It is measured by the receiver at the same end as the cable signal generator from one line to the other line to the energy transfer ratio.

Currently, the EIA/TIA TSB-36 and 568A draft list 11 NEXT frequencies. They are also known as "critical" frequencies, because in some encoding schemes, basic signals are also transmitted at these special frequencies. However, other important information is still transmitted over the bandwidth between these 11 frequency points. The signal strength of the information around the basic signal may be low. However, since NEXT is the ratio of power transfer, noise interference should be minimized to improve the signal resolution.

Measurement Technology

The following figure shows that the crosstalk track is composed of a series of peaks and troughs. The diagonal line is the specification for Category 5 cables. If the cable is measured only according to the 11 frequency points listed in the TSB-36, it has met the requirements. However, it should be noted that the equation in this requirement applies to the whole bandwidth.

Many cable manufacturers currently use network analyzer to measure electronic parameters. The Network Analyzer samples the bandwidth at a specified frequency, such as 201, 401, 801, and 1601. It does not specify the minimum number of data points that the manufacturer should select when testing cable crosstalk. In this way, the test may use a signal generator and power measuring instrument to generate 11 frequencies transmitted over a cable. In the draft 568A plan, it is recommended that at least 100 data points be detected every 10 hexadecimal notation. In decimal notation, the frequency increases by a factor of 10. For example, from 3 MHz to 30 MHz is a 10 hexadecimal notation ). The bandwidth of Category 5 cables is in the next 10-digit notation. Therefore, there are 300 data points for Category 5 cables. The current EIA/TIA 568 and TSB-36 documents do not specify the number of data points required for the exact test.

Evaluation Technology

With the increasing emphasis on ACR, some cable manufacturers have recently published information about the NEXT performance of their products. The measurement technology between major manufacturers is basically no different. However, they do not have a uniform standard on How to statistically analyze these numbers, so it is impossible to accurately measure the cable performance in different system configurations. Different vendors use different methods to represent the data. Due to the NEXT feature, the data may not accurately express the performance of the installed cable.

11 data points in the table

The first method to report crosstalk values is to use a list to list 11 frequencies and the measured values. This method has two inherent defects. First, each data point is generated in the middle window of the selected frequency. The window may only be Hz wide, but the customer may think of it as a much larger bandwidth. Return to Example 1. The cable in Example 1 does not meet the requirements of Category 5. However, if you only use the 11 data points, the cable has already met the requirements. Its second problem lies in the low slot or low point in the trajectory. These troughs are inherent in crosstalk measurement techniques and are affected by the sample length. The cable length varies, and its troughs also move at the frequency position. These troughs may reduce the crosstalk peak, or may be consistent with one of the 11 frequencies, and may obtain an ideal value of 30 dB less than the surrounding bandwidth.

Distribution Statistics

Another way to report crosstalk is to use a distribution Statistical Chart to display the expected NEXT margin performance of the cable. This method requires creating a font number from the cable sample. Use a selection process to generate a database from the database. The database then calculates the mean and standard deviation and generates a graph to simulate the distribution of cable performance. Theoretically, end users can analyze these charts to predict the percentage of cables that can pass the selected margin. This is a very useful tool, but if you cannot use the correct data for analysis, this tool cannot play a normal role. To prove this, we have used three different methods to analyze 30 samples of the Belden DataTwist 350 Product Model 1700A consisting of two different batches.

Method 1:

The statistical analysis uses 66 data points, which are related to line-to-line pairs.

The first method is to select 11 data points in each of the six possible cabling pairs to represent these "critical" frequencies. Input the worst case of each combination to the database regardless of the frequency to generate the standard deviation and average value. Assume a normal distribution, generate a cumulative distribution, and graphically represent it. In this case, the database is 180 data points, and each sample selects 6 points), and select 66 points for 30 samples from 1,980 points ). The mean and standard deviation were determined, and the result indicated in the Blue Line in the legend showed that 50% of the cables were dB higher than the minimum margin specified by Category 5.

This method is vulnerable to two major factors. First, the NEXT performance of the line-to-line pair combination may be 8 dB deviation. If cable manufacturers want to use only one set of cable pairs, this may be the result they want. However, if the combination of the special wire pair to the wire pair is not used, or if all the four wire pairs are in the network, the network will not benefit from this design improvement. The combination of special line pairs may artificially increase the mean remainder value. Secondly, this method does not take into account the frequency between data points. Therefore, it cannot accurately represent the worst condition of the cable in different system configurations.

Method 2:

The statistical analysis uses 66 data points, regardless of the line-to-line pair.

The second method can be used to avoid the influence of a good line on the measurement result of a combination. This time, 11 "critical" frequencies are measured, and the database contains 30 numbers extracted from 1,980 vertices. Each cable provides a worst case number regardless of the frequency and line pair combination ). Calculate the mean and standard deviation. The margin of 50 percentage points exceeds 8.2dB, which is indicated by the gray line in Figure 2. The remainder value of this 50-point reduction is 3.1dB. It is close to the real cable performance, but still ignores the frequency between the selected data points. According to the encoding scheme, there may be a lot of information.

Method 3:

More than 300 data points are used for statistical analysis, which is irrelevant to the line-to-line pair.

The green line in Figure 2 represents the same 30 cable samples, with each line pair having a frequency of over 300 tested. Each cable provides a worst case number in any frequency and line pair combination ). The database contains 30 numbers extracted from the 54,000 vertex set. Calculate the mean and standard deviation from the 30 selected points. At present, the margin of 50 percentage points exceeds 6.6dB, Which is 4.7dB lower than the result of method 1. Belden believes that this measurement method accurately represents the performance of the cable. It enables end users to determine the residual value of the expected worst case of the cable, regardless of the frequency, encoding scheme or wiring pin leads, such as wire-to-terminal configuration ).

Conclusion

With the increasing emphasis on ACR and Crosstalk in the industry, new methods for reporting cable performance have also been developed. In the past, the method of obtaining the list result from the 11 frequency point has little value. The 11 data points cannot accurately measure the overall bandwidth of the cable. Similarly, the distribution curve obtained from the macro data analysis process has lost value to customers. 11 The use of data points will produce the same error as the table. In addition, if each line pair combination provides a worst case point, some special line pairs will artificially increase the remainder value. Even if the number of data points is increased, a similar average value is obtained. The bandwidth is measured at 100 points every 10 times more, and the worst result is obtained for each cable, which accurately represents the performance of the cable, it is not affected by the combination of frequency and line. This more rigorous method is recommended by Belden.

Only by clarifying the cable evaluation process can you correctly select the required cable based on the data listed on the paper. It should be noted that "all frequency" refers to whether to measure more than 300 data points at a time or only 11 "critical" frequencies? What is the "worst case": indicates the most recent configuration of any line or the average margin of some special line pairs? The answer to these questions will directly affect the correct choice of cables.

1. When the link is too short, be careful about the crosstalk and ripple loss at the end.
2. Test parameters and Fault Analysis of Copper link in Integrated Wiring