Level III measurement accuracy of field testers for Twisted Pair Copper Cabling Explained in Technical Detail

Transcription

1 Level III measurement accuracy of field testers for Twisted Pair Copper Cabling Explained in Technical Detail 1 Outline Henriecus Koeman Fluke Networks July 9, 2001 This rather technical discussion contains detailed explanation of field tester measurement accuracy. It includes a discussion of test configurations, measurement accuracy related requirements, what the sensitivity of performance parameters is on overall measurement accuracy and worst case measurement performance per applicable standards. The following topics are addressed: What are the test configurations? (Permanent Link, Channel, Baseline) Parameters that have to be measured and over what frequency range. A reference to channel and permanent link pass/fail limits. Accuracy dependency on both field tester properties and link properties. The way accuracy is formally specified? 1. Individual performance parameters 2. Network analyzer comparisons What is every field tester performance parameter? Other aspects of field testers related to measurement accuracy. 1.1 Level III Field Tester Accuracy Summary First, the bottom line is shown. The worst case measurement accuracies for a Level III compliant field tester are shown in the following figures. The typical measurement accuracy is typically ½ of these values. 1

2 Insertion loss (attenuation) measurement accuracy at permanent link pass/fail limits for baseline and permanent link and channel pass/fail limits for the channel test configuration. 5 4 Accuracy in db 3 2 Channel Level III Accuracy Permanent link Level III Accuracy Baseline Level III Accuracy Frequency in MHz Figure 1: Insertion loss (attenuation) measurement accuracy NEXT loss, pair-to-pair, measurement accuracy at permanent link pass/fail limits for baseline and permanent link and channel pass/fail limits for the channel test configuration. 5 4 Channel Level III Accuracy Accuracy in db 3 2 Permanent link Level III Accuracy Baseline Level III Accuracy Frequency in MHz Figure 2: NEXT loss measurement accuracy 2

3 Return loss measurement accuracy at permanent link pass/fail limits for baseline and permanent link and channel pass/fail limits for the channel test configuration. 5 4 Permanent link Level III Accuracy Channel Level III Accuracy Accuracy in db Note: The 3 db rule generally applies at low frequencies. Baseline Level III Accuracy Frequency in MHz Figure 3: Return loss measurement accuracy ELFEXT, pair-to-pair, measurement accuracy at permanent link pass/fail limits for baseline and permanent link and channel pass/fail limits for the channel test configuration. 5 Channel Level III Accuracy 4 Permanent link Level III Accuracy Accuracy in db 3 2 Baseline Level III Accuracy Frequency in MHz Figure 4: ELFEXT measurement accuracy 3

4 2 Relevant standards The approved and draft standards that relate to field testing of twisted pair copper cabling are shown in Table 1. Standard Status Application Identification Freq. range IEC approved Class C and D cabling Level IIE MHz TIA/EIA-568B-2 approved Category 3 and 5E cabling Level IIE MHz IEC revision draft Class E added Level III MHz TIA/EIA-568B-cat 6 draft Category 6 added Level III MHz Table 1: Overview of field testing standards The standards that specify ISO/IEC Class E and TIA/EIA-568B Category 6 are still under development, and therefore the field testing requirements (Level III) are still under discussion. However, performance requirements for field testers are relatively stable and are being refined. 3 Test configurations The test configurations for which pass/fail limits are specified are: Permanent link Channel This is now true in both international and TIA specifications (starting with TIA/EIA-568B, approved in March 2001). Previously, the TIA recognized the Basic Link (as in TSB-67). This is now formally obsolete after the approval of TIA/EIA-568B. The baseline performance requirements are an indication of performance of the field tester by itself at the high performance reference plane of measurement, without the use of a suitable adapter to make a permanent link or channel measurement. As will become clear in this document: the overall measurement performance is highly dependent on the properties of the adapter that is used, as well as the signal processing capability of the field tester. 3.1 Channel Both international and US standard specify that the following connector and component segments can make up a channel configuration. Starting out at the equipment room or closet, where the jack on the equipment is mated with the plug of the user patch cord (see Figure 5 for a schematic diagram): Equipment in Floor Distributor a b c d e Terminal equipment in work area C2 C1 CP TO Figure 5: Schematic diagram of a channel configuration 4

5 A segment of patch cable a to a connection. This connection may be part of a cross-connect system (we label that C1 in that case) or a patch panel, which makes a connection to horizontal cable (in that case it will be directly to a connector labeled C2). The connection to either the cross-connect connector C1 or connection to the horizontal cable C2. A jumper cable b between C1 and C2, in case a cross-connect system is used. If there is no cross-connect, this jumper cable b is not present. We call a system without a cross-connect an interconnect cabling system. Connection C2, which is the entrance to the horizontal cabling. Horizontal cable c. This is generally the long segment of cable in a link. An optional consolidation point connector CP. This was previously called a transition point. This optional connection CP is sometimes used in Open Office Cabling systems. If there is this optional consolidation point CP, there is may be another segment of cable d to a telecommunication outlet TO. A telecommunications outlet connector TO. A segment of patch cable e to make the connection to the workstation (network interface card). There are really two versions of an Open Office Cabling system. This first type uses at the location of the CP an Multi-User-Telecommunication Outlet Assembly (MUTOA), and then a long patch cable, up to 20 m long to the work station. In this case, the MUTOA provides the function of a regular TO. The second type is as shown in the right portion of Figure 6: a CP is used, followed by segments of cable d to a telecommunications outlet to make the final connection to the work station. Often one finds 25 pair cabling between the equipment room or closet to the CP. At the location of the CP, the 25 pair cable transitions in individual 4-pair cables to the TO s near the workstation locations. A very important observation The plug of the user patch cord attached to cable segments a and e and which provides the connection to the equipment at either end is considered to be part of the equipment, and its impact on signal transmission is NOT considered part of the channel. Therefore, the channel starts or ends where the patch cable exits the plug mating with equipment. This is shown in a simplified schematic diagram in Figure 6. Equipment (hub) Beginning of channel user supplied test cords Simplified channel End of the channel Figure 6: Formal definition of the channel configuration Equipment (workstation) 5

6 The reason for this apparently odd definition is that the electrical performance of the equipment is tested with a plug mated with the equipment (you cannot test the equipment without anything connected to it!). The unfortunate consequence of this definition is that when testing the electrical performance of a channel, it is necessary to measure the channel through the standard modular 8-pin connector, but without the degradation of the measurement of this connection to the test equipment at both ends. This actually does have a considerable influence on measurement accuracy and field tester manufacturers to some degree have been successful in eliminating this connection from influencing a test result. It is a big deal and more about this later. 3.2 Permanent link Schematic diagram of the current ISO/IEC and TIA (per TIA/EIA-568B) permanent link is shown in Figure 7. Beginning of the (obsolete) basic link Local test unit Beginning of the permanent link Field test manufacturer supplied test cords High Quality Test port High Quality Test port Remote test unit Permanent link End of the permanent link End of the (obsolete) basic link Figure 7: Formal definition of the permanent link A number of observations of interest: The goal of the permanent link definition was to define a configuration, which was as permanent as possible, and subject to changes to the absolute minimum extent possible. Patch cords can be exchanged rather easily, so the main aspect was to minimize its impact by excluding it when acceptance testing the cabling. Therefore, the key difference between the (obsolete) basic link (see the next section for a detailed description) and permanent link definitions is that the reported result must not include the impact of the cable portion of the adapter cord. This is in addition to the mating connector with the field tester and its remote unit. However, the question remains on the plug that mates with the first jack of the permanent link. The plug of the test adapter to the first jack of the permanent link remains to be included in the definition of the permanent link, and its impact must be included in the reported results. Contrary to the basic link definition (see the next section for a detailed description), an optional connection (this is the consolidation point) is allowed in the permanent link. More about this later. Again, the performance of the plug that is part of the test adapter affects the reported result for the permanent link, and must be tuned to the cabling components of the permanent link. 6

7 The adapter of the field tester affects the performance of the permanent link that is measured. Possible permanent link configurations When referring to Figure 5 showing the channel, you can easily identify the portions which may be considered permanent (see Figure 8 for schematic diagrams). c C2 c TO C2 CP c d C2 CP TO Figure 8: Possible permanent link configurations The first option is the most simple one: you just have in the equipment room or closet a patch panel and in the work area you have a telecommunications outlet. This permanent link includes two connections and one segment of cable (which can be up to 90 m long). The second option is also like the top one. In this case, your installer only tests up to the consolidation point. Later you may add the Open Office cable to a telecommunication outlet. All connectors are subject to the same performance requirements and therefore from a performance point of view the second configuration is subject to the same test limits as the top configuration in figure 5. The last option shows the implementation with a consolidation point (CP). This causes an additional connector in the permanent link. During the measurement of some parameters this is a significant contributor (NEXT and return loss) and during other measurements its impact is minor. The pass/fail limit requirements for all permanent link configurations are the same and independent of the number of connectors in the permanent link. There are discussions within the ISO/IEC standardization working groups on making it possible to perform incremental testing: First test the cabling between C2 to the CP (this test configuration is called the CP link). Then extend the CP link to a permanent link (from C2 to the TO), or Extend the CP link to a channel test configuration. The difficulty is to provide consistent pass/fail limits for each test configuration. The CP link pass/fail limits need to be tighter than those for the permanent link and allow for variations in mated plug/jack performance. This topic is still being discussed. 7

8 3.3 Basic Link (baseline) The approved TIA standard TIA/EIA-568B no longer recognizes the basic link as a test configuration. The international ISO/IEC standard has never recognized the basic link. The obsolete TIA standards include TSB-67, TSB-95 and Addendum #5 to TIA/EIA-568-A for Enhanced Category 5 cabling. As long as return loss was not included as a measurement parameter, there was no real difference between testing a basic link and a permanent link, since virtually all test adapters by field tester manufacturers used cable with individually shielded wire pairs and thereby avoiding any amount of NEXT. However, return loss effects in adapter cable cannot be avoided and represents a potentially very significant source of measurement error. Major developments in the design of permanent link adapters have occurred recently. For reference purposes, the basic link test configuration is described here. There is an important difference between the basic link and permanent link test configurations, with a major impact on measurement accuracy in many cases. The configuration of the TIA basic link is shown in Figure 9. Local test unit Beginning of the (obsolete) basic link Field test manufacturer supplied test cords High Quality Test port Remote test unit Basic link End of the (obsolete) basic link High Quality Test port Figure 9: Formal definition of the TIA basic link A couple of observations of interest: Since the patch cord used to connect the field test instrument and its remote unit is supplied by the manufacturer of the field tester, there is no requirement that the mating connector to the field tester is of the modular 8-pin, RJ-45 type. Most often, the high quality test port of the field tester is also the location where the performance of the field tester is calibrated and therefore the highest in terms of accuracy. To really show off the measurement accuracy of the field tester, the accuracy in this test configuration is often emphasized. Standards now refer to this performance as baseline accuracy. The baseline accuracy has become essentially irrelevant for field testing purposes, because the measurement performance with permanent or channel adapter is critical. The cable portion and plug mating with the first jack (internationally often the word socket is used) is formally part of the basic link. The field tester manufacturer therefore influences the quality of the basic link as it is reported. By selecting patch cable with high performance, this enhances the overall outcome, in particular for the return loss property. Also, by selecting a plug that mates electrically well with the jack, that will enhance the reported performance, in 8

9 particular for the near-end-crosstalk (NEXT) measurement. Of major importance therefore is the quality of the adapter which the field tester manufacturers deliver with their test instrument. These adapters have to imitate to the degree possible, the highest quality of cabling used by the end user. The basic link definition does NOT allow for any connection in the middle (the consolidation point (CP) or transition point (which is the old name for consolidation point). 3.4 Special rules for Open Office cabling There are two versions of Open Office Cabling. In each case, often 25-pair cable is used to connect the equipment in the telecommunications outlet or room to a central location in the work area. Since the space needed for a single 25 pair cable is less than six (6) 4-pair cables, this is often considered an attractive solution. The one version uses a Consolidation Point (CP) (see Figure 10) and another segment of horizontal cable often called the CP cable to a Telecommunications Outlet (TO). HUB or other equipment Patch panel or crossconnect (not shown) Equipment Room/ Floor Distributor Connections to other equipment Cable bundle or 25-pair cable Consolidation Point Telecommunications Outlet Workstations Work Area Cable bundle or 25-pair cable Connections to other telecommunications outlets and workstations Figure 10: Principle of implementation of Open Office Cabling: use of Consolidation Point (CP). Testing this configuration is subject to discussions in the ISO/IEC working group. In this situation, the desired configuration to be tested is the permanent link from the equipment room/floor distributor to the TO, so that a compliant patch cord can be added with confidence that the channel performance is achieved. It will generally not possible to pre-test the CP link and expect no failures when the CP cable between CP and TO is added. 9

10 The configuration is subject to the same rules as for a permanent link or channel (the basic link does not allow a CP). The second version of Open Office Cabling uses a Multi-User Telecommunications Outlet instead of the CP and TO, and uses long (up to 20 m) patch cords directly to the workstation (see Figure 11). Same circuit arrangements as in Figure 10. Multi User Telecommunications Outlet Assembly (MUTOA) Workstations Work Area Cable bundle or 25-pair cable Connections to other workstations Figure 11: Principle of implementation of Open Office Cabling: use of Multi-User Telecommunication Outlet Assembly (MUTOA) Testing this configuration is subject to discussions in the ISO/IEC working group. This situation is addressed by extending the CP link to the channel end configuration. In this case, long patch cables may be used. The maximum length can be determined by subtracting the measured insertion loss (attenuation) from the channel pass/fail limit and dividing the difference by the insertion loss (attenuation) of patch cable per unit length. 10

11 4 Link parameters to be measured 4.1 Link parameters Table 2 contains an overview on the field parameters that need to be tested per ISO/IEC and TIA/EIA-568B and related draft standards. Standard: ISO/IEC TIA/EIA 568B ISO/IEC Status Approved Approved Draft Draft Class or Category Cl. C: 16 MHz Cat 3: 16 MHz Frequency range Cl. D: 100 MHz Cat 5e: 100 MHz Cl. C: 16 MHz Cl. D: 100 MHz Cl. E: 250 MHz Cl. F: 600 MHz Addendum to TIA/EIA 568-B Cat 3: 16 MHz Cat 5e: 100 MHz Cat 6: 250 MHz Wire Map x x x x Length x x Propagation delay x x x x Delay skew x x x x Insertion loss x x x x (attenuation) PP NEXT loss x x x x PS NEXT loss x x x x PP ACR x PS ACR x PP ELFEXT x x x x PS ELFEXT x x x x Return loss x x x x DC resistance x x Table 2: Overview of parameters that must be measured. 4.2 Pass/fail limits The pass/fail limits for each test configuration of the ISO/IEC and TIA standards can be found in the Expert Copper Cabling section of the Fluke Networks WEB site. They are regularly updated per the most recent drafts. 11

12 4.3 Brief description of link parameters Table 3 contains a summary description of each parameter to be measured. Parameter Description Wire Map Verifies proper connectivity of pairs 1,2 3,6 4,5 and 7,8 Length This is the physical length of the link. Propagation delay The travel time of the signal from the beginning to the end of the link. It is based on the phase shift of a 10 MHz sinusoidal signal. Delay skew The maximum difference of travel times between any two wire pairs. Insertion loss Loss of signal along the length of the link (attenuation) PP NEXT loss Pair-to-pair Near-end-crosstalk: the coupling from one wire pair to another measured at the beginning of the link. PS NEXT loss Power sum Near-end-crosstalk: the power sum of pair-to-pair NEXT couplings to a wire pair from all other wire pairs. This is a computed value from the pair-to-pair NEXT loss results. PP ELFEXT Pair-to-pair Equal level far end crosstalk: the coupling of a disturber pair to a disturbed pair measured at the remote end of the link relative to the insertion loss of the disturbed pair. PS ELFEXT Power sum Equal Level far end crosstalk: the power sum of all ELFEXT disturbers on a certain wire pair. This is a computed value from the pair-topair ELFEXT measurements. Return loss The reflected signal on a given wire pair. PP ACR Pair-to-pair attenuation-to-crosstalk ratio. This is a computed value from pair-to-pair NEXT loss and insertion loss. PS ACR Power sum attenuation-to-crosstalk ratio. This is a computed value from power sum NEXT loss and insertion loss. DC resistance DC loop resistance of a wire pair. Table 3: Summary description of each parameter to be measured. 12

13 4.3.1 Wire Map Measurement requirements The Wire Map test includes the test for proper connectivity. For the modular plug the correct connectivity has been specified as shown in figure 6. Two standard configurations have been called out: 568A and 568B. Electrically there is absolutely no difference: pair 1,2 is always connected to pair 1,2, etc. The color codes for the wires differ however. The Wire Map test contains a test for split pairs. Split pairs do have correct pin-to-pin connectivity, but the wires are not twisted together correctly in a pair. Refer to figure 7 for illustrations of correct and incorrect connectivity. TIA-568A shows 568A to be the preferred color coding. Modular jacks and punch down blocks generally show color coding. Plug color codes per 568A, locking tab is up 1 White/Green 1 2 Green 2 Pair White/Orange Blue 4 5 Pair White/Blue Orange White/Brown Brown 7 8 Pair 2 Pair 4 Plug color codes per 568B, locking tab is up 1 2 White/Orange Orange White/Green Blue White/Blue Green White/Brown Brown 7 8 Pair 2 Pair 1 Pair 3 Pair 4 Figure 12: Connectivity and color code assignments per 568A and 568B Note that it is not sufficient to measure DC resistance. An ac signal is needed in order to detect split pairs (connections to twisted pairs are mixed up at both ends). See Figure 13 for an example of a split pair situation. Other errors, like open circuits and shorts are readily detected by pulse reflections. 13

14 Figure 13: Examples of correct and incorrect wiring The reporting requirement for the Wire Map function is that the wire map is correct and as shown in Figure 13, graph at the left Length measurement requirements Length is expressly defined as the physical length of the cable. Field testers attempt to measure the propagation delay of an electrical signal in the cable and relate the measured delay to the length of the cable. Of course, the propagation delay time depends on the speed of the signal, which is expressed as a percentage of the speed of light. This quantity is called the Nominal Velocity of Propagation (NVP). Since the signal has to travel up and down the cable, the equation for length is: MeasuredTimeDelay* NVP* SpeedOfLight Length = 2 The Speed-of-Light is 300,000,000 meters/second. Practically the speed of the signal is approx. 0.2m/ns or 8 inches/ns (NVP is approximately 67%). Problems with the measurement of physical length by electronic means are: The speed of travel of electrical signals varies widely from lot to lot of cable. Differences of 10% are quite possible. The shape of a TDR pulse changes considerably as it travels to the end of the cable and back, and therefore it is not always easy to measure the time delay accurately. This is a problem for the tester and the length measurement accuracy requirement for the tester includes this factor. The pairs in a 4-pair cable all have different twist lengths in order to improve crosstalk performance, therefore the propagation delays are different. The (electrical) lengths of the pairs are based on propagation delay and are different as a result. Differences of 5% are quite common. However, the physical lengths of the pairs inside the jacket is the same. Standards specify that the shortest length must be used for NVP calibration and PASS/FAIL testing. The speed of travel is actually slightly dependent on frequency. For purposes of reporting the phase delay is measured at 10 MHz. Since the period of a 10 MHz signal is 100 ns, multiply the phase delay and 100 ns to obtain the propagation delay. 14

15 Calibrate Cable must be performed by the installer/user of test equipment Test standards recommend that the NVP of every new spool of cable is measured (use the calibrate cable function: measure 300 m or 1000 feet of cable and adjust the reported length to 300 m or 1000 ft). After cable from this spool has been installed, the length should be measured using the NVP of the spool that the cable came from and the NVP, as well as other measured data, be recorded in the cable administration system. NVP calibration improves the accuracy of the length measurement substantially (to a few percent). However, it is recognized that few users of test equipment actually maintain the discipline of properly measuring the NVP before installation and more often than not, the value of NVP and length of each link is not recorded in the cabling documentation of a building. The requirements for PASS/FAIL are therefore adjusted, so that anything less than the maximum length of the link + 10 % (110 m for a channel and 99 m for a permanent link) is a PASS. Anything in excess of the 10 % allowance for NVP uncertainty is a FAIL. The field tester adds an additional uncertainty of better than ±1 meter ±4 %. If the measured result is closer to the test limit than the accuracy of the tester, an asterisk is added to the PASS or FAIL result. This represents a warning to the user/installer, meaning that the outcome of the length test cannot be relied upon: a PASS may actually be a FAIL or a FAIL may actually be a PASS. Test standards also state that the pair with the longest twist length (therefore with the shortest electrical delay) shall be used to calibrate the cable. The test instrument must be able to measure all pairs, and report the length of the pair with the shortest delay (typically the same pair that was used when measuring the NVP). This length should apply to all wire pairs. (Again, propagation delays will differ!) The nominal test limits for length (TIA only) are: 100 m + 10% = 110 m for a channel 90 m + 10 % = 99 m for a permanent link 4.4 Minimum reporting requirements The minimum reporting requirements by field testers are specified in IEC , TIA/EIA- 568B.2 and applicable draft standards. The requirements include: Ability to download all datapoints Provide a minimum summary report output. The requirements for the minimum summary report output are shown in Table 4. It is generally sufficient to retain the summary data. The detailed data is very useful when 1) It includes diagnostic information, such as time domain responses (time domain NEXT and time domain reflections (return loss)). 2) A failure needs to be diagnosed. Natural variability, that the result of the mating different plugs (of a test adapter or user patch cord) with the jack (of the link under test) is generally high enough to limit reproducibility. Refer to section 5.4 for more information. 15

16 Function Wire Map Insertion loss (attenuation) Length (ISO/IEC: not required) NEXT loss pair-to-pair (ISO/IEC only: If PP NEXT loss fails, ACR must pass to obtain a passing test result) NEXT loss power sum (ISO/IEC only: If PS NEXT loss fails, ACR must pass to obtain a passing test result) ELFEXT pair-to-pair Measured from either end (if measurement from both directions is not required) All connectivity, including shields (if present) Pass/fail Worst case insertion loss (1 of 4 possible) Test worst case worst case Pair with worst case Pass/fail Length Test limit Pass/fail Worst case margin (1 of 6 possible) Test worst case margin worst case margin Pair worst case margin Pass/fail AND Worst case (1 of 6 possible) Test worst case worst case Pair combination with worst case Worst case margin (1 of 4 possible) Test worst case margin worst case margin Pair with worst case margin Pass/fail AND Worst case power sum NEXT loss (1 of 4 possible) Test worst case worst case Pair with worst case Worst case margin (1 of 12 possible) Test worst case margin worst case margin Pair worst case margin (disturber, disturbed) Pass/fail AND Worst case pair-to-pair ELFEXT Test worst case worst case Pair combination with worst case Measured from opposite end (if measurement from both ends is required) Worst case margin (1 of 6 possible) Test worst case margin worst case margin Pair worst case margin Pass/fail AND Worst case (1 of 6 possible) Test worst worst case Pair combination with worst case Worst case margin (1 of 4 possible) Test worst case margin worst case margin Pair with worst case margin Pass/fail AND Worst case power sum NEXT loss (1 of 4 possible) Test worst case worst case Pair with worst case Worst case margin (1 of 12 possible) Test worst case margin worst case margin Pair worst case margin (disturber, disturbed) Pass/fail AND Worst case pair-to-pair ELFEXT Test worst case worst case Pair combination with worst case ELFEXT power sum Worst case margin (1 of 4 possible) Test worst case margin worst case margin Pair combination with worst case margin (disturber, disturbed) Pass/fail AND Worst case power sum ELFEXT Test worst case worst case Pair combination with worst case Worst case margin (1 of 4 possible) Test worst case margin worst case margin Pair combination with worst case margin (disturber, disturbed) Pass/fail AND Worst case power sum ELFEXT Test worst case worst case Pair combination with worst case Table 4 Field tester summary reporting requirements 16

17 Function Return loss (pass/fail does not apply when insertion loss (attenuation) is less than 3 db) PROPAGATION DELAY Delay skew Measured from either end (if measurement from both directions is not required) Worst case margin (1 of 4 possible) Return loss Note worst case margin. Test worst case margin. worst case margin Pair with worst case margin Pass/fail AND Worst case return loss (1 of 4 possible) Test worst case Return worst case worst case Pair with worst case Worst case delay (1 of 4 possible) Test worst case Pair with worst case Pass/fail Worst case delay skew (1 of 1 possible) Test limit Pass/fail Measured from opposite end (if measurement from both ends is required) Worst case margin (1 of 4 possible) Return worst case margin. Test worst case margin. worst case margin Pair with worst case margin Pass/fail AND Worst case return loss (1 of 4 possible) Test limit at worst case Return loss at worst case Frequency at which worst case occurs. Pair with worst case DC resistance (TIA: not required) Worst case DC resistance (1 of 4 possible) Test limit Pass/fail Table 4 (continued) Summary reporting requirements for field testers 5 Measurement accuracy requirements 5.1 Types of requirements for field tester measurement accuracy There are two types of requirements specified in applicable standards: Specifications for field tester properties, that significantly affect measurement accuracy. The standards contain formulas how to obtain a computed measurement accuracy from the performance of each relevant property. Methods to compare the results obtained with field testers with results measured on the same link using laboratory equipment (i.e., network analyzers). This results in an observed measurement accuracy. The applicable standards stipulate that the computed accuracy must be in agreement with the observed measurement accuracy. This stipulation solves the following dilemna: There is no guarantee that there are no other properties than those specified in the standard that significantly affect measurement accuracy. This of course depends on the way the measurements are implemented. If an unspecified parameter affects measurement accuracy substantially, the results of comparing field tester results with those from laboratory equipment will likely be in disharmony with the computed measurement accuracy. When comparing the results of field testers with those of laboratory equipment, the question is how many links must be tested before concluding that a certain measurement accuracy is actually achieved. Requiring that critical performance parameters are met reduces the probability that insufficient links have been tested to establish compliance. 17

18 5.2 Computed measurement accuracy specifications Dependency of link properties on measurement accuracy The measurement accuracy not only depends on the properties of the field tester, the measurement accuracy depends also on properties of the link under test. The dependency of measurement accuracy on both link properties and field tester properties is shown by using a simple DC volts measurement example. A voltmeter with a 10 kω internal impedance is used to measure a battery that has an unknown internal impedance. This internal impedance may vary and can be as high as 1 kω depending on its state of charge. The accuracy of the voltmeter is better than 1 % when connected to a known voltage (with a very low source impedance!). What voltage does the voltmeter measure, assuming that the true battery voltage is 1 V? Ignoring the 1 % accuracy of the voltmeter, the indication may be as low as 0.91 V. This is a variation of 10 %. In addition, there is the accuracy of the voltmeter itself and the answer can therefore be as low as 0.9 V. So, while the accuracy of the voltmeter may be specified at 1 %, the measurement accuracy is far worse as a result of the unknown source impedance of the battery. < 1 kω voltmeter 10 kω V? battery 1 V Figure 14: Illustration of tester accuracy versus measurement accuracy For the RF measurements of cable testing, the situation is considerably more complex and more quantities than just the internal impedance of the voltmeter and the source impedance of the battery are involved Properties of links under test affecting measurement accuracy Since Level III measurement accuracy for testing ISO/IEC Class E or TIA Category 6 links is of highest interest, this discussion addresses only measurement accuracy for these conditions. The assumed link properties as defined in applicable standards are shown in Table 5. Parameter Return loss of permanent link Return loss of the channel Value (in db) 1 MHz 40 MHz: 26-5*log(f), 21 db max 40 MHz 250 MHz: 34-10*log(f) 1 MHz 40 MHz: 24-5*log(f), 19 db max 40 MHz 250 MHz: 32-10*log(f) Differential to common mode conversion gain 10 Common to differential mode conversion gain 5 Table 5: Properties of links affecting measurement accuracy 18

19 Return loss of the link under tests interacts with the source/load return loss properties of the tester, and thereby affects measurement accuracy. Normally crosstalk effects are measured by applying a differential mode signal to a wire pair and measuring the differential mode signal on a different wire pair. In reality, some portion of the signal that is applied couples as a common mode signal in that different wire pair and as a result of limited Common Mode Rejection (CMR) manifests itself as an additional differential mode signal. Similarly, if the signal is not applied in a perfectly balanced manner (the measure for this is Output Signal Balance), a common mode signal on one wire pair causes a differential mode signal (which is measured) on the other wire pair. Both coupling effects are stronger than the differential-to-differential mode coupling (estimated at 10 db and 5 db respectively), and thereby affect the measurement accuracy. 5.3 Observed measurement accuracy The standards define in great detail how to make the comparison of field tester results with those obtained with laboratory equipment. The principle is straightforward, but the implementation rather difficult as a result of differences in the connections. In each RF measurement a reference plane of measurement is present, and it is essential that the reference plane of measurement (for example, the locations where the permanent link or channel starts and where it ends). To conduct these tests, special adapters and test programs may be needed, which are generally not made available by field tester manufacturers. More details can be provided on this topic. 5.4 Natural variability of results, not included in accuracy specifications The sources of variability that are NOT included in accuracy specifications are the variations of mated performance when a jack of the permanent link under test is mated with the plug attached to the permanent link adapter (formerly the basic link adapter). From this discussion it will also become evident that the channel test results can and often will be highly dependent on the properties of the user patch cords. The variability depends on the parameter that is measured : Insertion loss Variability is small relative to measurement accuracy. NEXT loss Substantial, refer to the discussion that follows. ELFEXT Not evaluated, but considered less critical, since ELFEXT margins are generally very high and measurement accuracy is therefore not very important. Return loss Significant, refer to the discussion that follows. 19

20 5.4.1 NEXT loss variability The variability during NEXT loss measurement with different, compliant adapters can be estimated by recognizing the fundamental aspects of mating a plug and a jack, see Figure 15. The mated connection is 54 db = 1/8 of the plug by itself! The plug by itself is 36 db Figure 15: NEXT loss compensation in a mated connection Starting with a plug with MHz NEXT loss (which is the approximate de-embedded NEXT loss value of a reference test plug, as specified in applicable standards), the category 6 mated connection has to meet MHz NEXT loss. This is only 1/8 of the NEXT loss of the plug by itself. There can be an equal amount of overcompensation as undercompensation 180 out of phase, assuming that both mated connections are compliant with the requirement (e.g, MHz). When using the first measurement as a reference, the worst case change in the second measurement is twice the specification for the mated connection. This is worst case, of course. A more reasonable assumption is that the amount of change is reflected by a single connection. The impact of this additional connection on the MHz pass/fail limit level of a TIA category 6 or ISO/IEC Class E permanent link is approx. 1.9 db. This variability is even higher than the measurement accuracy of the field tester. However, this variability is currently not recognized in the standards for the purposes of estimating the accuracy band (commonly called grey zone ). Given typical NEXT loss measurement accuracy of a field tester with permanent link adapter as shown in section 3, it is very well possible that results obtained using different adapters do not agree within the accuracy specifications of the field tester (even exchanging the local and remote adapter may cause these changes) Return loss variability The considerations for variability of return loss are similar to those for NEXT loss. This time the variability is considerably less pronounced, although not insignificant relative to measurement accuracy that is often achieved by a field tester. See Figure 16. The mated connection is 24 db = 2.25 x the plug by itself! The plug by itself is 31 db Figure 16: Return loss compensation in a mated connection 20

21 In this case, the return loss of the plug has to be within 30 db and MHz (per TIA/EIA-568B.2). The return loss specification for the mated connection is MHz. The variability of the connection can therefore be represented by another connection with MHz return loss. This causes a change of approx db at the permanent link pass/fail limit Additional variability in channel test configurations The user patch cord is formally part of the channel that is tested, and must not be changed (no change at all is allowed after certifying a channel test result, not even reversing the orientation of the user patch cord!). One can understand the following changes in performance that might occur, if a change is made. The performance of the key transmission parameters NEXT loss and return loss is discussed. NEXT loss changes due to changing patch cords in a channel configuration Changes due to the different patch cable (just the cable portion of a patch cord assembly!) are minimal. The change at a nominally compliant channel, using compliant patch cable will be as low as MHz or less. Changes due to the different matings of plug at the end of the user patch cord and jack of the link are as discussed for the permanent link (approx MHz for a nominally compliant link; higher if the channel performs better than nominal). Return loss changes due to changing patch cords in a channel configuration Changes due to different patch cable (just the cable portion of a patch cord asembly!) can be substantial and is highly influenced by the characteristic impedance properties, as reflected in return loss specifications at low frequencies. In fact, a patch cord by itself can cause a return loss failure in the channel by itself, even if the remainder of the cable has excellent return loss performance. This is particularly often the case near 20 MHz, where the ¼ wavelength matches the typical length of a patch cable. This behavior is also the reason why permanent link testing is preferable over basic link testing: avoided is the variability that is caused by the return loss properties of basic link adapters. Changes due to the different matings of plug at the end of the user patch cord and jack of the link are as discussed for the permanent link (approx MHz for a nominally compliant link; higher if the channel performs better than nominal). 21

22 5.5 Properties of the field tester affecting measurement accuracy The properties of field testers that affect measurement accuracy and are recognized in applicable standards and Level III performance requirements are shown in Table 6. Parameters Baseline (db) Permanent Link (db) MHz MHz: 150 khz MHz: 250 khz 100 MHz 250 MHz: 500 khz Frequency range Frequency resolution (minimum resolution, all parameters except insertion loss; insertion loss resolution: 1 MHz) Dynamic accuracy Insertion loss 0.75 Dynamic accuracy NEXT loss 0.75 Dynamic accuracy FEXT loss Note Channel (db) Source/load return loss *log(f/100) 20 db max *log(f/100) 20 db max Random noise floor 75-15*log(f/100) Residual NEXT Note *log(f/100) test up to 85 db 54-20*log(f/100) test up to 85 db Residual FEXT Note *log(f/100) test up to 85 db *log(f/100) test up to 85 db Output Signal Balance 40-20*log(f/100) test up to 60 db 37-20*log(f/100) test up to 60 db Common Mode Rejection 40-20*log(f/100) test up to 60 db 37-20*log(f/100) test up to 60 db Tracking 0.5 Directivity 27-7*log(f/100) 30 db max 25-20*log(f/100) 25 db max Source Match *log(f/100) 20 db max Return loss of termination 20-15*log(f/100) 25 db max 16-15*log(f/100) 25 db max Note 1: It is assumed that the dynamic accuracy of insertion loss and FEXT loss measurements add up to 1 db total dynamic accuracy for ELFEXT. Note 2: The Level III residual NEXT requirements for the channel test configuration 54 20log(f/100) should take into consideration reflected FEXT (in the mated instrument connector) effects, which appear as residual NEXT. The standard needs to be clarified. Note 3: The current draft of the TIA Level III requirements shows a residual FEXT loss of 65-20log(f/100) requirement. This is to be corrected to match the FEXT loss requirement of compliant category 6 connecting hardware. The requirements and the method of verification of residual FEXT need to be clarified. Table 6: Field tester performance requirements as a function of test configuration (with adapter) 22

23 5.6 Length, Propagation Delay, Delay Skew and DC resistance measurement accuracy The measurement accuracy for length, propagation delay and delay skew are shown intable 7. Performance Parameter Length Propagation Delay Delay Skew DC resistance Range m 0 µs to 1 µs 0 ns 100 ns MHz Resolution 0.1 m 1 ns 1 ns 1 Ω Accuracy ± 1 m ± 4 % ± 5 ns ± 4 % ± 10 ns ± 1 Ω ± 1 % Table 7: Measurement accuracy for length, propagation delay, delay skew and DC resistance 5.7 Explanation of field tester performance parameters affecting measurement accuracy Dynamic accuracy All measurements involve applying a stimulus signal to the cabling under test, and measure the response to the stimulus signal. The gain ratio is the desired measure. The linearity of that ratio is called dynamic accuracy. It is more challenging to maintain linearity (dynamic accuracy) over a large dynamic range and over a wide frequency band. In the case of insertion loss the challenge to maintain good dynamic accuracy is the least as a result of a worst case 40 db (1 in 100) dynamic range. In the case of NEXT loss, the dynamic range is highest at low frequencies (65 db or approx. 1 in 2000), but declines at high frequencies. During FEXT loss measurements (to be able to compute ELFEXT), the dynamic range is highest at both low frequencies and high frequencies, and is therefore most challenging. In addition, stray coupling during dynamic accuracy verification make it very difficult to avoid measurement errors when performing testing dynamic accuracy for FEXT Source/load return loss Return Loss indicates the loss of signal due to reflections at the transmit output or the receive input of the tester. It is commonly known that if the impedance levels are not perfectly matched, that the signal energy that is measured is not optimized. The magnitude of the reflections are also influenced by the return loss properties of the link under test. Both source/load return loss of the tester and return loss of the link become worse as the test frequency increases. This property has a major impact at the measurement accuracy for insertion loss (attenuation), NEXT loss and ELFEXT at high frequencies Random noise floor The random noise floor is the signal that is measured when there is no stimulus applied to the link under test. In some measurements, the measured signals are extremely small, and the random noise that is generated in the measurement channel limits the measurement accuracy. This is most pronounced during the measurement of (EL) FEXT. There is virtually no impact from random noise during the insertion loss (attenuation) and return loss measurements. 23

24 5.7.4 Residual NEXT Residual NEXT is the NEXT loss measured by the field tester, when there is basically nothing connected to it. The tester measures the NEXT from its own internal measurement circuitry and the NEXT from the connector to the link to be tested. Per the definition of the links (which exclude the connection to the equipment itself), the NEXT contribution by this connector must be excluded. The Residual NEXT can add or subtract from the NEXT of the link that is measured, because the phase relationship is not known, but likely adds somewhere in the frequency range of interest. In the case of residual NEXT for the baseline and permanent link test configurations, this is not a critical problem, since the connection with the field tester involves a high quality connection, with very good NEXT properties. In case of a channel measurement residual NEXT is not a simple topic. The most serious error contribution is from the NEXT of the mated jack of the channel adapter and the plug of the user patch cord at the near end. Proprietary time domain techniques are used to reduce the impact of this modular 8-pin, RJ-45 connection. The NEXT loss from the mated connection with the channel adapter at the remote end and the remote user patch cord does have an influence at low frequencies with short links. The standards have not at all addressed this error contribution by assuming that the round trip insertion loss (attenuation) of the link reduces the impact to an insignificant value. In addition, at high frequencies, the FEXT loss performance is having an impact as well on the residual NEXT performance. Test signals are partly reflected (= return loss properties of the link) and arrive back in the measurement circuit as a result of the FEXT coupling in the channel adapter. The current test methods do not properly cover reflected FEXT at all, and in fact after correction of the local and remote residual NEXT contributions dominate the equivalent residual NEXT. Refer to section 7 for more information Residual FEXT FEXT is measured by applying a test signal to one wire pair at the beginning of the link and measuring the response on a different wire pair at the end of the link. If there were no FEXT in the link under test, residual FEXT would be the FEXT that is actually measured. The origin is from coupling inside the local and remote field tester units and connections to and from the field tester. For the baseline and permanent link test configurations, this is not a critical problem, since the connection with the field tester involves a high quality connection, with very good FEXT properties. In the case of a channel measurement, as in the case for NEXT, the FEXT loss of a modular 8-pin RJ-45 connection is having a significant impact. Any compensation for significant residual FEXT in the channel adapter is considerably more complex and the test procedure for residual FEXT does not address this complexity. Assuming that the mated FEXT is relatively independent of plug and jack matings, and no delay skew in the cable under test, all FEXT contributions remain in phase and theoretically can be subtracted out. However, practically there is delay skew (up to 50 ns or 5x 360 phase 100 MHz). An accurate compensation over the full length of a link requires a complex proprietary compensation scheme. 24