Installation Effects Upon Alien Crosstalk and Equal Level Far End Crosstalk

Transcription

1 Installation Effects Upon Alien Crosstalk and Equal Level Far End Crosstalk Paul Vanderlaan Product Development Engineer Belden Presented at BICSI January 20, 1999 Introduction New Parameters involving category 5 cabling have recently been introduced and are currently under development in the standards committee. Several parameters will be specified and regulated in an attempt to assure the end user that the installed system will perform as designed. Three documents are currently in draft status. They are presently identified as the TSB-95 for Category 5, TIA/EIA 568A Addendum-5, and the recently released TIA/EIA 568A Addendum-3. This presentation covers the new parameters Power Sum Near-End Crosstalk (PS-NEXT), Equal Level Far-End Crosstalk (ELFEXT), and Powersum Equal Level Far- End crosstalk (PS-ELFEXT). Additionally, Alien crosstalk will be discussed and the effect various bundling techniques have upon the parameter. Various methods of bundling cables via binders, cinch ties, and conduit will evaluated. Finally, recommendations will be provided to minimize possible cable damage while maximizing LAN performance, if bundled cables are employed in the installation. Near-End Crosstalk : Past and Present Near-end crosstalk (NEXT) is simply defined as the unwanted signal coupling from a near-end transmitter into a pair measured at the same end. In the case of Figure 1, the signal is transmitted on orange pair. The coupled power, as indicated by the orange wave, is measured at the same end on the blue pair. Near End Crosstalk (NEXT) Figure 1

2 Category 5 cables were originally specified in the ANSI/TIA/EIA-568-A document. This document was originally intended to assure that 2 pair networks (one transmitting pair, one receiving pair) would perform to their intended levels. However, with the creation of new encoding schemes utilizing all four pairs for data transmission, there is now a need to characterize and limit the crosstalk from more than one pair. This additional parameter is known as Powersum Near-End Crosstalk (PS-NEXT). It is defined in proposed TIA/EIA 568A Addendum-5 as the computation of the unwanted signal coupling from multiple transmitters at the near-end into a pair measured at the near-end. In the case of Figure 2, the signals are transmitted on orange, green and brown pairs. The coupled power, as indicated by the three orange waves, is measured at the same end on the blue pair. Power Sum Near End Crosstalk (PS-NEXT) Figure 2 This is important in such systems such as Gigabit Ethernet or 100BaseT4 systems, which utilize all four pairs for the transmission of data. Figure 3 is an example of a NEXT trace of a 100-meter sample of cable as measured on an analyzer. The three combinations of orange/blue, green/blue and brown/blue are represented by the green, blue, and black traces respectively. The calculated PS-NEXT performance is indicated by the red trace. Note that the red trace is always higher due to the summation of the three traces from below.

3 Power Sum Near End Crosstalk (PS-NEXT) Far-End Crosstalk Figure 3 Far-end crosstalk (FEXT) is defined as a measure of the unwanted signal coupling from a transmitter at the near-end into a neighboring pair measured at the far-end. Figure 4 is the illustrated example. The signal is transmitted on the orange pair and the red arrow indicates the FEXT coupling. The FEXT level, indicated by the orange waveform, is reduced by attenuation as it travels down the blue pair. In old networks, each end of the cable had one transmitter and one receiver. Any information that coupled onto the adjacent pair was not important, for that pair, at the opposite end, was a dedicated transmitter not listening for transmitted data. However, newer systems utilizing bi-directional transmission schemes, have four transmitters and receivers at each end. Each pair end is sending and receiving data. It becomes necessary to limit the coupled noise from adjacent pairs to ensure data integrity.

4 Far End Crosstalk (FEXT) Figure 4 Figure 5 is an example of a FEXT trace of a 100-meter sample of cable as measured on an analyzer. The red arrow #1 indicates a smooth straight line of coupled FEXT. Since the attenuation at low frequencies is low, the line is almost straight. However, as the frequency increases, the signal level becomes affected by the attenuation of the pair. This is indicated by the gradual roll-off of the trace at the red arrow #2. It is apparent that the FEXT measurement is dependent on attenuation. This, in turn, makes the measurement length dependent as well. This can create field testing problems. Longer lengths of cable appear to exhibit better FEXT than short lengths. Therefore, a length independent measurement is needed to assure the performance of the channel. Far-End Crosstalk Trace Figure 5

5 Equal Level Far-End Crosstalk Equal Level Far-End Crosstalk (ELFEXT) is defined as the measure of the unwanted signal coupling from a transmitter at the near-end into a neighboring pair measured at the far-end relative to the received signal level measured on that same pair. In simple terms, it puts the attenuation back into the FEXT signal. Figure 6 is the illustrated example. The signal is transmitted on the orange pair and the red arrow indicates the FEXT coupling. The FEXT level, indicated by the orange waveform, is reduced by attenuation as it travels down the blue pair. When the attenuation is removed from the measurement the resultant is a higher signal level as indicated by the yello w waveform. This is the calculated ELFEXT. Equal Level Far End Crosstalk (ELFEXT) Figure 6 Figure 7 exhibits the components of ELFEXT. Arrow #1 indicates the FEXT coupling from the orange pair to the blue pair. Arrow # 2 indicated the attenuation characteristic of the blue pair. The equation is simple: ELFEXT= FEXT - Attenuation. The blue trace, indicated by arrow #3, is the resultant calculated ELFEXT. With the attenuation removed, the ELFEXT trace is very linear and smooth.

6 FEXT - Attenuation = ELFEXT Figure 7 Power Sum Equal Level Far-End Crosstalk Finally, the last parameter discussed is power sum equal level far-end crosstalk (PS-ELFEXT). It is defined as the computation of the unwanted signal coupling from multiple transmitters at the near-end into a pair measured at the far-end relative to the received signal level measured on that same pair. Similar to PS- NEXT this is merely the summation of the three combinations of ELFEXT into one trace. Figure 8 indicates the three coupled FEXT disturbances onto the blue pair. The orange waveforms are then summed, the attenuation of the pair removed, and the resultant yellow trace is the computed PS-ELFEXT. Power Sum Equal Level Far End Crosstalk (PS- ELFEXT) Figure 8

7 Cable Bundling and Alien Crosstalk Several questions have been raised recently as to the feasibility of bundling cables to speed the cable pulling process. Is bundling cables permitted within the ANSI/TIA/EIA-568-A? Can bundling cable degrade the cables' internal electrical performance? Can bundling cable reduce the potential LAN performance? Should you be concerned? The answer to all these questions is yes. Old Bundled Cable specifications Originally, the bundled cable specifications were vaguely specified in ANSI/TIA/EIA-568-A, Section 13.1 Hybrid Cables. This only addressed overall jacketed cable groups (e.g. 6 jacketed cables with an overall jacket) Assemblies wrapped with a binder or mylar tape were not covered and therefore exempt from the requirements. The document specified cable to cable NEXT as 6+10log(n) db, where n = number of cables within the jacket assembly. This requirement was almost impossible to meet in cable counts higher than 3 cables. As an example, for two jacketed cables, the crosstalk from cable to cable was 9 db tighter than their respective internal NEXT specification. Six cables required a 13.9dB tighter NEXT specification. The high degree of difficulty in meeting these specifications in conjunction with the lack of a sound technological reason for the tight specification led to a more sensible approach. The new requirement is specified in the newly released ANSI/TIA/EIA 568-A-3 Addendum 3. This document creates a new definition to address wrapped or bound bundles of cables. It is aimed specifically at the manufacturers and assembly houses that generate this material. Currently, this document only addresses PS-NEXT from other cables in the assembly. Alien Crosstalk What is alien crosstalk and where is it regulated? There is no mention of alien crosstalk in any of the industry standards. The term alien crosstalk is not a recognized definition within the standards. Simply put, alien crosstalk is crosstalk noise from adjacent data cables as demonstrated in Figure 9. This phenomena falls under the TIA/EIA 568-A-3 Addendum 3, which addresses bundled cables' crosstalk performance. The requirement states that PS-NEXT from adjacent cables must be 3 db better than the cable's internal pair to pair requirements. This is an important parameter for network cards are unable to compensate for noise external to the cable so it is important to minimize the effects of adjacent cables. Alien Power Sum Near End Crosstalk (PSNEXT) Figure 9

8 Alien Near-End Crosstalk Testing In an attempt to generically characterize alien crosstalk performance, several bundling scenarios were examined. All cable samples were 100 meter, "high" performance category 5 cables. All samples were laid out and tested on the floor with bend radii well within the requirements of ANSI/TIA/EIA-568-A. Bundles of six cables were bound using various methods including: cabling and double wrapped with binders, cinched with plastic cinch ties, and pulled loosely into a 1 inch EMC conduit. The six cables in the conduit only represent a 22% fill factor. To more closely represent a typical 40% fill installation, 4 additional cables were pulled into the conduit and tested. Assemblies utilizing a type of mylar wrap were not considered due to the questionable ability to meet plenum smoke requirements. Figure 10 is a picture of the cabled and bound sample. It is an attempt to simulate a neat assembly that might be provided by a manufacture or third party assembly house randomly selecting cables and bundling them in an aesthetically pleasing manner without regard to the impact of cable performance. Figure 10 Figure 11 represents a particular cable's internal performance before cabling. Figure 12 indicates the subunit performance after the assembly process. Notice the 100 MHz performance has deteriorated from over 10 db of NEXT margin to less than 1 db. The NEXT performance, for which the customer has paid a significant amount of money, has been lost. Cable NEXT Performance Before Binding Figure 11

9 Cable NEXT Performance After Binding Figure 12 The alien, or cable to cable crosstalk, is depicted in Figure 13. The first red arrow indicates the specification for NEXT performance as specified in Addendum 3. It is 3 db tighter than the Category 5 specification. Arrows 2A and 2B indicate points of significant failure. The margin as indicated in the lower left hand corner is db. It is interesting to note that the NEXT from cable to cable actually exceeds Category 5 internal requirements. Not only has the process degraded the internal to the point of near failure, now there is crosstalk from the adjacent cable in excess of category 5 limits. Alien Power Sum NEXT Six Cables Double Bound

10 Cinch ties were used to bind the next sample. Figure 14 and 15 are provided to give an idea as to the frequency and tightness of the cinching. The sample was cinched every 48 inches in accordance with the minimum support distance in ANSI/TIA/EIA-569-A. The cinching slightly deformed the cables' jackets. Additionally, the cables were allowed to lie in their natural state. No attempt to spread the individual cables apart was made. Figure 16 indicates 4.6 db of margin to the Addendum-3 specification. Figure 14 Figure 15 Alien Power Sum NEXT Six Cables Cinched Every 48 Inches Figure 16 The conduit samples consisted of six 100m cables pulled into approximately 280ft of 1 inch EMT conduit. This left 24 feet of exposed cable on each end of the sample under test. This was an attempt to simulate an installation with cable on each end representing patch cable going to the desktop and through the wiring closet. The conduit consisted of two 140ft straight runs connected at the far end by two 90 degree bends to

11 form the shape of a 140 ft long letter U. Figure 17 represents the conduit at 22% fill. There is 3.5 db margin to the Addendum --3 specification. Alien Power Sum NEXT Six Cables - 22% Conduit Fill Figure 17 When four more cables were pulled in to raise the fill factor to 40%, the margin dropped to 3.0 db. It is important to note that due to the intensive and lengthy measurements needed in the verification process, only the two worst performing cables in the 22% fill trial were remeasured and plotted. It is believed that this is a valid assumption since the cable to cable performance was approximately equal on an adjacent cable to cable basis. This being the case, it is not expected that the addition of 4 more cables would radically change the alien NEXT performance of any given cable. Figure 18 indicates that the alien NEXT performance dropped.5 db to a 3.0 db margin to the respective specification. Alien Power Sum NEXT Ten Cables - 40% Conduit Fill Figure 18

12 Figure 19 is a summary of the margins for the respective bundling techniques. There are no surprises here in that the loosest bundled sample, the cinched bundle, exhibited the best alien crosstalk performance while the tightest, the cabled and double bound, exhibited the worst. However, it is possible that the deterioration of the individual cable's integrity, during the cabling assembly process, did exacerbate the alien NEXT failure. Alien Power Sum NEXT Margins to Addendum 3 Alien Equal Level Far End Crosstalk Testing Figure 19 Alien equal level far-end crosstalk (ELFEXT) is the same concept as mentioned earlier in the presentation. The only difference is the crosstalk is from cable to cable rather than pair to pair. Figure 20 illustrates this example. It is important to note that there is no current or proposed specification to regulate Alien ELFEXT. This was an attempt to characterize the performance and in no way indicates that any particular construction would pass or fail any future cable specification. The specification used for comparative purposes is the proposed category 5 Channel ELFEXT limit. Alien Power Sum Equal Level Far End Crosstalk (PS-ELFEXT)

13 Figure 20 Figure 21 is the cabled and double bound sample. The margin as indicated is only.2 db to the channel specification. There is almost no margin left for connectors or any other installation effects. Figure 22, 23 and 24 represent the cinched, 22% filled conduit and 40% filled conduit. Alien Power Sum ELFEXT Six Cables Double Bound Figure 21

14 Alien Power Sum ELFEXT Six Cables Cinched Every 48 Inches Figure 22 Alien Power Sum ELFEXT Six Cables - 22% Conduit Fill Figure 23

15 Alien Power Sum ELFEXT Ten Cables - 40% Conduit Fill Figure 24 Figure 25's table summarizes the performance of the various assemblies. As expected the cabled and double bound assembly had the worst alien ELFEXT performance. However, the cinched assembly no longer has the best performance. This may be due to the nature of ELFEXT, which depicts the coupling along the entire length of the cable unlike NEXT, which primarily represents only near end performance. The cinching of the ties may be producing periodic coupling, which is greater that that found in the cables enclosed loosely in conduit. Alien ELFEXT Margins to the Cat 5 Channel PS-ELFEXT Specification Figure 25

16 Conclusion There are new cable requirements, PS-NEXT, ELFEXT, and PS-ELFEXT which are under development or have been defined by Draft documents in TIA subcommittees. ANSI/TIA/EIA-568-A Addendum-3 specifically addresses the crosstalk from cable to cable, otherwise known as alien crosstalk. Cables are designed to meet their respective requirements. Care must be taken when cables are assembled by a third party. Unplanned twisting of the cable core can be detrimental to the internal performance. As an example, the cabled and double bound sample lost 10 db of internal crosstalk margin. Continuous tightly bound assemblies seem to have an inherent disadvantage for Alien crosstalk. The cabled and double bound sample failed Addendum -3 alien PS-NEXT by 4.5 db and only passed the category 5 channel PS-ELFEXT requirements by 0.2 db. Cinched cables appear to meet current requirements of alien PS-NEXT and exhibit good alien PS-ELFEXT, However, tight cinching may degrade the alien PS-ELFEXT at a higher rate than the alien PS-NEXT. Conduit does not appear to excessively degrade cable to cable performance. When compared to the cinched sample, the 22% filled conduit reduced alien PS-NEXT performance by 1.1 db but improved alien PS-ELFEXT performance by 1.9 db. The 40% filled conduit reduced the performance of alien PS-NEXT by 0.5 db but improved alien PS-ELFEXT by 1 db. It is important to realize that current generation hand held testers cannot verify alien crosstalk performance. In order to help assure maximum LAN performance bundled types of cable constructions should be avoided. If bundled assemblies are required, be sure to obtain the material from sources that have the capability to verify their finished assembly's performance to these new requirements.