12
IEEE802.3 Interim January 2003 10GBASE-T 1 Alien Crosstalk Measurements Alien Crosstalk Measurements Larry Cohen Solarflare Communications

cohen_1_0103

Embed Size (px)

DESCRIPTION

cohen_1_0103

Citation preview

  • IEEE802.3 Interim January 200310GBASE-T

    1

    Alien Crosstalk MeasurementsAlien Crosstalk Measurements

    Larry CohenSolarflare Communications

  • IEEE802.3 Interim January 200310GBASE-T

    2

    Alien Crosstalk OverviewAlien Crosstalk Overview

    How does installation practice affect alien NEXT?

    Measurement method and test setup NEXT Measurement results Summary

  • IEEE802.3 Interim January 200310GBASE-T

    3

    Alien Crosstalk Cable Test BedAlien Crosstalk Cable Test Bed

    Unbundled cable

    Bundled cable

  • IEEE802.3 Interim January 200310GBASE-T

    4

    Alien NEXT Measurement SetupAlien NEXT Measurement Setup

    Network Analyzer

    50

    6 dB Splitter

    50

    R A

    Cat 5e/6 test cable bundle(horizontal cable)

    50

    S

    50Attenuator

    Attenuator matched to lineinterface matching pads

    50:100 Interface

    Matchingpad

    Matchingpad

    50:100Interface

    100 OhmTerminations

    100 OhmTerminations

    Cat 5e/6 RJ45patch panels

    40 meters

    Isolated interconnect span(2 meters patch cord)

    1 meterpatch cords

    1 meterpatch cords

    Six disturber cables

    Victimcable

    Cable ties

    10 cm 10 cm

    VictimDisturber

    Tight bundle cross-section

  • IEEE802.3 Interim January 200310GBASE-T

    5

    Figure of MeritFigure of Merit

    Figure of merit used to compare effects independent of twist pitch

    Measure coupling between same wire pairs of victim and disturber cable (i.e. pair 1 disturber cable into pair 1 victim cable, etc) Same twist pitch pair coupling is dominant Reduces number of measurements (96 measurements become 24)

    Calibrate out insertion loss from balun (test fixture) and interconnect patch cord (between balun and patch panel).

    Calculate power sum of all same pair coupling measurements between the victim cable and six (surrounding) disturber cables

    Power sum result is integrated from DC to 500 MHz to obtain total power coupling figure of merit

    Figure of merit allows meaningful comparison of visually indistinguishable data plots.

  • IEEE802.3 Interim January 200310GBASE-T

    6

    Effects of Bundling and Cable TypeEffects of Bundling and Cable Type

    0 50 100 150 200 250 300 350 400 450 500-70

    -65

    -60

    -55

    -50

    -45

    -40

    -35

    -30

    -25

    -20

    Measured Power Sum Alien NEXT Coupling - 6 Disturber Cables

    Frequency (MHz)

    I

    n

    s

    e

    r

    t

    i

    o

    n

    g

    a

    i

    n

    (

    d

    B

    )

    Cat 5e bundle = -30.66 dBCat 5e unbundled = -34.41 dBCat 6 bundle = -33.08 dBCat 6 unbundled = -36.51 dBCat 5e PTP NEXT limit = -28.79 dB

    All test cases are six disturber cables; two disturbers in adjacent patch panel ports

    Unbundling reduces coupling about 4 dB

    Cat 6 has 2-3 dB advantage over Cat 5e.

    Variation in bundling may be more important than cable category.

  • IEEE802.3 Interim January 200310GBASE-T

    7

    Effect of Bundling ProximityEffect of Bundling Proximity

    0 50 100 150 200 250 300 350 400 450 500-80

    -70

    -60

    -50

    -40

    -30

    -20Measured Power Sum Alien NEXT - Three Disturber Cables

    Frequency (MHz)

    I

    n

    s

    e

    r

    t

    i

    o

    n

    g

    a

    i

    n

    (

    d

    B

    )

    Adjacent pairs = -35.08 dBOpposite pairs = -39.83 dBCat 5e PTP NEXT limit

    No victim/disturbers in adjacent patch panel ports.

    Opposite cables separated by width of one (center) cable over entire span.

    Separation reduces coupling 4-5 dB; correlates with unbundled test case.

    Disturber cable Victim cableDisturber cables

    Victim cable

    Opposite pairs Adjacent pairs

  • IEEE802.3 Interim January 200310GBASE-T

    8

    Effects of Patch Panel Port PlacementEffects of Patch Panel Port Placement

    0 50 100 150 200 250 300 350 400 450 500-80

    -70

    -60

    -50

    -40

    -30

    -20Measured Power Sum Alien NEXT Coupling into Victim Cable 4

    Frequency (MHz)

    I

    n

    s

    e

    r

    t

    i

    o

    n

    g

    a

    i

    n

    (

    d

    B

    )

    Cable 1 = -44.83 dBCable 2 = -44.61 dBCable 3 = -38.75 dBCable 5 = -39.38 dBCable 6 = -46.17 dBCable 7 = -46.58 dB

    Test case is unbundled Cat 5e

    Cables on adjacent plug ports of patch panel (cables 3 and 5) are dominant disturbers (> 5 dB)

    7 6 5 4 3 2 1

    Cat 5e/Cat6 Patch panel

    Victim cableDisturber cables Disturber cables

  • IEEE802.3 Interim January 200310GBASE-T

    9

    Adjacent vs. Separated Patch Panel PortsAdjacent vs. Separated Patch Panel Ports

    7 6 5 4 3 2 1

    Cat 5e/Cat6 Patch panel

    Victimport

    Separated ports

    Adjacent ports

    Test Configuration Measured Power Sum Alien NEXT Coupling(2 disturber cables)

    Adjacent ports Non-adjacent portsCat 5e bundled/Cat 5e patch panel -34.07 dB -36.13 dBCat 5e unbundled/Cat 5e patch panel -36.07 dB -42.62 dBCat 5e unbundled/Cat 5e Keystone panel -37.05 dB -44.28 dBCat 6 bundled/Cat 6 patch panel -37.13 dB -38.23 dBCat 6 unbundled/Cat 6 patch panel -39.53 dB -43.41 dBCat 6 unbundled/Cat 6 Keystone panel -44.49 dB -45.10 dB

    Patch panel coupling becomes very significant (> 4 dB) for unbundled cable; less significant (< 2 dB) for tightly bundled cable.

    Patch panel coupling depends on proximity, terminal layout geometry, and quality of installation.

    Keystone jacks in panel frames produce less coupling for Cat 5e installations.

    Alien NEXT effects from wiring installation may not be detected by cable analyzers.

  • IEEE802.3 Interim January 200310GBASE-T

    10

    Effect of NearEffect of Near--end Cable Isolationend Cable Isolation

    Add extra physical isolation (> 10 cm separation) between disturber and victim cables on short span between equipment and patch panel.

    Produces attenuated NEXT behavior

    Reduction better than 1 dB attenuation per meter isolated span length. 0 50 100 150 200 250 300 350 400 450 500-70

    -65

    -60

    -55

    -50

    -45

    -40

    -35

    -30

    -25

    -20Measured Power Sum Alien NEXT Coupling - Six Disturber Cables

    Frequency (MHz)

    I

    n

    s

    e

    r

    t

    i

    o

    n

    g

    a

    i

    n

    (

    d

    B

    )

    Cat 5e unbundled, no near-end isolation = -34.53 dBCat 5e unbundled, 2m isolated span = -37.17 dBCat 6 unbundled, no near-end isolation = -36.81 dBCat 6 unbundled, 2m isolated span = -39.46 dB

    Cat 5e PTP NEXT limit = -28.79 dB

  • IEEE802.3 Interim January 200310GBASE-T

    11

    Summary of All NEXT Coupling EffectsSummary of All NEXT Coupling Effects

    Test Configuration(Six disturber cables)

    Power Sum Alien NEXTCoupling

    Cat 5e Cat 6Bundled with adjacent patch panel ports -30.68 dB -33.22 dBBundled, no adjacent patch panel ports -31.67 dB -33.83 dBBundled, no adjacent patch panel ports, 2 meterisolated near-end span

    -33.28 dB -35.92 dB

    Unbundled with adjacent patch panel ports -34.53 dB -36.81 dBUnbundled, no adjacent patch panel ports -38.08 dB -38.47 dBUnbundled, no adjacent patch panel ports, with 2 meterisolated near-end span

    -40.34 dB -41.16 dB

    Unbundled with adjacent Keystone ports -35.61 dB -40.13 dBUnbundled, no adjacent Keystone ports -39.28 dB -40.22 dB

    Unbundling provides 3-4 dB reductionNot using adjacent patch panel ports provides 2-4 dB reductionIsolating the first two meters of the channel provides 2.5 dB reductionSpecific reductions are greater for Cat 5e installations than Cat 6 installationsKeystone jack panels provide 1 dB improvement for Cat 5e, 3-4 dB improvement for Cat 6 (no adjacency coupling with Cat 6 Keystone jacks)

  • IEEE802.3 Interim January 200310GBASE-T

    12

    Alien Crosstalk SummaryAlien Crosstalk Summary Worst-case alien crosstalk power sum may be as bad as Cat 5e

    pair-to-pair NEXT Alien NEXT variations from installation practice

    Affected by cable bundling, cable type (category), patch panel coupling, and near-end cable separation

    Patch panel coupling may be dominant effect for unbundled cables(> 4 dB degradation)

    Alien NEXT effects from wiring installation may not be detected by cable analyzers

    Different installation practices will cause significant variations in observed alien NEXT energy.

    Alien Crosstalk MeasurementsAlien Crosstalk OverviewAlien Crosstalk Cable Test BedAlien NEXT Measurement SetupFigure of MeritEffects of Bundling and Cable TypeEffect of Bundling ProximityEffects of Patch Panel Port PlacementAdjacent vs. Separated Patch Panel PortsEffect of Near-end Cable IsolationSummary of All NEXT Coupling EffectsAlien Crosstalk Summary