Cabling Ad Hoc Cat 5e...
Transcript of Cabling Ad Hoc Cat 5e...
IEEE802.3 Plenary March 200310GBASE-T
1
Cabling Ad Hoc Cat 5e Cabling Ad Hoc Cat 5e MeasurementsMeasurements
Larry CohenSolarflare Communications
IEEE802.3 Plenary March 200310GBASE-T
2
OverviewOverview
• Cabling Ad Hoc Test Plan Measurement – One Cat 5e horizontal cable sample, four test channel
configurations characterized to 500 MHz– Insertion loss– NEXT: pair-to-pair and power sum– ELFEXT: pair-to-pair and power sum– Return loss– Propagation delay– Delay skew– Data from both sides of channel
• Measurement of multiple cable samples – insertion loss test
• Conclusion
IEEE802.3 Plenary March 200310GBASE-T
3
Test Channel ConfigurationsTest Channel Configurations
BA C
Test Channel 2
B
Test Channel 1
A= 5 meters, B= 90 meters, C=5 meters
A and C = work area and equipment cord: B= 90 meters horizontal cable
B = 90 meters horizontal cable
A= 5 meters, B1= 5 meters of CP Cable cut from B, B2=85 meters of horizontal cable cut from B, D= 3 meters, E= 2 meters
B1= CP cable, B2= horizontal cable: A,D and E= work area, patch cord, and equipment cord
B2B1 D
Test Channel 4
A
WA CP C1
E
C2
CA
WA CP C1
CA
WA CP C1
Test Channel 3B2B1
A= 5 meters, B1= 5 meters of CP Cable cut from B, B2=85 meters of horizontal cable cut from B, C= 5 meters
B1= CP cable, B2 =horizontal cable ,A and C = work area and equipment cord:
IEEE802.3 Plenary March 200310GBASE-T
4
Measurement EnvironmentMeasurement Environment
• Measurements made at room temperature
• All patch cords and connectors are Cat 5e
• Test cable sample laid out in simple large loop, no stretching or excessive mechanical stress applied
• Channel pairs defined by T568B RJ45 jack pin/pair assignment
• Measurement performed by HP8753C Network Analyzer
• Measurements normalized for test fixture insertion loss• All test channel interface points are Cat 5e Keystone jack to
RJ45 plug (except channel 4 interface between A and B1 is inline Cat 5e coupler)
• No RJ54 interface on test fixture – direct soldered connection
IEEE802.3 Plenary March 200310GBASE-T
Channel Insertion Loss Measurement SetupChannel Insertion Loss Measurement Setup
Network Analyzer
50 Ω 50 Ω
6 dB Splitter
50 Ω
S R A
16.5
16.5
16.5
Test Channel
50:100Ω Interface
100Ω
100:50Ω Interface
Other channel pairs
Cat 5/5e/6 quad pair UTP
100Ω
Matchingpad
Matchingpad
50ΩAttenuator
Attenuator matched to lineinterface matching pads
IEEE802.3 Plenary March 200310GBASE-T
6
Insertion Loss MeasurementInsertion Loss Measurement
0 50 100 150 200 250 300 350 400 450 500-60
-50
-40
-30
-20
-10
0Meas ured Channel Ins ertion Gain - P air 3
Frequency (MHz)
Inse
rtion
gai
n (d
B)
Tes t channel 1 = 7.00 dBTes t channel 2 = 4.42 dBTes t channel 3 = 4.17 dBTes t channel 4 = 3.76 dBCat 5e limit
•Figure of merit is margin above Cat 5e channel limit at 100 MHz•Connectors increase loss slightly and cause insertion loss deviation (ILD)•Cat 5e margin (extrapolated limit) increases with frequency
IEEE802.3 Plenary March 200310GBASE-T
NEXT Measurement SetupNEXT Measurement Setup
Network Analyzer
Victim Channel
50Ω
6 dB Splitter
50Ω
R A
Disturber Channel
Cat 5/5e/6 quad pair UTP
50Ω
S
50:100Ω Interface
Matchingpad
50ΩAttenuator
Attenuator matched to lineinterface matching pads
50:100Ω Interface
Matchingpad
IEEE802.3 Plenary March 200310GBASE-T
8
PairPair--toto--Pair NEXT Measurement Pair NEXT Measurement –– Interior PairsInterior Pairs
0 50 100 150 200 250 300 350 400 450 500-80
-70
-60
-50
-40
-30
-20
-10Meas ured P air-to-P air NEXT Coupling Gain - NEXT31
Frequency (MHz)
Inse
rtion
gai
n (d
B)
Tes t channel 1 = -44.74 dBTes t channel 2 = -26.31 dBTes t channel 3 = -26.31 dBTes t channel 4 = -20.07 dBCat 5e limit
•NEXT at equipment room termination
•Figure of merit is integrated (average) power coupling loss to 500 MHz
•Adding connectors significantly increases interior pair NEXT coupling
•Connectors are dominant NEXT source
IEEE802.3 Plenary March 200310GBASE-T
9
Power Sum NEXT Measurement Power Sum NEXT Measurement –– Interior PairsInterior Pairs
0 50 100 150 200 250 300 350 400 450 500-80
-70
-60
-50
-40
-30
-20
-10Meas ured P ower S um NEXT Coupling Gain - P air 3
Frequency (MHz)
Inse
rtion
gai
n (d
B)
Tes t channel 1 = -37.80 dBTes t channel 2 = -25.17 dBTes t channel 3 = -25.33 dBTes t channel 4 = -19.01 dBCat 5e limit
•Connector-dominated NEXT has different slope characteristic than cable-dominated NEXT•Connector NEXT shows smooth curve or large periodic fluctuations – cable NEXT is “noisy”
IEEE802.3 Plenary March 200310GBASE-T
10
PairPair--toto--Pair NEXT Measurement Pair NEXT Measurement –– Exterior PairsExterior Pairs
0 50 100 150 200 250 300 350 400 450 500-80
-70
-60
-50
-40
-30
-20
-10Meas ured P air-to-P air NEXT Coupling Gain - NEXT24
Frequency (MHz)
Inse
rtion
gai
n (d
B)
Tes t channel 1 = -39.64 dBTes t channel 2 = -42.51 dBTes t channel 3 = -42.81 dBTes t channel 4 = -29.57 dBCat 5e limit
•Connectors have less effect on exterior pair NEXT coupling – physical separation
•Added connector in test channel 4 is an inline coupler –worse than ordinary connector.
IEEE802.3 Plenary March 200310GBASE-T
11
Power Sum NEXT Measurement Power Sum NEXT Measurement –– Exterior PairsExterior Pairs
0 50 100 150 200 250 300 350 400 450 500-80
-70
-60
-50
-40
-30
-20
-10Meas ured P ower S um NEXT Coupling Gain - P air 2
Frequency (MHz)
Inse
rtion
gai
n (d
B)
Tes t channel 1 = -37.52 dBTes t channel 2 = -34.69 dBTes t channel 3 = -35.01 dBTes t channel 4 = -24.23 dBCat 5e limit
•One bad connector can significantly increase NEXT coupling (added inline coupler for channel 4)
IEEE802.3 Plenary March 200310GBASE-T
FEXT/ELFEXT Measurement SetupFEXT/ELFEXT Measurement Setup
Network Analyzer
Disturber Channel
100Ω
50Ω
6 dB Splitter
50Ω50Ω
S R A
Victim Channel
100Ω
Cat 5/5e/6 quad pair UTP
50:100Ω Interface
Matchingpad
100:50Ω Interface
Matchingpad
50ΩAttenuator
Attenuator matched to lineinterface matching pads
LNA
IEEE802.3 Plenary March 200310GBASE-T
13
PairPair--toto--Pair FEXT MeasurementPair FEXT Measurement
0 50 100 150 200 250 300 350 400 450 500-90
-80
-70
-60
-50
-40
-30Meas ured P air-to-P air FEXT Coupling Gain - FEXT32
Frequency (MHz)
Inse
rtion
gai
n (d
B)
Tes t channel 1 = -69.10 dBTes t channel 2 = -65.03 dBTes t channel 3 = -61.57 dBTes t channel 4 = -59.78 dB
•FEXT is much less than NEXT on long lengths
•Measured FEXT becomes stronger on short lengths
•FEXT increased by connectors
IEEE802.3 Plenary March 200310GBASE-T
14
ELFEXT MeasurementELFEXT Measurement
0 50 100 150 200 250 300 350 400 450 500-70
-60
-50
-40
-30
-20
-10
0Meas ured P air-to-P air ELFEXT Coupling Gain - ELFEXT32
Frequency (MHz)
Inse
rtion
gai
n (d
B)
Tes t channel 1 = -40.14 dBTes t channel 2 = -28.97 dBTes t channel 3 = -27.20 dBTes t channel 4 = -23.12 dBCat 5e limit
•FEXT is measured directly•ELFEXT is FEXT normalized by channel loss•Greater margin to limit than NEXT•ELFEXT increased by connectors
IEEE802.3 Plenary March 200310GBASE-T
15
Return Loss Measurement SetupReturn Loss Measurement Setup
-20 dB
-20 dB
Network Analyzer
Channel Under Test
50Ω
6 dB Splitter
50Ω50Ω
S R A
Other channel pairs
Test cable/channel Cat5e/6 quad pair UTP
50:100Ω
Matchingpad
100Ω
100Ω
100Ω
Matchingpad
100Ω ReturnLoss Bridge
Test Port
100:50Ω
IEEE802.3 Plenary March 200310GBASE-T
16
Return Loss MeasurementReturn Loss Measurement
0 50 100 150 200 250 300 350 400 450 5000
5
10
15
20
25
30
35
40Meas ured Channel Return Los s - P air 3
Frequency (MHz)
Ret
urn
Loss
(dB
)
Tes t channel 1Tes t channel 2Tes t channel 3Tes t channel 4Cat 5e limit
•Cable return loss is excellent•Return loss is degraded by adding connectors•Connector effects reduced by distance
IEEE802.3 Plenary March 200310GBASE-T
17
Multiple Sample Test Multiple Sample Test –– 100 meter channel100 meter channel
BA C
Test Channel
A= 2 meters, B= 93 meters, C=4.5 meters
A and C = Cat 5e work area and equipment cord: B = 93 meters horizontal cable
•Compare different cable samples from different manufacturers
•Measurements made at room temperature
•No stretching or excessive mechanical stress applied to test cable
•No RJ54 interface on test fixture – direct soldered connection
IEEE802.3 Plenary March 200310GBASE-T
18
Insertion Loss from Cable SamplesInsertion Loss from Cable Samples
•All sample channels are significantly better than the extrapolated Cat 5e/Class D channel limit line•Cat 6 test channel better than all Cat 5/5e test channels
0 50 100 150 200 250 300 350 400 450 500-60
-50
-40
-30
-20
-10
0Meas ured Wors t-Cas e P air Ins ertion Los s 100 Meter Channel
Frequency (MHz)
Inse
rtion
gai
n (d
B)
Cat 5e s ample #1Cat 5e s ample #2Cat 5e s ample #3Cat 5 s ample #1 Cat 5e s ample #4Cat 6 s ample #1 Cat 5e s ample #5Cat 5e limit
IEEE802.3 Plenary March 200310GBASE-T
19
Worst Measured Cat 5/5e Loss vs. Cat 6 LimitWorst Measured Cat 5/5e Loss vs. Cat 6 Limit
•Worst Cat5/5e channel approximated by extrapolated Cat 6 channel limit line
0 50 100 150 200 250 300 350 400 450 500-60
-50
-40
-30
-20
-10
0Meas ured Wors t-Cas e P air Ins ertion Los s 100 Meter Channel
Frequency (MHz)
Inse
rtion
gai
n (d
B)
Cat 5 s ample #1 Cat 5e s ample #4Cat 5e s ample #5Cat 6 limit
IEEE802.3 Plenary March 200310GBASE-T
20
Propagation Delay and Delay SkewPropagation Delay and Delay Skew
• Propagation delay derived from time-domain conversion (IFFT) of measured network analyzer data
• Propagation delay varied from 450 to 500 nsec over different cable samples (be careful using propagation delay to measure length!)
• No correlation between loss and propagation delay across different cables brands
• Delay skew less than 15 nsec over various cable samples
IEEE802.3 Plenary March 200310GBASE-T
21
Cat 5e Measurement SummaryCat 5e Measurement Summary
• Lots of Cat 5e cable performs much better than specified TIA/ISO limits• Most significant channel degradations are due to connectors• Poor connectors can significantly increase internal crosstalk and reduce
return loss• TIA/ISO limits are designed for worst-case pass/fail limit bounds
– Never intended as a typical channel characterization– Provide margin for test equipment imperfections and measurement noise– Use of extrapolated TIA/ISO insertion loss limits as a channel model is very
pessimistic with respect to a typical Cat 5e channel• Typical Cat 5e channel insertion loss at room temperature can be
approximated by the extrapolated Cat 6 channel limit line• Significant channel degradations can be mitigated by replacing
connectors