Alien Crosstalk Measurements
and Performance in Conduit
Configuration
Larry Cohen
Carlos Aldana
SolarFlare Communications
cohen_5_21_2003_conduit.pdf
IEEE802.3 Interim May 2003
10GBASE-T
[email protected]
[email protected]
1
Overview
• Purpose:
– Evaluate 10GBase-T capacity in a conduit
alien crosstalk environment
•
•
•
•
Measurement method and test setup
Measurement results
Channel capacity results
Summary
IEEE802.3 Interim May 2003
10GBASE-T
2
Alien NEXT Measurement Setup
Network Analyzer
S
R
50Ω
A
50Ω
50Ω
Attenuator matched to line
interface matching pads
Matching
pad
50Ω
Attenuator
50:100Ω Interface
Matching
pad
50:100Ω
Interface
Seven cables through conduit are
unbundled (no tie-wraps). Fill factor is
30-40 % in standard 1 inch EMT
6 dB Splitter
Disturber
Victim
Isolated interconnect span
(2 meters UTP patch cord)
Conduit Test Bed
140 feet
100 Ohm
Terminations
1" EMT
100 Ohm
Terminations
119 feet
Exposed cable
1 meter UTP
patch cords
1 meter UTP
patch cords
Cat 5e Keystone
jack panel
IEEE802.3 Interim May 2003
10GBASE-T
3
Alien Crosstalk Conduit Test Bed
•Actual setup used
“closed course”
configuration; very
large number of
bends.
•7 cables in 1” EMT
provided 40% fill
factor
•5 meters of
exposed cable at
near end tiewrapped every 3 ft
IEEE802.3 Interim May 2003
10GBASE-T
4
Test Procedure
• Measure pair-to-pair coupling between all wire
pairs of victim and six disturber cables (96 total
measurements)
• Calculate channel capacity of victim cable using
power sum coupling into each victim pair from the
six surrounding disturber cables
• Test case is 140 feet of cable in 119 feet of
conduit; 5 meters of exposed cable tie-wrapped at
3 foot intervals
IEEE802.3 Interim May 2003
10GBASE-T
5
Measurement Results
Measured Power Sum Alien NEXT Coupling - Worst Pair
-20
•Result
significantly better
than hybrid bundle
specification
Insertion gain (dB)
-30
-40
•Worst-case pair
(Pair 3, connector
pins 3-6) are
much worse than
other pairs.
-50
-60
Preliminary Cat 5e alien NEXT model
Cat 5e unbundled conduit w/ 16 feet tie-wrap
-70
-80
0
50
100
150
200
250
300
350
400
450
500
Frequency (MHz)
IEEE802.3 Interim May 2003
10GBASE-T
6
Capacity Calculations
• 10GBase-T Cabling Ad Hoc Channel Models:
– Established models for Category 5e Insertion
Loss, Return Loss, self NEXT and self FEXT (April
17, 2003).
• Alien crosstalk model for Cat5e based on
coupling measurements in the conduit
environment
• Background Noise
– Upper bound noise level of -150 dBm/Hz appears
consistent with actual data center noise.2,3
• Launch power of 10 dBm.
IEEE802.3 Interim May 2003
10GBASE-T
7
Capacity Calculations
• Self-Impairment Cancellation:
– 55 dB Return Loss Cancellation (15 dB Hybrid, 40 dB
PHY)
– 40 dB NEXT Cancellation
– 25 dB FEXT Cancellation
• No Established Level for Alien NEXT Mitigation.
Mitigation of 0 dB to 25 dB Realized through
Combination of One or More Alternatives:
– PHY (0 dB to 10 dB)4
– Improved Installation Practices (0 dB to 10 dB)1
– Enhanced Cable Design/Specification (0 dB to 25 dB)5
IEEE802.3 Interim May 2003
10GBASE-T
8
Channel Capacity Results
18
Capacity for Unbundled Conduit with 16 ft tie-wrap
16.4
16
15.4
13.9
Capacity (Gb/s )
14
11.9
12
10
16.8
9.82
8
6
4
0
5
10
15
20
25
Alien Mitigation (dB)
IEEE802.3 Interim May 2003
10GBASE-T
9
Summary
• Capacity greater than 10 Gbps can be achieved
with:
– Ad hoc Cat 5e models and alien crosstalk
measurement models in the conduit
environment
– Reasonable levels of impairment cancellation
IEEE802.3 Interim May 2003
10GBASE-T
10
References
1Cohen,
L., “Alien Crosstalk Measurements,” IEEE802.3
Contribution, Jan. 2003.
2Pagnanelli, C., “Data Center Background Noise
Measurements,” IEEE802.3 10GBASE-T Contribution,
May 2003.
3Cobb, T., “Background Noise,” IEEE802.3 10GBASE-T
Contribution, May 2003.
4Solarflare Communications, “10GBASE-T Tutorial,”
IEEE802.3 Contribution, Nov. 2002.
5Cobb, T., “Experimental Cat6 Cable Developed with
Improved Alien NEXT,” IEEE802.3 Contribution, March
2003.
IEEE802.3 Interim May 2003
10GBASE-T
11