Upload
ngongoc
View
225
Download
2
Embed Size (px)
Citation preview
k i ib iBack to Basics: Fiber Optic Fundamentals and Best PracticesFundamentals and Best Practices
Rodney Casteel RCDD/NTS/OSP, CommScope, Chair TIA FOTC
Adrian Young, Fluke, FOTC Standards Co-ChairAdrian Young, Fluke, FOTC Standards Co Chair
Robert Reid, Panduit
Matt Brown, JDSU
Lee Kellett AFLLee Kellett, AFL
Agenda• Who Is FOTC – Rodney Casteel
• Standards Update – Matt Brownp
• High Speed Fiber Plant for Data Centers:– High Speed Channels and Drivers – Rodney Casteel
– Cabling & Application Standards – Robert Reid
• Fiber Cleaning & Inspection – Lee Kellett
• Fiber Testing & Troubleshooting– Adrian Young
Matt Brown– Matt Brown
• Final Questions
Fiber Optics Technology ConsortiumOverview:
f h l i i d i i• Part of the Telecommunications Industry Association (www.tiaonline.org)
• Until last year, we had been known as the Fiber Optics LAN Section (FOLS). Our new name was chosen to reflect our expanding charter.
• Formed 20 years agoy g• Mission: to educate users about the benefits of deploying fiber in
customer-owned networks• FOTC provides vendor neutral information• FOTC provides vendor-neutral information
Fiber Optics Technology Consortium
Current Members
• 3M• AFL
• Leviton• OFS
• Corning• CommScope
Fl k N t k
• Panduit• Sumitomo Electric Lightwave
S i E• Fluke Networks• General Cable• JDSU
• Superior Essex• TE Connectivity
www.tiafotc.org TIA Fiber Optics Technology Consortium
Fiber Optics Technology Consortium• Maintain a website with Fiber FAQs, White Papers and other
resources – www.tiafotc.org.
• Developed and maintain a free Cost Model that allows users to compare installed first costs of several architectures.
• Host a webinar series throughout the year with all webinars available on demand.
• Speak at industry conferences like BICSI• Speak at industry conferences like BICSI
• Contribute to industry publications – check out our article on Making Networks Greener in BICSI News.Making Networks Greener in BICSI News.
• Conducting market research
Fiber Optics Technology Consortium• Recent Webinars Available on Demand
– LAN Standards, News & Trends: 2014 UpdateP Li k T i f MPO C bl Pl f Hi h S d– Permanent Link Testing of MPO Cable Plant for Higher Speed Channels
– Fiber Optic Connectors, Designs, Applications & Choices
• Visit www.tiafotc.org or our channel on BrightTalk
Webinars are eligible for CEC credit for up to two years after they are first broadcast. Email [email protected] if you have completed a webinar and want to receive your CECwebinar and want to receive your CEC.
www.tiafotc.org TIA Fiber Optics Technology Consortium
Standards Update
Matt Brown
JDSU
Standards Update (TIA)
568 3 O ti l fib bli d t t d d• 568.3 – Optical fiber cabling and component standard• Being updated to revision “D” – along with 568.0 and 568.1• Transmission performance and test requirements will be in Clause 7Transmission performance and test requirements will be in Clause 7• Annex D will provide guidelines for field testing• Task group established to review and update Clause 7 and Annex D
Higher Speed Channels
Rodney Casteel RCDD/NTS/OSP CommScope
Market Drivers
• 60% increase in Global Internet users by 2018.
• 75% increase in Global networked devices by 2018.– (Approximately 3 devices and/or connections per person on the(Approximately 3 devices and/or connections per person on the
planet)
• Fixed broadband speeds will increase 2.6x Globally by 2018.p y y
• IP video will represent 79% of all traffic by 2018.
Cisco Visual Networking Index (VNI):Forecast and Methodology, 2013-2018June 2014
Global Mobile Data Traffic Growth
18
16 15.9 EB16
14
1210 8 EB
EB = Exabyte1,000,000,000,000,000,000
10
8 7.0 EB
10.8 EB
“Cisco VNI: Forecast and Methodology,
2013-2018"February 2014
6
42.6 EB
4.4 EB
2
01.5 EB
2013 2014 2015 2016 2017 2018
Market Drivers
• Global data center IP traffic by 2017 will reach 7.7 zetabytes or 644 exabytes per month.
• Global Cloud IP traffic will reach 5 3 zettabytes in 2017Global Cloud IP traffic will reach 5.3 zettabytes in 2017 which is a 4.5x increase
Gl b l Cl d IP t ffi ill t f t thi d f t t l• Global Cloud IP traffic will account for two-thirds of total data center traffic by 2017.
Cisco Global Cloud Index: Forecast and Methodology, 2012 - 2017
Storage RequirementsGl b l i f h h 2020
40
351 ZB = 1 billion TB1,000,000,000,000,000,000,000
Global storage capacity forecast through 202040.00
30
25
ytes
(ZB
)
By 2020 the average person will maintain
20
15
10
Zeta
b
7 23By 2020 the average person will maintain 130 terabytes of personal data. 5
02012 2020
2.59
7.23
2017
Source: Cisco IBSG
Source: IDC
2012 20202017
Data center emerging trends
Cloud Virtualized Green
Big Data ConvergedDCIMSDN
EthernetInfinibandFCoE
iSCSi Fibre Channel
C bli I f t t i t th f i l t tiCabling Infrastructure impacts the success of implementation
Changing infrastructure requirements for the data center
Density ScalabilityBandwidth Manageability
The cloud and physical infrastructure
L3
L2
The cloud and physical infrastructure
L3
L2
The cloud and physical infrastructure
• High bandwidth, low latency required to support L2 switch links
• Scalable infrastructure essential
Traditional Data Center Cabling Infrastructure
Core Layer
Aggregation Layer
Access LayerAccess Layer
Reference Architecture
Fiber/CopperPatching
(includes FC network)
ConsoleServer
Fiber PatchingAggregation
Switch Access Switch
Laser Optimized MM Structured CablingCategory Copper CablingSFP+ Twin-Ax Copper
(includes FC network) Switch Access Switch
Core Switch Main Distribution Area
Core SwitchServer Row Equipment Distribution Area Main Distribution Area
Newer Data Center Architecture• Spine-Leaf offers a lower
latency option for server to server communications
• Offers more redundancy
• Requires higher density connections between leaf and spine switchesand spine switches
• Any to any concept
Pre-terminated cabling: Scalable and Quick
• Typical pre-term copper install is eighttimes faster than field term
Installation time dramatically reduced
times faster than field term
• Reduces deployment risk
• “Migration” to 40/100GbE on fiber is• Migration to 40/100GbE on fiber is much less disruptive
• Minimal packaging and waste on site -GREEN
IEEE 802.3ba: 40/100G EthernetApproved (June 2010)pp ( )
10G 40G 100GApproach 10G x4 10G x1010G
Laser Type
Fiber Type
Connector
VCSEL Array
OM3/OM4
MPO
VCSEL Array
OM3/OM4
MPO x 2
VCSEL
OM3/OM4
LC x2Connector
# of Fibers12
MPO x 2
24
C
2
Transceiver Tolerances
MaximumDistance
Relaxed(to lower cost)
OM3: 100+ m*OM4: 125 – 150 m*
Relaxed(to lower cost)OM3: 100+ m*
OM4: 125 – 150 m*
Tight
OM3: 300mOM4: 550m
* 150 meters with OM4 requires low loss connectorsExtended reach out to 300m on OM3 and 400m on OM4 possible with alternate transceivers
40GBASE-eSR4 QSFP+Emerging Defacto Standardg g
• “Extended Reach” transceivers now available– Cisco / Dell
• Operates as 4 x 10G– QSFP+ has 2.5X edge-density for 10GBASE-S
• Operates as 1 x 40G– 300m (OM3) or 400m (OM4) vs. 100/150 for Std –SR4 device
• Lower cost alternative to SM (40GBASE-LR4 QSFP+)– Lower CAPEX – Estimated 75%
f d ( )– Lower OPEX – 50% of power dissipation (1.5W vs. 3.5W)
Cisco QSFP40G BiDi
• 2 Wavelengths at 20G line rate– 850nm & 900nm
– < 3.5W Power Dissipation
– Utilizes duplex LC connectivity
– Compatible with Nexus 9000 series
• Low Loss connections (with 4LC & 4MPO connections)OM3 100– OM3 – 100m
– OM4 – 125m
Serial Duplex Cable PlantTransmission Example - 10GBASE-SRp
Structured Cabling - 10G Ethernet Cross Connect Model with MPO Cassettes
“Parallel Optics” Cable Plant - Parallel Transmission Example - 40GBASE-SR4p
Type-B:1-1 array patch cord
ositi
on:
ost R
x Type-B:1-1 array patch cord
Type-B:1-1 array patch cord
ositi
on:
ost R
x
Type-B1-1array cable
Type-B:1-1 array patch cord
ral s
igna
l tra
nspo
mos
t Tx
to ri
ghtm
o
Type-B1-1array cable
Type-B:1-1 array patch cord
Type-B1-1array cable
Type-B:1-1 array patch cord
ral s
igna
l tra
nspo
mos
t Tx
to ri
ghtm
oLa
tele
ft mLa
tele
ft m
Example optical pathExample optical pathExample optical path
Structured Cabling - 40G Ethernet Cross Connect Model with MPO Cassettes
“Parallel Optics” Cable PlantMulti-Row Parallel - 100GBASE-SR10
Key-up to key-upmated connections
PUSH
PULL
Position 1
BPUSH
PULL
PUSH
PULL
Position 1
P iti 12
Position 12
P iti 1
Transceiver
:
RxRx
RxRx
:
Key-up to key-upmated connections
PUSH
PULL
Position 1
BPUSH
PULL
PUSH
PULL
Position 1
P iti 12
Position 12
P iti 1
Transceiver
:
RxRx
RxRx
:
Key-up to key-upmated connections
PUSH
PULL
Position 1
PUSH
PULL
Position 1
BBPUSH
PULL
PUSH
PULL
Position 1
P iti 12
Position 12
P iti 1
PUSH
PULL
PUSH
PULL
PUSH
PULL
Position 1
P iti 12
Position 12
P iti 1
Transceiver
:
RxRx
RxRx
:
Transceiver
:
RxRx
RxRx
::
RxRx
RxRx
:
PUSH
PULL
Position 1
Position 12
Position 12
Type-B:1-1 array patch cords
B
Position 12 Position 1
PUSH
PULL
PUSH
PULL
Position 1
Position 12
Position 12
Position 1
Rx
:
TxTx
TxTx
:
ositi
on:
mos
t Rx
PUSH
PULL
Position 1
Position 12
Position 12
Type-B:1-1 array patch cords
B
Position 12 Position 1
PUSH
PULL
PUSH
PULL
Position 1
Position 12
Position 12
Position 1
Rx
:
TxTx
TxTx
:
PUSH
PULL
Position 1
Position 12
PUSH
PULL
Position 1
Position 12
Position 12Position 12
Type-B:1-1 array patch cords
BB
Position 12 Position 1Position 12 Position 1
PUSH
PULL
PUSH
PULL
Position 1
Position 12
Position 12
Position 1
PUSH
PULL
PUSH
PULL
PUSH
PULL
Position 1
Position 12
Position 12
Position 1
Rx
:
TxTx
TxTx
:
RxRx
:
TxTx
TxTx
::
TxTx
TxTx
:
ositi
on:
mos
t Rx
Type-B:1-1array cables
array patch cords
Type-B:1-1ral s
igna
l tra
nspo
mos
t Tx
to ri
ghtm
Type-B:1-1array cables
array patch cords
Type-B:1-1
Type-B:1-1array cables
array patch cords
Type-B:1-1ral s
igna
l tra
nspo
mos
t Tx
to ri
ghtm
PUSH
PULL
Position 1
Position12
Key-up to key-upmated connections
Position 1
B
B
PUSH
PULL
PUSH
PULL
Position 1
Position 12
Position 12
Position 1
PP H
Position 1 Position 12
Transceiver
:
RxRx
RxRx
:
TxTx
yparray patch cords
Late
leftm
PUSH
PULL
Position 1
Position12
Key-up to key-upmated connections
Position 1
B
B
PUSH
PULL
PUSH
PULL
Position 1
Position 12
Position 12
Position 1
PP H
Position 1 Position 12
Transceiver
:
RxRx
RxRx
:
TxTx
yparray patch cords
PUSH
PULL
Position 1
Position12
PUSH
PULL
Position 1
Position12
Key-up to key-upmated connections
Position 1Position 1
BB
BB
PUSH
PULL
PUSH
PULL
Position 1
Position 12
Position 12
Position 1
PUSH
PULL
PUSH
PULL
PUSH
PULL
Position 1
Position 12
Position 12
Position 1
PP H
Position 1 Position 12
PP PP H
Position 1 Position 12
Transceiver
:
RxRx
RxRx
:
TxTx
Transceiver
:
RxRx
RxRx
::
RxRx
RxRx
:
TxTxTxTx
yparray patch cords
Late
leftm
PUSH
PULL
Position12
Example optical channel
BPUSH
PULL
PUSH
PULL
Position 12 Position 1
:
TxTx
:
PUSH
PULL
Position12
Example optical channel
BPUSH
PULL
PUSH
PULL
Position 12 Position 1
:
TxTx
:
PUSH
PULL
Position12PU
SH
PULL
Position12
Example optical channelExample optical channel
BBPUSH
PULL
PUSH
PULL
Position 12 Position 1
PUSH
PULL
PUSH
PULL
PUSH
PULL
Position 12 Position 1
:
TxTx
::
TxTx
::
TxTx
:
“Parallel Optics” Cable PlantMulti-Row Parallel - 100GBASE-SR10
Higher Speed IEEE RoadmapOngoing 802.3 Effortsg g
Parallel Optics Single-mode Duplex• 100GBASE-SR4
– 0 to 106m: 100G over OM4, Parallel multimode fiber (850nm)
4 25G QSFP+ ith MPO
• 40GBASE-ER4– 0 to 40000m: 40G Ultra-long
reach over single-mode fiber– 4x25G QSFP+ with MPO
• 100GBASE-UR4– 0 to 20m: 100G Ultra-short reach,
g– In support of Metro Area Networks
– Extended reach option to 40GBASE-LR4
– Same CWDM wavelengths, 20km and Un-retimed parallel optics
– 4x25G QSFP+ with MPO
• 100GBASE-PSM4
40km options
100GBASE PSM4– 0 to 500m: 100G Over single-mode
fiber (1310nm window)
Cabling & Application StandardsAssuring Application Complianceg pp p
Robert Reid Panduit
Ethernet Loss Budgets“Where did my power budget go?”
10 Gb/s MM Cabling SystemIEEE Link Model 850nm Serial, 300m, 2000MHz.km MMF
System Designer Uncontrolled Power Penalties
Deterministic Jitter Noise
Power Budget(7.3dB) C
8
7
Reflection Noise
Relative Intensity Noise (RIN)
Mode Partition Noise (MPN)
CMargin6
5 ( )
Modal Noise (MN) = 0.3dB (function of CIL)
Margin (Headroom) = 0.8dBBCILCIL
4
System Designer Controlled Power Penalties
Channel Insertion Loss (CIL) = 2.6dB
= 1.5dB (connectors) + 1.1dB (fiber)AISIISI
3
2
Inter-symbol interference (ISI) = 3.02dB
• Both can be controlled and changed
• 75% of the Budget
AISIISI1
0
10 Gb/s MM Channel ModelISI Power Penalty - Different Fiber Grades
10GbE model allows ISI Power Penalty of 3.018dB @ 300m ISI scales with DMD - lower DMD means lower power penalties
10 Gb/s MM Channel ModelCabling Vendor Models - “Engineered Channels”
Application standards can be ambiguous - Industry perception of a hard budget limit of 2 6dB for 10GBASE SR channel for example2.6dB for 10GBASE-SR channel for example
Response to current situation - Many requests for design help from customers (10G links are new to some) Typically vendors have Excel-based calculators or tabulations oflinks are new to some). Typically vendors have Excel based calculators or tabulations of reach and power budget for different fiber types (bandwidth), connector styles and total insertion loss.
Implications: • ‘Overspecification’ of fiber media without tools/tables• Design choices - 2.6dB for SR limit narrows component selection and can limit g p
flexibility choices for cable plant (cross-connect)• Erroneous/unrealistic test limit specification
10 Gb/s MM Channel ModelCabling Vendor Models - “Engineered Channels”
10 Gb/s MM Channel ModelCustomer Example - Value Proposition for OM4
317 meter design goal - Beyond 300 SR channel
10 Gb/s MM Channel ModelCustomer Example - Value Proposition for OM4
System ‘A’ - OM3 Fiber throughout with standard pigtails (0.3dB)
System ‘B’ – OM3 Fiber throughout with ‘optimized’ pigtails (0.2dB)y g p p g ( )
System ‘C’ – OM4 Fiber throughout with standard pigtails (0.3dB)
System ‘D’ – OM4 Fiber throughout with ‘optimized’ pigtails (0.2dB)
OM4 value:0.7dB additional headroom at t t h dditi l htarget reach, or additional reach beyond the target
40/100G Fiber Cabling40GBASE-SR4/100GBASE-SR10 Channel Budget
Channel Insertion Loss (CIL) = 1.9dB
Power Budget(8.3dB)
100 meter= 1.5dB (connectors) + 0.4dB (fiber)
100 meter Channel
S IEEESource: IEEE
40/100G Fiber CablingLink Power Budgeting for Cabling
100 meter OM3 channel with two 0.75dB (Max.) connectors (1.5dB connector
1 301.401.501.60
insertion loss total)
150 meter OM4 channel with two 0.50dB (Max.) connectors (1.0dB connector insertion loss total)0.80
0.901.001.101.201.30
or L
oss
(dB)
insertion loss total)
“Engineered Link”
0 200.300.400.500.600.70
Tota
l Con
nect
o
OM3OM4
0.100.20
100 110 120 130 140 150 160 170 180 190 200
Maximum Reach (m)Source: Panduit extrapolation from IEEE model
• Trade-off between SCS ‘wants’ and IEEE requirements
‘Flexible’ Application Loss Budgets10G MM Cross-Connect Cable Plants
‘Flexible’ Application Loss Budgets40/100G MM Cross-Connect Cable Plant
‘Flexible’ Application Loss Budgets“10G/40G” Cable Plant Reach (Ethernet)
IEC Definition of Loss BudgetISO/IEC 11801 - Backbone/Horizontal Links
Connector Loss Max = 0.75dB;Splice Loss max = 0.30dB
Fiber Attenuation Max = 3.5dB/km
Link CertificationThe Importance of Link Measurement CAPABILITY
Gage R&R (Gage Repeatability and Reproducibility) is the amount of measurement variation introduced by the LSPM system, which consists of the LSPM itself and the individuals using the instrument(s). A Gage R&R study quantifies three things:
1. Repeatability - variation from the LSPM(s)2. Reproducibility - variation from the individuals using the LSPM(s)2. Reproducibility variation from the individuals using the LSPM(s)3. Overall Gage R&R, which is the combined effect of (1) and (2)
The overall Gage R&R is normally expressed as a percentage of the loss budget limits, d l f 20 30% G R&R l i id d t bl i tand a value of 20-30% Gage R&R or less is considered acceptable in most cases.
Example: for a 1.85dB loss budget (2, 0.75dB connectors & 100m of 3.5dB/km MM fiber), an acceptable Gage R&R value would be 30 percent of 1.85dB (0.6dB) or less (error of measurement standard deviation of about 0 1dB is then required)(error of measurement standard deviation of about 0.1dB is then required).
Link CertificationCAPABILITY (Precision) vs BIAS (Accuracy)
A Gage R&R study quantifies the inherent variation in the measurement system, but measurement system bias (accuracy) must be verified through a external calibration process.
Types of Measurement ErrorsRepeatability and Reproducibility of Gauge
Vertical Axis is Probability of Link AcceptanceHorizontal Axis is Link LossTotal Gauge
R&R
epta
nce
mit
mit
of L
ink
Acc
e
Test
Lim
Test
Lim
Pro
babi
lity
o
About 10% chance of failing a good link @ about 1.5dB (but there are practically no links here)
About 10% chance of passing a failing link @ about 2.1dB (and some links are here)
P
RED = “Ideal Gauge”GREEN = Actual GaugeBLUE = Link Loss Distribution
False Positive False Negative
Link Loss RecommendationsContractor Test Error Types
False positive…link indicates fail but truly a pass• Impacts the customer’s ability to deploy links in a timelyImpacts the customer s ability to deploy links in a timely
fashion - “Profitability Issue”False negative…link indicates pass but truly a failg p y
• Presents link reliability issues and potential warranty claims - “Day two issue”
Both types of errors are minimized by providing excellent measurement capability for application loss budgets that are
tti ti ht d ti ht d t t th t d i tgetting tighter and tighter and to support the customers desire to deploy more connectors in the channel.
Link Certification“One Jumper” Method
Several link test configurations exist as defined by standards. The goal of testing any link should be that the impact of the tester referencing cables minimized so h h l f h i bi dthat the result of the test is not biased.
There are three standard methods of completing a link loss test:
• One Jumper Method (TIA Method ‘B’)One Jumper Method (TIA Method B )
• Two Jumper Method (TIA Method ‘A’)
• Three or “Golden” Jumper Method (TIA Method ‘C’)
Both methods ‘A’ & ‘C’ have measurement ‘artifacts’ that cannot be effectively subtracted out & overall yield higher link measurement uncertainty & BIAS
ll h d “ f l ” h d d d hAll methods use “reference quality” patch cords and adapters. This ensures accurate, repeatable and reproducible measurements.
Link Loss Method Comparison
• Test links were built with various lengths and numbers of connectors
• Several personnel were used to measure perm links
• Reference cords were used for all methods
• 5.15 sigma’s (99%) of measurement error for ‘A’ and ‘C’ > 1.0dB
• Method ‘A’ and Method ‘C’ are NOT recommended in links with tight• Method A and Method C are NOT recommended in links with tight application budgets (where variability presented above is a significant fraction of the link budget specification)
Link CertificationCustomer Example #1 - Non-use of Reference cords (Time Based Variability)
Re-reference Events -taken from lag in time stamps of Tester data (not all events shown)
‘Commissioning’ Testing ‘Witness’ or Audit Testing New Testing(Contractor #1) (Customer) (With Reference Cords)
Link CertificationCustomer Example #1 – Reproducibility of Measurements
Significant Diff. in average Headroom between Audit and Commissioning permanent link loss testsloss tests
Extremely poor reproducibility > 1dB
Consumes Headroom spec.
Link CertificationCustomer Example #1 - Correlation of Measurements
Expectation - Strong relationship between audit tests and commissioning tests and the ability to predict one from the otherpredict one from the other
Result - Poor relationship between tests (random) and no ability to
ff ti l di t f th theffectively predict one from the other
Link CertificationCustomer Example #2 - Contamination (Location Based Variability)
Large DC account indicated that they were encountering such a high failure rate of link failures for pre-terminated, cassette-based 10G multimode plug and play fiber product at their data center, that testing was halted until root case was found and rectified (50-60% failure rate of li k b f t ti t d)links before testing was stopped).
Contingency Analysis of ‘Status’ By ‘Rack Unit ID’(Failure rate as a function of test location)
Link CertificationCustomer Example #2 - Contamination (Time Based Variability)
Out of Control In Control
Distribution Analysis of Headroom by Date Tested(Failure rate as a function of test time)
Link CertificationCustomer Example #2 - Contamination (Location Based Variability)
• Highly unlikely that products could have been supplied that would produce such a linearly increasing failure rate (conclusion is that this is not ‘nature’ or related to natural variation of the product)nature or related to natural variation of the product)
• Systemic testing issues at play here (possibly damaged reference cords or the like)
• All discrepant links were retested with the best practices in inspection• All discrepant links were retested with the best practices in inspection, cleaning and proper use (and care) of reference patch cords
• All of links that were that previously failed (as indicated in the plot), passed ith significant headroom to the standard hen retestedwith significant headroom to the standard when retested
• This customer has since adopted these cleaning and inspection practices on reference patch cords and links under test, and this has had significant i t th i t bilit d t bilit f t iimpact on their measurement capability and stability of measurements in particular
Best PracticesConnector Inspection and CleaningConnector Inspection and Cleaning
L K ll tt AFLLee Kellett, AFL
Why Do We Care?
• Connector contamination and damage is the leading root cause of fiber optic network failuresroot cause of fiber optic network failures.
• Network failures cause downtime and truck rolls.
• These cost money... lots of money.
• Inspecting and cleaning before connecting saves p g g gtroubleshooting costs and downtime and improves performance. Period!p
How Dirty Can It Be?
Let’s Do The Math...
What Happens?
• Dust and dirt can literally block the light
• Dirt and oils can cause light to refract and be lost at• Dirt and oils can cause light to refract and be lost at the connection
• Particles can prevent proper mating of connectors
• Dirt can damage connector end face when mating g gand cause permanent damage – cleaning will no longer helpg p
Why Inspect, Clean, Inspect?
• Inspect first to determine need for cleaning
• Dry cleaning is quite effective, but is not perfect – so inspect after clean
• Many customers now require proof of inspection to certify installationscertify installations
• Verifies pre-connectorized products have been supplied as neededsupplied as needed
• Saves time and money in the long run
What Equipment Do I Need?
• A good inspection scope – Manual/view only is ok; auto pass/fail is best
– stand alone or connected to your other test equipment
• Cleaning supplies– Dry is okDry is ok
– Wet and dry options are prefered
– Make sure they are designed for fiber – tissues don’t– Make sure they are designed for fiber tissues don t work!
Reality Check
WHAT WE HEAR...
I h t h d i i k
REALITY...
YIKES! Hi h d t k f• I have not had issues – a quick rub on my shirt works
• YIKES! High speed networks of today are not forgiving – NEED clean, low loss connections
• I cleaned – no need to inspect • If there are issues – how will you prove it was not you? How do you know for sure?
• It takes too much time – not h i
you know for sure?
• How much does it cost to replace connectors? How much
worth itp
does a truck roll cost?
It’s Not Just Us!
• There are IEC standards that define pass/fail criteria
• Cisco has a 20+ page document detailing cleaning and inspection procedures for fiber connectors
• AT&T has their own pass/fail criteria and a 112 page document on inspecting and cleaningp g g
• All of us on this panel, and many more at this conference agree - this is a fundamental requirementconference agree - this is a fundamental requirement for today’s networks.
Best Practices - Summary
• Inspect, Clean and Inspect every connectorAssures optimum performance– Assures optimum performance
– Prevents damage
S ti d i th bi i t l d ti– Saves time and money in the big picture – less downtime, fewer truck rolls, less damage and replacement
Ass res performance needs ill be met– Assures performance needs will be met
– Provides a better product to your customers
TIER 1 CERTIFICATIONTI R C RTIFICATION
M tt B JDSUMatt Brown, JDSU
What is Tier 1 Fiber Certification?
• Tier 1 Fiber Certification:• Measure Length• Measure Length
• Measure Loss
Ch k P l it• Check Polarity
“ ”• Ensure Loss does not exceed a “limit”(AKA loss budget)
l• Document results
A Consistency Challenge
T h A T h BTech ATester A
Tech BTester A
If different results –different best practices
Tech ATester B
Tech BTester BTester B Tester B
A Consistency Challenge
T h A T h BTech ATester A
Tech BTester A
If different results –different tester specifications
Tech ATester B
Tech BTester BTester B Tester B
Leading Causes of Inconsistent Results1. Fiber end-face condition
Covered already– Covered already2. Reference method selected matches
h i l fi ti d lphysical configuration and was properly performed
3. Multimode Transmitter Launch Condition – The dreaded Encircled Flux!
A Brief Note on StandardsThe wonderful thing about standards…
…is there are so many to choose from!
Relevant TIA Standards• TIA-568.3: “Optical Fiber Cabling and Components
Standard”– Section 7: “Optical fiber transmission performance and test
requirements”– Annex C (Informative): “Guidelines for field-testing length,Annex C (Informative): Guidelines for field testing length,
loss, and polarity of optical fiber cabling”• TIA-526-14-B: “Optical power loss measurements of
i t ll d lti d fib bl l t”installed multimode fiber cable plant• TIA-526-7: “Measurement of optical power loss of
installed single-mode fiber cable plant”installed single mode fiber cable plant
Channels and Links – Applies to Fiber as Well
Connections and splices possibleEquipment Cord Equipment Cord
Optical Patch Panel
Optical Patch P l
Equipment EquipmentPanel Panel
Link
Channel
dB vs. dBm dBm = an ABSOLUTE measurement of power
• (1mW = 0dBm) dB = a RELATIVE measurement Loss is a Reference Measurement (not an Absolute Measurement) First step in an accurate loss measurement is performing a reference!First step in an accurate loss measurement is performing a reference! Purpose of a reference is to “zero out” any test cables and connectors
Tx Rx
2 dB 5 dB1 mW = 0 dBm
.5 dB-7.5 dBm
Loss = 7.5 dB
Test Reference Cords (TRCs)• Use high performance connectors
– Optimal optical and geometrical characteristics• Numerical aperture (NA)
• Core/ferrule concentricity
• When mated with other TRCs produce near zero loss and reduces When mated with other TRCs produce near ero loss and reducesuncertainty
• Called for in various standards for loss measurements of installed fiber cablingcabling
Setting Reference – Three options:• 1 Fiber Reference
Connect the OLTS together w/reference jumper – reference power t ( t t 0dB)
Light Source Power MeterTest Jumper
meter (set to 0dB)
Disconnect the fiber at the power meter. Connect a test jumper to the power meter. Add couplers (channel testing) or connect to bulkhead (link testing) and connect to fiber system under test
Light Source Power Meter
Test Jumper
connect to fiber system under test
Test JumperFiber System under Test
Coupler/Bulkhead Coupler/Bulkhead
OLTS = Optical Loss Test Set. Typically has Light Source and Power Meter at both ends. Simplex shown for clarity.
Setting Reference – Three options:
Connect the OLTS together using two test jumpers and a coupler –f t ( t t 0dB)
• 2 Fiber Reference
Light Source Power MeterTest Jumper
reference power meter (set to 0dB)
Test JumperCoupler
Disconnect the fibers at the coupler and connect the system to be tested (link testing). Need to add one coupler for channel testing
Light Source Power Meter
Test Jumper Test JumperFiber System under Test
Coupler/Bulkhead Coupler/Bulkhead
Setting Reference – Three options:
Connect the OLTS together with two test jumpers, two couplers AND a third test jumper – reference power meter (set to 0dB)
• 3 Fiber Reference
j p p ( )
Light Source Power Meter
T t JTest Jumper
Couplers Couplers
Disconnect the fibers at the couplers, remove the third test jumperand connect system to be tested. Leave couplers in for channel t ti f li k t ti
Test Jumpers Test Jumpers
Light Source Power Meter
testing, remove for link testing.
Fiber System under Test
Coupler/Bulkhead Coupler/Bulkhead
Test Jumpers Test Jumpers
y
Summary of Reference Methods Difference is the number of bulkhead (coupler) connections referenced out of
the loss measurement. Use the method recommended by your local standards OR by your vendor!
One Fiber
Use the method recommended by your local standards OR by your vendor! For link testing, 1 jumper method is universally recommended
Two Fiber
Three Fiber
Light Source Power Meter
Test Jumpers Test JumpersFiber System under Test
Always check your reference!Connect test jumpers together and measure lossEnsure no “gainers”Save result for proof of good reference
Multimode Launch Conditions• Different multimode light sources = different modal power distributions
(commonly referred to as launch conditions)L h diti di tl i t li k l t• Launch conditions directly impact link loss measurements accuracy– LED overfills a multimode fiber tending to overstate loss – Laser underfills a multimode fiber tending to understate loss
Higher modesLower modes Lower modes
CoreOverfilled
( )
Cladding
CladdingCore
Underfilled
Cladding
Claddingsource (LED) Claddingsource (Laser) Cladding
Encircled Flux (EF)• Ratio between the transmitted power at a given radius
of the fiber core and the total injected power0,6
0,7
0,8
0,9
1
d flu
x
• Defined in IEC 61280-4-1 standard to characterize the launch conditions of MMF test sources
I d t th l h d t NOT t th0
0,1
0,2
0,3
0,4
0,5
0 5 10 15 20 25 30
radius (µm)
Enci
rcle
d
• Is measured at the launch cord connector – NOT at the source output
• Replaces older “launch condition” requires such as• Replaces older launch condition requires such as Coupled Power Ratio (CPR)
• Can be achieved by using a universal or matched modal y gcontroller (TSB-4979)
TSB-4979
• Universal ControllerFor legacy sources
Legacy Source Black Box
– For legacy sources
– Adds a “black box” to the output of the legacy source
M h d C ll
Universal Controller
• Matched Controller– Specific source matched with specific launch cord
– Launch cord may have additional conditioning
Specific Source Black Box
Matched Controller
Launch Cord
Measuring Length
• Use feet/meter markings on fiber jacket
• Physically measure
• Use a tester that measures lengthUse a tester that measures length– Typically using propagation delay and refractive index
Loss Limits• Acceptable loss is based on
several factors:• Maximum allowable losses
(TIA)– Number of connections – Number of splices– Loss per Km (at specific
( )– Loss per connection = 0.75 dB– Loss per splice= 0.3dBLoss per Km (at specific
wavelengths)– Regional or vendor
requirements
– Loss per Km (slope)• 850nm = 3.5 dB• 1300nm = 1 5 dBrequirements 1300nm 1.5 dB• 1310 nm = 1.0 dB• 1550 nm = 1.0 dBFor Tier 1 Certification the user must
tell the OLTS how many connectorstell the OLTS how many connectors and splices are in the fiber system under test
Tier 1 Fiber Certification Example• One Fiber Reference• 2 connections (default for one fiber reference)
N li• No splices• 300 meters of MMF• Loss limit at 850nm:
– 0.75 dB per connector 1.5dB– 300 meters (3.5 dB per km) 1.05dB– Total 2.55 dB
• Loss limit at 1300nm:– 0.75 dB per connector 1.5dB– 300 meters (1.5 dB per km) 0.45dB
Limit is based on settingsLoss is measuredMargin in calculated( p )
– Total 1.95dB
Will My Application Actually Work?• In this context, application is the protocol that will “ride” on the fiber.
– Typically Ethernet or Fiber Channel• What is the connection between the “limit” on the previous slide and
what the application requires?– Very littleVery little…
Cable Type 1GbE 10GbE 40 /100GbE
Loss (dB) Length (m) Loss (dB) Length (m) Loss (dB) Length (m)( ) g ( ) ( ) g ( ) ( ) g ( )
OM3 4.5 1000 2.6 300 1.9 100
OM4 4.8 1100 3.1 1100 1.5 150
Loss and Length Limits at 850nm
Compliant Networks• Most Enterprise Optical Loss Test Sets will report
“Compliant Networks” based on loss measurementmeasurement
• Cautions! –– Can “PASS” generic limit, but have too much
loss for specific applicationloss for specific application– Most testing performed is on links – but
applications run on channels• If the Application to be carried on the fiber is• If the Application to be carried on the fiber is
known, use Application (Network) limit
Ensure Your Results Are Accurate and Consistent1. Treat your test reference jumpers AND the fiber under test with respect
– inspect and clean ALL fibers ALL the time• Inspect Before You ConnectSMspect e o e ou Co ect• IEC 61300-3-35 Certification
2. Understand reference methods and their impact on limit, loss, and margin
• Reference method chosen in tester setup is correct and matches actual physical setup
• Check the reference often3 U d t d lti d l h diti d h l t3. Understand your multimode launch condition and have a plan to move
to Encircled Flux• Standard modal power distribution = consistent loss results between testers
4 Complement your Tier 1 certification with Tier 2 certification4. Complement your Tier 1 certification with Tier 2 certification
Top 5 errors in OTDR (Tier 2) testingTop 5 errors in OTDR (Tier 2) testing
Ad i Y Fl k N t kAdrian Young, Fluke Networks
Top five in no specific order
1. Failing to use a launch fiber
2 Addi h t d t bl t th l h fib2. Adding a short adapter cable to the launch fiber
3. Using only a launch fiber
4. Failing to verify launch fiber
5 Incorrect test limit5. Incorrect test limit
Failing to use a launch fiber
• The OTDR receiver needs time to settle after the OTDR port
• If you use a patch cord 2 1 m (7 ft )If you use a patch cord 2.1 m (7 ft.)– The first event/connection will be missed
– The OTDR may or may not complain
Adapter cable added
• You’re testing LC links and all you have is an SC to SC launch fiber
• So the easiest solution is to add a short SC to LC cord on the end of the launch fiber
SC SC LC LC LC
Two connections will be measured as a single loss event, which can result in failing a good link
LaunchFiber
Using only a launch fiber
• As a guide, the launch fiber should be– 100 m (328 ft.) for multimode( )
– 130 m (427 ft.) for singlemode (good for links to ≈ 27 km / 16.8 miles)
-0.30 dB ? dB
LaunchFiber
Failing to verify the launch fibers
• If you only use a launch fiber, how do you know if it is good?
• Poor launch fibers represent the majority of support callsPoor launch fibers represent the majority of support calls
How good is this connector? You don’t know!? dB? dB
LaunchFiber
Failing to verify the launch fibers
• If you use a launch and tail fiber, you can verify them before testingg
• Poor launch fibers represent the majority of support calls
LaunchFiber
Tail (Receive)Fiber
Using a tail fiber
• With a tail fiber, the connection at the far end is now characterized
• Requires a technician to be at the far end– Most common objection to doing thisj g
-0.30 dB 0.80 dB
LaunchFiber
Tail (Receive)Fiber
Testing in one direction only?
• Is that really a fail at connection ?– Event limit set to 0.75 dB
-0.30 dB 0.80 dB
LaunchFiber
Tail (Receive)Fiber
Testing in one direction only?
• Tested in the other direction, it now fails at connection !– Event limit set to 0.75 dB
0.90 dB -0.37 dB
LaunchFiber
Tail (Receive)Fiber
Bi-directional averaging
• When bi-directional averaging is implemented– Mismatches in backscatter etc. between the launch/tail fibers and the /
fiber under test are taken out, mathematically speaking
0 30 dB 0 80 dB
-0.30 dB 0.80 dB
0.90 dB -0.37 dB
Bi-directional averaging
• When bi-directional averaging is implemented– Mismatches in backscatter etc. between the launch/tail fibers and the /
fiber under test are taken out, mathematically speaking
0.30 dB 0.22 dB
Wrong test limit• OTDR loss event measurements heavily rely on good reflectance
• Poor reflectance can result in– Optimistic / negative loss readings
– Errors when the application runs
• Agree on a reflectance limit
• As a guide (talk to your vendor)– -35 dB for multimode
– -40 dB for singlemode
– -55 dB for APC singlemode S li k t t d
No reflectance limit Reflectance limit -35 dB
– -55 dB for APC singlemode Same link tested
Questions?
• Rodney Casteel ([email protected])
• Robert Reid ([email protected])
• Adrian Young ([email protected])
• Matt Brown([email protected])
105
• Lee Kellett ([email protected])