2.7 simplying fiber cert

8/13/2019 2.7 Simplying Fiber Cert

http://slidepdf.com/reader/full/27-simplying-fiber-cert 1/26

Fiber Installation

Simplifying the Complexity of Fiber Certification

Eric Lie, PsiberData



Standards: What’s Hot

Encircled Flux Standard in Multimode testing

The new ISO standards has EF compliance as a normative requirement

Enforcing the standard in the field

Qualifying/maintaining light sources for EF compliance in the field

The new IEC 14763-3 Standard

TIA and ISO/IEC14763-3 standard provide significantly different limit budgets

Application limits different from Premise limits and Link Vs Channel Testing Uncertainty in measurements

Negative Loss Measurements

Test usually shown as a pass in the tester but results not meaningful to end-user

Standards should enforce a failure result if the negative loss is greater than uncertainty

MPO/MTP Testing

With the limits getting tighter for 10GbE and 40GbE uncertainty plays significant part

in pass/fail decision

How do you calculate loss/budget?



Encircled Flux Standard in

Multimode testing

Encircled Flux: What you need to know



EF Basics: What does it mean

-300 -200 -100 0 100 200 3000

500

1000

1500

2000

2500

3000

3500

4000

I n t e n s i t y A . U .

Pixels along x-axis Pixels along y-axis Position of pixels

I n t e n s i t y A . U .

©Psiber Data

Output of CCD (Red to blue transition indicates intensity distribution)

Measurement Setup

Calculated EF template

um (Core of MM Fiber)

VCSEL Source for

comparison



Encircled Flux: Field Qualification?

Issue is to guarantee the EF compliance at the end of launch cord Solution 1: Use Modally transparent cords to remove Modal distortion due to the ref.

cords

Solution 2: Use matched cords with light sources to guarantee compliance

Solution 3: Use external mode conditioner with a non-EF light source

Laboratory grade equipment available commercially to verify EF launch Eg: Arden Photonics Modal Explorer, Photon Kinetics Launch Analyzer



Encircled Flux: Benefits

Reduction of uncertainty in Fiber loss measurements

Between different manufacturers

Improved repeatability of measurements

Apart from standards compliance, very meaningful in 10GbE and

going forward in 40GbE installations due to their tight insertion lossrequirement



IL using EF launch, how different?

Test Setup (Tested with both non EF, EF Adapters)

Tx Rx Tx Rx

Δconn1 Δconn4Loss Measured = Δconn1+ Δconn2+ Δconn3Δconn4

Δconn3Δconn2

2m Link

Tx Rx Tx Rx

Δconn1 Δconn4Loss Measured = Δconn1+ Δconn2+ Δconn3Δconn4

Δconn3Δconn2

20 m Link

Test Methodology

Measured a 2m link and a 20m link

Tested with both non EF and EF compliant adapters

Every measurement was taken immediately after reference

Tested with both 1-Jumper and 3-Jumper reference EF light source qualified with Arden Photonics, Modal Explorer



IL using EF launch, how different?

Results

Difference between EF and non EF (average of 28 measurements on same link at850 nm)

1 Jumper Reference : 1.14 dB (link loss 6 dB, 19% less

with non EF)

3 Jumper Reference : 0.7 dB

Standard Deviation of measurements between non EF and EF EF Launch (28 samples)

1 Jumper method : 0.14 dB


Non EF Launch: (28 samples)



Conclusion and Findings: The EF launch measures higher loss of the link, possibly closer to the real loss with a

difference of 1.14 dB

1 Jumper method results in an uncertainty slightly lower than 3 jumper




Multimode testingThe new IEC Standard14763-3



Referencing: One Jumper

Fiber1 Loss measured by instrument is

Reference Value subtracted = Lconn1 + Lconn4

Optical loss limit should be calculated as follows,Lconn1 + Length of Fiber * Loss/km (dB) + Lconn2 + Lsplice1 + Lsplice2 + Lconn3 + Lconn4

What is special about Lconn2 and Lconn3 ?



Optical Budget: Refresher (MMF)

Fiber1 Loss measured by instrument is

Lconn1 + Lconn2 + Lsplice1 + Lsplice2+ Lconn3+ Lconn4 + Lfiber

But what we need to certify is:Lconn1 + Length of Fiber * Loss/km (dB) + Lconn2 + Lsplice1 + Lsplice2 + Lconn3 + Lconn4

Typical DeploymentPatch Panel Patch Panel

Permanent Link

Reference gradeLaunch cord

Reference grade

Tail cord



Referencing: One Jumper contd…

Reference Connector to Reference Connector mated loss : 0.1 dB

Reference Connector to Random Connector : 0.3 dB

Random to Random Connector :0.75 dB

Limit14763-3= Length of Fiber * Loss/km + 0.3 + Lsplice1 + Lsplice2 + 0.3 dB

= 0.6 dB + Length of Fiber * Loss/km (dB) + Lsplice1 + Lsplice2

LimitTIA/11801= Length of Fiber * Loss/km + 0.75 + Lsplice1 + Lsplice2 + 0.75 dB

Ref plug mated to random jack



Referencing: Three Jumper

Reference Connector to Reference Connector mated loss : 0.1 dB

Reference Connector to Random Connector : 0.3 dB

Limit14763-3

= Fiber * Loss/km + (0.3 – 0.1) + Lsplice1

+ Lsplice2

+ (0.3 – 0.1) dB

= 0.4 dB + Length of Fiber * Loss/km (dB) + Lsplice1 + Lsplice2

LimitTIA/11801= Length of Fiber * Loss/km + 0.0 + Lsplice1 + Lsplice2 + 0.0 dB

Ref-ref mated




Multimode testingMPO/MTP Testing



Testing issues in 40GbE Fiber

Polarity Issues

Multiple types of MPO cords in the market,

causing a lot of confusion

Need for a graphical framework to provide

“wiremap” like features

•

TYPE A TYPE B TYPE C



MPO Testing: Breakout and

Cassettes

Interoperability between 10G->40G linksMost installations use break-out cables to get 10G links

Test Instrument should be able to support LC on one end and MPO on the remote unit

MPO/MTP

LIGHT SOURCE

LC Adapter

f



Testing for Insertion Loss:

Methodology

Broadly two different solutions are available Testing using test adapters with MPO/MTP connectors

Pros

Quick way to test all 12 fibers in a matter of seconds

Ability to test breakout links, cassettes

Ability to test polarity of the fiber and continuity quickly

Less prone to errors

Cons

No reference grade MPO launch cord specified in standard

No loss limits defined in the cabling standards yet

Testing with a LC Adapter + MPO fan-out or a breakout reference cable

Pros Can guarantee a EF launch at the end of launch cord

Cheaper way to test MPO trunks by re-using LC adapters

Cons

Too many to mention here



Testing for Insertion Loss: Methodology

Current state of Standards Standards don’t yet define a reference grade MPO launch cable

Loss of ref-ref mated connector, ref-random not defined yet

3 Jumper reference recommended assuming a channel test for MPO trunks

Pass/Fail Criteria for MPO trunks Can use IEEE limits to test for 40GbE specification

1.5 dB for OM4 and 1.9 dB of OM3

Use premise cabling standards like ISO/IEC 14763-3, TIA 568 C.3 etc.



MPO Testing with the IEEE Limit

(sample workflow)

Select 40G/100G limit Based on the fiber type, OM3/OM4 and

the application selected, the loss limit is

applied.

The channels are appropriately selected when

the test limit is selected.



Testing with premise limits and MTP light source/

Power Meter - Referencing

Using the 3-Jumper test reference scheme

Launch cord (Type A) Tail Cord (Type A)

Ref Cord (A/B/C)

Should be same as trunk type

MPO/MTP

Light SourceMPO/MTP

Power Meter



Testing with premise limits and MTP light source/

Power Meter - Referencing

Using the 1-Jumper test reference scheme

with Gender/type interchangeable cords

MPO/MTP

Light Source

MPO/MTP

Power Meter

Unpinned- Unpinned cord(A)



Testing with gender/type interchangeable cords.

Launch cord (Type A) New Tail Cord (Type A)Trunk under test

MPO/MTP

Light SourceMPO/MTP

Power Meter

Gender changed to male (pinned)on one side



Uncertainty between 1 and 3 Jumper reference

Setup Tested a 6x20m trunks 4 times with both one jumper and three jumper reference

Results Std Deviation with 1 Jumper

Reference : 0.08 dB Std Deviation with 3 Jumper

Reference : 0.19 dB

Average difference between

1 and 3 jumper reference : 0.28 dB



Conclusion and learnings

IEEE limits by far accepted by many large datacenters

(eg: Facebook)

Can re-use 14763-3 limit for MTP testing barring compliance to EFlaunch condition

Uncertainty with the MTP/MPO adapters quite low and provides amuch better method to qualify trunks (see next slide)

Most of the issues for trunk cables observed are no-connect, incorrect polarity or unclean connectors.

Using MPO/MTP adapters to qualify the links still the most practical

way to enforce qualification in the field



Thanks for listening.

Happy testing!

[email protected]

2.7 simplying fiber cert

Documents