Reliability Performance of TE SubComs Generation-3 DPSK HPOE - - PowerPoint PPT Presentation
conference & convention enabling the next generation of networks & services Reliability Performance of TE SubComs Generation-3 DPSK HPOE Generation-3 DPSK HPOE Barbara Dean and Charles Breverman Tyco Electronics Subsea
conference & convention
Barbara Dean and Charles Breverman
Tyco Electronics Subsea Communications
conference & convention
Presenter Profile
Barbara joined Bell Laboratories after receiving her
and reliability analysis of optical transmitters, receivers and passive components for both terrestrial and undersea applications. Her engineering team was responsible for the qualification and certification
Place picture here
responsible for the qualification and certification strategy for undersea pump lasers. Barbara later joined the undersea system division (now TE SubCom) and is currently responsible for Quality Management implementation and Reliability Management and assessment.
Barbara Dean Managing Director Email: bdean@subcom.com Tel: (+1) 732 578 7874 Mobile Tel: (+1) 732 856 4511
conference & convention
Outline
conference & convention
Introduction
communication systems has roughly doubled in the past five years.
possibility of over 100 transmit/receivers (High possibility of over 100 transmit/receivers (High Performance Optical Equipment or HPOE) per line-pair in a station.
due to outage mitigation, sparing and station staffing.
conference & convention
Subcom’s Gen-3 HPOE
providing bidirectional transmission
the Common Shelf
transmission laser
conference & convention
Key Considerations
Where possible, choose parts from suppliers with a proven record of excellent reliability in both terminal and submerged equipment. Push these suppliers to meet performance specifications, rather than immediately opting for an unproven rather than immediately opting for an unproven supplier that claims to meet the specification. Take advantage of improvements in monolithic integration, with more of the required functionality executed in fewer ICs. Eliminate adhesives in the optical path – especially important for new entrants.
conference & convention
Types of Components for Each Functionality
FEC Processing
Client Interface
Increased Integration FEC Processing
Line Transmitter
Integration
conference & convention
Types of Components for Each Functionality
Line Receiver
High Reliability Suppliers Other
conference & convention
Initial Estimate of Component Failure Rate
Telcordia aging tests
– These estimates are likely to be very conservative due to limited sample size
experience of similar parts from same wafer fab and packaging house
– Must understand if parts are likely to be returned upon failure and the FMA capability of the supplier
technology
conference & convention
Early Estimate of HPOE Reliability
Module Functionality FITs @90% c.l., 65° ° ° °C
Client Interface 2,100 FEC Processing 690 Line Transmitter 3,080 Line Receiver 5,150 Optical Amplifiers 4,550 Optical Jumpers 450 PWB, passive electrical components, firmware 3,000 Estimated Total FITs ~19,000
conference & convention
Risk Issues in Design and Start of Manufacture
Typical issue – migration of IC package from ceramic to plastic occurred during Models phase causing qualification approval of component to be simultaneous with start of production simultaneous with start of production Atypical issue – discovery of counterfeit parts
conference & convention
Types of Counterfeit Parts
Dummy packages Parts with labeling that misrepresent the part’s form, fit, function or origin (including Pb-free solder) Parts that are labeled as having undergone Parts that are labeled as having undergone additional reliability screening, but have not Parts rejected by the manufacturer’s test and inspection process Used parts represented as new
conference & convention
Counterfeit Parts Encountered
Very high in-process failure rate of a chip capacitor – Suspect used part sold as new A few in-process failures of an IC from a respected supplier
– Request for FMA led to discovery that the Serial Number – Request for FMA led to discovery that the Serial Number was bogus – Suspect relabeled or purpose-built counterfeit part World-wide supply of the true part was very much smaller than assumed, leading to a potential schedule risk
Flash memory chip query returned the information that the chip was from Supplier A
– Package labeled as Supplier B – Purpose built counterfeit part
conference & convention
Countering Counterfeit Parts
Treat as Intellectual Property theft and inform the victimized supplier and any counterfeit data repositories Purchase from Authorized Distributors or from Distributors open to audits and supplying life Distributors open to audits and supplying life traceability of the part Use EMS with a Counterfeit Part plan and ability to test subtle features All countermeasures put in place before start of manufacture
conference & convention
00 2000 2500 3000 HPOEs
Number of Deployed Active HPOEs
Field Experience
deployed in systems or upgrades that have completed the acceptance process
Time 500 1000 1500
12/31/2006 12/31/2007 12/31/2008 12/31/2009# Active HP
Dec ‘09 Dec ‘07 Dec ‘08 Dec ‘06
process
conference & convention
Distribution of Size of HPOE Installations
llations 15 20
active HPOEs ordered for a network build or upgrade
installation size is N~24
# HPOEs per Installation # Installat 100 200 300 400 500 600 5 10
installation size is N~24
a confidence that the failure rate of a given installation is the same as predicted scales as
N / 1
conference & convention
Evolution of Return Rate with Time
Return Rate is the point estimate
– Ratio of total number of returns to total active HPOEs
15000 20000 turn Rate (FITs)
Time Evolution of HPOE Return Rate
Component-Based Failure Estimate
total active HPOEs deployed at the end
– No “confidence level” attached – represents what the systems actually experienced
Time 5000 10000
12/31/2006 12/31/2007 12/31/2008 12/31/2009Estimated Retur
Dec ‘06 Dec ‘07 Dec ‘08 Dec ‘09
conference & convention
Observations on Return Rate
At all times, the population return rate is lower than the original conservative estimate There is a peak after the first year of deployment followed by a steady decrease Increase in population size will make the estimate Increase in population size will make the estimate more robust, but not by itself lead to the observed decrease Is a very low return rate for newer installations masking continued higher return rate for earlier installations?
conference & convention
10000 15000 rn Rate (FITs)
Return Rate by Quarter Year
All HPOEs Deployed First 500 HPOEs Deployed Last 500 HPOEs Deployed
Latest Return Rates
deployed are compared to the last 500 HPOEs deployed in 2008
contributing to the lower
Time 5000 10
03/31/2009 07/31/2009 11/30/2009 03/31/2010Estimated Return
Mar ‘09 Mar ‘10
contributing to the lower return rate
‘turn-on’ issues were dealt with quickly and effectively
conference & convention
Remaining Contributors to Return Rate
Electronic / Board Mechanical
issue resolved
no trends with respect to component or suppliers Points to an opportunity for additional focus on ‘low tech’ components and Electronic Manufacturing
conference & convention
Conclusions
During development of the Gen-3 HPOE, SubCom instituted measures designed to improve the overall reliability At all times since the initial deployment, the return rate of the population was lower than the conservative component- based estimate The return rate for the past 5 quarters has been below 5,000
deployed early in the product cycle and those deployed most recently The remaining cause for returns is random failure of electronic parts These techniques, with additional emphasis on Electronic Manufacture, are being used in the next generations under development
conference & convention
Pacifico Convention Plaza Yokohama & InterContinental The Grand Yokohama 11 ~ 14 May 2010 www.suboptic.org The 7th International Conference & Convention