Reliability issues associated with achieving a 25 year warranty

1. © 2004 - 2007
2010
– 2010
Overcoming 25 Year Life Reliability Challenges in Solar Electronics
DfR Solutions Webinar
September 27, 2012

2. © 2004 - 2007
2010
Solar Panel “Scams” in the News…
http://www.bbb.org/blog/2012/06/dont-fall-for-a-solar-paneling-scam-this-summer/

3. © 2004 - 2007
2010
Leading Causes of “Hard” Photovoltaic (PV) System Failures
oCentral Inverter: 37% from 2009 Sandia Study
oIGBT most common component
o3 basic fail categories
oManufacturing Quality
oInadequate Design
oDefective Electronic Components
J. Granata, Sandia; 2009 PV
Reliability Conference
IGBT = insulated gate bipolar transistor

4. © 2004 - 2007
2010
Leading Causes of “Hard” PV System Failures
oCentral Inverter: 51% from Sun Edison 2008- 2010 study
oCommunications: 11%
oWeather Station: 7%
“Owner/Operator Perspective on Reliability Customer Needs and Field Data”, Sandia National Laboratories, Utility-Scale Grid-Tied PV Inverter Reliability Workshop, January 2011.

5. © 2004 - 2007
2010
Leading Causes of Calls for PV Inverter Failures
o“Other” is largest
oControl software: 16%
oPrinted Circuit Board: 11%

6. © 2004 - 2007
2010
Leading Root Causes of “Hard” PV Inverter Failures
oParts & Materials: 57%
oSoftware: 16%

7. © 2004 - 2007
2010
Reliability by Inverter Type

8. © 2004 - 2007
2010
PV System Reliability & Availability
oEvery alert must be addressed
oMost software & communication issues can be handled quickly
oHardware takes longer to resolve
oPotentially easier to improve availability by focusing on software & communication!
oTend to be hardware biased!

9. © 2004 - 2007
2010
PV System Failures: Utility Perspective
oLeading failure causes in US:
oPoor Design
oPoor Installation
oCode Infraction
oAmbiguous Instructions
oLack of information and documentation
oLeading failure causes in developing countries:
oSelling PV systems to users and financing them privately or by soft loans
oEncouraged purchase of cheap, poor quality components and minimal maintenance
oLed to early failures and poor long term sustainability
Evaluation of the PREP Component : PV
Systems for Rural Electrification in Kiribati &
Tuvalu (7 ACP RPR 175)
DESIGN REVIEW AND APPROVAL OF GRID- TIED PHOTOVOLTAIC SYSTEMS

10. © 2004 - 2007
2010
Inverter Overview
oInverters perform two key functions
oConverts the direct current (DC) coming from the panels to the alternating current (AC) used by the electric grid
oPerform algorithms to maximize the power produced by the system.

11. © 2004 - 2007
2010
Types of PV Inverters
Graph courtesy of Doron Shmilovitz, Tel Aviv University, Israel

12. © 2004 - 2007
2010
Central Inverters
oPV modules feed into a central dc-dc converter stage that has:
oMaximum power point tracking (MPPT)
oDC-AC inverter to connect to the utility grid.
oCan’t maximize energy extraction from all modules

13. © 2004 - 2007
2010
Micro-Inverters vs Central Inverters
oBetter Reliability & Availability
oLack of single failure point
oLonger warranty: 15-25 years versus 5-10 years
oLower DC Voltages, less vulnerable to arcing
oOptimized Maximum Power Point Tracking (MPPT) per module
oPV Module level real-time monitoring
Images courtesy of Paul Parker, SolarBridge

14. © 2004 - 2007
2010
Micro-Inverters
oThe electronic components used in a micro-inverter are commercial off-the-shelf (COTS)
oParts designed for consumer electronics but need to survive 25 years in solar installations
oOutdoor/Partially Protected & Temp Not Controlled

15. © 2004 - 2007
2010
Distributed Inverters
oSimilar advantages as Micro
oDesigned to work with power optimizers
oStandard 10-12 year warranty, some extended to 20-25 years
oBuilt-in module-level monitoring receiver with communication to internet via broadband or wireless
oBecause MPPT & voltage management are handled separately for each module by the power optimizer, the inverter is only responsible for DC to AC inversion. So potentially less complicated, expensive & more reliable

16. © 2004 - 2007
2010
Micro vs Distributed Inverters
oFrom a system reliability and maintainability perspective, both offer advantages over Central inverters.
oAll inverter types face the same environmental and electronics failure modes.
oBoth regarded as more cost effective for commercial applications vs utilities
oBoth add significant software complexity for optimizing and monitoring functions

17. © 2004 - 2007
2010
Micro vs Distributed Distinctions
oComponent Selection
oLife limits of Electrolytic Caps versus Ceramic Caps
oUse of ASICs to reduce discreet parts count
oDerating
oPower optimizers are low voltage devices and more easily highly derated
oEfficiency & Heat Dissipation Differences
oDistributed/string benefits are true only if designed for the specific optimizer
oOne company doing this as a complete system
oNo current standard for DC Optimizers, so other string inverter companies may be hesitant to design a product around a specific vendors’ optimizer.

18. © 2004 - 2007
2010
Micro vs Distributed Distinctions
oCeramic versus electrolytic capacitors
oDC/DC converters rely on ceramic capacitors which have a low, fixed rate of aging.
oPossible due to the relatively high switching frequency used in converters.
oMicro inverters require large input capacitance due to the grid low frequency.
oIn some cases, this is implemented with electrolytic capacitors which have a significantly shorter lifetime.
Graphic courtesy of SolarEdge

19. © 2004 - 2007
2010
Component Failures
Image Courtesy of SolarBridge

20. © 2004 - 2007
2010
Micro vs Distributed Distinctions
oASICs allow embedding many electronics into the chip, thereby
oReduces number of discreet components and potential points of failure.
oReduced component count on the circuit board also allows manufacturers of power optimizers to decrease the overall size of the product and raise MTBF

21. © 2004 - 2007
2010
Micro vs Distributed Distinctions
oDerating
oPower optimizers are low voltage devices and more easily highly derated
oA lower derating factor may shorten the lifetime of the product due to increased stress under operating conditions.

22. © 2004 - 2007
2010
Micro vs Distributed Distinctions
oEfficiency & Heat Dissipation Differences
oMicro-inverters have lower efficiencies than power optimizers. The highest known efficiency of micro- inverter brands is 96%, meaning 4% heat dissipation to the module.

23. © 2004 - 2007
2010
Micro vs Distributed Distinctions
oDistributed/string benefits are true only if designed for the specific optimizer
oOne company doing this as a complete system
oNo current standard for DC Optimizers, so other string inverter companies may be hesitant to design a product around a specific vendors’ optimizer.

24. © 2004 - 2007
2010
Inverter Field Failure Mechanisms
oSolder joint fatigue failure
oPlated through hole fatigue failure
oConductive anodic filament formation (CAF)
oShock or Vibration (shipping and in use)
oComponent wear out
oPotting Induced Failure

25. © 2004 - 2007
2010
oSolder joints “wear out” or fatigue and fail under the long term influence of temperature cycling and mechanical stresses.
Solder Fatigue

26. © 2004 - 2007
2010
Plated Through Hole Fatigue
oWhen a printed circuit board experiences temperature cycling, expansion/contraction in the z-direction is much higher than that in the x-y plane
oHigh stress can build up in the copper via barrels resulting in cracking near the center of the barrel as shown in the cross section photos below.

27. © 2004 - 2007
2010
Conductive Anodic Filament Formation (CAF)
oCAF formation is a risk when Plated Through Hole (PTH) vias are so close together that damage from drilling can open up a pathway between vias.
oCopper from the via can migrate along the pathway and eventually cause shorting.

28. © 2004 - 2007
2010
28
Failure after Exposure to Vibration
oMechanical shock and vibration also leads to solder joint failures
oCan occur during transportation, installation or use

29. © 2004 - 2007
2010
Potting Electronic Assemblies
oPotting is the process of filling an electronic assembly with a resin compound
oProvides resistance to shock and vibration, and excludes moisture and corrosives.

30. © 2004 - 2007
2010
Printed Circuit Board Warpage due to Potting Shrinkage

31. © 2004 - 2007
2010
oThe use of underfills, potting compounds and thick conformal coatings greatly influences the failure behavior under temperature cycling
oAny time a material goes through its glass transition (Tg) temperature problems tend to occur
oUnderfills designed for enhancing shock robustness do not enhance thermal cycling robustness
oPotting materials can cause PCB warpage and tensile stresses on electronic packages that greatly reduce time to failure
Potting Rules of Thumb

32. © 2004 - 2007
2010
Some Questions You Should Ask Your Inverter Supplier
oHow have you evaluated reliability?
oNot MTBF, certifications or specifications but evaluation under stress conditions to failure
oWhat is your field failure history & repair rate?
oWhat fails & why?
oHow long and how many units in the field?
oWho installs them?
oWhat software monitoring options are available?

33. © 2004 - 2007
2010
– 2010
An Effective Means to Model the Life of Inverter Electronics

34. © 2004 - 2007
2010
Micro-Inverter Requirements
oThe electronic components used in a micro-inverter are off-the-shelf which means they were designed for consumer electronics.
oHow will such electronic assemblies survive the demands of the solar industry?
Consumer Electronics Micro-Inverter Electronics Expected Life 5-7 years 20-25 years User Environment Indoor/Protected Outdoor/Partially Protected Temp Controlled Temp Not Controlled

35. © 2004 - 2007
2010
What can be done?
oHow can a micro-inverter supplier design the product to meet the requirements AND convince the customer of this?
oThis talk will discuss a novel new method to model the reliability of an electronic assembly in a variety of conditions based on the design (before building anything).
oDesign for Reliability (DfR) concepts and Physics of Failure (PoF) are used. .
oA comprehensive software package was developed to simplify this modeling, thus making it available to design engineers.

36. © 2004 - 2007
2010
Design for Reliability (DfR)
oDfR: A process for ensuring the reliability of a product or system during the design stage before physical prototype
oReliability: The measure of a product’s ability to
o…perform the specified function
o…at the customer (with their use environment)
o…over the desired lifetime

37. © 2004 - 2007
2010
DfR & Physics of Failure (PoF)
oDue to some of the limitations of classic DfR, there has been an increasing interest in PoF (aka, Reliability Physics) – to improve on DfR techniques.
oPoF Definition: The use of science (physics, chemistry, etc.) to capture an understanding of failure mechanisms and evaluate useful life under actual operating conditions

38. © 2004 - 2007
2010
Micro-Inverter Environment
oExtreme hot and cold locations (AZ to AK)
oPossible exposure to moisture/humidity
oLarge diurnal thermal cycle events (daily)
oLargest temp swings occur in desert locations where it can reach 64C in the direct sun down to 23C at night (Δ41C)

39. © 2004 - 2007
2010
NREL – Solar Panel Data
oDiurnal Cycles for Each Month

40. © 2004 - 2007
2010
Possible Inverter Failure Mechanisms
oSolder joint fatigue failure
oPlated through hole fatigue failure
oConductive anodic filament formation (CAF)
oShock or Vibration (during shipping)
oComponent wear out.

41. © 2004 - 2007
2010
Is There a Method to Model these Failure Mechanisms?
oYes
oAlgorithms exist to estimate the failure rate from solder joint fatigue for different types of components.
oIPC TR-579 models PTH reliability
oRisk for CAF can be determined
oFinite Element Analysis can be used for Shock & Vibration risk.
oMTBF calculations can be performed to estimate component failure rates.

42. © 2004 - 2007
2010
Leverage in Product Design
70% of a Product’s Total Cost is Committed by Design
http://www.ami.ac.uk/courses/topics/0248_dfx/index.html

43. © 2004 - 200107
Why is modeling reliability early
important?
Architectural Design for Reliability, R. Cranwell and R. Hunter, Sandia Labs, 1997

44. © 2004 - 2007
2010
Earlier is Cheaper
Reduce Costs by Improving Reliability Upfront

45. © 2004 - 2007
2010
What is the cost impact of poor reliability?
Aberdeen Group, Printed Circuit Board Design Integrity: The Key to Successful PCB Development, 2007 http://new.marketwire.com/2.0/rel.jsp?id=730231

46. © 2004 - 2007
2010
Solder Joint (SJ) Wearout
oElimination of leaded devices
oProvides lower RC and higher package densities
oReduces compliance
Cycles to failure -40 to 125C
QFP: >10,000
BGA: 3,000 to 8,000
QFN: 1,000 to 3,000
CSP / Flip Chip: <1,000

47. © 2004 - 2007
2010
The Newest DfR Tool - Sherlock
oSherlock – a new tool that models all the circuit card assemblies and provides predicted life curves from many common failure mechanisms.
oIt is a Semi-Automated CAE knowledge based program
oThe Semi-Automated Features simplify model creation and analysis
oEliminates the long, complicated, model creation process and the need for a PhD level expert in PoF, FEA and CFD numerical modeling.

48. © 2004 - 2007
2010
Software Coverage
oThis software modeling tool predicts failures from
oSolder joint wear-out from thermal cycling (SAC305 or eutectic SnPb)
oPlated through hole fatigue
oConductive anodic filament formation
o217 MTBF calculations are also generated
oIn addition the software uses FEA to determine
oBoard deflection and SJ failure from mechanical vibration
oThe natural frequencies for the board based on the mount points.
oBoard deflection due to shock events

49. © 2004 - 2007
2010
oODB files are preferred since they contain all the data necessary for modeling
oComponent details and placement
oPCB outline & stack-up
oDrill hole file with mount points
oMetal layers, silkscreen, solder mask layers
Data Import
Gerber Data can also be imported

50. © 2004 - 2007
2010
PCB Details Required for Modeling

51. © 2004 - 2007
2010
Establish Part Parameters
oComponents identified along with packaging properties.
oMinimizes data entry through intelligent parsing and embedded package and material databases

52. © 2004 - 2007
2010
Defining The Durability & Reliability Objectives
oDefine the Expected Service Life
oSets The Analysis Range And The Scale For Reliability Over Time Plots
oLife Cycle Phases
oShows relative time spent at each phase

53. © 2004 - 2007
2010
Identify Field Environment
oApproach 1: Use of industry/military specifications
oMIL-STD-810,
oMIL-HDBK-310,
oSAE J1211,
oIPC-SM-785,
oTelcordia GR3108,
oIEC 60721-3, etc.
oAdvantages
oSometimes very comprehensive
oAgreement throughout the industry
oDisadvantages
oMost more than 20 years old
oAlways less or greater than actual (by how much, unknown)

54. © 2004 - 2007
2010
Field Environment (cont.)
oApproach 2: Based on actual measurements of similar products in similar environments
oDetermine average and realistic worst-case
oIdentify all failure-inducing loads
oInclude all environments
oManufacturing
oTransportation
oStorage
oField
Container and Ambient Temperature15.025.035.045.055.065.075.0050100150200250300350400450Hours Temperature (°C) Container Temp (°C) Outdoor Temp (°C)

55. © 2004 - 2007
2010
Input Field (or Test) Environment
Thermal Cycle Profile
Vibration Profile
Shock Profile.
Handles very complex environments

56. © 2004 - 2007
2010
Thermal Cycle Modeling
oThe thermal cycle conditions the product will experience can be modeled (Minor’s rule applied for multiple temperatures).
oPowered components can be assigned an added value above ambient temperature.
Example of Diurnal TC Conditions in a hot climate

57. © 2004 - 2007
2010
TC Fatigue Analysis – Example Plot

58. © 2004 - 2007
2010
Highest Risk Components
oLarge Resistors provide the weakest solder joints in this example.

59. © 2004 - 2007
2010
Plated Through-Hole Reliability Modeling
When a PCB experiences thermal cycling, the expansion/ contraction in the z- direction is much higher than that in the x-y plane.
The glass fibers constrain the board in the x-y plane but not through the thickness.
As a result, a great deal of stress can be built up in the copper via barrels resulting in eventual cracking near the center of the barrel as shown in the cross section photos.

60. © 2004 - 2007
2010
PTH Fatigue Results - Example

61. © 2004 - 2007
2010
Combined (SJ & PTH) Lifetime Prediction
oCombines analysis results into overall failure prediction curve
PTH
Solder Joint
Combined

62. © 2004 - 2007
2010
Risk for Conductive Anodic Filament Formation (CAF)
oCAF formation becomes a risk when plated through hole vias are so close together that damage from drilling can open up a pathway between vias.
oCopper from the via can migrate along the pathway and eventually cause shorting.

63. © 2004 - 2007
2010
CAF Analysis
oThe primary variables that effect the probability of CAF formation are:
oDistance between vias
oDamage during drilling process
oTemperature and humidity conditions
oVoltage differential between vias
oThe analysis takes into account the first two variables only (measures distance between all PTH pairs).
oVias identified as being too close are flagged.

64. © 2004 - 2007
2010
CAF Analysis
oSoftware will flag vias at high risk for CAF formation

65. © 2004 - 2007
2010
Finite Element Analysis
oPCBA Example with Mesh Outlined

66. © 2004 - 2007
2010
Natural Frequencies are Calculated
Select the number of natural frequencies to look for within the desired frequency range.
Components in high strain regions are at risk.
•Move the components
•Move/add mounting points.
MPs

67. © 2004 - 2007
2010
Vibration Environment
Complex vibration profiles can be modeled.
• Qualification test parameters
• Shipping/Transportation
• Field conditions

68. © 2004 - 2007
2010
Random Vibration Strain
Board strain from random vibration is shown (calculated for x, y, and z directions)
Vibration set to simulate shipping

69. © 2004 - 2007
2010
Vibration Results – Component Breakdown
Components most at risk for vibration fatigue damage are listed first.

70. © 2004 - 2007
2010
Software Shock
oImplements Shock based upon a critical board level strain
oWill not predict how many drops to failure
oEither the design is robust with regards to the expected shock environment or it is not
oAdditional work being initiated to investigate corner staking patterns and material influences

71. © 2004 - 2007
2010
Shock Results - Example

72. © 2004 - 2007
2010
Shock Results – Component Breakdown
Components listed in order of maximum strain experienced.

73. © 2004 - 2007
2010
Constant Failure Rate Module – Components (Mil- HNBK-217F)
This takes into account the failure rate of the components themselves.

74. © 2004 - 2007
2010
Combined Failure Rate is Provided

75. © 2004 - 2007
2010
Additional Uses for Modeling
oUse Sherlock to determine thermal cycle test requirements.
oUse to modify mount point locations
oUse to determine ESS conditions
oComponent Replacement
oDetermine impact of changing to Pb-free solder
oDetermine expected warranty costs

76. © 2004 - 2007
2010
Warranty Cost Estimate: Example
•Solder joint fatigue was the dominant wearout mechanism, where the failure rate is predicted ~2% after 20 yrs.
•Warranty from SJ failure = failure rate * motherboard replacement cost * units sold.

77. © 2004 - 2007
2010
Summary
oIt is important to eliminate design flaws early in development.
oMicro-Inverters must survive a challenging environment for long periods of time.
oA software tool is now available to model the primary failure mechanisms so that inverter electronics can be made more reliable.
oSherlock modeling will enable a number of “what if” scenarios.
oChanging package types
oChanging location of components
oChanging the mount point locations
oChanging laminate type, etc.
oThe software can also be used to determine the TC test conditions that best simulate the field use conditions.
oMicro-Inverter designs can be built with more confidence that they will survive the challenging environments where they are placed.

78. © 2004 - 2007
2010
Questions
Thank you for your attention.
Any questions?
ctulkoff@dfrsolutions.com
gcaswell@dfrsolutions.com