Part quality-presentation-how-to-test-when-what-it-all-means

© 2004 - 2007
2010
9000 Virginia Manor Rd Ste 290, Beltsville MD 20705 | 301-474-0607 | www.dfrsolutions.com
– 2010
Part Quality: How to Test, When to Test, and What Does It All Mean?
DfR Open House
March 18, 2013
Presented by: Greg Caswell Ed Wyrwas

© 2004 - 2007
2010
oDetermining the appropriate testing methodology for any evaluation requires a requisite understanding of the component/part, the stresses it may see during its use environment, and the potential failure modes that a testing activity might uncover.
oDfR is skilled at numerous testing activities
oThis talk will identify several and explain their respective merits.
How To Test

© 2004 - 2007
2010
oWear-out is caused by electrolyte diffusion through the end seal
oIncreased temperature increases rate of diffusion
oAs electrolyte volume decreases, ESR increases
oDfR Solutions used temperature dependent rate of weight loss testing and critical weight loss dependence on % ESR increase(failure identified as 200%) to predict characteristic life of capacitors
Electrolytic Capacitor Wearout Behavior
y = 0.0239x R² = 0.9975
0
2
4
6
0
100
200
300
Weight Loss (mg)
Time (hrs)
Average Weight Loss Over Time 2
105C=0.0258 mg of electrolyte/hour
y = 0.0062x R² = 0.9825
0
0.5
1
1.5
0
200
Weight Loss (mg)
Time (hrs)
y = 0.004x R² = 0.9904
0
0.5
1
0
100
200
300
Weight Loss (mg)
Time (hrs)

© 2004 - 2007
2010
oBased on rates at 105°C, 85°C, and 76°C
oExpected rate at 45°C 0.0008 mg of electrolyte/hr
oBalance only reads to tenths of a mg
oModeled using an exponential function
oData fits well to model
Results: Rate of Weight Loss Temperature Dependence
y = 7E-05e0.053x R² = 0.9919
0
0.005
0.01
0.015
0.02
0
20
40
60
80
100
Rate of Weight Loss (mg/hr)
Temperature (°C)
Rate of Weight Loss Temperature Dependence

© 2004 - 2007
2010
oExponential function fits relationship between mass loss and % increase in ESR
oModels a sharp increase in ESR after a given mass loss
oSharp increase in ESR is seen around critical weight loss of 1500 mg
Results: % Increase in ESR with Weight Loss
y = 3.4859e0.0027x R² = 0.88
0
200
400
600
800
1000
0
500
1000
1500
2000
2500
% Inc ESR
Weight Loss (mg)
% Increase in ESR with Weight Loss

© 2004 - 2007
2010
oIf failure is defined as a 200% increase in ESR, then characteristic life for the aluminum electrolytic capacitors tested is:
o58,100 hours at 105°C
o1,870,000 hours at 45°C, based on rate of weight loss temperature dependence
Discussion: Characteristic Lifetime Estimates

© 2004 - 2007
2010
oAnalysis estimates characteristic life is ~60,000 hrs
oCaps are rated to 10,000 hrs
oCharacteristic life (η) is 63.2% failure
oAssuming β = 4
oConservative estimate based upon previous life studies
o<0.1% unreliability at 10,000 hours
Discussion: Characteristic Life vs. Time to First Failure

© 2004 - 2007
2010
oCustomer’s reliability goal was an annualized failure rate (AFR) of <0.5%
oAFR should be calculated using disk drive wearout parameters and not an arbitrary MTBF number
oDrive constituent and wearout mechanism
oBearings, platter, axle – mechanical shock and high temperatures
oBearing lubrication – high temperatures, low temperatures, and humidity
oPlatter – electro-magnetic field
oArmature, head, slider – mechanical shock, wearout from use (load-unload process)
Hard Drive Testing

© 2004 - 2007
2010
oUtilization inefficiency can be defined as the excess number of times the disk drive performs this 7-step load-unload routine:
1.Motor acceleration
2.Slider loading
3.Track following
4.Armature sweeping
5.Track following
6.Slider unloading
7.Motor deceleration
oReducing the number of times the disk drive performs this routine extends the drive’s life. Most high reliability disk drives are spec’d for 500k-600k load-unload (LUL) cycles.
oCustomer stated that the disk drive will contain their system’s operating system and act as a data logging storage device. This means that they can control, with software, how often the disk drive spins down by enabling a “capacity assessment” or “lookup” routine.
HDD Testing Approach

© 2004 - 2007
2010
oAFR was calculated using two utilization criteria: Inefficiency and Data logging interval
oInefficiency consists of the percentage of time the operating system starts and stops the hard drive during normal, non-data logging use
oArray analyzed: 100%, 50%, 25%, 10%, 5%, 2.5%, 1%, and 0.5%
oData logging interval considers a LUL cycle to write data to the drive
oArray analyzed: 1hr, 30min, 15min, 10min, 5min, 1min, 30sec, and 10sec
oA combination of these two criteria, High-to-low inefficiency and long-to-short intervals, were weighted against a total LUL cycle count of 600k cycles for the drive
HDD Results
Total Load-Unload Cycles6.00E+050.01%0.01%0.03%0.04%0.09%0.44%0.87%2.59% 0.5% Inefficency43.887.6175.2262.8525.62628525615768Yearly:876017520350405256010512052560010512003153600Hourly:12461260120360Datalogging Cycle:1 hr30 min15 min10 min5 min1 min30 sec10 secUtilization Breakdown by Load-Unload CyclesAnnualized Failure Rate (AFR) by Load-Unload Routine Utilization

© 2004 - 2007
2010
Ion Chromatography
DfR’s recommended limits are derived from experience and experiments using printed circuit boards
Contaminant
Upper Control Limit**
(μg/in2)
Maximum Level
(μg/in2)
Bromide
10
15
Chloride
2
4
Fluoride
1
2
Nitrate
4
6
Nitrite
4
6
Phosphate
4
6
Sulfate
4
6
Total Weak Organic Acids
50
100

© 2004 - 2007
2010
oA DfR customer wanted to know if the flux in the flux pens they were using would induce low surface insulation resistance (SIR) and asked DfR for assistance in performing an electomigration test following IPC-TM-650 2.6.14-1.
Electrochemical Migration Testing (ECM)
As the purpose of this test was to see the potential for unactivated flux to contribute to electromigration, the test vehicles were not sent through reflow or wave solder process. The samples were set up for monitoring and stabilized in a 65C, 85%RH environment in accordance with IPC-TR- 476A recommendations for 96 hours. Following the 96 hour stabilization period, the samples were energized and resistance monitored to date for one week.

© 2004 - 2007
2010
oThere are two pathways that migration can take. The first is over an external surface, often referred to as electrochemical migration (ECM) or dendritic growth. The second is through an internal path which can be created by problems within the laminate such as delamination and is usually called a conductive anodic filament (CAF).
oA liquid medium is required for several purposes. It must dissolve the metal ions and allow electrical conduction via ion migration. Water from humidity is the most common medium (either adsorbed as mono-layers or condensed).
oThe bias applies a force on any positive ions present (such as those on the surface of some metals), driving them to migrate through the medium from anode to cathode. Such ions will deposit on the surface of the cathode, reducing the distance and resistance between those electrodes.
oThere were no signs of migration or reduced surface resistance as shown in the graphs.
Electrochemical Migration Testing (ECM)
Control Boards – No Flux
Fluxed Test Boards

© 2004 - 2007
2010
o22 nm technology, while displaying an impressive combination of leading edge computation power and off-the-shelf pricing, has a severe risk factor: its primary markets have design environments of home / office and design lifetimes of between 18 months and five (5) years. While the manufacturer tries to provide some guidance on reliability, these reports are not available to all customers, they provide no insight on how adjustments in electrical or environmental parameters could improve or reduce reliability, and they provide no prediction of performance beyond five (5) years.
oHow do we test to a thirty year lifetime and induce semiconductor wearout mechanisms on a COTS microprocessor?
Advanced Microprocessor Testing (IC Wearout)

© 2004 - 2007
2010
oBaseline testing to stabilize the processors was conducted under the following thermal-electrical loading. It is worth noting that the clock speed is nearly 16% above the published maximum clock speed of 3.8 GHz.
oWhy would we test at these conditions?
oAppropriate to drive Dielectric Breakdown and Hot Carrier Injection mechanisms
oDfR has extensive knowledge
on the mechanisms that lead to
failure of integrated circuits
Advanced Microprocessor Testing (IC Wearout)

© 2004 - 2007
2010
Test Conditions – Design of Experiment
oVoltage, frequency, and operational stress are controlled through burn-in and benchmarking software. Initial characterization was performed to determine the combinatorial limits of voltage, frequency and operational states for the microprocessor.
oThe ‘hot side’ of the test is controlled by COTS radiator setups (one per system). These temperatures are controlled by modifying the fan speed on the radiator block.
oThe ‘cold side’ of the test is controlled by a closed loop water block with external chiller. The chiller is set to -25°C.
oBoth sides of the test are using propylene glycol as the transfer solution in the heat exchanger.
From Chiller
To Chiller
Chiller
Cold Side
Hot Side
To Chiller
From Chiller
Water block
Water block

© 2004 - 2007
2010
oThe product was subjected to a customer directed HALT process to uncover design and/or process weaknesses. During the test process, the product was subjected to progressively higher stress levels brought on by thermal dwells, vibration, rapid temperature transitions and combined environments
HALT Testing (Avionics Case Study)

© 2004 - 20019070 0 0 Virginia Manor Rd Ste 290, Beltsville MD 20705 | 301-474-0607 | www.dfrsolutions.com
HALT Testing
Step Stress Temperature Testing
-100
-50
0
50
100
Time
Temperature (°C)
T.setpoint
T.product
T.ambient
T1
T2
T3
T4
Vibration Stepping within Thermal Operating Limits
-60
-40
-20
0
20
40
60
80
100
120
Time
Temperature (°C)
0
5
10
15
20
25
30
35
40
45
50
Grms
T.setpoint
T.products
T.ambient
T1
T2
T3
T4
V.setpoint
V.product
Thermal Operating Limit Plot Vibration Combined Stress Plot

© 2004 - 2007
2010
oA DfR customer was interested in dielectric withstanding voltage testing for polymeric conformal coating per IPC-TM-650 2.5.7.1 which involves 0 to 1500VAC over 15 seconds at 100VAC per second ramp followed by 1500 VAC for 1 minute.
Dielectric Withstanding Voltage Testing
All six PCBs arced and shorted during the initial voltage ramp
Arcing and shorting occurred consistently between 1.0 and 1.5kV

© 2004 - 2007
2010
oA customer asked DfR for assistance in performing an elevated temperature power cycle fan test.
oThe fans of a particular lot were failing at an elevated rate during burn in testing that subjected the product to elevated temperatures and power cycling.
oInitial failure analysis reports from the manufacturer suggested that the failures were due to foreign material intrusion into the fan which caused wear between the stator and impeller magnet. The initial conclusion made by the manufacturer was that the failures were caused by the customer.
Fan Testing

© 2004 - 2007
2010
oThis failure theory was deemed implausible by our customer and DfR for the following reasons:
oThe failure mode has been duplicated under test in a relatively clean environment
oThe amount of dust accumulated on the fan blades appears typical and does not indicate usage in a harsh environment
oComposition of the foreign material contains elements that are typically contained in metals and filled plastics:
oSilicon – source, filler material used in plastic
oIron (Ferrite) – impeller magnets
oZinc – corrosion inhibitor used on metal (galvanization) stators
oDfR set up a special thermal/power cycling test of the fans to verify operation and reliability
Fan Testing

© 2004 - 2007
2010
Fan Testing
DfR created a custom circulating thermal chamber constructed to simultaneously test 32 fans. The chamber is designed to regulate the air to a set temperature and provide even and consistent air resistance when the fans are powered.
Thirty two fans were placed in the circulating thermal chamber in two banks of 16, as shown. The banks of fans are facing opposite directions such that the air circulates in the chamber. Heater elements mounted in the ends of the chamber provide additional heating to maintain the chamber temperature at 70°C. The fans are wired to the power supplies with current sense resistors in series with the return line of each fan. The voltage drop across the sense resistor is monitored by the data logger and recorded at 30 second intervals.

© 2004 - 2007
2010
oA DfR customer requested that DfR perform development testing for a 12x12mm QFN to include mechanical robustness (shock, vibration) for laptop applications
oPrimarily two vibration tests are used in the technical specifications for desktop and laptop applications: IEC 68-2 and MIL-STD-810F. The MIL-STD is primarily limited to ‘ruggedized’ versions of desktop and laptop products. When faced with multiple test standards, most component manufacturers who have worked with DfR Solutions have selected the most rigorous to ensure the widest acceptance among their potential customers. Therefore, DfR recommended vibration testing be performed as per MIL-STD- 810F (Method 514, Proc I, Cat. 24). Board design was in accordance with JESD22B113 for an eight layer construction populated with 15 QFNs in very specific locations.
Vibration

© 2004 - 2007
2010
oThe maximum measured shock produced by the DfR tower is 3000g. The shock load selected for the QFN test activity was 1500g. This allowed for a 100% engineering margin in the equipment capability and ensured repeatability.
Shock

© 2004 - 2007
2010
Ball Shear
“Slow” ball shear (<800 μm/s)
“Impact” ball shear (>800μm/s, 0.5m/s common)

© 2004 - 2007
2010
Ball Shear Fractography
JEDEC diagram of brittle, interfacial fracture in shear. Solder pad or intermetallic compound with no plastic deformation should be partially or fully visible.
JEDEC diagram of ductile fracture in shear. Plastic deformation of solder is observed.
Slow shear
Impact shear

© 2004 - 2007
2010
Slow shear
Impact shear
Ball Shear Force Results
Impact shear is a more appropriate method of testing solder joints for variability in intermetallic compound. In this case, no major anomalies were observed between the 5 reballers under investigation, however statistical analysis revealed moderate differentiation.

© 2004 - 2007
2010
oA 1000 hour combinatorial test was designed to meet JEDEC JESD 219 industry specification for solid state drive endurance testing
o95% input/output (IO) utilization
oPower cycling and ‘environmental interrupt’
oSteady state or temperature cycling
oThe target MTBF of this RDT was 2 million hours
oDue to the large number of failures in this reliability test, the failure rate of this SSD population is 266,360 FIT
oThe corresponding MTBF is 3,754 hours
oRoot cause analysis identified a manufacturing defect and issues with quality and workmanship in the failed population
Solid State Drive Reliability Demonstration Testing

© 2004 - 2007
2010
oTest population: 396
oTotal failures: 128
oTotal Device Hours: 317226
oConfidence Level: 60%
oFailure Rate: 266k FIT
oMTBF: 3754 hours
oNorris-Landzberg Acceleration factor (Temperature cycling)
oNo acceleration factor due to static temperature exposure
oArrhenius Acceleration factor (Test Duration)
oCalculated as 1.56X
Failure Rate and MTBF

© 2004 - 2007
2010
oWalk-in Chamber capability
oChambers 1 and 2 (6’x8’), Chamber 3 (6’x10’)
oChamber 3 is currently set up with a testing capacity of 1000 solid state or hard disk drives
o5 C/min ramp rates, thermocouples inside and outside of chamber, air flow control
oCustomer-side remote access to test monitoring (web interface)
DfR’s BIG Chambers

© 2004 - 2007
2010
Optocoupler Testing
2 chamber test set up (left), DUT and control board in chamber (top and right)
Control Board
DUT

© 2004 - 2007
2010
oCurrent Transfer Ratio (CTR), is the gain of the optocoupler. It is the ratio of the phototransistor collector current compared to the infrared emitting diode (IRED) forward current expressed as a percentage (%). CTR=(IC/IF)*100
oThe CTR depends upon the current gain (hfe) of the transistor, the supply voltage to the phototransistor, the forward current through the IRED and operating temperature.
CTR Definition

© 2004 - 2007
2010
Optocoupler Testing
T = 100°C
T = 125°C
If1 = 50 mA
20 optocouplers
20 optocouplers
If2 = 60 mA
20 optocouplers
20 optocouplers
Little or no change in the Current Transfer Ratio (CTR) was observed on two different manufacturer’s Optocouplers after exposure to 110C or 150C at 10/20 mA current
Raising current levels provided insight into degradation

© 2004 - 2007
2010
oHow robust is a power supply’s design?
oElectrical and EMI analysis
oImpedance Loading
oPhysical construction overview
oThermal review
oElectrical derating
oQuality control
Power Supply Testing

© 2004 - 20019070 0 0 Virginia Manor Rd Ste 290, Beltsville MD 20705 | 301-474-0607 | www.dfrsolutions.com
Counterfeit Device Testing
Suspect capacitor - group 2
0.0E+00
1.0E-04
2.0E-04
3.0E-04
4.0E-04
5.0E-04
6.0E-04
7.0E-04
8.0E-04
9.0E-04
1.0E-03
-60 -40 -20 0 20 40 60 80 100 120
Temperature (°C)
Dissipation factor
Cap 4 - Bag 10
Suspect capacitor - group 3
0.0E+00
2.0E-04
4.0E-04
6.0E-04
8.0E-04
1.0E-03
1.2E-03
-60 -40 -20 0 20 40 60 80 100 120
Temperature (°C)
Dissipation factor
C7 SN2854 - 0632

© 2004 - 2007
2010
oCyclic Bend
oTemperature/Humidity/Bias
oFlammability
oSweat
oSalt fog
oSulfur
oConnector mating
oDielectric
oFourier Transform Infra-Red
oCircuit Board SIR
oDigital Image Correlation
Other Testing

© 2004 - 2007
2010
oCritical component strategy
oDesign review
oCritical component identification
oComponent qualification reliability testing
oLifetime limiting components
oLifetime results of test coupled with DfR’s Sherlock software results in aggregate reliability assessment
oAlternate use environment than previously tested or you have no experience (where does institutional knowledge apply?)
oNew designs
oNew parts and materials technologies
When to Test?

© 2004 - 2007
2010
oCan facilitate lowering of warranty costs
oCan determine life expectancies of devices in your design
oCan help meet customer or industry requirements
oCan also determine manufacturing defects
oAn determine design weak points, margins, functional limitations
oLet DfR work with you to establish a viable and cost effective test program.
What Does it All Mean?

Part quality-presentation-how-to-test-when-what-it-all-means

More Related Content

What's hot (8)

Similar to Part quality-presentation-how-to-test-when-what-it-all-means (20)

More from Greg Caswell (12)

Recently uploaded (20)

Part quality-presentation-how-to-test-when-what-it-all-means