Checkpoint 1: MTBF Estimated (MIL-HDBK-217 or Similar) Major
Mean Time Between Failures (MTBF) provides a quantitative reliability prediction for the product. While imperfect, it enables comparison between design alternatives, identifies reliability weak points, and satisfies customer/contractual requirements.
MTBF Calculation Methods
Basic MTBF Relationship:
MTBF = 1 / λ_system
λ_system = Σ λ_component (sum of all component failure rates)
Where λ = failure rate in failures per million hours (FPMH) or FIT (failures in 10⁹ hours)
1 FIT = 1 failure per 10⁹ hours = 0.001 FPMH
MIL-HDBK-217F Component Failure Rate Model:
λ_component = λ_base × π_T × π_S × π_Q × π_E
Where:
λ_base = Base failure rate (from tables, depends on component type)
π_T = Temperature factor (Arrhenius model: increases with temperature)
π_S = Stress factor (electrical stress ratio)
π_Q = Quality factor (military vs. commercial grade)
π_E = Environment factor (ground fixed, mobile, airborne, etc.)
MTBF Calculation Example
Example: Industrial Controller Board
Environment: Ground, Fixed (π_E varies by component type)
Temperature: 55°C ambient, component temps 60-85°C
Quality: Commercial grade (π_Q = 2-10 depending on type)
Component breakdown:
50× Ceramic caps: λ = 50 × 0.012 FPMH = 0.6 FPMH
10× Electrolytic caps: λ = 10 × 0.15 FPMH = 1.5 FPMH
80× Resistors: λ = 80 × 0.003 FPMH = 0.24 FPMH
5× Linear ICs: λ = 5 × 0.05 FPMH = 0.25 FPMH
3× Digital ICs (complex): λ = 3 × 0.12 FPMH = 0.36 FPMH
1× Microcontroller: λ = 1 × 0.30 FPMH = 0.30 FPMH
2× Voltage regulators: λ = 2 × 0.08 FPMH = 0.16 FPMH
4× Diodes: λ = 4 × 0.005 FPMH = 0.02 FPMH
6× MOSFETs: λ = 6 × 0.06 FPMH = 0.36 FPMH
3× Connectors: λ = 3 × 0.02 FPMH = 0.06 FPMH
1× PCB: λ = 0.001 FPMH per layer × 4 = 0.004 FPMH
200× Solder joints: λ = 200 × 0.0005 = 0.10 FPMH
λ_total = 0.6+1.5+0.24+0.25+0.36+0.30+0.16+0.02+0.36+0.06+0.004+0.10
λ_total = 3.954 FPMH
MTBF = 1,000,000 / 3.954 = 252,908 hours = 28.9 years
Note: Electrolytic caps are 38% of total failure rate!
Alternative Prediction Methods
| Method | Standard | Best For | Notes |
|---|
| MIL-HDBK-217F | US Military | General electronics, contractual requirements | Conservative, widely recognized |
| Telcordia SR-332 | Telecom | Telecom equipment, more realistic | Three prediction methods |
| FIDES | French Def. | Process-quality-sensitive prediction | Includes manufacturing quality factors |
| IEC 62380 | International | General commercial electronics | Based on field data |
| NSWC-11 | US Navy | Mechanical/electromechanical parts | Connectors, relays, switches |
MTBF prediction performed using Telcordia SR-332 Method I (parts count). Result: MTBF = 185,000 hours (21 years). Top failure contributors identified: 1) Electrolytic capacitors (35%), 2) Power MOSFETs (20%), 3) Microcontroller (10%). Design actions: Replaced 85°C/2000hr electrolytics with 105°C/10000hr types (MTBF improves to 290,000 hours). Added heatsink to MOSFETs to reduce junction temperature by 20°C.
No MTBF prediction performed. Product warranty is 3 years. Field failure rate reaches 8% in year 2, primarily due to electrolytic capacitor dry-out at 65°C operating temperature. If MTBF had been calculated, the cap lifetime issue would have been caught at design stage. Cost: $2M in warranty claims vs. $0.30/board for better capacitors.
- MTBF is not lifetime: MTBF describes random failures in the flat part of the bathtub curve. Wear-out failures (caps, fans, batteries) are NOT captured by MTBF and must be analyzed separately.
- Temperature dominates: A 10°C increase in operating temperature typically doubles the failure rate. Thermal management is the most effective reliability improvement.
- Quality grades matter: Military-grade components can have 5-10× lower failure rates than commercial equivalents. Automotive grade (AEC-Q) is intermediate.
- Solder joint reliability: In harsh environments (vibration, thermal cycling), solder joints become dominant failure mode. BGA solder joints fatigue faster than leaded packages.
Checkpoint 2: Solder Joint Fatigue Life (Thermal Cycling) Critical
Solder joints fail due to thermo-mechanical fatigue caused by repeated temperature cycling. The CTE mismatch between the IC package and the PCB substrate creates shear strain in solder joints, leading to crack initiation and propagation.
Coffin-Manson Solder Fatigue Model
Coffin-Manson Equation for Solder Fatigue:
Nf = C × (Δγ)^(-n)
Where:
Nf = Number of cycles to failure
Δγ = Shear strain range in solder joint
C = Material constant (≈ 0.5 for SAC305)
n = Fatigue exponent (≈ 1.9 for SAC305, 2.2 for SnPb)
Shear strain calculation:
Δγ = (ΔCTE × ΔT × DNP) / h
Where:
ΔCTE = CTE mismatch between package and board (ppm/°C)
ΔT = Temperature cycling range (°C)
DNP = Distance from Neutral Point (mm) -- farthest solder joint
h = Solder joint height (standoff) (mm)
Example: BGA-256 on FR4 PCB
Package: 17mm × 17mm plastic BGA, CTE_pkg = 15 ppm/°C
PCB: FR4, CTE_board = 16 ppm/°C (x-y direction)
ΔCTE = |16 - 15| = 1 ppm/°C (well matched!)
ΔT = 100°C (cycling -40 to +60°C)
DNP = √(8.5² + 8.5²) = 12.0mm (corner ball to center)
h = 0.5mm (BGA standoff height)
Δγ = (1×10⁻⁶ × 100 × 12.0) / 0.5 = 0.0024 = 0.24%
For SAC305 (C=0.5, n=1.9):
Nf = 0.5 × (0.0024)^(-1.9) = 0.5 × 127,551 = 63,776 cycles
At 1 cycle/day: 175 years. Very reliable.
Example: Large Ceramic Capacitor (1210 package) on FR4
CTE_ceramic = 6-7 ppm/°C, CTE_board = 16 ppm/°C
ΔCTE = |16 - 6| = 10 ppm/°C (large mismatch!)
ΔT = 80°C (cycling 0 to 80°C daily)
DNP = L/2 = 3.2mm/2 = 1.6mm (end of component)
h = 0.05mm (thin solder fillet height for cap)
Δγ = (10×10⁻⁶ × 80 × 1.6) / 0.05 = 0.0256 = 2.56%
Nf = 0.5 × (0.0256)^(-1.9) = 0.5 × 1,894 = 947 cycles
At 1 cycle/day: 2.6 years. RELIABILITY CONCERN!
Mitigation: Use flex-term capacitors (polymer termination)
Flex-term reduces strain by 50% → Nf increases ~4×: 3,788 cycles = 10.4 years ✓
Solder Joint Life Improvement Strategies
| Strategy | Improvement Factor | Application |
|---|
| Increase standoff height | Proportional to h increase | BGA: Use larger balls; Chip: fillet shape |
| Underfill (BGA) | 3-10× life improvement | Critical BGAs in harsh environments |
| Corner staking (BGA) | 2-3× life improvement | Less expensive than full underfill |
| Flex-term capacitors | 3-5× life improvement | Large MLCC (1206, 1210, 1812, 2220) |
| Reduce ΔT (thermal management) | Proportional to ΔT reduction | All components benefit |
| CTE-matched substrate | Eliminates mismatch | Ceramic substrates, metal-core PCB |
Automotive ECU with -40°C to +125°C cycling: BGA processor with underfill specified (Henkel Loctite 3563). Large ceramic caps (1210, 1812) use Murata NFM series (flexible termination). MOSFET packages (D²PAK) have compliant leads. Predicted solder fatigue life: >5000 thermal cycles. AEC-Q100 qualification requires 1000 cycles. 5× margin.
Outdoor telecom equipment with large 2220 ceramic capacitors (47µF/25V X5R). Standard termination (rigid). PCB flex during temperature cycling (daily -20°C to +60°C). Capacitors crack after 800 thermal cycles (2.2 years). Root cause: CTE mismatch strain exceeds ceramic fracture limit. Field replacement cost: $500 per unit × 5000 units.
Checkpoint 3: Conformal Coating Specified if Needed Major
Conformal coating protects PCBs from moisture, dust, chemicals, and temperature extremes. It must be specified for products operating in harsh environments or requiring high reliability.
Coating Types and Selection
| Coating Type | Standard | Moisture Resistance | Temp Range | Application |
|---|
| Acrylic (AR) | IPC-CC-830B Type AR | Good | -65 to +125°C | General purpose, easy rework |
| Polyurethane (UR) | IPC-CC-830B Type UR | Excellent | -65 to +125°C | Chemical resistance, humidity |
| Silicone (SR) | IPC-CC-830B Type SR | Excellent | -65 to +200°C | Wide temperature range, flexible |
| Epoxy (ER) | IPC-CC-830B Type ER | Excellent | -65 to +150°C | Hardest, chemical resistance, no rework |
| Parylene (XY) | IPC-CC-830B Type XY | Outstanding | -65 to +150°C | Thinnest, most uniform, vapor deposited |
Coating Thickness Requirements (IPC-CC-830B):
Acrylic: 25-75µm (0.001-0.003")
Polyurethane: 25-75µm (0.001-0.003")
Silicone: 50-200µm (0.002-0.008")
Epoxy: 25-75µm (0.001-0.003")
Parylene: 5-25µm (0.0002-0.001")
Keep-out areas (must NOT be coated):
- Connectors and test points
- Switches and pushbuttons
- Heat sinks and thermal interfaces
- Adjustment potentiometers
- LEDs (unless coating is optically clear)
- Battery holders
- Grounding/shielding contact areas
Marine electronics board: Polyurethane conformal coating (HumiSeal 1A33) specified at 50µm thickness. Keep-out areas defined on assembly drawing: connectors masked with peelable tape, test points with removable dots, heatsink mounting area clear. IPC-CC-830B qualification testing performed: 500hr 85°C/85%RH humidity exposure, 10 thermal shock cycles (-55°C to +125°C). Insulation resistance maintained >100MΩ after testing.
Outdoor sensor PCB deployed without conformal coating. After 6 months, moisture ingress causes electrochemical migration between 3.3V and ground traces (0.15mm spacing). Dendritic growth shorts out the power rail. Failure rate increases to 15% per year in humid climates.
Checkpoint 4: HALT/HASS Testing Planned Minor
Highly Accelerated Life Testing (HALT) finds design weaknesses by subjecting products to progressively increasing stress levels. Highly Accelerated Stress Screening (HASS) is the production screen derived from HALT findings.
HALT Process Overview
- Cold step stress: Decrease temperature in 10°C steps (20°C → -60°C or lower). Hold 10 minutes at each step. Monitor for functional failures at each step. Record the lower operating limit (LOL) and lower destruct limit (LDL).
- Hot step stress: Increase temperature in 10°C steps (20°C → 120°C or higher). Same protocol. Record upper operating limit (UOL) and upper destruct limit (UDL).
- Rapid thermal cycling: Cycle between LOL and UOL at maximum ramp rate (40-70°C/minute). Start with 10 cycles, increase range if no failures.
- Vibration step stress: Increase 6-DOF random vibration in 5G steps (5G → 50G or until failure). Record operating and destruct limits.
- Combined environment: Apply thermal cycling + vibration simultaneously. This typically reveals failures not found by either stress alone.
- Analyze all failures: For each failure, determine root cause, implement corrective action, and retest.
HALT Typical Results (Electronics):
Temperature limits (typical industrial product):
LOL (Lower Operating Limit): -40 to -60°C
LDL (Lower Destruct Limit): -60 to -80°C
UOL (Upper Operating Limit): +100 to +130°C
UDL (Upper Destruct Limit): +120 to +160°C
Vibration limits:
Operating limit: 20-40 Grms
Destruct limit: 30-60 Grms
Design margin = Operating limit - Specification limit
Example: Spec = -20 to +70°C, LOL = -55°C, UOL = +115°C
Cold margin = -55 - (-20) = 35°C (good)
Hot margin = +115 - 70 = 45°C (good)
Target: ≥20°C margin on each side for robust design
HASS (Production Screening) Profile Design
HASS Profile Guidelines:
Derived from HALT results (use ~60-80% of operating limits):
Temperature range: 60-80% of (UOL - LOL)
Vibration level: 50-70% of operating vibration limit
Duration: 5-20 thermal cycles + vibration periods
Example HASS Profile:
HALT found: LOL=-55°C, UOL=+115°C (range=170°C)
HASS thermal: 70% × 170 = 119°C range → -25°C to +94°C
HALT vibration operating limit: 35 Grms
HASS vibration: 60% × 35 = 21 Grms
Duration: 10 thermal cycles at 40°C/min + 5 min vibration at each extreme
Total HASS time: approximately 45 minutes per unit
HALT performed on 5 prototype units. Found: 1) Crystal oscillator fails at -50°C (replaced with wider-temp unit), 2) BGA solder joints crack at 45G vibration (added corner staking), 3) Electrolytic cap ESR rises above limit at +110°C (verified lifetime adequate at +85°C max operating). All issues corrected before production. HASS profile screens 100% of production units in 30 minutes, catching infant mortality failures.
No HALT performed. Product shipped based on functional test only at room temperature. Field failures begin at 6 months: cold-start issues in northern climates (-30°C), solder joint failures on boards mounted in vibrating industrial equipment (15G), electrolytic cap dry-out in warm warehouses. Each failure type requires separate field investigation costing $50K+ in engineering time and customer goodwill.
Checkpoint 5: Vibration/Shock Resistance Verified Major
Products in automotive, industrial, military, and transportation applications must withstand vibration and mechanical shock. The PCB design must prevent resonance-induced component failures and ensure solder joint integrity.
PCB Natural Frequency
Natural frequency of a simply-supported rectangular PCB:
f_n = (π/2) × √(D / (ρ×t)) × (1/a² + 1/b²)
Where:
D = Flexural rigidity = E×t³ / (12×(1-ν²))
E = Young's modulus of FR4 ≈ 22 GPa
t = Board thickness (m)
ρ = Density ≈ 2000 kg/m³
ν = Poisson's ratio ≈ 0.15
a, b = Board dimensions (m)
Example: 100mm × 80mm × 1.6mm FR4 board
D = 22×10⁹ × (0.0016)³ / (12 × (1-0.0225))
D = 22×10⁹ × 4.096×10⁻⁹ / 11.73 = 7.68 N·m
f_n = (π/2) × √(7.68 / (2000×0.0016)) × (1/0.1² + 1/0.08²)
f_n = 1.571 × √(2.4) × (100 + 156.25)
f_n = 1.571 × 1.549 × 256.25 = 623 Hz
Rule: Board natural frequency should be >2× the excitation frequency
If vibration spec covers 10-500Hz: f_n = 623Hz > 1000Hz target? No!
Board may resonate within spec range. Add stiffeners or supports.
Design for Vibration Resistance
| Strategy | Effect on f_n | Implementation |
|---|
| Increase board thickness | f_n ∝ t | 2.0mm instead of 1.6mm (+25%) |
| Add mid-board support | f_n × 4 (center support) | Standoff or boss at center |
| Reduce unsupported span | f_n ∝ 1/a² | Additional mounting points |
| Stiffener bars | Increases effective D | Aluminum bar epoxied to board edge |
| Component staking | Prevents relative motion | Adhesive on heavy/tall components |
| Strain relief on leads | Prevents fatigue | Compliant leads, stress relief bends |
Automotive ECU (vibration spec: 10-2000Hz, 10G): PCB (160×100mm) mounted at 6 points (4 corners + 2 center edges). Calculated f_n = 1450Hz (with center supports). All components <2g glued with silicone adhesive. Large connectors have strain-relief mounting brackets. BGA corners staked with underfill. Passed IEC 60068-2-64 random vibration testing for 8 hours without failure.
Industrial control board (150×100mm) mounted at 4 corners only in a vibrating motor cabinet. Board f_n = 450Hz, within the excitation spectrum. After 3 months, crystal oscillator (tall, heavy package) breaks free. Tantalum caps crack from board flexing. Through-hole connector pins fatigue at solder joints. 30% field failure rate within first year.
Checkpoint 6: Corrosion Protection for Harsh Environments Major
For products operating in humid, chemical, salt-spray, or outdoor environments, corrosion protection measures must be implemented to prevent electrochemical degradation of conductors and solder joints.
Corrosion Mechanisms in Electronics
Electrochemical Migration (ECM):
Occurs when: Moisture + Voltage bias + Close conductor spacing
Rate increases with: Higher voltage, closer spacing, more humidity
Critical threshold: >60% relative humidity enables ECM
Dangerous spacing: < 0.3mm between conductors at >5V
Galvanic Corrosion Risk (dissimilar metals):
High risk pairs (>0.5V galvanic potential):
- Gold/Copper → Steel/Aluminum
- Tin/Solder → Copper (when contaminated)
- Silver → Aluminum
Prevention hierarchy:
1. Conformal coating (barrier to moisture)
2. Adequate clearance (>0.5mm at <50V per IPC-2221)
3. Compatible surface finishes (minimize galvanic potential)
4. Enclosure sealing (IP65 or better for outdoor)
5. Desiccant packs (absorb trapped moisture)
Protection Strategies by Environment
| Environment | Primary Risk | Protection Method | Test Standard |
|---|
| Indoor, controlled | Minimal | Standard solder mask sufficient | N/A |
| Indoor, industrial | Chemical fumes, dust | Conformal coating (acrylic) | IPC-CC-830B |
| Outdoor, sheltered | Humidity cycling | Conformal coating + sealed enclosure | IEC 60068-2-30 |
| Outdoor, exposed | Rain, UV, temperature | Potting or sealed enclosure (IP67) | IEC 60529 |
| Marine/coastal | Salt spray | Conformal coating (UR/XY) + IP66 | IEC 60068-2-11 |
| Automotive underhood | Salt, oil, heat, vibration | Conformal coating + sealed housing | ISO 16750 |
| Subsea/underwater | Continuous immersion | Full potting (epoxy/polyurethane) | MIL-STD-810H |
Agricultural sensor PCB: Environment = outdoor, rain/dust/chemicals. Protection: 1) PCB surface finish: ENIG (most corrosion-resistant). 2) Conformal coating: Parylene-C (25µm, vapor deposited for uniform coverage). 3) Sealed enclosure: IP67 rated aluminum housing with silicone gaskets. 4) Breather vent with Gore-Tex membrane (pressure equalization without moisture entry). 5) Silica gel desiccant inside enclosure. Passed: 1000hr salt spray (ASTM B117) and 85°C/85%RH for 1000hr without insulation resistance degradation.
Pool/spa controller PCB in plastic enclosure with vent holes (for heat dissipation). Chlorine vapor from pool chemistry enters enclosure. Within 18 months, copper traces corrode, solder joints develop green oxide, and relay contacts become intermittent. PCB surface finish (bare copper OSP) provided no protection against corrosive chlorine atmosphere.
