
The 7 main PCB testing methods are: AOI (optical inspection, $0.10-0.20/board, catches 85% visual defects), ICT (in-circuit test, $0.30-0.80/board, tests components electrically), FCT (functional test, $0.50-2.00/board, verifies operation), X-Ray (BGA inspection, $1-3/board), Flying Probe (prototype-friendly, no fixture needed), SPI (solder paste inspection, pre-reflow), and Burn-in (stress testing, 24-72 hours). Most manufacturers combine AOI+ICT+FCT for optimal coverage.
The $2 Million Lesson I Learned About Testing
True story: Early in my career, a customer shipped 50,000 units of a medical monitoring device with "100% tested" boards. Three months later, field failure rate hit 8%. The root cause? We tested functionality (FCT) but skipped AOI. A batch of capacitors had hairline cracks invisible to the eye but perfectly detectable with optical inspection.
Total recall cost: over $2 million. The AOI that would've caught it? About $0.15 per board.
Testing isn't a cost—it's insurance. And like insurance, the right combination matters more than any single method.
Quick Reference: All 7 Testing Methods
| Method | Type | What It Catches | Speed | Cost/Unit | Best For |
|---|---|---|---|---|---|
| **SPI** | Inspection | Solder paste defects | Fast | $0.02-0.05 | Pre-reflow QC |
| **AOI** | Inspection | Visual defects | Fast | $0.05-0.15 | Post-reflow screening |
| **AXI (X-Ray)** | Inspection | Hidden solder joints | Medium | $0.20-0.50 | BGA, QFN boards |
| **ICT** | Electrical | Component values, shorts | Fast | $0.10-0.30 | High-volume |
| **Flying Probe** | Electrical | Component values, shorts | Slow | $0.50-5.00 | Low-volume, prototypes |
| **FCT** | Functional | System operation | Varies | $0.50-5.00 | Final validation |
| **Burn-in** | Stress | Infant mortality | Slow | $5-50+ | High-reliability |
**Hommer's Rule of Thumb**: Most projects need at least 3 of these. The question is which 3.
1. SPI: Solder Paste Inspection
What Is SPI?
Solder Paste Inspection happens *before* components are placed. A 3D scanner measures the volume, height, and area of solder paste deposits on each pad.
Why It Matters
Here's a stat that surprised me when I first learned it: 70-80% of SMT defects originate from solder paste printing. Not pick-and-place. Not reflow. Paste printing.
| Defect | Cause | SPI Detection |
|---|---|---|
| Bridging | Excess paste | ✅ Yes |
| Opens | Insufficient paste | ✅ Yes |
| Tombstoning | Uneven paste volume | ✅ Yes |
| Cold joints | Wrong paste height | ✅ Yes |
When to Use SPI
- Always for high-reliability products ([medical](/industries/medical), [automotive](/industries/automotive))
- Fine-pitch components (0.4mm BGA, 0201 passives)
- New product introduction (NPI)
- When yield is below target
The Reality
SPI catches problems *before* they become expensive. A board caught at SPI costs pennies to rework. The same defect caught at FCT? Could mean scrapping the entire assembly.
2. AOI: Automated Optical Inspection
What Is AOI?
AOI uses high-resolution cameras (often 3D now) to photograph every component and compare it against the expected placement. Think of it as a very fast, very consistent human inspector that never gets tired.
What AOI Catches
Excels at: - Missing components - Wrong components (wrong value, wrong part) - Polarity errors (caps, diodes, ICs) - Placement offset - Solder bridges (visible) - Solder insufficiency (visible) - Tombstoning
Struggles with: - Hidden solder joints (BGA, QFN) - Intermittent connections - Wrong component values (same package) - Functional defects
Real-World Coverage
In our SMT production lines, AOI typically catches 85-90% of assembly defects. That's excellent for surface-visible issues, but it's not 100%.
The Speed Advantage
| Inspection Method | Time per Board (typical) |
|---|---|
| Manual visual | 5-15 minutes |
| AOI | 15-45 seconds |
| 3D AOI | 30-60 seconds |
For high-volume production, AOI isn't optional—it's mandatory.
3. AXI: Automated X-Ray Inspection
What Is AXI?
AXI shoots X-rays through your board to see what cameras can't: the solder joints hidden under BGAs, QFNs, and other bottom-terminated components.
Why X-Ray Exists
Modern components are increasingly hidden. A BGA might have 500 solder balls—none visible from above. If even one fails, the product fails. AOI literally cannot see these joints.
What X-Ray Reveals
| Defect | Description | Detection |
|---|---|---|
| **Voids** | Air bubbles in solder | ✅ Excellent |
| **Bridging** | Shorts between BGA balls | ✅ Excellent |
| **Head-in-Pillow** | Partial BGA connection | ✅ Good |
| **Open joints** | Missing connection | ✅ Excellent |
| **Insufficient solder** | Weak joints | ✅ Good |
When X-Ray Is Essential
- BGAs (especially fine-pitch < 0.8mm)
- QFN/DFN packages
- [Automotive electronics](/industries/automotive) (IATF 16949 often requires it)
- Aerospace and defense
- Medical implants
The Cost Reality
X-ray inspection adds $0.20-0.50 per board for inline AXI. Manual X-ray for sampling is cheaper but slower. For boards with multiple BGAs, it's worth every cent.
**Hommer's Experience**: We once caught a systematic BGA soldering issue using AXI that would've caused 100% field failure. The reflow profile was *just* off enough to create head-in-pillow defects invisible to AOI. X-ray caught all of them.
4. ICT: In-Circuit Testing
What Is ICT?
In-Circuit Testing uses a "bed of nails" fixture—hundreds of spring-loaded probes that contact test points on your board simultaneously. It then measures every component value and checks every net for opens/shorts.
The Power of ICT
ICT is brutally thorough for manufacturing defects:
| Test | What It Checks |
|---|---|
| Analog | Resistor/capacitor values |
| Digital | IC presence and orientation |
| Connectivity | Opens and shorts |
| Power | Voltage rails |
Defect coverage: Up to 98% for *manufacturing* defects.
ICT Economics
Here's the catch: ICT requires a custom fixture for each board design.
| Cost Factor | Typical Range |
|---|---|
| Fixture cost | $5,000-$25,000 |
| Test time | 30 seconds - 2 minutes |
| Cost per test | $0.10-0.30 |
| Break-even volume | 500-2,000 units |
When ICT Makes Sense
✅ Use ICT when: - Production volume > 1,000/year - Design is stable (no frequent revisions) - Board has dedicated test points - High defect cost (automotive, medical)
❌ Skip ICT when: - Prototype or low-volume - Frequent design changes - No test point access - Budget doesn't support fixture
Our testing services include ICT for qualifying production runs.
5. Flying Probe: The Flexible Alternative
What Is Flying Probe?
Instead of a custom fixture with hundreds of probes, flying probe testers use 4-8 motorized probes that move to each test point sequentially. Same electrical tests as ICT, no fixture required.
Flying Probe vs ICT: Head-to-Head
| Factor | ICT | Flying Probe |
|---|---|---|
| Setup cost | $5,000-25,000 | $500-2,000 |
| Setup time | 2-4 weeks | 1-3 days |
| Test time | 30 sec - 2 min | 5-15 minutes |
| Cost at 100 pcs | ~$30/unit | ~$5/unit |
| Cost at 10,000 pcs | ~$0.15/unit | ~$5/unit |
| Design changes | Fixture modification | Software update |
The Sweet Spot
Flying probe wins for: - Prototypes (absolutely) - Production under 500-1,000 units - Boards with frequent revisions - Emergency production runs
ICT wins for: - High-volume (>2,000/year) - Stable designs - When test time matters
My Recommendation
For most of our customers at PCB Portugal, I recommend flying probe for NPI and early production, then transitioning to ICT once the design stabilizes and volumes justify the fixture cost.
6. FCT: Functional Circuit Testing
What Is FCT?
Functional testing powers up the board and verifies it actually *works*—not just that components are present and connected, but that the system performs its intended function.
FCT vs Structural Tests (AOI/ICT)
| Aspect | Structural Tests | Functional Tests |
|---|---|---|
| Question answered | "Is it built right?" | "Does it work?" |
| Defects found | Assembly errors | Design + assembly errors |
| Speed | Fast | Varies |
| Coverage | Manufacturing defects | Real-world operation |
Types of Functional Tests
- **Power-on test**: Does it boot?
- **Communication test**: Can it talk (UART, SPI, I2C)?
- **Peripheral test**: Do LEDs, buttons, sensors work?
- **Performance test**: Does it meet specifications?
- **Calibration**: Adjust for production variation
FCT Reality Check
FCT catches what other tests miss: - Firmware bugs triggered by hardware variation - Timing-sensitive failures - Noise/interference issues - Thermal-related problems
But FCT is only as good as the test coverage. A 30-second FCT that only checks "power LED on" isn't the same as a 5-minute FCT that exercises every function.
**Hommer's Insight**: I always ask customers: "What would you test if you only had 60 seconds with each board?" That answer should be your minimum FCT.
7. Burn-in Testing: The Stress Test
What Is Burn-in?
Burn-in subjects boards to elevated temperature (and sometimes voltage) for extended periods—typically 24-168 hours. The goal: accelerate "infant mortality" failures that would otherwise occur in the field.
The Bathtub Curve
Electronics failures follow a predictable pattern:
- **Infant mortality** (early failures): Weak components fail quickly
- **Useful life**: Low, random failure rate
- **Wear-out**: End-of-life failures increase
Burn-in eliminates category 1 *before* shipping.
When Burn-in Is Required
| Industry | Typical Burn-in | Why |
|---|---|---|
| Military/Aerospace | 168+ hours | Mission-critical |
| Medical implants | 72-168 hours | Life-critical |
| Automotive ECU | 24-48 hours | Safety + warranty |
| Telecom infrastructure | 48-72 hours | Uptime requirements |
| Consumer | Rarely | Cost-sensitive |
The Cost of Burn-in
Burn-in is expensive: - Equipment (ovens, monitoring) - Time (days, not minutes) - Energy - Floor space
But for a pacemaker? A satellite? An aircraft system? Absolutely justified.
Building Your Test Strategy
The Minimum Viable Test Plan
For most commercial products, I recommend:
Essential (non-negotiable): 1. AOI after reflow 2. FCT before shipping
Add based on risk: - SPI for fine-pitch assembly - X-ray for BGAs - ICT/Flying Probe for electrical verification
Test Strategy by Industry
| Industry | Recommended Tests |
|---|---|
| **Consumer electronics** | AOI → FCT |
| **Industrial control** | AOI → Flying Probe → FCT |
| **[Automotive](/industries/automotive)** | SPI → AOI → ICT → AXI → FCT |
| **[Medical](/industries/medical)** | SPI → AOI → ICT → AXI → FCT → Burn-in |
| **Aerospace/Defense** | Everything + environmental |
Test Strategy by Volume
| Annual Volume | Recommended Electrical Test |
|---|---|
| 1-100 | Flying Probe |
| 100-1,000 | Flying Probe (consider ICT transition) |
| 1,000-10,000 | ICT |
| 10,000+ | ICT + boundary scan |
Cost-Benefit Analysis
Testing ROI Example
Let's do the math for a 10,000-unit production run:
Scenario: Skip AOI to "save money" - AOI cost saved: $0.10 × 10,000 = $1,000 - Typical defect escape rate without AOI: 2% - Defective units: 200 - Field failure cost per unit: $150 (including shipping, repair, reputation) - Total cost of escapes: $30,000
Net "savings": -$29,000
That's why we never skip AOI.
Where to Invest First
If budget is limited, prioritize in this order:
- **AOI** - Best cost/coverage ratio
- **FCT** - Catches functional issues
- **ICT/Flying Probe** - For electrical verification
- **X-Ray** - If you have BGAs
- **SPI** - For yield improvement
- **Burn-in** - For high-reliability applications
Common Testing Mistakes
Mistake 1: Testing Only at the End
"We'll catch everything at FCT."
No, you won't. FCT tells you the board doesn't work, but not *why*. Without earlier tests, debugging is painful and expensive.
Mistake 2: Over-Testing Low-Risk Products
A simple LED driver doesn't need X-ray, ICT, burn-in, and FCT. Match the test strategy to the risk.
Mistake 3: No Test Points in Design
If you want ICT, you need test points. Design them in from the start—adding them later means a board spin.
Mistake 4: Ignoring Test Coverage Reports
Your test coverage isn't 100% just because you have all the equipment. Review coverage reports. Improve test fixtures. Iterate.
FAQ
How much does a complete test suite cost per board?
For a typical industrial board with BGA: - SPI: $0.03 - AOI: $0.10 - ICT: $0.15 - AXI (sample): $0.05 - FCT: $0.50
Total: ~$0.83/board
Compare to a single field failure at $100-500. Testing is cheap.
Can testing guarantee zero defects?
No. Testing reduces defects but can't eliminate them entirely. Even 100% testing catches 95-99% of defects. The goal is an acceptable defect level, not perfection.
What's boundary scan testing?
Boundary scan (JTAG) tests digital ICs through their debug interface. Useful for dense digital boards where physical probe access is limited. We consider it a supplement to ICT, not a replacement.
How do I choose between ICT and Flying Probe?
Rule of thumb: If you'll test more than 1,000 identical boards per year, ICT. Otherwise, flying probe.
Conclusion: My Testing Philosophy
After 15 years in this industry, here's what I believe:
- **No single test catches everything**. Use multiple methods.
- **Catch defects early**. SPI catches what becomes unfixable later.
- **Match testing to risk**. Not every product needs aerospace-level QA.
- **Design for test**. Test points aren't optional.
- **Trust but verify**. Review coverage data regularly.
Need help designing a test strategy for your product? Contact our engineering team or explore our testing capabilities.
Related Reading
Explore more technical guides that complement your testing knowledge:
- **[Turnkey vs Consignment Assembly](/blog/turnkey-vs-consignment-pcba)** - Testing is included differently depending on your assembly model. Understand the implications.
- **[PCB Materials Comparison](/blog/pcb-materials-comparison)** - Material choice affects test accessibility. Aluminum boards need X-ray for thermal via inspection.
- **[Wire Harness & Cable Assembly Guide](/blog/wire-harness-cable-assembly-guide)** - Cable assemblies have their own testing requirements (continuity, hi-pot, pull testing).
References
- [IPC-A-610](https://www.ipc.org/) - Acceptability of Electronic Assemblies
- [IPC-9252](https://www.ipc.org/) - Requirements for Electrical Testing of Unpopulated Printed Boards
- [JEDEC JESD22-A108](https://www.jedec.org/) - Temperature Cycling
*Written by Hommer Zhao, founder of PCB Portugal. Based on testing strategies developed over 15+ years of PCBA production experience. Last updated: December 2024.*

Fundador & Especialista Técnico
Fundador da WellPCB com mais de 15 anos de experiência em fabrico de PCB e montagem eletrónica. Especialista em processos de produção, gestão de qualidade e otimização da cadeia de fornecimento.
Ver todos os artigos deste autor →Artigos Relacionados

Turnkey vs Consignment PCBA: Which Model Saves You Money?
Which PCB assembly model is right for you? Compare full turnkey, consignment, and partial turnkey PCBA services. Cost analysis, pros/cons, and real-world recommendations.

SMT vs THT: Guida al Montaggio PCB [2025]
Confronto tra montaggio SMT e THT. Costi, velocità e applicazioni per ogni tecnologia.

Top 5 Metodi Test PCB: ICT, FCT, AOI [2025]
Confronto metodi test PCB. Quando usare ciascuno e come combinarli.