Why a working prototype isn’t enough
If you’ve ever built a GPS or temperature logger and watched it perform flawlessly on your desk, only to see it stumble in production, you’re not alone. Hardware prototypes often gloss over the messy details of manufacturing, testing and supply chains. To deliver a low‑power IoT tracker that ships by the tens of thousands and survives years in the field, you must design for repeatability and resilience from day one.
This post walks through the stage‑gated journey from engineering validation to mass production (EVT → DVT → PVT), unpacks the nuts and bolts of production testing and burn‑in, and explains why diversifying manufacturing between China and Vietnam strengthens supply chains. While the examples focus on asset‑tracking devices, the principles apply to any battery‑powered IoT hardware.
Stage gates: EVT, DVT & PVT at a glance
Manufacturing disciplines use three gates to move a design from “looks good” to “builds repeatedly”:
EVT (Engineering Validation Test). This is the first time the design is built using production‑intent materials and processes. EVT builds range from 100 to 1 000 units and must be fully functional. The goal is to validate the core architecture and catch design flaws early. All functional test stations are present to collect data.
DVT (Design Validation Test). At DVT you commit to a single production‑worthy design: hard tools and mass‑production capable processes are used, and thousands of units may be built. The focus shifts to yield and reliability; test limits are tightened and corrections must be verified. Exit criteria require high confidence that the design will yield at acceptable levels.
PVT (Production Validation Test). PVT units are the first ones intended for sale. This gate tests the factory itself: fixtures, cycle time and yield must all run at mass‑production rates. Builds typically range from thousands to tens of thousands, and production must achieve mass‑production yields on at least one line.
These gates aren’t just forms to check off: they expose problems when it’s still cheap to fix them. Skipping them usually means paying for the same work later, plus rework and delays.
Designing test systems that scale
Testing isn’t the last step of a build – it’s a design activity. The devices you ship should be the ones you’d buy yourself. Here’s how to build a test strategy that scales.
Separate board integrity from system function
Production lines run in layers:
SMT inspection & in‑circuit test (ICT). During surface‑mount assembly, automated optical inspection and X‑ray machines catch tombstones and solder bridges. ICT can be used to validate power rails and key nets.
Functional Test (FCT). In a custom fixture, your board powers up, reports over its radio and exercises sensors. FCT fixtures must provide stable power, known loads and reliable mechanical contact via pogo pins or micro‑coax. Inputs are simulated and outputs are captured; anything outside tolerance causes the board to fail.
System test. Once assembled into its enclosure, the complete tracker should operate in its intended environment (temperature, vibration, shock). This test often runs in parallel with burn‑in.
Burn‑in: screening out latent defects
Burn‑in testing accelerates ageing by subjecting boards to elevated temperatures (and sometimes voltages) to stress weak components. For example, smartphone boards may be baked at 100 °C for 16 hours. The idea isn’t to destroy boards, but to force early failures to show up on the line rather than in the field. In a low‑power tracker, burn‑in might include powering the device with its typical duty cycle (GPS fix, sleep, transmit) while it bakes. When burn‑in is complete, any boards that still pass functional test are likely to remain reliable.
Don’t forget calibration and firmware paths
Many IoT devices rely on sensors with some calibration (e.g. temperature offset, accelerometer bias). Plan to write calibration constants during test, either into EEPROM or with fuses. Provide a manufacturing firmware mode that boots quickly, bypasses battery‑powered sleep and includes hooks for test and calibration. After test passes, update to release firmware.
Build traceability into the process
When field returns happen (and they will), you’ll want to trace them back to a lot, BOM version and test data. At minimum, log:
- Serial/IMEI and lot or date code
- Firmware and hardware revisions
- Key parametric measurements (current draw, RF power, sensor outputs)
- Pass/fail results at each station
Traceability turns guesswork into root‑cause analysis. Without it, you’re left swapping boards until the failure reproduces.
Yield and reliability metrics
During EVT and DVT you’ll gather data. Use it. Track the following:
- First pass yield (FPY) at each station (ICT, FCT, system test).
- Final yield (boards that pass all stations divided by total). Use this to decide when you can exit DVT.
- Failure Pareto. What are the top failure modes? Solder issues, component tolerance, firmware? Fix the big ones and iterate.
- Power budget compliance. Measure average current in realistic duty cycles. A design that hits battery targets in the lab may drift when you change suppliers.
Dual manufacturing: resilience through diversification
No test plan can save you from supply‑chain shocks. Pandemic disruptions, tariffs, natural disasters and geopolitical tension have made “China Plus One” a common strategy. Here’s why.
China’s strengths. China remains the world’s largest manufacturing base and accounts for a huge share of global high‑tech exports. It offers sophisticated logistics – goods can reach the United States in weeks – and mature supply chains for high‑tech components. Its labour productivity and scale make it the default for complex, high‑volume products.
Vietnam’s strengths. Vietnam has become a key “China Plus One” destination thanks to its young, educated workforce and competitive labour costs: labour averages about US$3 per hour versus higher rates in China. Major brands like Samsung, LG and Foxconn have invested there. Vietnam’s shipping times to the U.S. – 24 to 41 days – are comparable to China’s. The country has aligned a large portion of its national standards with international norms, improving quality and compliance.
Resilient supply chains. By splitting production between China and Vietnam, companies can mitigate risks. Tariff changes, port closures or local outbreaks can be absorbed. Labour‑intensive assemblies can move to Vietnam for cost savings, while complex high‑tech modules stay in China. This diversification is part of a broader China Plus One approach: firms invest in Southeast Asia (Vietnam, Malaysia, Thailand) to reduce dependency on a single country and to manage costs when labour in China is no longer cheap.
Summary: engineering discipline meets supply‑chain strategy
Building a production‑ready IoT tracker isn’t about hacking together a prototype – it’s about engineering a repeatable system. Stage gates (EVT, DVT, PVT) de‑risk the design; layered testing, burn‑in and traceability ensure reliability; and yield metrics reveal when a design is mature. Finally, dual manufacturing in China and Vietnam provides cost flexibility and resilience. If you want to ship thousands of low‑power trackers that just work, pay attention to process and supply chain as much as you do to schematics.
Top comments (0)