You tested 50 boards. All passed. You placed the 5,000-unit order.
Then the problems started.
This is the moment every hardware engineer dreads. The design worked. The prototypes passed. But something happened between “works in the lab” and “works on the line.” That something is what an NPI pilot run PCBA process is supposed to catch—but only if you’re looking for the right things.
Here are three real examples of what can go wrong when the gap between pilot and production isn’t properly closed.
A client designed a 4-layer board with two modules—an EG912U on the top side and an ESP32 on the bottom side. The modules were placed so that their footprints overlapped diagonally at the corners.
The prototypes passed. The pilot run passed. Then the first production batch came off the line, and 30% of the boards failed functional test.
The issue: during reflow soldering, the overlapping copper patterns created a localized thermal mass concentration. One side of the board absorbed heat faster than the other, causing uneven temperature distribution. Some BGA balls reached full reflow temperature. Others didn’t. The result was cold solder joints that passed AOI inspection but failed under load.
The modules were electrically correct. The footprints were per the datasheets. But the layout—specifically, placing two large modules on opposite sides with overlapping footprints—created a thermal imbalance that no one had anticipated.
What an NPI pilot run PCBA process should have caught: A proper thermal analysis during the pilot phase would have identified the uneven heat distribution. Instead, the problem was only discovered when 1,500 boards needed rework.
The fix: The client relocated the modules to eliminate the footprint overlap. No BOM changes. No component substitutions. Just a layout adjustment that should have happened before production started.
A client needed a small product encapsulated with hot-melt potting compound. They had been doing the potting manually in their UK lab—small batches, careful work, no issues.
When it came time to scale to 5,000 units, they assumed the same process would work. They were wrong.
Manual potting at 3 minutes per unit doesn’t sound bad until you multiply it by 5,000. That’s 250 hours of labor, assuming no breaks, no errors, no rework. Their delivery timeline simply couldn’t absorb it.
The solution: We designed a custom potting fixture that held the enclosure in place, controlled the liquid level, and included an integrated clamping mechanism—so workers didn’t have to hold the parts while the compound cured.
We built nine fixtures for the production line. The result: cycle time dropped from 3 minutes per unit to 10 units per minute. One worker could manage 4-5 fixtures simultaneously, monitoring multiple units at once.
What an NPI pilot run PCBA process should have caught: The client’s assumption that “works in the lab” equals “works on the line” was the real failure. A proper pilot run using production-scale tooling—not lab-scale manual work—would have revealed the throughput gap before the 5,000-unit order was placed.
A client developed functional test firmware for their product. It was comprehensive—designed to simulate real-world usage, validate every feature, and catch every possible failure mode.
It took 5 minutes per board.
In the lab, that was fine. The engineer ran the test on a handful of prototypes, captured data, and moved on. But on a production line, 5 minutes per board is an eternity. For a 5,000-unit order, that’s 417 hours of testing time.
The client hadn’t considered that production testing has different requirements than engineering validation. Production testing needs to be fast, repeatable, and focused on catching assembly defects—not simulating every possible user scenario.
We flagged the issue during our pre-production review. We showed the client the math: 5 minutes × 5,000 units = no chance of meeting their delivery deadline.
The solution: The client went back and developed a streamlined test firmware specifically for production—focused on critical checks, optimized for speed, and designed to run on our test fixtures. The new test took 30 seconds per board.
What an NPI pilot run PCBA process should have caught: The client assumed the test firmware was “done.” But “done for engineering” and “done for production” are two different things. A pilot run using production test fixtures would have exposed the test time issue immediately, giving the client weeks of lead time to develop the streamlined version instead of emergency-revising it at the last minute.
All three were avoidable. All three were discovered during the NPI pilot run PCBA phase—but only because we were looking for them.
The purpose of an NPI pilot run isn’t to prove the design works. The prototype already proved that. The purpose is to prove that the design can be manufactured consistently, at volume, on schedule.
That means testing with production tooling, production processes, and production timelines—not lab conditions.
If you’re moving from prototype to production, ask yourself these questions before you place the order:
If the answer to any of these is “no,” you’re not ready for production.
An NPI pilot run PCBA is your last chance to find problems before they find your customers. Don’t waste it.
Ready to run a pilot that actually tells you something? Contact us and we’ll help you identify what’s missing before it becomes a production issue.
