At 11 p.m., Jason’s phone vibrated violently on his bedside table. On the other end of the line, the factory manager’s voice was a note higher than usual: “President Jason, we’ve got a major problem!
We can’t fit all 5,000 sets of casings—the capacitor on the circuit board is 0.3 millimeters taller than the space allocated in the design.”
Jason is the hardware partner at a smart wearable company.
They had just landed a major overseas order for a smart ring capable of monitoring heart rate—a device designed to be as compact as possible.
Structural and hardware engineers were in a heated argument in the group chat:
The structural engineers put forward their viewpoint.
They stated that they reserved a 1.2 mm clearance in accordance with the specification sheet.
On the other hand, the hardware engineers held a different opinion.
They pointed out that they chose a capacitor model with a thickness of 1.0 mm.
Therefore, they believed sufficient assembly space was available.
The root cause was found after investigation. The capacitor package in the schematic library is a general model.
This model fails to consider one key feature of the SMD capacitor.
The solder pad on the top of the surface-mount capacitor has minor protrusion.
The actual assembly height was 1.5 millimeters.
Because of those 0.3 millimeters, 300,000 in mold costs, the cost of the first batch of materials, and the penalty for late delivery all went down the drain.
This true story reveals a harsh reality: component package selection in PCB design is never just an engineer’s responsibility—it’s a risk thread that runs through R&D, procurement, manufacturing, and after-sales service.
Executives may not need to know how to design PCBs, but they must be able to identify the source of the explosion caused by package control failure.
Below, drawing on the six ironclad rules compiled by Jason’s team during their post-incident review, I’ll dig up and reveal to you, one by one, all the landmines buried in this incident.
Iron Rule #1
A package is not just a two-dimensional pad; it is a three-dimensional spatial contract.
The first pitfall Jason fell into was treating the package as nothing more than an electrical connection and a flat-surface dimension.
Many managers think the same way: as long as the pads are provided and the component can be soldered onto them, isn’t that the end of the story?
This is completely wrong. A complete definition of a package must include mechanical dimensions along the X, Y, and Z axes.
Engineers must consider the height of polarized capacitors and treat the outline dimension of inductors as another key factor.
Connectors require adequate insertion and extraction clearance. In addition, designers need to confirm if wiring can pass under components.
Engineers must evaluate all these factors, as they directly influence assembly feasibility and determine whether designers can place all components inside the housing.
In the smart ring project, failing to perform a comprehensive 3D interference check turned a “height-compliant” capacitor into an assembly nightmare.
Nowadays, mainstream EDA tools all come with 3D preview capabilities.
From then on, Jason established a rule: designers must generate 3D models for all board-level components before mechanical design freeze and conduct fit-and-assembly verification together with the enclosure.
Spending an extra half-day saves 300,000—it’s a calculation that executives understand all too well.
Iron Rule #2
Through-hole versus surface-mount—it’s not just a matter of cost; it’s a matter of having the leeway to experiment.
After that incident, Jason completely overhauled the hardware team’s decision-making process.
Originally, in pursuit of miniaturization, they had exclusively used ultra-small surface-mount components—0402 resistors and capacitors, and BGA control chips.
It seemed to save space, and the calculations showed it was even cheaper.
But when hand-soldering the first batch of prototypes, it took two days just to fix cold solder joints, and they had no idea where to even start debugging the jumpers.
During the post-mortem, they established a principle:
During the prototype validation phase, prioritize slightly larger SMD packages or even through-hole components to allow for operational flexibility during manual debugging.
Switch to miniaturized packages simultaneously during mass production.
Prototypes are meant to “find problems,” not to “show off perfection.”
This point is especially important for executives: the bulk of R&D costs typically lies not in material price differences, but in troubleshooting time.
Leaving a safety net for the debugging phase is equivalent to purchasing a time insurance policy for the entire project.
If a suitable package cannot be found in the standard library, it is absolutely worth spending some time to customize one—this is not a task where “good enough” is acceptable.
Iron Rule #3
The ground plane and decoupling capacitors are the invisible foundations of performance.
But the story doesn’t end there.
After Jason and his team finally resolved the assembly issues, the first batch of prototypes was sent for EMC certification—only to fail the radiated emissions test again.
Engineers traced the root cause after three days of troubleshooting: partial fragmentation of the ground plane and a decoupling capacitor placed 1.5 cm away from the power supply pin.
The engineers thought, “It’s all ground anyway—just connect it,” but they overlooked the fact that high-frequency circuits require a complete, low-impedance return path.
Improperly selected decoupling capacitors and haphazard placement directly turned the circuit board into a miniature noise broadcast station.
Jason later added a note to the hardware design review checklist that senior management could easily understand:
“If anyone fragments the ground plane for the sake of routing convenience, or places a decoupling capacitor haphazardly just to get it done, please let me know.
This could force the product into another round of EMC remediation, delaying the schedule by at least three weeks.”
Since then, no one has dared to tamper with the ground plane carelessly.
Iron Rule #4
Virtual components are hidden time bombs in the BOM.
Just when everyone thought the worst was finally over, the procurement department stormed into the R&D department with the BOM:
“Why are there five items on this list with no supplier, no part number, and not even any packaging information?”
Upon investigation, it turned out that “virtual components” in the schematic were to blame.
Power symbols, ground symbols and ideal sources are commonly used for simulation.
These symbols can float freely on schematic diagrams. However, none of them ought to appear in the Bill of Materials.
The BOM is used for physical component procurement and assembly.
Even one virtual component remaining in the BOM will trigger risks.
In mild cases, the procurement department cannot issue formal purchase orders.
In severe cases, it will cause production line stoppages caused by missing materials.
This risk may unexpectedly consume one to two weeks of the project schedule.
From then on, Jason’s team implemented a strict rule:
Engineers shall filter out all virtual components line by line (except power and ground symbols) before finalizing the BOM and replace them with real components equipped with complete supplier information.
The management logic behind this is straightforward: any break in the data flow becomes a drain on cash flow.
Iron Rule 5
The Ordering of the Bill of Materials Holds the Key to Supply Chain Success or Failure
After that incident, Jason personally reviewed the BOM and discovered another, even more hidden pitfall: the part numbers were in disarray.
Some component designations were incomplete, some lacked specification descriptions, and others were missing manufacturer part numbers.
The procurement team had no choice but to verify each one manually—a process that was both slow and highly prone to errors.
A full BOM features continuous item numbers and complete required information.
Such a BOM brings convenience for engineers to carry out reviews.
More importantly, it exerts a direct impact on subsequent work. It decides whether procurement can rapidly find qualified suppliers.
It also determines whether contract manufacturers can mount components accurately.
If tolerance specifications are missing for any component, risks will emerge.
Operators may assemble the entire batch of circuit boards with resistors of improper precision.
This problem will eventually cause product performance drift. Consequently, continuous customer complaints will follow.
In the eyes of senior executives, missing information for even a single component represents a potential recall cost.
Iron Rule No. 6
Leave unused gate circuits floating—simulation software won’t tell you just how dangerous that is.
This last story sounds most like a ghost story.
After the product was launched, a very small number of users reported that the ring would occasionally “freeze” and require a reset.
For a month, we couldn’t find the cause.
That is, until a senior engineer suddenly asked, “How is the input of that unused channel in the op-amp handled?”
We reviewed the design. The related pin was left in a floating state.
Floating pins tend to oscillate randomly under noisy working conditions.
They are prone to coupling interference signals. In extreme cases, they may trigger abnormal operation of the chip.
Nevertheless, simulation models always display stable and ideal signal status.
Designers can implement an extremely straightforward solution: they shall ground all unused logic gate inputs or connect them to a fixed voltage level.
Engineers shall configure unused op-amp channels as voltage followers or set them to a fixed state; they must never leave these channels floating.
Summary
Jason later often repeated a phrase to the team—one that executives would do well to include in their management principles:
“Passing simulation does not mean the design is safe; floating pins are risks waiting to happen.”
From the series of incidents that nearly derailed a smart ring project, Jason and his team distilled these six ironclad rules, all of which point to the same truth:
The selection of component packages is, from the very beginning, a cross-functional management decision—not merely a technical choice.
It is inextricably linked to structural space, procurement data, manufacturing processes, signal integrity, material costs, and even post-sales quality.