The production process can reveal things that are difficult to see from product files alone. A product can look straightforward on paper and still behave differently once it enters production.
We have seen this in several very different products.
Sometimes the issue appears during assembly. Sometimes a test sequence turns out to be much slower than expected. Sometimes the product works normally, but the measurement result changes with the conditions under which it is calibrated.
These are not necessarily design failures.
They are simply things that became visible when the products were actually being built and tested.
One vehicle connectivity product used a plastic enclosure filled with hot-melt adhesive to provide a potting-like effect.
The PCB also formed part of the enclosure structure, so it had to remain level during the filling process.
A fixture helped with positioning, but the operator still had to control the filling speed and judge the final adhesive level. During assembly, the operator sometimes needed to press the PCB slightly to adjust its angle.
The trial production revealed an unexpected problem.
The product used an external coaxial cable. A small section of the metal shielding layer had not been completely covered during cable stripping.
Normally, this was difficult to notice.
But when the operator pressed the PCB into position, the exposed shielding could touch components on the PCB and create a short circuit.
There was another complication: the enclosure had very little space around the cable, so adding a heat-shrink tube afterward was not practical.
The cable supplier had been specified by the customer, and the product itself was relatively simple. At the beginning, sampling seemed reasonable.
The trial production showed why the actual assembly process mattered.
The problem was not simply the cable.
It was the combination of:
For this production run, we therefore maintained 100% inspection.
The useful information came from seeing how the parts actually interacted during assembly, rather than from looking at any single component specification.
Another project involved two PCBs used in an industrial refrigeration system.
The boards needed to communicate with each other.
The customer’s test sequence simulated the actual operating process. The main board activated a specific combination of relays, while the other board received the signals and converted them into the required communication output. The final data was sent through USB for verification.
The sequence made sense from a functional perspective.
The interesting part appeared when the same procedure was used repeatedly in production.
The final stage required approximately four minutes before the complete result could be confirmed.
During those four minutes, the operator essentially had to wait for the test to finish while monitoring the data.
For one engineering sample, this was hardly important.
Repeated across a production order, it became a very different consideration.
The product was working as intended.
The test was also working as intended.
But the test station had become a significant part of the production cycle.
This was not a failure of the product. It was simply something that became much more obvious when the test was repeated unit after unit.
A different situation appeared on a sensor-based product using an optical sensor for signal conversion.
The basic functional check was straightforward.
The unit powered up and performed its intended function.
During the final calibration process, however, we noticed that the measurement result was sensitive to ambient temperature.
A unit could pass the basic functional check while its final measurement result still shifted when calibration was performed under different temperature conditions.
We reported the observation to the customer.
The calibration process was subsequently carried out under controlled temperature conditions.
There was no obvious component failure involved.
The PCB worked.
The sensor worked.
The basic functional test worked.
The difference appeared in the relationship between the measurement result and the environment in which calibration was performed.
Again, this was something that became much easier to see during actual production than from the individual product itself.
These three projects had little in common from a product perspective.
One was a vehicle connectivity product.
One was an industrial refrigeration system.
One was an optical sensing product.
The manufacturing findings were equally different.
In the first case, a small variation in cable preparation interacted with the physical assembly process.
In the second, a functionally representative test sequence created a significant production-time requirement.
In the third, the measurement result was affected by the conditions under which calibration was performed.
None of these findings came simply from checking whether the PCB had been assembled correctly.
They appeared when the products were being:
That is one of the differences between looking at a product as an engineering file and manufacturing it as a physical product.
The product files describe what is supposed to be built, while the production process shows what happens when that product is actually manufactured.
A manufacturing file can contain a great deal of information: PCB data, BOMs, drawings, assembly requirements, test procedures, and specifications.
But some information only becomes visible through repeated production.
A cable that is acceptable on paper may interact unexpectedly with the assembly process.
A test procedure that works perfectly for one unit may create a four-minute production cycle.
A measurement that appears stable during a basic functional check may respond differently to the conditions used during calibration.
These are not necessarily problems with the original product documentation.
They are simply observations that emerge when the product moves from documents into physical production.
And sometimes, those observations are the most useful information to come out of a production run.
