What it really costs to leave certification until the end
The cheapest certification failure is the one you design out in month one. Here is what the expensive version looks like.
Certification is the last gate before a product can be sold, and that position on the timeline is exactly what makes it dangerous. By the time a product reaches a test house, the tooling is cut, the components are bought and the launch date has been promised to somebody. A failure at that point is not a technical problem. It is a commercial one.
The uncomfortable part is that most certification failures are not surprises. They are decisions made months earlier by somebody who did not know they were making them.
What actually fails
In our experience the same handful of causes come round again and again, and almost all of them are design decisions rather than manufacturing defects:
- Radiated emissions from a switching supply that was laid out for cost rather than for loop area.
- A plastic housing with no route to ground, chosen before anyone asked how the product would pass an electrostatic discharge test.
- A radio module placed against a metal bracket, which passes on the bench and fails in the chamber.
- Creepage and clearance distances on a mains board that were correct for one standard and not for the market the product is actually being sold into.
- A power supply certified for a different enclosure, which means its certification does not transfer.
Why it costs so much when it lands late
The test itself is rarely the expensive part. What costs money is everything that has to move behind it. A layout change means a new board spin and new assembly. A housing change may mean modifying or recutting a mould. Components already bought for the first production run may no longer be the right ones. And every one of those is on a lead time that does not care about your launch date.
There is also a quieter cost. A team that has failed once tends to over-correct, adding shielding, filtering and parts that are not needed, because nobody wants to fail twice. That weight stays in the bill of materials for the life of the product.
Designing for the marks from the start
The fix is not complicated, it is just early. Decide which markets the product is for before the first schematic, because that decides which marks apply — CE and UKCA behave differently since divergence, FCC governs the radio, UL and CCC govern safety in their respective markets, and Energy Star governs consumption. Then design to the strictest of them rather than to each in turn.
Pre-compliance testing during development is the other half. A rough scan on a development board tells you where you stand while a fix is still a layout change rather than a recall of parts. It will not certify anything, and it is not meant to.
This is why we run verification and validation inside the project rather than at the end of it. Not because it is more rigorous, but because it is cheaper — and because a product that was designed for its certification usually passes it.
Working on something like this?
We design and manufacture hardware end to end. The first call is with an engineer and costs nothing.
