I've reviewed many design history files over the years, and this gap shows up almost every time.
How a prototype gets built
The best technician in the building builds the prototype, on equipment that will never see the production floor, in a room where nobody tracks assembly time. It works. It gets photographed. It goes in the deck.
Then someone has to scale a manufacturing line to make ten thousand of them, built by recently hired operators, using tooling that's never run more than a single shift. And the process has to hold up under a quality system.
Two different questions
"Does the design work?" is one question. "Can this be built the same way, correctly, by someone other than the person who invented it, at volume, forever?" is a different one.
Design verification answers the first question. It doesn't answer the second — and a lot of programs don't find that out until tooling is already cut.
Where the gap usually hides
The fixes are rarely dramatic. It's usually one of a handful of things:
- Features that are hard to hold to tolerance in production. A dimension the prototype shop could hit by careful hand work becomes a yield problem at volume.
- A fixture that only works in the hands of one very patient engineer. If the result depends on who's at the bench, the process isn't under control yet.
- A step that depends on judgment instead of a controlled parameter. "Until it looks right" is not a process specification, and it's very hard to validate.
Caught early, it's a design tweak
Caught late, the same issue is a validation redo. The cheapest time to find a manufacturability problem is before design freeze, while a feature can still be changed on paper rather than in hardened tooling and validated processes.
- Who built the prototypes, and could a newly trained operator build them the same way?
- Which dimensions have the least margin between capability and tolerance?
- Which assembly steps rely on operator judgment rather than a set parameter or a fixture?
- What does the production tooling and equipment plan look like, and how different is it from what built the prototypes?
What's the latest you've seen a manufacturability issue surface in a program: design freeze, tooling, or worse?
Chris Steadham is the founder of Truestead Advisors and has spent 30+ years in medical device product development, manufacturing scale-up and general management. Working through a question like this on a deal or a program? Start a conversation.