AEE Engineering8 min read
Every shipping electronic product went through roughly the same journey, whether it took eighteen months or five years. Knowing the stages in advance does two things for a founder: it makes quotes and schedules legible, and it tells you which corners are safe to cut at your stage and which ones always collect their debt later.
Here is the journey as we run it, stage by stage, with the artifact each one should produce and the mistake most often made in it.
1. Validate the problem, not the product
Before any engineering: evidence that someone will pay for what the device does. Talk to buyers, price the alternatives they use today, and write down the three capabilities that would make them switch. Hardware iterations are expensive; conviction is cheap to test first.
Artifact: a one-page problem statement with target price and the must-have capabilities. Common mistake: validating that people like the idea instead of validating what they would pay for it.
2. Write requirements someone could build from
Translate the problem into engineering language: what the device senses and does, how long the battery lasts, what it costs at volume, what environment it survives, which certifications it needs. Good requirements are testable. "Long battery life" is a wish; "30 days between charges at 10 readings per hour" is a requirement.
Artifact: a requirements document with numbers in it. Common mistake: skipping straight to features and leaving cost, certification, and environment to be discovered mid-design, where they force restarts.
3. Check feasibility before committing
A short, focused study: can the requirements be met at the target cost with parts that exist and physics that cooperates? This is where an experienced engineer earns their fee by killing a doomed architecture in week two instead of month eight.
Artifact: a feasibility assessment with the major risks named and either retired or priced. Common mistake: treating feasibility as a formality. The projects that skip it are the ones that later discover the sensor they need does not exist under forty dollars.
4. Architect the system
Choose the bones: processor class, connectivity, power source and budget, sensor approach, and how the product splits between hardware, firmware, and any companion software. Every later decision inherits from these, which is why they deserve real deliberation instead of defaulting to whatever the last project used.
Artifact: a system architecture document and block diagram, plus a first bill-of-materials cost estimate. Common mistake: letting a development kit choose your architecture. Dev boards are for learning, not for inheriting.
5. Design in detail
The long middle: schematics, PCB layout, firmware, enclosure CAD, all advancing together with reviews at each milestone. Together matters — boards designed without their enclosure, or firmware written after pin assignments are frozen, generate exactly the integration surprises this stage exists to prevent.
Artifact: a complete design package under version control: schematics, layout, firmware source, mechanical CAD, BOM. Common mistake: siloed disciplines. The most expensive sentence in hardware is "the firmware team didn't know the hardware did that."
6. Prototype and iterate
Build real hardware early and let it embarrass the design while changes are still cheap. Expect more than one revision; two or three board spins is a normal, healthy count for a product of moderate complexity. Each spin should answer named questions, not just "let's see if it works."
Artifact: working prototypes and a bring-up report per revision — what passed, what failed, what changed. Common mistake: demanding the first prototype be perfect, which quietly pressures engineers into conservative designs and hides problems until production.
7. Validate and certify
The design gets proven against the stage-2 requirements: performance testing, environmental stress, battery life measured rather than calculated, and the regulatory gauntlet — FCC for anything that radiates, safety marks where they apply, plus industry-specific standards. Staged validation builds (the EVT, DVT, and PVT sequence) structure this phase; we cover how those gates work on the concept-to-production page.
Artifact: test reports traceable to requirements and certification paperwork. Common mistake: meeting regulatory requirements after design freeze. Pre-compliance testing during stage 6 costs little and prevents the classic launch-delaying EMC failure.
8. Prepare manufacturing
The design becomes buildable at volume: design-for-manufacturing refinements, production test fixtures, assembly documentation, supplier qualification, and first articles inspected against the drawings. If your design firm and factory are different companies, this handoff is the riskiest seam in the whole journey — budget real time for it, or pick a partner where design and production share a roof.
Artifact: a released production package and a passed first-article inspection. Common mistake: treating manufacturing as a purchasing task instead of an engineering stage.
9. Launch, then keep engineering
Shipping is not the finish line. Field data arrives, components go end-of-life, costs want reduction, and firmware needs updates. Products that stay profitable have someone doing this sustaining work deliberately; products that don't have it quietly become unbuildable.
Artifact: a sustaining plan — who watches the BOM, who owns field issues, how changes get qualified. Common mistake: disbanding the project the week it ships.
How long and how much?
Realistic ranges for a product of moderate complexity: twelve to twenty-four months from stage 1 to stage 9, with engineering cost depending mostly on how many disciplines the product touches. The electronic product design page carries our cost guidance, and a conversation about your specific product is free.