Before your first board spin

Ordering the first prototype boards is a milestone. It is also the point where a hardware project starts spending real money and real calendar time on every mistake. A respin costs the boards, the assembly, the parts, and usually several weeks.

The goal of a first spin is not perfection. It is to learn as much as possible from one round of boards, so the next spin moves toward production instead of repeating the first. This checklist is built around that goal. For the builds that follow, see From prototype to production.

1. Write down what the board must do

One page is enough. It should cover:

What the board does, and what it does not do.
How it is powered, and the worst-case current.
The environment it lives in, and the enclosure it fits in. Outdoors, that includes sunlight on the parts.
Interfaces to the outside world: connectors, buttons, displays, radios.
Where the product will be sold.
Target cost and expected volume.

Without this page, design decisions get made by default, and reviews have nothing to check against.

2. Settle the architecture before the schematic

Draw the block diagram and the power tree before schematic capture, and for a battery product, a first power budget. Choose the main processor, the radio approach if there is one, and the power source deliberately, because they are the hardest decisions to change later. Module or chip-down and Battery-powered design cover two of them in depth. If you are unsure between two approaches, this is the cheapest moment to ask for a second opinion.

3. Check part availability now

Every part on the board should be:

In stock in the quantities you will need, at more than one distributor if possible.
Active in its lifecycle, not “not recommended for new designs.”
Available in a package your contract manufacturer can place.
Matched to its footprint and pinout, checked against the datasheet for the exact part number you are buying.

Parts with a single source or long lead times should have an alternate identified now, even if you do not use it yet.

4. Design for bring-up

The first boards will not work perfectly. Make it easy to find out why:

Test points on every power rail and on reset, and a way to observe key clocks, such as a clock-output pin, rather than probing a crystal directly.
A way to power sections of the board separately and measure each one’s current, such as zero-ohm links or jumpers.
Debug and programming headers that are easy to reach.
A serial log output or a status LED that firmware can drive on day one.
Space to add or remove a part where you are unsure of a value.

Some of this will come off the board later. On the first spin, it is cheap insurance.

5. Plan for firmware from the start

Firmware work often starts the day boards arrive, and stalls on missing access. Make sure firmware can:

Program and debug the processor.
Read a board revision, so later code can support several hardware versions.
Recover from a bad firmware load.
Set or override boot and strapping pins without rework.

If possible, start firmware on a development kit with the same processor while the boards are being built.

6. Check the fit before ordering

Compare the board outline, mounting holes, connector positions, and tall parts against the enclosure model. Print the board outline on paper at full scale and hold it up against the enclosure or a 3D print. It takes a few minutes and catches problems that are easy to miss on a screen.

7. Look ahead to compliance

You will not test for certification on the first spin, but the first spin decides how hard certification will be, and its boards are good for early bench checks such as near-field scans and ESD. Before ordering, check for:

Ground planes and short return paths under fast signals.
Filtering on cables that leave the enclosure.
ESD protection on exposed connectors and buttons.
Radio modules placed exactly as their integration guide requires.
Room for a shield or a filter you may need later, and, on a chip-down radio, footprints to tune the antenna match and a way to measure conducted RF power.

For more, see Plan for compliance from day one.

8. Write a bring-up plan

Before the boards arrive, write down the order you will test them in: power rails first, then clocks and reset, then the processor, then each subsystem. Note what “working” means for each step, and what equipment you need. Before applying power, inspect the board and measure each rail’s resistance to ground. Then power the first board from a current-limited bench supply, so a short shows up as a reading rather than smoke. Make sure the rest of the equipment is on hand before the boards arrive: a scope with enough bandwidth, a logic analyzer or protocol decoder for the buses you use, and the probes, adapters, and cables to reach the board. Losing days to a missing probe is an easy delay to avoid. A plan turns the first week with new boards from improvising into a checklist.

9. Order enough boards

Order more boards than you think you need. Some will be damaged in bring-up, some will go to firmware, some to mechanical fit checks, and some to early testing. Building a few extra in the first run costs far less than a second small run later. Bare boards cost much less than assembled ones, so order extra bare boards and keep the stencil, even if you only assemble a few.

10. Get a review before you order

When the boards arrive, the first board bring-up checklist walks through power-up and testing step by step.

A second set of eyes on the schematic and layout before the order goes in costs little next to a respin. A reviewer who has not lived inside the design will see what the team has stopped noticing. What I look for in a hardware design review covers what a good one checks. Send your board fab and assembler one consistent set of files at a single revision, and ask them to run their own manufacturability checks before they build. What a complete manufacturing package includes lists the files and the gaps that most often hold up a first order.

If your first spin is coming up and you want that review, get in touch.