Designing a custom circuit

A custom circuit is really several library blocks, chosen and combined to hit your requirements. You describe the whole product — with its constraints — and Forge assembles it. Your job is to direct and review, not to accept blindly.

Give Forge the full picture in one description:

• the inputs (which sensors, what they measure),
• the brain (does it need Wi-Fi? Bluetooth? just local logic?),
• the outputs (screen, LEDs, motor, buzzer),
• the power (battery + runtime, or plugged in),
• and the cost target in naira.

Forge gives you back more than a parts list. You get a bill of materials and a wiring diagram — the actual pin-to-pin connections between the parts, drawn from real component pinouts. And it runs an ERC (electrical rule check) over those connections: it flags a pin wired to the wrong place, a part whose pinout it can't confirm (shown as a dashed wire), and other connection mistakes — before you ever buy anything.

The more of your requirements you state, the more that first design already fits. Then you read it critically: is every part pulling its weight? Do the connections make sense? Is the power path right for the runtime? Is it within budget?

That review — not the generation — is the skill. The design is a strong first draft to interrogate, not a final answer to trust.

Two custom-circuit skills separate a design that works on paper from one that works in the field.

Power budgeting

A battery product lives or dies on its power budget. The idea is simple: add up how much current each part draws, and multiply by how long it must run.

battery capacity ≈ total current × runtime (plus margin)

A Wi-Fi node that transmits often draws far more than one that sleeps most of the time. If the battery is too big or too expensive, you have levers: sleep the controller between readings, report less often, or use a lower-power radio. Naming your runtime ("about a week") lets Forge size the battery and pick a power path that can actually hit it.

Honest trade-offs

When the BOM overshoots the cost or size target, you trade, deliberately:

• swap a part for a cheaper one that's good enough,
• drop a non-essential feature,
• or accept a limit (shorter runtime, smaller screen) on purpose.

Good product development is a series of conscious trade-offs — not pretending there are none.

The first design is a starting point — and Forge is a copilot, not just an autopilot. You don't have to re-describe the whole thing to change one connection. You can steer the design directly:

• Edit the wiring canvas. Open the wiring diagram and edit it: add or remove a part from the component palette, and draw or redraw a connection pin-to-pin. The ERC re-checks live as you edit, so a mistake shows up immediately instead of on the bench.
• Ask for a change in plain English. Or tell Forge — "drive the buzzer from a different pin," "add a second temperature sensor on the same I²C bus" — and it revises the wiring for you. Direct it or edit it yourself; use whichever is faster for the change.
• Export to KiCad. When the design is right, download it as a real KiCad 9 project — a schematic with your parts' real footprints from the Microscale library. Open it in eeschema, then Tools → Update PCB from Schematic to start the board layout.

That export is the bridge from idea to real hardware: the same design flows on to a fabricated PCB and, through Build, to assembled units. Autopilot gets you a strong first draft fast; the copilot tools are how you make it yours and take it all the way to a board.

Design a real custom circuit with constraints, then interrogate and edit the result.

Send the prompt — an outdoor sensor node with three sensors, Wi-Fi, a week of runtime, and a naira budget — then work the design like an engineer:

• Read the BOM critically: is every part needed, or could one be dropped?
• Open the wiring diagram and check the ERC — are the connections sound? Any dashed (unverified) wires?
• Edit it: add or swap a part from the palette, or redraw a connection, and watch the ERC re-check.
• Does the power path (battery + charging + converter) suit a week of runtime? Is the total within the target price?
• When it looks right, export the KiCad project — that's the file you'd take to a fabricated board.

Where would you substitute, cut, or rewire to hit the budget and the requirements? That judgement — directing and editing, not just accepting — is the advanced skill.