Write the spec

YOU WILL BE ABLE TO

Turn written requirements into a parts list drawn from the library.
Budget pins in a table before writing a line of JSON.
Choose between two parts that satisfy the same requirement, and say why.
State the library's power limits for driven loads and where real designs get their power.

A board does not start as JSON. It starts as REQUIREMENTS: sentences about what the thing must do. This unit designs one board start to finish — a reaction timer: it shows a score from 0 to 9, it has a button you slap when the light demands it, and it beeps. Three sentences. The design process is the walk from those sentences to a spec the validator blesses.

Step one: requirements to parts. Read each sentence and shop the library — you have seventeen parts and the palette shows sixteen. "Shows a score 0 to 9" is one digit: sevenseg_cc. "A button you slap": pushbutton. "It beeps": buzzer_passive. Some sentences admit two answers — a knob could be potentiometer (smooth, analog, needs an ADC pin) or rotary_encoder (clicky, digital, three ordinary GPIOs) — and choosing IS design.

Requirement	Part	GPIOs	Components it generates
show 0-9	`sevenseg_cc`	7	8 (module + 7 resistors)
react button	`pushbutton`	1	1
beep	`buzzer_passive`	1	2 (buzzer + resistor)

Requirements to parts, with cost: what each sentence buys, and what it spends.

Step two: the PIN BUDGET. The Pico gives you 26 assignable GPIOs, and only three of them — GP26, GP27, GP28 — can read analog. Every part in your table bids for pins, one peripheral per pin, and greedy parts bid hard: the seven-segment display alone takes seven. Write the pin plan as a table BEFORE the spec, exactly the way you check nets before trusting a schematic. The engine's own demo board runs 14 parts and consumes 21 of the 26 — running out is not a beginner mistake, it is the normal condition of real boards.

Signal	Pin	Note
segments a-g	GP2-GP8	seven consecutive, easy to read in code
button	GP14	any digital pin
buzzer	GP13	any digital pin; PWM makes the tone
free for later	GP10-GP12, GP26-GP28, and more	encoder or knob slots in Lesson 4

The reaction timer's pin plan — written before the spec, the way nets are checked before the schematic is trusted.

NOTE

A word on POWER before you drive anything heavier than an LED. The transistor_driver wires its load between the 3V3 rail and the collector — and the Pico's 3.3 V regulator has roughly 300 mA to share with everything on the board, the radio included. Real designs power motors and relays from VBUS, the USB 5 V — a limitation of this library today, and a teaching point: reading where the power COMES FROM is part of reading a design. The driver's 1k base resistor sets about $I_B = 2.6/1000 \approx 2.6\,\text{mA}$ of base current (3.3 V minus a 0.7 V junction, across 1k); with a small-signal transistor's $\beta \approx 50$ , plan on loads of about 100 mA or less. Heavier loads want a smaller base resistor — or a MOSFET, which is PS-3 talk.

WORKED EXAMPLE

Budget check

A classmate's board wants: a rotary encoder, an RGB LED, a pot, and two plain LEDs. Total pins, and any constraints?
Encoder 3 + RGB 3 + pot 1 + LEDs 2 = 9 GPIOs.
The pot's one pin MUST be one of GP26/GP27/GP28; every other pin is an ordinary GPIO.
Nine of 26 — comfortable. The constraint is the KIND of pin, not the count. (The engine agrees: that board validates clean, 11 components, 23 traces.)

9 GPIOs, one of which must be an ADC pin.

PRACTICE

Requirements: "a knob that scrolls a menu and clicks to select." One part or two?

Why write the pin plan before the spec, when the validator would catch conflicts anyway?

A 250 mA motor on the transistor driver: work the base-current math and give a verdict.

Which requirement CANNOT be met by this library, and what is the honest answer: "amplify a microphone signal"?