Brochage

Brochage STM32 Blue Pill (F103C8T6)

Broches

Microcontrôleur	STM32F103C8T6, Arm Cortex-M3
Fréquence d'horloge	72 MHz (8 MHz crystal)
Mémoire flash	64 KB (many chips actually have 128 KB)
SRAM	20 KB
Alimentation	5 V pin or USB, onboard 3.3 V regulator
Niveau logique	3.3 V (FT pins are 5 V tolerant)
Courant GPIO	25 mA absolute max per pin, 150 mA total
PC13–PC15	3 mA sink max, 2 MHz max
ADC	2 × 12-bit, 10 channels
Périphériques	3 × USART, 2 × SPI, 2 × I2C, CAN, USB full-speed
USB	Micro-USB, native (no USB-serial chip)
Programming	SWD header or serial bootloader on USART1

BOOT0 = 0 (its normal resting position): boots from the main flash — where it should sit for everyday use.
BOOT0 = 1 with BOOT1 = 0: boots the chip’s built-in serial bootloader instead, which flashes new firmware over USART1 (PA9/PA10) through a USB-to-serial adapter — no ST-Link needed.
BOOT0 = 1 with BOOT1 = 1 boots from embedded SRAM instead of flash, rarely what you want. Most boards expose BOOT1 as a second small jumper right next to BOOT0 — check both before assuming a board is stuck.
After flashing through the serial bootloader, move BOOT0 back to 0 and reset or power-cycle the board — the new firmware will not run while BOOT0 still points at the bootloader.
An SWD debugger (ST-Link) ignores BOOT0/BOOT1 entirely and can flash or debug regardless of jumper position — the jumpers only matter for the USB-to-serial flashing path.

Logic is 3.3V. Most GPIOs are NOT 5V tolerant — only PA8–PA15, PB3/PB4, PB6–PB15 and PC6–PC15 are FT (5V-tolerant); every ADC-capable pin (PA0–PA7, PB0, PB1, PB5, PC0–PC5) is 3.3V-only and can be damaged by 5V.
PC13, the onboard LED pin, is active-LOW and — like the whole PC13–PC15 group — can only sink about 3mA at up to 2MHz through the backup-domain power switch; never use it as a current source for anything beyond that one onboard LED.
Every other GPIO is rated 20mA per pin (a relaxed spec pushes this to 25mA, at reduced output voltage), but the whole package is capped around 150mA combined — much less headroom to spread across many pins at once than an AVR board’s 200mA, despite the similar-looking per-pin figure.
Some boards genuinely carry a different, non-ST chip (CS32F103 and similar clones) that can misbehave on USB or DMA-heavy peripherals — check the reported Device ID in your programmer if something is unreliable. That is separate from a genuine ST die reporting 128KB instead of the datasheet’s 64KB, which is usually just relabeled surplus silicon and not a fake.
There is no standard numbered header position on this board either: every reference names pins PA0…PC15 by their MCU name, never by a position number.