The core board is a highly integrated embedded hardware module, which usually contains core components such as processor (CPU/MPU/MCU), memory (RAM/ROM), power management, and basic peripheral interfaces.

Ⅰ. main features of the core board

1. High level of integration
It integrates core components such as main control chips, memory, Flash, and PMIC (power management) to reduce hardware design complexity.
2. Standardized interfaces
Provide universal expansion interfaces (such as PCIe, USB, MIPI, GPIO, etc.) to facilitate connection to the backplane for function expansion.
3. Low power consumption design
Optimize power consumption for mobile devices and IoT scenarios, and support sleep mode and dynamic frequency modulation.
4. Rapid development
Provide complete BSP (Board Support Package) and Linux/Android drivers to shorten the time to market.
5. Miniaturization
High-density PCB design for space-constrained applications (e.g., wearables, drones).

Ⅱ. the typical composition of the core board

Module Description:
Main control chips: ARM Cortex-A/M series (e.g., i.MX6/8, RK3588, STM32MP1), RISC-V (e.g., K210), etc.
Memory DDR3/DDR4/LPDDR4, the capacity is usually 512MB~8GB.
Storage eMMC (4GB~128GB), SPI Flash, QSPI NOR Flash, etc.
Power management PMICs, such as the NXP PF series, support multiple voltage outputs.
Peripheral interfaces USB, UART, I2C, SPI, MIPI-DSI/CSI, Ethernet, PCIe, etc.
Wireless module optional Wi-Fi/BT (e.g., ESP32), 4G/5G (e.g., Quectel EC series).

Ⅲ. Core board vs development board vs single board computer (SBC)

Type Features Applicable scenarios
Core board contains only core components and needs to be used with the base plate for flexible customization. Industrial control, mass production equipment.
Development board Fully functional, integrated peripherals (e.g., HDMI, USB Host) for validation and prototyping. Learning, evaluation, early prototyping.
Single-board computers run independently (such as Raspberry Pi, Jetson Nano) with rich interfaces but weak scalability. Education, maker, lightweight applications.

Ⅳ. Mainstream core board plan

1. Low-power MCU level (IoT/embedded control)

STM32MP157 (Cortex-A7 M4 Dual-Core)
Applicable: Industrial HMI, PLC, Edge Computing.
Manufacturer: Wildfire, Spot Atom.
ESP32-S3（Wi-Fi 6   Bluetooth 5）
Applicable to: smart home, wireless sensing node.

2. Mid-to-high-end application processors (AI/multimedia)

Rockchip RK3588 (Cortex-A76 A55, 6TOPS NPU)
Applicable: Edge AI, NVR, advertising machine.
Manufacturer: Firefly, Rockchip official core board.
NXP i.MX8M Plus（Cortex-A53   2.3TOPS NPU）
Applicable: Machine vision, intelligent gateway.

3. High-performance computing (server/autonomous driving)

NVIDIA Jetson Orin NX (Ampere architecture, 100TOPS)
Applicable: Autonomous driving, robots.
TI AM68A (Cortex-A72 DSP, multi-camera support)
Applicable: Industrial vision.

Ⅴ. How to choose a core board?

1. Performance requirements
For compute-intensive (such as AI), choose the NPU-supported RK3588/Jetson; Control class selection STM32MP1.
2. Peripheral interfaces
Need multiple cameras? Optional models with MIPI-CSI interface (e.g., i.MX8M Plus).
3. Operating system support
Linux/Android requires ARM Cortex-A series, RTOS optional Cortex-M/RISC-V.
4. Cost and mass production
It is recommended to choose domestic solutions (such as Allwinner and Rockchip) for mass production to avoid the risk of production stoppage.

Ⅵ. Typical application scenarios

Industry 4.0: PLC, HMI, motion control (STM32MP1 EtherCAT).
Smart hardware: smart speaker, face recognition terminal (RK3588 Linux).
Edge Computing: AI Inference Box (Jetson Orin/NXP i.MX8M).
IoT terminal: LoRa/Wi-Fi sensor node (ESP32-S3).