High performance edge computing for AI and machine vision

phyCORE®-AM67x

Root Part #: PCM-080

High performance machine vision and edge computing

  • 55 mm x 39 mm
  • 4x Cortex®-A53, 1x Cortex®-R5F
  • 2x C7x DSPs, 3D GPU
  • Secure Boot

Product Overview

A high-performance System on Module designed for advanced vision systems, AI workloads, and real-time industrial control. The phyCORE®-AM67x combines powerful processing, integrated deep learning acceleration, and high-speed connectivity, making it ideal for robotics, medical imaging, factory automation, and next-generation edge devices.

When standard embedded platforms cannot meet performance or processing demands, the phyCORE®-AM67x provides the headroom required for complex, data-intensive systems. Built on the Texas Instruments AM67x platform, this SOM combines quad Cortex-A53 application cores, a dedicated Cortex-R5F real-time core, and dual C7x DSPs with integrated deep learning acceleration, enabling advanced AI inference, high-speed analytics, and deterministic control in a single platform.

Designed for machine vision and sensor-driven applications, it supports multiple camera inputs, high-speed interfaces including PCIe 3.0 and USB 3.1, with flexible connectivity for demanding embedded systems.

With PHYTEC manufacturing expertise and hands-on engineering support, teams can reduce development risk, simplify integration, and move from prototype to production faster.

Contact Sales for more information about this product.

When to Choose an AM67x System On Module

Designed for high-performance applications where other application processors cannot meet processing, AI, or real-time requirements.

Choose this module if your application requires:

  • High-performance processing for complex, data-intensive workloads
  • On-device AI acceleration for machine vision and deep learning applications
  • Multiple camera inputs for advanced imaging and vision systems
  • High-speed interfaces for real-time data processing and analytics
  • Deterministic real-time control for industrial and robotics systems
  • A scalable SOM platform for advanced OEM and embedded products

Typical applications include:

  • Robotics controllers and autonomous systems
  • Smart vision cameras and AI-enabled imaging devices
  • Automated inspection and quality control systems
  • Medical imaging and diagnostic platforms
  • Industrial edge gateways and analytics systems
  • Autonomous mobile equipment and vehicles

Development and Integration Benefits

  • Compact 55 mm x 39 mm form factor to support high-performance designs in limited space
  • Dual DSP architecture to accelerate AI and signal processing workloads
  • PCIe 3.0 and USB 3.1 for high-speed data transfer and connectivity
  • Secure boot and trusted architecture for secure embedded deployments
  • Access to local engineering support from concept through to production
  • Faster route to market compared to fully custom high-performance designs
Expand All

Features

Advanced Vision and AI Capabilities

3x Display (DSI, MIPI DPI, OLDI) with 4K video codec for HMI. 4x MIPI CSI Cameras with USB 3.1 and PCIe 3.0

Deep Learning Acceleration

Two Deep Learning Accelerators with on chip C7x DSP with scalar and vector cores, dedicated “MMA” deep learning accelerator combined with a large 2.25MB L2 memory enabling performance up to 4 TOPS

Security

Work with PHYTEC to add Secure Boot to your application. All phyCORE-AM67x SOMs have the ability to burn security keys to the processor and establish root-of-trust by signing software images to ensure only trusted code is run on your device.

Real-time Capable

Due to the heterogeneous architecture of the TI AM67x processor, you can run the majority of your application using Linux and off load critical components to the specialized low latency best-in-class real-time Cortex-R5 core.

Specifications

Processor

Processor TI AM6754
Architecture Arm Cortex-A53 / Arm Cortex-R5F
Frequency 1.4 GHz
Co-Processors 2x C7x DSP

Memory

RAM 4 GB default / 8 GB max LPDDR4
eMMC 32 GB default / 128 GB max
NOR 64 MB up to 256 MB (Octal SPI/Dual SPI Flash)
EEPROM 2x 4kB I2C
SD/SDIO/MMC Interfaces 3
External Memory Bus 1x GPMC

High-Speed Interfaces

USB 1x 2.0 Dual Role, 1x 3.1 Dual Role
PCIe 1x PCIe 3.0 1-lane

Peripherals

CAN 4x CAN-FD
eCAP 3
eQEP 3
I2C 7
PWM 3
SPI/SSP 5
UART 9
GPIO Yes

Connectivity

Ethernet 1x 10/100/1000, 1x RGMII

Multimedia

Audio 5x McASP
Camera 4x MIPI CSI-2, 1x MIPI CSI 2.0 Transmitter
Display 3x OLDI/LVDS, DSI or DPI
Graphics 3D GPU - OpenGL 3.x/2.0/1.1, Vulkan 1.2
Video 3840x1080 at 60fps

Miscellaneous

JTAG Yes
RTC Yes
Security Accelerators AES, DRBG, PKA, RSA, SHA-2, HW-enforced RoT, Arm TrustZone© based TEE

Mechanical

Dimensions 39 mm x 55 mm
Connector 2x 160-pin 0.5 mm pitch

Operating Conditions

Operating Temperature -40° to +85° C
Power (Max) TBD
Power (Typical) 2.722W
Supply Voltage 5.0V

Hardware Documentation

SOM

Hardware Manual

phyCORE-AM67x Hardware Manual

Component Placement
phyCORE-AM67x System on Module

phyCORE-AM67x Component Placement 

Schematic Symbol & Footprint

phyCORE-AM67x

Carrier Board

Component Placement
phyBOARD-AM67x Reference Design

phyBOARD-AM67x Component Placement

PHYTEC PinMux Tool

Pinmuxing tool 

Connectors

SOM Connectors
Samtec ASP-205225-06 (PHYTEC Article # VB408)

ASP-205225-06.zip

Carrier Board Connectors
Samtec REF-178709-03 (BTH-080-01-L-D-A-KTR, PHYTEC Article # VM240)

REF-178709-03.zip

The AM67x Processor

The AM67x scalable processor family is based on the evolutionary Jacinto™ 7 architecture, targeted at Smart Vision Camera and General Compute applications and built on extensive market knowledge accumulated over a decade of TI’s leadership in the Vision processor market. The AM67x family is built for a broad set of cost-sensitive high performance compute applications in Factory Automation, Building Automation, and other markets.

The Texas Instruments Sitara™ AM67x processor
AM62

Software

Linux

BSP-Yocto-Ampliphy-AM67x-PD25.1.1
2026-04-17

Yocto Scarthgap 5.0.16

BSP-Yocto-Ampliphy-AM67x-PD25.1.0
2025-09-26

Yocto Scarthgap 5.0.12 (Linux 6.12.35, U-Boot 2025.01)

BSP-Yocto-Ampliphy-AM67x-ALPHA1
2025-03-31

Yocto Scarthgap 5.0.7 (Linux 6.6.58, U-Boot 2024.04)

Related Case Studies

INCH Pro Embedded EVSE Charging Station for Prototyping

Etrel is a leading manufacturer of interactive charging stations for electric vehicles and provider of IT platform for charging management.

Learn More

chargeIT Mobility – Base Charging Box

chargeIT Mobility is a leading intelligent charging solutions provider for EVs.

Learn More

Ready to get started?

Let’s go! Learn more about PHYTEC SOMs