SHW3G & SHF3G (3MP Global)

Overview

SHW3G (HFOV 122°) and SHF3G (HFOV 173°) cameras are designed for embodied intelligence applications. Built with a 3MP Global Shutter sensor, they can be used in humanoid robot head and torso positions for robot perception and recognition scenarios. The main difference is the lens field of view: SHW3G provides 122° HFOV for wide-angle perception, while SHF3G provides 173° HFOV for ultra-wide perception. Other platform specifications are the same unless noted in the lens option table.

SHW3G (HFOV 122°)

SHF3G (HFOV 173°)

Key Features and Application

Features:
• Output RAW data
• Global Shutter
• Low latency
• Support external trigger
• Multi-camera synchronization

Application:
• Humanoid Robots
• Data collection

Getting Started

Specification

Parameter	Value
Sensor	3MP Global Shutter
Image Size	1/3.1 inch CMOS
Output Pixels	2064 (H) × 1552 (V)
Pixel Size	2.25um*2.25um
Frame Rate	MAX 125fps
Shutter	Global Shutter
Output data	RAW@12bit/10bit/8bit
Serializer	MAXIM MAX96717
Camera Interface	GMSL2
Power Supply	9~16V POC
Current	Less than 200mA@12VDC
Connector	Amphenol Fakra (Z Code)
Operating temp. range	-40~+85℃
Dimensions	W: 25mm, L:25mm, H:32.13mm (SHW3G) / 31.82mm (SHF3G)
Weight	Less than 50g

Dimensions

[SHW3G (HFOV 122°)]

[SHF3G (HFOV 173°)]

Hardware Overview

Block Diagram

Frame Sync usage requirements

	Requires two trigger input signals	PWM trigger signal requirements	Within the Camera Serializer GPIO ---> Sensor GPIO	Notes
1	XVS signal	Frequency: 30 Hz, Duty cycle: 10%	MFP7 ---> XVS	If multiple cameras are used for synchronized triggering (i.e., slave mode), both XVS and XHS trigger signals must be input simultaneously and meet the corresponding requirements in order to achieve synchronized triggering.
2	XHS signal	Frequency: 83 kHz, Duty cycle: ≥ 90%	MFP3 ---> XHS

I2C Address Information

Component	Parameter	Value
Serializer	Model	MAX96717
	I2C Address	0x80 (8bit address)
	GMSL Rate	GMSL2 (6G bps)
Sensor	Model	RGGB
	I2C Address	0x6C (8bit address)
	Frame Sync	Controlled by MAX96717 MFP7 and MFP3
	Master/Slave Control	Controlled by MAX96717 MFP5 (Low:Master, High:Slave-default)
	Reset	Controlled by MAX96717 MFP0

Lens Specifications

Model	HFOV	VFOV	F.No	Depth of Field	Water-proof	Lens Mount
SHW3G	122°	90°	2.2	0.34m - INF@Focus at 1M	IP67	AA
SHF3G	173°	134°	2.0	0.2m - INF@Focus at 0.4M	IP67	AA

Employing Camera

1. Adaptation to NVIDIA® Jetson™ platform

SHW3G / SHF3G Camera Connect to Nvidia Jetson AGX Orin

Step 1: Installation Steps

Quick Setup

Connect the SHW3G or SHF3G camera to the SG10A-AGON-G2M-A1 board using the coaxial cable
Mount the SG10A-AGON-G2M-A1 board onto the Jetson AGX Orin module
SHW3G or SHF3G camera Connect the power supply
SG10A-AGON-G2M-A1 board Connect the power supply
Power on the system

Step 2: Software Preparation

SDK Download

-Select the appropriate driver package based on your camera type and JetPack version:

-Copy the full link address to DownGit to download

NO.	JetPack Version	Camera	NVIDIA Jetson Devices	Adapter Board	Download Link
1	JP6.2	SHW3G / SHF3G	Jetson AGX Orin Developer Kit	SG10A-AGON-G2M-A1	Download

JetPack Versions

NVIDIA JetPack (Jetpack 5.1.2 or Jetpack 6.0 ) is the official software development kit (SDK) for the Jetson series of development boards. It includes the operating system, drivers, CUDA, cuDNN, TensorRT, and other development tools and libraries. Each JetPack version typically corresponds to a specific Jetson Linux version (formerly known as L4T - Linux for Tegra).

36.4.3: L4T R36.4.3 (Jetpack 6.2)
36.4: L4T R36.4 (Jetpack 6.1)
36.3: L4T R36.3 (Jetpack 6.0)
35.4.1: L4T R35.4.1 (Jetpack 5.1.2)

For more information, visit NVIDIA's official Jetson Download Center.

2. Camera Integration with Customer's Self-developed Platform

For customers with their own deserializer who want to adapt our camera (serializer) to their platform, detailed technical coordination is required.

The diagram illustrates the communication architecture between a camera and controller system. It shows how data flows from the Sensor/ISP through the Serializer on the Camera side, across to the Deserializer and SOC on the Controller side. The system utilizes Fsync signals for synchronization and MFP7 and MFP3 interfaces for control. This architecture is essential for proper integration of SENSING cameras with customer-developed platforms.

Step 1: Link Register initialization

SENSING will provide:

Serializer and Deserializer Configuration
- Register configuration for the camera module-Getting Camera Information
- I2C communication protocol details
Link Status Troubleshooting Guide
- Link training parameters
- Error detection settings

tip

Please refer to the software flow and demo code below to develop your driver code.

Software Development

Driver Development:

/* Example code for MAX9296 I2C initialization */
#define MAX9296_I2C_ADDR 0x90 // 8-bit address

int max9296_init() {
    // Initialize I2C bus
    i2c_init();
    
    // disable MIPI output
    i2c_write(MAX9296_I2C_ADDR, 0x0313, 0x00);
    delay_ms(100);
    // Configure link settings for GMSL2 (6Gbps)
    i2c_write(MAX9296_I2C_ADDR, 0x0001, 0x02);

    // Configure linkA and linkB settings for GMSL2 selection (default value)
    i2c_write(MAX9296_I2C_ADDR, 0x0006, 0xC0);
    
    // Configure MIPI rate 1200Mbps
    i2c_write(MAX9296_I2C_ADDR, 0x0320, 0x2C); 
    
    // enable MIPI output
    i2c_write(MAX9296_I2C_ADDR, 0x0313, 0x02);
    
    return 0;
}

Camera Configuration:

/* Example code for   initialization */

#define MAX9295_I2C_ADDR 0x80 // 8-bit address

int camera_init() {
    // Initialize deserializer first
    max9296_init();
    
    // Reset ISP through MAX9295A
    i2c_write(0x80, 0x02BE, 0x10); // MFP0 high
    // 
    i2c_write(0x80, 0x0057, 0x12); 
    i2c_write(0x80, 0x005B, 0x11); 
    //  Configure datatype  YUV422 8bit
    i2c_write(0x80, 0x0318, 0x5E); 

    //  camera trigger  MFP7  low to  high
    i2c_write(0x80, 0x02D3, 0x00); // MFP7 low
    delay_ms(300);
    i2c_write(0x80, 0x02D3, 0x10); // MFP7 high

    return 0;
}

Integration Steps

BSP Integration:
- Modify the device tree to include the CSI interface configuration
- Add camera driver to kernel build configuration
- Configure media controller pipeline for the camera
Application Development:

/* Example code for capturing camera frames */
#include "camera_api.h"

int main() {
    // Open camera device
    int fd = open("/dev/video0", O_RDWR);
    if (fd < 0) {
        perror("Failed to open camera device");
        return -1;
    }
    
    // Configure video capture format
    struct v4l2_format fmt = {0};
    fmt.type = V4L2_BUF_TYPE_VIDEO_CAPTURE;
    fmt.fmt.pix.width = 1920;
    fmt.fmt.pix.height = 1536;
    fmt.fmt.pix.pixelformat = V4L2_PIX_FMT_UYVY;
    
    if (ioctl(fd, VIDIOC_S_FMT, &fmt) < 0) {
        perror("Failed to set format");
        close(fd);
        return -1;
    }
    
    // Request and map buffers
    // ... (buffer setup code) ...
    
    // Start streaming
    // ... (streaming code) ...
    
    // Capture and process frames
    // ... (frame processing code) ...
    
    // Cleanup
    close(fd);
    return 0;
}

Step 2: Data Processing

After receiving the module data through the MIPI CSI interface:

Data Reception
- MIPI CSI-2 protocol implementation
- Data rate configuration
Image Processing
- YUV422 data parsing
- Image format conversion
- Display configuration

Technical Support

Documentation
- Detailed register descriptions
Engineering Support
- Technical consultation
- Debug assistance
- Performance optimization

tip

SENSING Technology provides technical support for integration with most platforms. For detailed documentation, sample code, and technical assistance, please contact our support team.

Overview​

Key Features and Application​

Getting Started​

Specification​

Dimensions​

Hardware Overview​

Block Diagram​

Frame Sync usage requirements​

I2C Address Information​

Lens Specifications​

Employing Camera​

1. Adaptation to NVIDIA® Jetson™ platform​

Step 1: Installation Steps​

Step 2: Software Preparation​

2. Camera Integration with Customer's Self-developed Platform​

Step 1: Link Register initialization​

Software Development​

Integration Steps​

Step 2: Data Processing​

Technical Support​