978-474-0044
The Gidel FantoVision40-CXP12 Edge Computer is a high-performance AI vision edge computing platform designed for high-bandwidth, low-latency machine vision and imaging applications. Featuring a 4x CoaXPress-12 interface that captures images from up to 4 x CoaXPress 2.1 machine vision cameras, a NVIDIA Jetson Orin NX embedded AI computer, and Intel Arria 10 FPGA, this AI Vision system delivers real-time image acquisition, preprocessing, compression, and AI inference in a single integrated platform.
Built for industrial machine vision, inspection, medical imaging, and defense applications, the FantoVision40-CXP12 combines unmatched I/O bandwidth, deterministic FPGA processing, and GPU-accelerated AI in a cost-effective and scalable edge computer.
The integrated NVIDIA Orin NX platform provides a powerful GPU-accelerated AI engine for deep learning inference, computer vision algorithms, multi-camera analytics, and real-time defect detection and classification. Software Engineers can use CUDA C/C++, NVIDIA libraries, and popular AI frameworks to build and deploy advanced vision applications directly on the edge computer.
FantoVision vision systems are available as a a plug-and-play FPGA-based frame grabber, or as part of a complete AI vision system.
FantoVision40-CXP12 Datasheet
FantoVision40 Family Getting Started User Guide
Compare FantoVision edge computers
How FantoVision overcomes I/O and bandwidth limitations commonly faced by NVIDIA Jetson-based embedded systems
FantoVision edge computers are available as:
The FantoVision40-CXP12 is optimized for high-speed CoaXPress Vision camera systems requiring real-time processing at the edge. With up to 50 Gb/s bandwidth of total acquisition bandwidth, the system can capture, process, and stream multiple high-resolution image sources without dropped frames or latency bottlenecks. The architecture eliminates the need to send raw, uncompressed data to servers or the cloud.
By combining FPGA-based image preprocessing with GPU-accelerated AI inference, the system enables:
At the core of the system is an Intel Arria 10 FPGA running Gidel’s ProcVision™ image processing suite, enabling:
By performing these operations before data reaches the GPU or network, the FantoVision40-CXP12 significantly reduces system load, bandwidth usage, and storage requirements.
FantoVision's hybrid FPGA + GPU architecture provides the best combination of deterministic real-time processing and flexible AI compute that include:
Need help selecting the right CoaXPress CXP12 Vision AI edge computer for your application? Contact us today to discuss your requirements and see how the FantoVision40-CXP12 can power your next high-performance vision system.
General Specifications
| Camera Interface | 4x CoaXPress-12 links 2x CoaXPress-12 links 1x CoaXPress-12 link |
| Compliant Interfaces | CoaXPress 2.1 |
| Camera Input Connectors | 4x Micro-BNC |
| Supported Down-Connection Speeds |
1.25 Gb/s (CXP-1) 2.5 Gb/s (CXP-2) 3.125 Gb/s (CXP-3) 5 Gb/s (CXP-5) 6.25 Gb/s (CXP-6) 10 Gb/s (CXP-10) 12.5 Gb/s (CXP-12) |
| Supported Up-Connection Speeds |
20.83 Mbps (CXP-1 to CXP-6) 41.66 Mbps (CXP-10, CXP-12) |
| Maximum Stream Packet Size | 8,192 bytes |
| Advanced Features |
On-the-fly selective ROI acquisition Link Aggregation mode Chunk Data |
| Host Interfaces |
1x GbE 1x USB 3.2 Gen 2 (single lane) 1x USB 2.0 1x HDMI Out 1x Recovery 1x Restart 1x UART Max USB 3.x power allocation: ≤ 4 W |
| Storage |
NVMe interface: Gen 4 ×2 Up to 2TB Contact us if you need a larger capacity. |
| Image Formats |
Mono, Bayer, RGBA (8, 10, 12, 14, 16 bits/color) RGB (8, 10, 12 bits/color) |
| Maximum Resolution |
Horizontal: 16K pixels (64-bit) Vertical: 65K lines Higher resolutions available |
| Compression & Image Enhancements |
On Jetson Compression: H.264,H.265
On FPGA Compression: JPEG, Lossless, Quality+ On FPGA Enhancements: HDR correction |
| Maximum Acquisition Throughput | 40 Gb/s |
| Camera Types | Area, Line Scan |
| GPIO |
8x RS422 differential pairs 4x Opto-Isolators 2x High-Drive outputs 6x LVTTL GPIO power out: 2x 12V @ 1.0A aggregate |
| Streaming & Communication Protocols |
Custom streaming using NVIDIA tools RTSP streaming option |
| Dimensions | 134 × 90 × 60 mm (L × W × H) |
| Weight | 750 g |
| Cooling |
Passive cooling Active cooling (fan) option |
| MTBF | >580,000 hours (Passive Cooling, 40°C, 60% confidence) |
| Certifications | CE, FCC, CCI |
Processors
| Embedded Computer | NVIDIA Jetson Orin NX 16GB |
| Jetson Computer Resources | JetPack Version Up to 8-core Arm Cortex-A78AE, 2MB L2 + 4MB L3 1024-core NVIDIA Ampere GPU with 32 Tensor Cores Up to 157 TOPS AI computation Up to 16GB @ 102.4 GB/s Video encode: 1x 4K60, 3x 4K30, 6x 1080p60, 12x 1080p30 (H.265), H.264, H.265, AV1 Video decode: 1x 8K30,2x 4K60, 4x 4K30, 9x 1080p60, 18x 1080p30 (H.265), H.264, H.265, VP9, AV1 |
| FPGA Types | Altera Arria 10: 160 GX, 270 GX, 660 GX |
| GPU-FPGA interconnectivity | PCIe Gen 3 x4 @ up to 26.4 Gb/s |
Environmental Conditions
| Temperature Range | Commercial: 0 °C - 65 °C case temperature Industrial: -25 °C - 65 °C case temperature. (Limited FPGA options) |
| IP Compliance |
IP51-compliant For configurations meeting higher IP standards, contact us. |
| Environmental Compliance |
RoHS Conflict Minerals ISO REACH |
| NVIDIA Jetson Model |
Orin NX 16GB Nano NX (Subject to minimum order quantity) |
| FPGA |
Arria 10 160 GX: 2GB / 4GB Arria 10 270 GX: 2GB / 10GB and 4GB (Industrial Temperature) Arria 10 660 GX: 9GB DRAM Throughput: 2GB – 10GB DDR4 @ up to 24GB/s |
| Camera Interface | Customized user protocol option |
| Storage |
NVMe interface: Gen 3 ×1 / Gen 4 ×2 (Depends on Jetson model) 100GB 1TB (Subject to minimum order quantity) 2TB Contact us if you need a larger capacity. |
| Modular Real-Time Image Processing |
Gidel FPGA flow can integrate advanced image processing algorithms, including: Compression encoders: JPEG | Lossless | Quality+ High Dynamic Range (HDR) correction from a single exposure White Balance – Maintains color accuracy across variable lighting conditions Dynamic Luminance Balance – Preserves consistent brightness under changing illumination Morphological operations such as Open/Close using a round structuring element |
| User FPGA Code | Users can integrate their own FPGA code with Gidel’s IPs (see FPGA Processing tab) |
| Streaming & Communication | RTSP streaming option |
| Cooling |
Passive cooling Active cooling (fan) option |
| Ruggedized | Extended vibration resistance (recommended for harsh environments) |
| Temperature Range | Industrial: -25°C to 65°C case temperature (limited FPGA options) |
| IP Compliance | For custom IP ratings or ratings above IP53, contact us. |
FantoVision modular edge computers are customizable, allowing you to select from a range of Jetson SoM, FPGA, storage, and cooling options. Contact us to get an edge computer customized for you!
AI/Algorithm Development
The FantoVision edge computers are supported by NVIDIA’s libraries enabling software engineers to develop algorithms/AI on the Jetson via CUDA C/C++.
ProcVision Developer’s Suite
The ProcVision Suite offers a comprehensive set of tools, including:
