Specifications
Platform
BitFlow Claxon CoaXPress frame grabber family (BitFlow, Inc., an Advantech business unit)
Models
Claxon CXP1 (single-link), Claxon CXP2 (dual-link), Claxon CXP4 (quad-link), Claxon FXP / Fiber (CoaXPress over Fiber)
Part Numbers
CLX-PC3-CXP1, CLX-PC3-CXP2, CLX-PC3-CXP4; Advantech PCIE-1162-AE / PCIE-1164-BE
Standard
Full CoaXPress 1.x / 2.x compliance, complete GenICam support
Link Speed
3.25 Gb/s to 12.50 Gb/s per link (CXP-12)
Aggregate Bandwidth
Up to 50 Gb/s across four links; BitFlow systems support aggregate bandwidths exceeding 100 Gb/s
Host Interface
Half-size x8 PCI Express (Gen 2.0 on CXP2, Gen 3.0 on CXP4); compatible with x8/x16 slots Gen 1.0-5.0
Power Over CoaXPress
Up to 13 W per link with Safe Power full fault protection
Cable Length
Up to 100 m on standard coaxial cable; hundreds of metres over fiber with QSFP+ transceivers
Low-Speed Uplink
41.6 Mbps uplink for camera control on every link
GPU Direct
GPUDirect RDMA (NVIDIA) and DirectGMA (AMD) with sub-5 microsecond host-to-GPU latency
DMA Architecture
Scatter/gather DMA with up to 128 independent DMA streams, zero CPU cycles and no DDR dependency
FPGA Pre-Processing
Debayering, flat-field correction, LUTs and ROI extraction at full pixel clock rates
Camera Support
One quad-link, one dual-link or two single-link CXP-12 cameras with simultaneous multi-camera capture
Triggering / I/O
Separate I/O per camera, quadrature encoder support and a programmable signal generator
StreamSync
StreamSync acquisition maximises throughput while minimising image latency
Software
BitFlow SDK (120+ code examples) for Windows and Linux; HALCON, LabVIEW, VisionPro and MATLAB integration
Connectivity
Micro BNC connectors; QSFP+ transceivers on Claxon Fiber and the QFi copper-to-fiber converter
Compliance
GigE Vision 3.0 RDMA, GenICam, RoHS compliant, PCB rated UL 94 V-0
Roadmap
CoaXPress 3.0 (CXP-25, 25 Gb/s per link) support targeted before year-end 2026
Overview
The BitFlow Claxon is a CoaXPress 2.0 frame grabber family built for AI-driven machine vision, where multiple high-resolution cameras must feed deep-learning inference pipelines without frame loss. Each board implements the full CoaXPress 2.0 specification rather than a subset, delivering 12.5 Gb/s per link and scaling to 50 Gb/s across four links.
Unlike frame grabbers built around commodity DDR memory, Claxon uses scatter-gather DMA with small FIFO buffers and SRAM control tables, streaming image data at full sustained throughput directly into host or GPU memory. This eliminates frame-transfer latency and keeps the family immune to the DRAM shortages affecting buffer-based competitors.
The Claxon platform spans the single-link CXP1, dual-link CXP2, quad-link CXP4 and long-reach Fiber variants, letting system integrators migrate from the earlier Cyton generation without major software changes. Combined with Advantech edge AI systems and NVIDIA TensorRT, Claxon enables zero-latency, zero-frame-loss Industrial Vision AI.
Key Benefits
Zero-latency acquisition: scatter-gather DMA and GPUDirect RDMA deliver sub-5 microsecond host-to-GPU latency. Full CoaXPress 2.0 with 12.5 Gb/s links, 13 W Power-over-Coax and a 41.6 Mbps control uplink. No DDR dependency keeps Claxon in stock while DRAM-based grabbers face 20-30 week lead times. Flexible scaling from one to four CXP-12 cameras with deterministic triggering.
Applications
AI automated optical inspection (AOI), semiconductor and electronics inspection, EV battery and electrode metrology, robotic bin-picking, 3D stereo and laser-triangulation vision, and high-throughput multi-camera lines requiring real-time GPU inference.
Request a Quote — BITFLOW CLAXON CXP-12 FRAME GRABBERS — COAXPRESS 2.0 IMAGE ACQUISITION FOR AI MACHINE VISION
QS Compute — global B2B supply of AI computing hardware, edge AI systems and accelerators. Volume pricing, 15-day sample lead time.
Get Your Quote →