Specifications
MPN
MPF200-ETH-SENSOR-BRIDGE-R2 (Rev 2.0, production-ready board)
FPGA
MPF200T PolarFire FPGA — low-power, non-volatile programmable logic
Ethernet
2x 10 Gigabit Ethernet (GbE) SFP ports
MIPI Inputs
Dual MIPI CSI-2 receiver ports for direct camera or sensor connection
FMC Expansion
FMC slot enabling protocol expansion to SLVS-EC, CoaXPress, JESD204B and SDI
Frame Buffering
DDR4 memory for sensor frame buffering
Configuration
SPI Flash for in-field upgrades
Host Platforms
NVIDIA Jetson AGX Orin, NVIDIA IGX Orin and NVIDIA IGX Thor
Software Framework
NVIDIA Holoscan SDK with the Holoscan Sensor Bridge (HSB) framework
FPGA IP
NVIDIA IP core processes sensor data inside the FPGA before 10GbE transmission
Reference Demo
12 MP sensor streamed at 4K over MIPI CSI-2 into the FPGA and out over 10 Gigabit Ethernet
Application Note
AN5522 — PolarFire Ethernet Sensor Bridge for NVIDIA Jetson AGX Orin Developer Kit AI Applications
Design Goal
Low power, low latency edge sensor-to-AI connectivity
Ecosystem Role
Hardware half of the NVIDIA Holoscan hardware platform for streaming AI pipelines
Rev 2.0 Changes
Smaller, production-ready board for sensor integration using NVIDIA HSB
Bandwidth
Scalable multi-sensor data aggregation over Ethernet
Target Applications
Industrial cameras, high-performance edge computers and medical devices
Overview
The Microchip PolarFire FPGA Ethernet Sensor Bridge solves a specific architectural problem in edge AI: getting many high-speed sensors into an NVIDIA Jetson or IGX platform without running a long custom ribbon cable or a proprietary frame grabber. It is built on the MPF200T PolarFire FPGA and exposes a two-part front end — dual MIPI CSI-2 inputs for direct sensor attachment and an FMC slot that can be adapted to SLVS-EC, CoaXPress, JESD204B or SDI — then aggregates that data and transmits it over dual 10 Gigabit Ethernet SFP ports.
On the FPGA side an NVIDIA IP core processes the sensor data before it leaves the board, and DDR4 memory provides frame buffering so bursts from high-resolution sensors do not have to be consumed in lockstep by the host. SPI Flash holds the configuration and allows field upgrades. The result is a low-power, low-latency bridge designed specifically for the NVIDIA Holoscan SDK and its Holoscan Sensor Bridge framework, where the platform appears as a networked sensor source rather than a PCIe device.
The board targets NVIDIA Jetson AGX Orin, NVIDIA IGX Orin and NVIDIA IGX Thor, and Microchip pairs it with an application note (AN5522) that walks through connecting the hardware to a Jetson AGX Orin Developer Kit. A reference demonstration streams a 12 MP sensor at 4K over MIPI CSI-2 into the FPGA and out across 10 Gigabit Ethernet. Revision 2.0 shrinks the design into a production-ready board, which is what makes it usable beyond the lab.
Key Benefits
Decouples sensors from the host — Ethernet transport lets cameras sit metres away instead of being tethered by a short MIPI ribbon. Dual 10GbE plus dual MIPI CSI-2 handles high-resolution, high-frame-rate sensors without host bandwidth bottlenecks. FMC slot for protocol expansion to CoaXPress, SLVS-EC, JESD204B and SDI keeps one board usable across several sensor standards. Built for NVIDIA Holoscan with an NVIDIA IP core, HSB framework support and an official application note. Low-power non-volatile PolarFire FPGA suits fanless and space-constrained edge enclosures.
Applications
Medical imaging and surgical vision pipelines built on NVIDIA Holoscan; industrial camera aggregation for machine vision; high-throughput scientific and industrial sensor capture; edge AI systems where sensor and compute must be physically separated; multi-camera inspection systems requiring deterministic sensor timing.
Request a Quote — MICROCHIP POLARFIRE FPGA ETHERNET SENSOR BRIDGE — NVIDIA HOLOSCAN PLATFORM
QS Compute — global B2B supply of AI computing hardware, edge AI systems and accelerators. Volume pricing, 15-day sample lead time.
Get Your Quote →