In machine vision, embedded displays and other automation systems, image sensors and processors do not always share the same electrical interface or video protocol. The CrossLink FPGA is designed to address this integration challenge by combining hardened MIPI D-PHY interfaces, configurable I/O and programmable logic in a compact bridging device. Its architecture lets engineering teams build pathways between supported camera, display and video interfaces without treating the bridge as a fixed-function connection.
For industrial automation engineers and system integrators, the key consideration is how the device’s physical interfaces, logic resources and clocking fit the requirements of a particular design. CrossLink supports interface translation and stream operations such as aggregation, splitting and data conversion, while leaving implementation details to the configured logic and IP.
CrossLink FPGA architecture for video bridging
CrossLink combines three principal building blocks: up to two hardened MIPI D-PHY blocks, two banks of programmable I/O, and an FPGA fabric for implementing bridging functions. It also includes system resources such as embedded I2C, a power management unit, configuration interfaces and clocking resources.
The fabric contains 5,936 lookup tables, 180 Kb of embedded block RAM and 47 Kb of distributed RAM. These resources can be configured for the logic and buffering needs of a design. The programmable logic is built from functional units containing lookup tables and registers, connected through a configurable routing network. This allows designers to implement application-specific operations rather than relying only on a fixed bridge function.
Examples identified for the family include CSI-2 aggregation, repeater designs, display splitting, and translation between MIPI DSI or CSI-2 and interfaces such as CMOS, OpenLDI, FPD-Link and LVDS. These are design possibilities, not automatic capabilities: the selected IP, device configuration and system requirements determine the implemented behavior.
MIPI D-PHY and configurable I/O options
The hardened D-PHY blocks support transmit and receive operation for MIPI camera and display applications. They provide high-speed and low-power modes, with support for continuous-clock and low-power-clock operation. Each quad supports up to 6 Gb/s in total, with a stated maximum data rate of 1,500 Mb/s per lane. The blocks include dedicated transmit frequency synthesis and serializer and deserializer functions for interfacing with the FPGA fabric.
CrossLink also provides programmable I/O in Banks 1 and 2. These banks support differential standards including LVDS transmit and receive, and receive support for subLVDS, SLVS200 and MIPI. Single-ended options include LVCMOS and LVTTL standards. The banks support single-data-rate and double-data-rate operation, with input/output gearboxes and programmable delay cells for interface timing alignment.
Bank 0 serves a different role: its dedicated CMOS GPIOs support single-data-rate interfaces and do not provide differential signaling. This distinction matters when assigning pins and choosing a package. The available I/O count varies by package, and the 36-ball WLCSP package has only one hardened D-PHY available; device and package selection should therefore be checked against interface and pin requirements.
Clocking, timing and signal integrity
Video bridges often need to accommodate different clock relationships between source, processing logic and destination. CrossLink provides a general-purpose PLL with programmable output phase adjustment, dynamic output enables and clock-input selection. Its primary clock network can distribute clocks from sources such as PLL outputs, I/O pins, clock dividers, the internal oscillator and divided D-PHY clocks.
For source-synchronous interfaces, the programmable I/O delay and gearbox features can support edge-aligned or center-aligned timing arrangements. Banks 1 and 2 also provide programmable differential termination for supported inputs; MIPI D-PHY termination is dynamically enabled according to the high-speed select signal. These features give designers configuration options, but they do not remove the need to verify timing constraints, signal integrity and electrical compatibility at the system level.
Configuration and implementation considerations
CrossLink designs are implemented with Lattice Diamond design software, which uses synthesis results and design constraints to place and route logic. The tools extract routing timing for verification. The device supports several configuration approaches, including on-chip one-time-programmable non-volatile memory, booting from external SPI flash, and I2C or SPI programming. Dual-image booting and TransFR I/O for field updates are also listed among the family’s configuration features.
Before committing to a design, engineering teams should assess:
- Interface direction and protocol: Identify which links transmit and receive, and confirm that the required protocol and signaling standard are supported.
- Lane rate and timing: Compare the required data rate with the device’s documented limits and validate timing using the relevant electrical and switching specifications.
- Pin and package constraints: Check I/O bank capabilities, voltage requirements, differential-pair availability and package-specific resources.
- Logic and memory use: Estimate the fabric, RAM and clocking required for conversion, buffering, arbitration or stream handling.
Practical implications for automation designs
CrossLink offers a configurable way to connect image sensors, processors and displays that use different supported interfaces. Its combination of hardened MIPI D-PHY, programmable I/O and FPGA logic can help consolidate protocol bridging and custom video handling within one device. For industrial machine-vision and HMI designs, the practical value depends on matching the exact data path, lane rate, electrical standard, package and configuration method to the application.
Successful implementation still requires design-specific verification. Use the device documentation and applicable interface guides to confirm electrical limits, timing, power and pin assignments, and test the completed design in its intended system. Treating those checks as part of device selection—not as a final integration step—helps determine whether the CrossLink FPGA is an appropriate bridge for the application.