Source: Tiger Says Chip
Original Author:Tiger Says Chip
This article summarizes the core principles and processes for selecting FPGA chips, aiming to provide decision-making support for designers to ensure project success.1. Core Principles of FPGA Chip SelectionChoosing an FPGA is like selecting an engine and chassis for a car; it must match performance requirements while considering cost, maintenance, and availability. The ideal selection is a comprehensive balance of performance, resources, development difficulty, and supply security.
1. Clarify Functional Requirements
The first step is to outline the system objectives. It is essential to understand precisely what tasks the FPGA will perform in the system, such as high-speed data processing, protocol interface conversion, signal acquisition and control, or algorithm acceleration. Clarifying requirements determines all subsequent decision directions.
2. Evaluate Logic and Storage Resources
Logic Units (LUT, FF): Assess the complexity of the logic circuit to ensure the FPGA has sufficient logic units, leaving some margin for future functional adjustments. It is generally recommended that resource usage does not exceed 80%.On-chip Memory (Block RAM, etc.): Calculate the required memory capacity based on data buffering, FIFO, image processing, etc. Pay attention to physical distribution and the minimum configurable unit to avoid fragmentation waste.
3. Clock and PLL Resources
Calculate the number of PLLs and the different clocks that can be generated based on the required clock frequencies and whether synchronization is needed. If multiple independent synchronous domains are required, the FPGA must have sufficient clock trees and PLL support.
4. I/O Interfaces and Pin Resources
Count the required number of pins based on actual peripheral interfaces, communication protocols, debugging, and expansion interfaces, reserving 10-20% margin to avoid limitations due to later changes.Check the I/O standards supported by the FPGA, such as LVDS, LVCMOS, differential signals, etc., to adapt to external connections.
5. Performance Indicators: Operating Frequency and Speed Grade
A higher frequency is not always better; it must be combined with design timing constraints, process limitations, and the final actual compilation results. The theoretical maximum frequency is for reference only; the actual operating frequency must be adjusted based on timing analysis results and signal integrity.Different manufacturers have different ways of marking speed grades, so be careful to distinguish them when purchasing.
6. Special Hard Core Resource Requirements
This includes on-chip high-speed transceivers (SerDes), DSP multipliers, hard-core processors, embedded memory controllers, etc. These resources can significantly optimize the performance and power consumption of specific algorithms or interfaces.If the design relies on a certain hardware acceleration unit, such as requiring a large number of parallel multiplications, ensure that the FPGA model integrates enough DSP blocks.
7. Package Type and PCB Design Difficulty
QFP packages are suitable for low pin counts, simple PCBs, and easy manual soldering. BGA is suitable for high pin density and high board-level performance requirements in miniaturized products, but it has greater challenges in wiring, soldering, and testing, requiring high PCB process standards.Package size and pin spacing directly affect wiring efficiency, cost, and actual production capacity.
8. Supply and Market Availability
It is recommended to choose mainstream series and models with high market circulation for easier procurement and project maintenance, ensuring price transparency and resource sustainability.Be cautious with new, niche, or discontinued products, as they can easily impact project timelines due to shortages.
2. Suggested Selection Process
Requirement Analysis Phase: Communicate and outline, draw block diagrams, and list functional and resource requirements.Initial Specification Screening: Use the manufacturer’s online selection tools to preliminarily filter series and models that meet the requirements.Resource Matching and Secondary Optimization: Simulate in the development environment, attempt resource mapping, reserve reasonable margins, and optimize level and interface distribution.Evaluate Packaging and Manufacturing Capabilities: Combine the company’s PCB process capabilities, expected yield, assembly, and soldering to select feasible packaging.Confirm Market Availability: Verify the model’s supply cycle, price, and after-sales support with the supply chain.Comprehensive Weighing and Final Decision: Make the final chip model decision based on performance, cost, and risk.3. Common ConsiderationsDo not simply pursue ultra-high resources or maximum frequency; focus on actual needs;Maintain the scalability and upgradability of the design to avoid selecting just enough;Pay attention to the support of development tools, richness of IP resources, community technical documentation, and other “soft” resources;Lock in the chip early in the project and procure a small number of samples for feasibility verification.Conclusion: FPGA selection is the cornerstone of project success, integrating system engineering, logic design, hardware implementation, and supply chain management. A scientific and rigorous selection process can effectively mitigate project risks, control costs, and ensure product development efficiency and future sustainability.
END
Reprinted content only represents the author’s views
Does not represent the position of the Institute of Semiconductors, Chinese Academy of Sciences
Editor: Schrödinger’s Cat
Responsible Editor: Six Dollar Fish
Submission Email: [email protected]
Previous Recommendations1. Huang Kun: The “First Master” of Chinese Semiconductors2. He is rooted at No. 35 Tsinghua East Road, dedicating his life to the “National Heavy Weapon”; he resists unfair competition and does not allow opportunism…3. The Institute of Semiconductors has made significant progress in space communication laser technology4. The Institute of Semiconductors has made significant progress in the theory and experiment of Raman scattering in layered semiconductor materials5. The Institute of Semiconductors has made progress in the research of bionic covering neuron models and learning methods6. The Institute of Semiconductors has made significant progress in inverted structure perovskite solar cells
