当前位置:网站首页>硬件电路设计学习笔记1--温升设计
硬件电路设计学习笔记1--温升设计
2022-07-28 05:23:00 【Tsd-Xu】
在硬件电路设计中,我们需要考虑电子元器件的发热及温升,如果不考虑温升,当温度达到一定高度时,部分元器件将失去功能,不能工作,严重的将引发火灾等,所以温升设计必不可少。
通常PCB板温升设计有以下几点:
1,发热器件布局要合理,如发热器件布局要靠近板边,更利于散热(该点不绝对,要根据实际情况布局).
2,功率器件提供良好的散热,如MOSFET单独加散热片
3,PCB开窗设计,通常dcdc下方的需要开窗敷铜,用于散热。
4,增大功率线的线宽及铜厚。
5,发热器件尽可能要均匀分布在PCB板上
6,等等。
在PCB设计中,有些情况下,即使注意了布局也会出现过温问题。下图为最近做一款PCB温升设计数据,布局合理的情况下依然有部分点温升(测量温度-室温)大于50℃。该PCB铜厚为1oZ。
增加PCB铜厚到2oZ后,温升有明显改善,但还有部分点温升大于50℃,如下图。
大功率PCB布线开窗后加2mm锡后,温升有所改善,但还有超标的温度点,如下图。
上图中,超标的温度点通过增加连接线束的线径等方式使温升问题有所改善。
可以看到,温升问题解决的过程中,都在增加PCB的制作成本,所以成本和温升的考量也需要平衡,一般只要能刚刚好达到要求为最佳。
边栏推荐
- Deep learning (self supervision: simpl) -- a simple framework for contractual learning of visual representations
- ESXi社区版网卡驱动再次更新
- Deep learning - metaformer is actually what you need for vision
- Notes on the thinking of software analysis and library acquisition of a Taobao customer
- What are the detailed steps of wechat applet development?
- Reinforcement learning - Strategic Learning
- Adaface: quality adaptive margin for face recognition image quality adaptive edge loss for face recognition
- 基于直方图修改的可逆数字水印方法
- ESXi Arm Edition version 1.10更新
- 强化学习——价值学习中的DQN
猜你喜欢

后门攻击与对抗样本攻击的比较研究

What are the advantages of small program development system? Why choose it?

一、语音合成与自回归模型

深度学习(一):走进机器学习与深度学习理论部分

卷积神经网络

机器学习之聚类

深度学习数据窃取攻击在数据沙箱模式下的威胁分析与防御方法研究阅读心得

Distributed cluster architecture scenario optimization solution: distributed scheduling problem

CVE_ 2017_ 11882 vulnerability recurrence (Metasploit opens NT remote desktop to add an account)

Deep learning (incremental learning) -- iccv2021:ss-il: separated softmax for incremental learning
随机推荐
深度学习(自监督:CPC v2)——Data-Efficient Image Recognition with Contrastive Predictive Coding
Tf.keras build neural network function expansion
Internet of things interoperability system: classification, standards and future development
Deep learning - metaformer is actually what you need for vision
五、视频处理与GStreamer
Deep learning (II) into machine learning and deep learning programming
Deep learning (self supervision: CPC V2) -- data efficient image recognition with contractual predictive coding
Byte Android post 4 rounds of interviews, received 50k*18 offers, and successfully broke the situation under the layoff
一、AMD - OpenVINO环境配置
物联网互操作系统:分类、标准与未来发展方向综述
Reinforcement learning - Multi-Agent Reinforcement Learning
Deep learning (self supervision: simpl) -- a simple framework for contractual learning of visual representations
Research on threat analysis and defense methods of deep learning data theft attack in data sandbox mode
USB Network Native Driver for ESXi更新到支持ESXi7.0.1
强化学习——多智能体强化学习
Interviewer: let you design a set of image loading framework. How would you design it?
四、模型优化器与推理引擎
Common CTF encryption methods JS
Reinforcement learning - dqn in value learning
C语言EOF的理解