基于SIP的FPGA驱动电压补偿测试研究  

Research on FPGA driver voltage compensation testing based on SIP

作　　者：黄健 陈诚[1] 王建超[1] 李岱林 杜晓冬 HUANG Jian;CHEN Cheng;WANG Jianchao;LI Dailin;DU Xiaodong(China Electronic Technology Group Corporation No.58 Research Institute,Wuxi 214035,China)

机构地区：[1]中国电子科技集团公司第五十八研究所,江苏无锡214035

出　　处：《现代电子技术》2025年第4期30-33,共4页Modern Electronics Technique

摘　　要：在基于SIP的现场可编程门阵列(FPGA)性能参数验证测试时,驱动电压测试会受到多种因素的影响,如PCB线阻、插座信号损耗以及测试温度等,这些因素导致ATE测试的实测值与真实值之间存在偏差。为了提高驱动电压的测试精度,提出一种基于卷积神经网络(CNN)与长短时记忆(LSTM)网络的误差补偿方法。将PCB线长、测试温度等参数作为特征输入到CNN-LSTM模型中,模型经过训练迭代后能够预测出驱动电压的误差值;再将预测的误差值应用于ATE测试机中,对实测值进行补偿和修正,从而使得测试结果更加接近真实值。实验结果表明,所提方法能够有效地减小测试误差,提高FPGA驱动电压测试的准确性。In the performance parameter verification testing of field-programmable gate arrays(FPGA)based on system in package(SIP),the driver voltage testing is influenced by various factors,such as PCB line resistance,socket signal loss,and testing temperature,which lead to discrepancies between the measured values and the true values in automated test equipment(ATE)testing.In order to enhance the testing accuracy of the driver voltage,a method of error compensation based on convolutional neural networks(CNN)and long short-term memory(LSTM)networks is proposed.By inputting parameters such as PCB line length and testing temperature into the CNN-LSTM model,the model can predict the error value of the driver voltage after training and iteration.The predicted error value is applied to the ATE tester to compensate and correct the measured values,so as to make the testing results closer to the true values.The experimental results demonstrate the proposed method can effectively reduce testing errors and improve the accuracy of FPGA driver voltage testing.

关 键 词：驱动电压测试 误差补偿 系统级封装(SIP)技术 现场可编程门阵列 卷积神经网络 长短时记忆网络 

分 类 号：TN407-34[电子电信—微电子学与固体电子学]