Please use this identifier to cite or link to this item:
標題: Reliability+Assessment+of+Ball+Grid+Array+Packages+Using+Norris-Landzberg+Acceleration+Model
作者: 鍾官榮
關鍵字: Accelerated life testing;加速壽命測試;BGA;Norris-Landzberg model;Temperature cycling tests;球閘陣列封裝;Norris-Landzberg模型;溫度循環試驗
出版社: 國立中興大學工學院;Airiti Press Inc.
Project: 興大工程學刊, Volume 21, Issue 2, Page(s) 81-87.
Norris-Landzberg加速壽命模型由於其簡單易用的優勢,廣?工業界用來預測電子產品因溫度產生的疲勞破壞時間。但是許多研究證實其模型中的參數會因焊點材料與封裝型式的差異而產生不同的參數值。在現今有各種型式之球閘陣列封裝(Ball Grid Array, BGA)被大量使用的情況下,如繼續使用原始之模型,必增加其預估壽命的誤差,必需修正原來的N-L關係式。因此,本論文首先詳述Norris-Landzberg加速壽命模型,以不同焊料、封裝體與錫膏?依據,探討模型參數的變異;接著說明Salmela與Dauksher等兩種N-L修正模型;最後針對一些球閘陣列封裝進行溫度循環試驗,比較原始的N-L關係式與Salmela修正模型預估準度,結果證明:與使用原始的N-L關係式比較,Salmela的修正模型在其引用的數據中平均預估誤差由52.4%減少至17.8%;而在我們針對BGA所做的實驗數據,其平均預估誤差由41%減少至32%。因此,Salmela修正模型提供了對球閘陣列封裝可靠度壽命一種較精確的預估方法。

Norris-Landzberg acceleration model has been widely used to predict the lifetimes of electronic products due to its simplicity, only considering the conditions of the temperature cycling test. However, a lot of research shows that the model exponents are not constant depending on the package types and solder materials (ex. SnPb vs. SnAgCu). It will increase the average errors of prediction if the original model is continually used without modification for Ball Grid Array (BGA) packages, which they nowadays are in the majority with respect to quite different types by the industry. In this paper, the variation of these exponents in the original N-L model was demonstrated using the data of different solder materials and package types from some research sources, and different solder pastes from ours. The modified N-L models from Salmela and Dauksher were introduced respectively. Compared to the original model developed from Norris and Landzberg, Salmela demonstrated that his model is able to reduce the average errors of prediction from 52.4% to 17.8% using the data with respect to different package types and solder materials. Finally, the temperature cycling tests of different types of BGAs were performed. The average errors of prediction decrease from 41% to 32% using Salmela's model. In a word, the modified N-L model developed from Salmela is certainly to provide the better accuracy in predicting the lifetimes of BGAs.
ISSN: 1017-4397
Appears in Collections:第21卷 第2期

Files in This Item:
File SizeFormat Existing users please Login
10174397-201007-201008210093-201008210093-81-87.pdf906.59 kBAdobe PDFThis file is only available in the university internal network   
Show full item record

Google ScholarTM


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.