Presentation is loading. Please wait.

Presentation is loading. Please wait.

Advisor : Dr. Hsu Presenter : Ai-Chen Liao

Similar presentations


Presentation on theme: "Advisor : Dr. Hsu Presenter : Ai-Chen Liao"— Presentation transcript:

1 Visualization of machine clustering for a Taiwanese IC packaging foundry
Advisor : Dr. Hsu Presenter : Ai-Chen Liao Authors : Hsu-Hao Yang, He-Yau Kang Tzu-Chiang Liu 老師,各位實驗室的成員大家好,今天要為大家介紹的第一篇是有關於機械分群的視覺化,在台灣ic封裝廠的應用。 ESWA . Page(s) :

2 Outline Motivation Objective Background The data and results
The k-means algorithm SOM MST The data and results Conclusion Personal Opinions 這是我的outline,會帶過一些我們曾學過的方法,最主要強調實驗及結果的部分,以及它是怎麼拿來做應用的。

3 IC Manufacture Flow IC後段製程 IC前段製程 晶片製造 氧化 光罩校準 蝕刻 雜質擴散 離質植入 氣相沈積 晶片針測
晶圓製造 晶棒成長 切割 晶片製造 氧化 光罩校準 蝕刻 雜質擴散 離質植入 氣相沈積 晶片針測 光罩 光罩製作 IC設計 邏輯設計 線路設計 圖形設計 導線架 蝕刻沖壓 IC封裝 晶圓切割 黏晶 焊線 封膠 剪切成形 印字 檢測 IC測試 進料檢驗 晶片測試 外觀檢驗 烘烤 入庫檢驗

4 An IC packaging process
IC Chip Humidity _ ESD Heat Transfer IC 封裝的功能與目的 提供電氣傳導路徑(包括金線、Lead frame及銅導線),讓微細的IC電路彼此做連結 首先在我們進人之前先讓大家看一下ic封裝流程,ic的封裝是半導體產業的其中一環,ic製造步驟之一,在台灣做ic封裝的公司像是日月光就是其中一家, 保護IC晶片不受外力的破壞及避免溼氣滲透到IC內部 提供IC晶片散熱路徑

5 晶圓切割(Die Sawing): 依據客戶所指定的晶粒尺寸,將晶片切割至封裝製程所需要的尺寸,以利下一製程(黏晶)作業.

6 將晶片固定於導線架(Leadframe)上.
黏晶(Die Bounding): 將晶片固定於導線架(Leadframe)上.

7 銲線(Wire Bounding): 銲線的目的是將晶粒上的接點以極細的金線,連接到導線架上之內引腳,藉而將IC晶粒之電路訊號傳輸到外界。

8 封膠(Molding): 封膠之目的有以下數點: 1、防止濕氣等由外部侵入。 2、以機械方式支持導線。 3、有效地將內部產生之熱排出
於外部。 4、提供能夠手持之形體。

9 電鍍(Plating): 油墨/雷射蓋印: (Marking)
於外引腳表面鍍上一層純錫或錫合金,以增加其導電性及抗氧化性,並使其於表面黏著(SMD)作業時,易於與PC Borad上之錫膏結合. 油墨/雷射蓋印: (Marking) 於膠體表面蓋印,以作業產品識別用.

10 剪切/成形(Trim/Form): 檢測 (Inspection)
剪切之目的,乃是要將整條導線架上已封裝好之晶粒,每個獨立分開。同時,亦要把不需要的連接用材料及部份凸出之樹脂切除(dejunk)。剪切完成時之每個獨立封膠晶粒之模樣,是一塊堅固的樹脂Borad上 檢測 (Inspection)

11 Motivation Despite the increasing popularity worldwide, an IC packaging foundry is exposed to the following challenges:

12 Objective To help the foundry more effectively assign a family of machines to orders that appear in various forms such as emergent or quality-demanding and are highly dynamic in nature.

13 Background─ The k-means algorithm
1. Given : {2,4,10,12,3,20,30,11,25} k=2 2. Randomly assign means : m1=3 ; m2=4  k1={2,3} , k2={4,10,12,20,30,11,25} ,m1=2.5 , m2=16  k1={2,3,4} , k2={10,12,20,30,11,25} , m1=3 , m2=18  k1={2,3,4,10} , k2={12,20,30,11,25} , m1=4.75 , m2=19.6  k1={2,3,4,10,11,12} , k2={20,30,25} , m1=7 , m2=25 …..

14 Background ─ SOM & MST

15 The data and results We collect 366 records of data from a Taiwanese foundry, and six attributes as follows: (1)Prod_Qty [ ] (2)Prod_Time [648.92] (3)Small_Stops [280.54] (4)Standby [123.21] (5)Change-over_Time [30.12] (6)Breakdowns [46.43]

16 The data and results According to Yang et al., the result of using the SOM to cluster data is shown in Fig. 3. Size : 19*19 Neighborhood function : 8 Learning rate : 0.1 Learning epoch : 70 Form Fig. 3, it is determined that the number of cluster is seven.

17 The data and results

18 The data and results

19 Conclusion Given the clusters obtained and taking advantage of a visualization, we plot centers of clusters using parallel coordinates that provide an easier outlook for managers to assign. The draw back remains because given the graphs we still need to find the original values to know more about the details.

20 Personal Opinions Advantage Drawback Application … Other foundry
優點:實際運用於產業中 缺點:在於對圖分析的時候,管理者需要對於該領域很有經驗,才能夠下決策下的準確


Download ppt "Advisor : Dr. Hsu Presenter : Ai-Chen Liao"

Similar presentations


Ads by Google