Center for Electronics Manufacturing and Assembly
Rochester Institute of Technology, Rochester, New York 14623
http://smt.rit.edu
目的
研究KIC2000的Navigator Power功能,并找出最好的炉温设置来降低焊接中的电能消耗。
方法
              使用KIC Navigator 找出最省电的炉温设置
              •      用下面的方法对比电能消耗:
              o       使用当前的炉温设置,测出回流炉的电能消耗量
              o       使用KIC Navigator优化出的炉温设置,再次测量回流炉的电能消耗量
 
KIC 2000 – Navigator
在KIC 2000 Navigator Power软件中有三种优化模式:
              •      最小的PWI:
              o       这个模式下软件会优化链速及炉温以降低PWI (制程窗口指数)
              •      最大的链速:
              o       这个模式下软件会调整炉温以达到最大链速
              最少的电能消耗:
 
o      这个模式下软件会降低炉温设置和链速来降低电能消耗,同时保证曲线的合格
KIC 2000 术语解释
制程窗口指数 (PWI)
PWI可以量化出当前曲线在制程界限中处于什么位置,并用百分比表示出来. PWI越低, 就越接近制程目标. 制程的中心点位0,制程窗口的另外两端为±99%.  PWI等于或大于100%表示曲线超出了制程范围。
In-spec Profile
PWI <100%的曲线表示合格.  在图1中,最大圆圈(100%)内的曲线都是合格的, 任何超出最大圆圈区域的曲线都不合格.
[Process Windows Index (PWI)]
Figure 1:  Process Window Index (PWI) and In-spec Profile
图中的每个点代表具体的PWI数值. 离中心最远的点 (最大PWI) 表示这个曲线最差。
实际测试
•   我们选择了以下3家公司.
公司 A: Surmotech CMS是一家从事医疗/工业/军事方面的制造商. 地址位于纽约, Surmotech通过了ISO 9001:2000认证,它提供全方面的解决方案,包括设计, 打样,工程,物料管理,测试及售后服务.
公司 B: Marquardt是一家环球的电子/机电制造商,提供手工具及自动化工业品.
公司 C: SenDEC 包括2个事业群: Contract Electronics Manufacturing (CEM)公司和SenDEC's Products公司.  SenDEC’s CEM 公司提供电子代工服务(EMS)包括设计,打样, PCB贴装, 电子装配, 工程测试, 返修,物料管理及全面服务. SenDEC Products公司 生产自有品牌的电子监视器和控制器.
•     在这3个公司中,用不同的产品做了以下测试:
·          
 

  Sn-Pb Pb-Free
Company A X  
Company B X  
Company C X X (2)
*括号内的数字表示产品测试了几次
•     我们安装了一个电能表来测量电能消耗,记录每个产品和曲线的电能消耗.
测试程序
 把每个地点的测试产品看成是相同的.
 使用客户处的炉温设置及锡膏.
 使用现有的炉温设置,测出曲线是否合格
 如果不合格,客户使用他们的方法来调整到合格
 曲线合格后,至少运行3个小时
 接下来的3个小时内, 用KIC Navigator 找出一个省电的炉温设置。
 让回流炉使用这个设置,在炉温稳定后, 经过公司内的QC部门同意后,把这个测试产品放入炉内。
 接下来的3小时会不断地进行产品测试。
2种曲线的电能消耗量会被连续的记录下来,然后使用t-Test来分析和统计它们的不同
·          
数据收集
公司 A:
·         Oven name: Speedline Technologies Electrovert Bravo 8105
·         No of Zones: 8
·         No of products run: 1 Sn-Pb
Sn-Pb Product Zone #1 Zone #2 Zone #3 Zone #4 Zone #5 Zone #6 Zone #7 Zone #8 Conveyor Speed PWI Temperature set point difference
Existing set points 120 140 170 170 180 205 230 245 30 in/min N/A 13C
KIC set points 112 130 160 182 182 194 245 242 25 in/min 22%
公司 B:
·         Oven name: Rehm V8 Nitro 32 B
·         No of Zones: 7
·         No of products run: 1 Sn-Pb
Sn-Pb Product Zone #1 Zone #2 Zone #3 Zone #4 Zone #5 Zone #6 Zone #7 Conveyor Speed PWI Temperature set point difference
Existing set points 120 140 165 180 240 200 165 27 in/min N/A 107C
KIC set points
(Paste Specs)
94 129 154 193 234 166 133 24 in/min 93%
·          
Sn-Pb Product Zone #1 Zone #2 Zone #3 Zone #4 Zone #5 Zone #6 Zone #7 Conveyor Speed PWI Temperature set point difference
Existing set points 120 140 165 180 240 200 165 27 in/min N/A 21C
KIC set points
(JEDEC Specs)
120 139 161 170 237 187 175 24 in/min 53%
公司 C:
·         Oven name: Vitronics Soltec XPM2
·         No of Zones: 8
·         No of products run: 1 Sn-Pb, 2 Pb-Free
Pb-Free Product
#1
Zone #1 Zone #2 Zone #3 Zone #4 Zone #5 Zone #6 Zone #7 Zone #8 Conveyor Speed PWI Temperature set point difference
Existing set points 110 120 160 180 200 240 270 275 25 in/min N/A 8C
KIC set points 110 120 160 180 199 234 270 274 23 in/min 63%
·          
Pb-Free Product
#2
Zone #1 Zone #2 Zone #3 Zone #4 Zone #5 Zone #6 Zone #7 Zone #8 Conveyor Speed PWI Temperature set point difference
Existing set points 135 145 155 185 220 245 285 275 23 in/min N/A 45C
KIC set points 97 124 151 183 222 247 286 290 22 in/min 72%
·          
Sn-Pb Product Zone #1 Zone #2 Zone #3 Zone #4 Zone #5 Zone #6 Zone #7 Zone #8 Conveyor Speed PWI Temperature set point difference
Existing set points 90 130 180 180 180 220 240 220 28 in/min N/A 50C
KIC set points 71 117 181 170 159 209 249 234 23 in/min 69%
数据分析
公司 A:
下面是有铅产品的数据分析:
:
[Mean Power Consumption Comparison - Company A]
2: 平均电能消耗对比公司A
图中我们可以明显看出,使用KIC曲线产生的电能消耗量低于工厂原本曲线的消耗. 从图中的平均值和最高点的正态分布位置也可以得出 (分别是11122 and 11412 ) , 公司A的曲线 (395.4瓦时) 比KIC的曲线要大 (318.1瓦时)。与A公司的炉温设置相比,KICNavigator power推荐的曲线具有更少的电能消耗量。
 
[Boxplot for Power Consumption in Company A]
3:  A公司的电能消耗柱形图
图3中可以看出KIC曲线的电能消耗的中间值低于A公司的现有消耗,并且电能的上下浮动量更加的小。
  图中可以看到电能消耗量的差异, KIC曲线已经比A公司曲线减少了电能消耗量,但还是需要知道这个差异是否有意义.  下面来对比一下.
我们取这2个样本来计算以确定是否有明显不同.  这里假设2者之间差异为0(H0: µ1-µ2 = 0),然后相对于另一种假设进行测试, 可以进行左尾检测 (µ1-µ2 < 0) 或右尾检测 (µ1-µ2 > 0). 如果概率少于选择的指标水准 (α=0.05), 那假设就错误了.
Two-Sample T-Test and CI: Power Consumption, Profile
Two-sample t-Test for Power Consumption
 
Profile N Mean StDev SE Mean
KIC 121 11122 318 29
Company A 121 11412 395 36
Two-sample t-Test for Power Consumption
Difference = µ (KIC) - µ (Company A) Estimate for difference: -290.995% upper bound for difference: -214.7t-Test of difference = 0 (vs <): T-Value = -6.31; p-Value = 0.000; DF = 229
t-Test的结果 (假设值 < 0.05) 表明了2者之间的差异是有效的.  而且,KIC推荐的曲线减少了2.55%的用电量 (290.9 Watt-Hour).
 
Company B:
在做有铅产品测试时,B公司使用JEDEC来代替锡膏制程.  JEDEC的制程范围比较窄, 因此很窄的制程窗口带来的挑战就是在窄的窗口里怎样来减少产品的不合格率. 为了更方便的分析, 我们用KIC Navigator 为JEDEC和锡膏制程找出曲线设置. 以下是数据分析:
 
[Mean Power Consumption Comparison - Company B]
Figure 4: Mean Power Consumption Comparison – Company B
KIC的2个曲线 (KIC JEDEC and KIC Paste Specs) 明显要比B公司的曲线要省电, 并且使用KIC锡膏制程的曲线时最好的.  图中的平均值也可以看出来 (12725, 11216 和13042瓦时)。
B公司曲线的标准偏差 (2052瓦时)略小于 KIC JEDEC的曲线, 但是大于KIC锡膏制程(1969瓦时).  这就导致了KIC锡膏制程的曲线紧凑和陡峭.
 
[Boxplot for Power Consumption in Company B]
5: B公司的电能消耗柱形图 
图5 可以看出B公司电能消耗的中心值(13039.1瓦时) 比KIC JEDEC (12999.1瓦时) 和 KIC Paste Specs (11165.7 瓦时) 要高.
 
Comparison between Company B (JEDEC Specs) and KIC JEDEC Specs Profiles:
Two-Sample T-Test and CI: Power Consumption, Profile
Profile N Mean StDev SE Mean
KIC JEDEC Specs 121 12725 2064 188
Company B
(JEDEC Specs)
121 13042 2052 187
Two-sample t-Test for Power Consumption
Difference = µ (KIC JEDEC Specs) – µ (Company B JEDEC Specs)
Estimate for difference:  -317
95% upper bound for difference:  120
t-Test of difference = 0 (vs <): t-Value = -1.20; p-Value = 0.116; DF = 239
t-Test的结果 (p-value > 0.05) 显示出2个曲线的平均耗电量差不多.  但是使用KIC Navigator power曲线减少了2.43% (317 瓦时) 的电能消耗,而用JEDEC制程的曲线则无明显减少.
 B公司 (JEDEC Specs) KIC锡膏制程比较:
Two-Sample T-Test and CI: Power Consumption, Profile
Profile N Mean StDev SE Mean
KIC JEDEC Specs 121 12725 2064 188
Company B
(JEDEC Specs)
121 13042 2052 187
Two-sample t-Test for Power Consumption
Difference = µ (KIC Paste Specs) – µ (Company B JEDEC Specs)
Estimate for difference:  -1826
95% upper bound for difference:  -1399
t-Test of difference = 0 (vs <): t-Value = -7.06; p-Value = 0.000; DF = 239
t-Test的结果 (p-value < 0.05)表明了2者之间的差异是有效的.  而且,KIC推荐的曲线减少了14%的用电量
 
以下是第一个无铅产品的数据分析:
[Mean Power Consumption Comparison - Company C Pb-Free Product 1]
6: C公司无铅产品1的平均电能消耗量比较
KIC的曲线比C公司使用的曲线更加省电.  从平均值和最高点的正态分布位置也能看得出 (分别为10800和11061瓦时)  C公司曲线的标准偏差(1178瓦时) 大于KIC曲线(1065瓦时).  所以C公司的曲线看起来短而宽。
[Boxplot for Power Consumption in Company C for Pb Free Product 1]
7: C公司无铅产品1的柱形图
图7可以看出C公司的电能消耗量 (11328.5 瓦时)要大于KIC的曲线 (10766.4瓦时).
 
Two-Sample T-Test and CI: Power Consumption, Profile
Profile N Mean StDev SE Mean
KIC 121 10800 1065 97
Company C
(Product #1)
121 11061 1178 107
Two-sample t-Test for Power Consumption
Difference = µ (KIC) – µ (Company C Product #1)
Estimate for difference:  -261
95% upper bound for difference:  -22
t-Test of difference = 0 (vs <): t-Value = -1.81; p-Value = 0.036; DF = 237
t-Test的结果 (p-value < 0.05) 表明了2者之间的差异是有效的. 
而且,KIC Navigator Power推荐的曲线减少了2.36%(261瓦时)的用电量.
以下是第二个无铅产品的数据分析:
[Mean Power Consumption Comparison - Company C Pb Free Product 2]
Figure 8: Mean Power Consumption Comparison – Company C Pb Free Product 2
KIC的曲线耗能比C公司的要少. 从平均值和最高点的正态分布位置也能看得出 (分别是11192和11237瓦时).尽管C公司曲线的标准偏差(984.2瓦时)要小于KIC曲线(1151瓦时).  所以KIC的曲线看起来短而宽.
.
[Boxplot for Power Consumption in Company C Pb Free Product 2]
Figure 9: Boxplot for Power Consumption in Company C Pb Free Product 2
2个曲线的能耗中间值(图9) 几乎差不多
 
Two-Sample T-Test and CI: Power Consumption, Profile
Profile N Mean StDev SE Mean
KIC 121 11192 1151 105
Company C
(Product #2)
121 11237 984 89
Two-sample t-Test for Power Consumption
Difference = µ (KIC) – µ (Company C Product #2)
Estimate for difference:  -45
95% upper bound for difference:  182
t -Test的结果 (p-value > 0.05)显示2者几乎没有差异. 这次KIC Navigator Power推荐的曲线相对减少了45瓦时的能耗.
公司 C: Product #3 (Sn-Pb)
以下是有铅产品的数据分析:
 
[Mean Power Consumption Comparison - Company C Sn-Pb Product]
Figure 10: Mean Power Consumption Comparison – Company C Sn-Pb Product
KIC曲线的能耗略比C公司使用的曲线能耗要少一些.  从平均值和最高点的正态分布位置也能看得出 (分别是9037和 9250瓦时).  C公司曲线的标准偏差值(1966瓦时)要大于KIC曲线(1544瓦时). 所以C公司的曲线看起来比较宽。
[Boxplot for Power Consumption in Company C Sn-Pb Product]
Figure 11: Boxplot for Power Consumption in Company C Sn-Pb Product
C公司曲线的中间值 (9731.28 瓦时)要比KIC曲线的 (8739.35 瓦时)要大。
Two-Sample T-Test and CI: Power Consumption, Profile
Profile N Mean StDev SE Mean
KIC 121 9037 1544 140
Company C
(Product #3)
121 9250 1966 179
Two-sample t-Test for Power Consumption
Difference = µ (KIC) – µ (Company C Product #3)
Estimate for difference:  -213
95% upper bound for difference:  162
t-Test of difference = 0 (vs <): t-Value = -0.94; p-Value = 0.175; DF = 227
t-Test的结果 (p-value > 0.05)显示2者差异不明显. 但是使用KIC推荐的曲线还是减少了2.3% (213瓦时)的能耗。
 
不同公司的平均能耗对照表
[Comparison of Mean Power Consumption for All Profiles]
图12 所有曲线的平均电量消耗对比
在所有的案例中,KIC软件推荐的曲线比这些公司现有的曲线要省电很多.
 
不同公司能耗偏差的对照表
[Comparison of Std. Dev. of Power Consumption for All Profiles]
图13: 所有曲线的电量消耗的标准偏差比较
所有KIC曲线的能耗标准偏差要小于这些公司现有的曲线
测试中的一些问题
A公司
 o     回流炉不能在KIC的推荐设置下保持稳定(由于温区间串温). 由于这个原因, KIC推荐的第一个区设置71度改成了110度.
 
C公司
·         软件在预测时出现了问题
结论
对比现有的曲线,KIC Navigator 推荐出的曲线更加省电,见图14:
[Power Consumption Reduction for all Profiles]
Figure 14: Power Consumption Reduction for all Profiles
 
此外, 在大多数案例中也证明了, 使用KIC Navigator可以降低2%或更高的能耗,见图15
·         :
[Percentage Reduction in Power Consumption]
Figure 15: Percentage Reduction in Power Consumption
 
KIC的Navigator Power可以在保证产品质量的前提下降低能耗。
·          
 
相对于回流炉的一些约束,KICNavigator Power显示出更多的灵活性
·          
 
KIC软件收集了大量的锡膏参数 (有铅和无铅).  可以自动得到对应锡膏的制程规格.