Reducing Power Consumption by Selecting Optimal Oven Recipes

By Abhinav Ajmera, Sreekanth Varma Penmatsa and Prof. S. Manian Ramkumar for KIC

The goal of this article is to study the efficiency of the KIC 2000’s Auto-Focus Power option to select an optimal oven recipe for reduced power consumption within Sn-Pb and Pb-free reflow soldering.

The authors’ objective was to use the new KIC Auto-Focus Power option to generate and select oven recipes for reduced power consumption, and to measure the reduction in energy consumption by comparing the following:

  • The energy consumed for an existing in-spec oven recipe for a production run
  • The equivalent energy consumed after changing the oven recipe to the settings suggested by the new KIC Auto-Focus Power option.

KIC2000 Auto-Focus Power Software and Terminology

The new KIC2000 Auto-Focus Power software features the following three profile optimization settings:

  • Minimize Process Window Index (PWI): This option searches for different set point temperatures and conveyor speeds that will position the profile towards the center of the process window. The PWI is a measure of how well a profile performs relative to critical process limits. It is specified as an absolute value in percent. The lower the PWI, the closer the profile is to the specification targets. The center of the process window is zero and the extreme ends of the process window are ±99 percent. A PWI of 100 percent or more will indicate that the profile is not in specification (Figure 1).
  • Maximize Conveyor Speed: This option searches for the set point temperatures that will maximize conveyor speed for an in-spec profile.
  • Minimize Energy Consumption: This option searches for an in-spec profile that will minimize power consumption and maintain the PWI below a specified maximum.

pwi-bullseye-tcs-small

Experimental Study

Three different companies were selected for this project:

Company A: Surmotech CMS is a full-service contract manufacturer specializing in high reliability medical, industrial and military applications. Located in Victor, NY, Surmotech is an ISO 9001:2000 registered company that focuses on providing complete turnkey solutions that include design, prototyping, engineering, materials management, testing and field service.

Company B: Marquardt is a global manufacturer of electromechanical and electronic components, supplying the hand tool and automotive industries.

Company C: SenDEC is organized into two main business units: the Contract Electronics Manufacturing (CEM) Group and the Products Group. SenDEC’s CEM Group provides electronics manufacturing services (EMS) including design, prototype, PCB assembly, electromechanical assembly, test engineering, rework, material management and turnkey box build services. The SenDEC Products Group manufactures its own family of digital monitoring, display and control devices for numerous markets across the globe.

The different assemblies considered for the experimental runs in the three companies are shown:

* The number in the parenthesis indicates number of products run

* The number in the parenthesis indicates number of products run

Power consumption measurement was carried out by installing a meter on the oven at each site to record the amount of power being consumed for the production runs for each profile.

Experimental Procedure

At each site, the product that was to be used for the production runs was identified. Reflow profile set points and paste details for that product were taken from the customer. The product then was run using the existing profile set points to check if the reflow profile was in-spec with respect to the paste manufacturer specifications or the company’s approved specifications.

If the profile was not in-spec, the customer was asked to bring the process in-spec, using its current method. Once the profile was in-spec, the production run was executed for at least a three-hour period. Following this period, the new KIC Auto-Focus Power option was used to identify a recipe (oven set points) for reduced power consumption.

The new recipe was loaded in the oven. After the oven stabilized, a test board was run for approval by the quality control department within the company. Subsequent to the approval from the quality department, the production run was executed for the next three-hour period. The power consumption was recorded continuously for both profiles and then analyzed for difference and statistical significance of the difference, using a t-Test.

Data Collection

Data collection for the following three companies — Company A, Company B and Company C — is below.

Company A

Oven name: Speedline Technologies Electrovert Bravo 8105

Number of Zones: 8

Number of products run: 1 Sn-Pb

chart 2Company B

Oven name: rehm V8 nitro 32 B

Number of Zones: 7

Number of products run: 1 Sn-Pb

chart 3

chart 4Company C

Oven name: Vitronics Soltec XPM2

Number of Zones: 8

Number of products run: 1 Sn-Pb, 2 Pb-Free

chart 5chart 6chart 7Data Analysis

Company A: The data analysis for the Sn-Pb product run is discussed in Figure 2

Figure 2. Mean Power Consumption Comparison - Company A

Figure 2. Mean Power Consumption Comparison – Company A

From Figure 2 it is evident that the KIC profile appears to be consuming less power than Company A’s profile. This is indicated by the means (11122 and 11412 Watt-Hour respectively) shown in the table within the graph, as well as the relative position of the peaks for the fitted normal distribution. The standard deviation for Company A’s profile (395.4 Watt-Hour) is greater than the KIC profile (318.1 Watt-Hour), indicating a much tighter power consumption profile for KIC recommended set points, when compared to Company A’s set points.

Figure 3. Boxplot for Power Consumption in Company A

Figure 3. Boxplot for Power Consumption in Company A

The box plot (Figure 3) reveals that the median power consumption for the KIC recommended profile was lower than Company A’s profile and also the power consumption distribution for the KIC recommended profile was considerably tighter than that for Company A’s profile. Even though, graphically, there seems to be a difference in the power consumption, the KIC recommended profile provides less power consumption than Company A’s profile. It was necessary to determine if the difference observed was statistically significant. This was carried out using the 2-Sample t-Test.

The 2-Sample t-Test is a hypothesis test for two population means to determine if they are significantly different. This procedure uses the null hypothesis that the difference between two population means is equal to a hypothesized value (H0: μ1-μ2 = 0), and tests it against an alternative hypothesis, which can be left-tailed (μ1-μ2 < 0) or right-tailed (μ1-μ2 >0). If the t- Test’s p-value is less than the chosen significance level (a=0.05), the null hypothesis will be rejected.

Two-Sample t-Test and C1: Power Consumption, Profile Chart-8

Difference = μ (KIC) – μ (Company A)

Estimate for difference: -290.9

95% upper bound for difference: -214.7

t-Test of difference = 0 (vs <): T-Value = -6.31; p-Value = 0.000; DF = 229

The results of the t-Test (p-value < 0.05) indicate that the difference between the average power consumption for the two profiles is significant. Furthermore, there is a reduction in power consumption of 2.55% (290.9 Watt-Hour) with the KIC recommended profile.

Company B: Company B had been using JEDEC specifications instead of the paste specifications for the Sn-Pb product considered for the experiment. JEDEC specifications are much wider than the paste specifications, thereby posing the challenge of shrinking the process window within tighter specifications without compromising the quality of the output. For a better analysis, the KIC Auto-Focus Power option was used to generate profiles for both the JEDEC and the paste specifications. The data analysis followed the identical format to the above referenced company A.

Both KIC profiles (KIC JEDEC and KIC Paste Specs) appear to be consuming less power when compared to Company B’s profile, with the KIC Paste Specs being the best. This is indicated by the means (12725, 11216 and 13042 Watt-Hour), as well as the relative position of the peaks for the fitted normal distribution.

The standard deviation for Company B’s profile (2052 Watt-Hour) is slightly less than the KIC JEDEC profile, but greater than the KIC Paste Specs profile (1969 Watt-Hour). This has resulted in a much steeper and more tightly fitted distribution for the KIC Paste Specs Profile.

The box plot comparing the profiles for Company B shows that the median power consumption is higher for Company B’s profile (13039.1 Watt-Hour) as compared to the KIC JEDEC (12999.1 Watt-Hour) and KIC Paste Specs (11165.7 Watt-Hour) profiles.

Comparison between Company B (JEDEC Specs) and KIC JEDEC Specs Profiles

The results of the t-Test (p-value > 0.05) indicate that the difference between the average power consumption for the two profiles is not significant. However, there is a reduction in power consumption of 2.43 percent (317 Watt-Hour) with the KIC recommended profile. When using the JEDEC specifications, the profile difference was not statistically significant.

The results of the t-Test (p-value < 0.05) indicate that the difference between the average power consumption for the two profiles is significant. Furthermore, there is a reduction in power consumption of 14 percent (1826 Watt-Hour) with the KIC recommended profile.

Company C— Product #1 (Pb-Free): The data analysis followed the identical format to the above referenced company A and B.

The results of the t-Tests (p-value < 0.05) indicate that the difference between the average power consumption for the two profiles is significant. There is a reduction in power consumption of 2.36 percent (261 Watt-Hour) with the KIC recommended profile.

Company C— Product #2 (Pb-Free): The data analysis followed the identical format to the above referenced company A.

The results of the t-Test (p-value > 0.05) indicate that the difference between the average power consumption for the two profiles is not significant. There is very minimal reduction in power consumption (45 Watt-Hour) with the KIC recommended profile.

Company C— Product #3 (Sn-Pb): The data analysis followed the identical format to the above referenced company A.

The results of the t-Test (p-value > 0.05) indicate that the difference between the average power consumption for the two profiles is not significant; however, there is a reduction in power consumption of 2.3 percent (213 Watt-Hour) with the KIC recommended profile.

Comparison between the Average Power Consumption for different companies

Figure 4. Comparison of Mean Power Consumption for All Profiles

Figure 4. Comparison of Mean Power Consumption for All Profiles

In all cases, the KIC recommended profiles have lower power consumption than the existing company profiles.

Comparison between the Standard Deviation of the Power Consumption for different Companies

Figure 5. Comparison of Std. Dev. of Power Consumption for All Profiles

Figure 5. Comparison of Std. Dev. of Power Consumption for All Profiles

Moreover, the Standard Deviation of Power Consumption for all KIC Profiles is less than the existing company profiles.

Conclusion

The recipes recommended by the new KIC Auto-Focus Power feature have been observed to have either reduced the power consumption when compared existing recipe or maintained it (Figure 6 and 7).

Figure 6. Power Consumption Reduction for all Profiles

Figure 6. Power Consumption Reduction for all Profiles

Figure 7. Percentage Reduction in Power Consumption

Figure 7. Percentage Reduction in Power Consumption

The new Auto-Focus Power option has the capability to generate a reduced power consumption recipe without compromising the productivity and quality of the output. Additionally, the software provides the flexibility to incorporate the restrictions offered by the reflow oven. The updated software version has an large solder paste menu (both Sn-Pb and Pb-Free) to select from, which automatically feeds the appropriate solder paste specifications.

###

Abhinav Ajmera, Sreekanth Varma Penmatsa and Prof. S. Manian Ramkumar may be contacted at the Center for Electronics Manufacturing and Assembly, 78 Lomb Memorial Dr., Rochester Institute of Technology, Rochester, NY 14623; 585-475-6934; Fax: 585-475-7167; Web site: http://smt.rit.edu.