One way to address these issues is to implement a statistical process control (SPC) program to certify an oven's temperature stability and repeatability. Oven users want to maximize production and throughput, and they demand assurances about the process repeatability for a specific piece of equipment, or consistency between several machines. By implementing a certification program, oven manufacturers can prove the stability and calibration of their ovens for ISO 9001 and customer quality requirements as well as reduce or eliminate process variances between manufacturing lines.
For these reasons, one reflow oven manufacturer initiated a statistical process control (SPC) program to certify oven performance. The first step was to select a data collection device. Once selected, the data collection device was used to develop a calibrated test board that responds to even the smallest of variations in the oven. With the hardware in place and the initial testing completed, the CPK index was established. CPK index is a measure of process capability with a K index. The K index is the target values of the process. The acceptance criteria was as follows: Achieve a peak temperature of 422oF (217oC) with a change in temperature (gT) of no more than 4oC, and provide a time above liquidous (TAL) of 70 sec, +/-5 sec.
Selecting the Testing ToolsA pass-through thermal profiler was selected to monitor and record process test data. Before selecting the thermal profiler, a specific set of requirements was established:
- The datalogger must be capable of withstanding defined process temperatures through extended exposure times.
- The datalogger must provide accurate data.
- A sufficient number of data channels must be provided to record data from each of the thermocouples used on the test vehicle.
- The datalogger must provide this data at a resolution of 0.1oC with a minimum sample rate of 1 data point every 0.5 sec.
- The datalogger's software must support "cut and paste" functions to simplify transfer of collected data to a calculation spreadsheet, minimizing data input requirements and reducing the risk operator error.
- Calibration of the thermal profiler must be verified through the manufacturer.
Once a thermal profiler that met all of the specified criteria was selected, the next step was to find an optimal test vehicle.
Several types of thermocouple boards, ranging from copper clad and stainless steel to other materials with exposed thermocouple tips, were tested with a wide variety of effects. The material of choice proved to be 0.062" thick, Class 1 fiberglass laminate. The selected laminate responds quickly to variations in the oven but masks noise typically associated with bare thermocouples. In order to attach the thermocouples so all are thermally weighted evenly, the 0.011" thermocouple tips were embedded halfway into the fiberglass laminate. Combined with strain relief on each thermocouple, this provides a test vehicle capable of more than 200 runs through a reflow oven.
Test vehicle accuracy was established by connecting it to a calibrated datalogger, submerging it into an isothermal bath, and recording the data (figure 1). At 32oF (0oC), each channel was verified as functioning accurately. The test vehicle then was plunged into a 185oF (85oC) hot bath. Maximum rising slopes and peak temperatures were recorded to confirm temperature stability. Also, the maximum rising slope was analyzed to ensure that thermal reaction was the same for all thermocouples (within +/-1.5oC) and that the maximum slope numbers fell into the expected reaction slope window. With the datalogger still recording, the process was reversed, and the test vehicle again was plunged into the cold water. The maximum falling slope was recorded. Both maximum rising and falling slopes are significant factors to measure when determining the stability of a test vehicle.
After examining the data, it appeared that the ultraviolet-cured glue, used to attach the thermocouples to the boards' surface, began to deteriorate between 200 and 300 profiling runs. The boards were visually inspected, then placed into an isothermal bath for rigorous testing.
SoftwareAfter selecting a data collection device and test vehicle, an Excel spreadsheet program was created to calculate test results from collected data. For the reflow oven, a peak temperature value of 422oF (217oC) and a TAL of 70 sec was specified. These values were selected because they represented typical values for most solder pastes, and because most standard reflow ovens are capable of achieving the same target values.
When the Excel spreadsheet is opened, the user is prompted for machine configuration. Once initiated, the program prompts the user with an auto-open feature, which asks for the iterations of the machine (table 1):
- Type of machine.
- Number of zones.
- Number of cooling zones.
- Placement of the center board support.
- Other oven-specific information.
The SPC program is initiated by preheating the oven. When the oven has stabilized, the test vehicle is sent through for a total of sixteen passes. The data is downloaded and copied to the Excel spreadsheet, and the CPK values are calculated. A "passing" CPK value of 1.33 or greater is required for both peak temperature and time at liquidous (TAL). If the oven is more than a few degrees or seconds away from the target peak temperature or TAL, respectively, or has a gT greater than 4oC peak/10 sec TAL, then the CPK value will be less than 1.33.
Using predetermined engineering data specifications, the limit is defined as a 4oC gT. Multiplying the engineering data specifications by three provides the upper and lower specification limits required for CPK. This sets a process control limit of 4oC gT, with an upper and lower specification limit around the target of 422oF (217oC), or around a TAL value of 70 sec. Alpha Metals generally specifies a 30 sec TAL gT. For the purposes of this study, the reflow oven manufacturer cut that value by half, yielding a 15 sec TAL gT, with the upper and lower specification limits set at three times the engineering data specifications.
If the process standard deviation is tight enough to be within the set gT range and target specifications are met, the CPK will be 1.33 or greater. If the CPK value is too high, the specification limits will be too far apart. CPK values of 3, 4 or 5 indicate that the specification limits are too wide to fail and are not realistic or useful. Target CPK values range from 1.34 to 2.0.