Introduction:
The phrase “statistical quality control” (SQC) refers to the application of statistical methods to monitor and evaluate systems and to determine whether changing key input variable (KIV) settings is appropriate. The intent is to familiarize participants with a disciplined approach to documenting, evaluating, and improving product and manufacturing approaches.
Case Study: Printed Circuit Boards:
1. Printed circuit board (PCB) assemblies are used for sophisticated electronic equipment from computers to everyday appliances.
2. Manufacturing printed circuit boards involves placing a large number of small components into precise positions and soldering them into place. Due to the emphasis on miniaturization, the tendency is to reduce the size of the components and the spacing between the components as much as electrical characteristics will allow. Therefore, both the multiplicity of possible failures and also the number of locations in the circuit boards where failures could occur continue to increase.
3. Also, identifying the source of a quality problem is becoming increasingly difficult. One says that a unit is “nonconforming” if at least one of its associated “quality characteristics” is outside the “specification limits”. These specification limits are numbers specified by design engineers. For example, if voltage outputs of a specific circuit are greater than 12.5 V or less than 11.2 V we might say that the unit is nonconforming.
4. As usual, the company did not typically use the terms “defective” or “defect” because the engineering specifications may or may not correspond to what the customer actually needs. Also, somewhat arbitrarily, the particular company in question preferred to discuss the “yield” instead of the fraction nonconforming. If the “process capability” or standard fraction nonconforming is p0, then 1 – p0 is called the standard yield.
5. Typical circuit board component process capabilities are in the region of 50 parts per million defective (ppm) for solder and component nonconformities. However, since the average board contains over 2000 solder joints and 300 components, even 50 ppm defective generates far too many boards requiring rework and a low overall capability.
6. In early 1998, an electronics manufacturing company with plants in the Midwest introduced to the field a new advanced product that quickly captured 83% of the market in North America, as described in Brady and Allen (2002).
7. During the initial production period, yields (the % of product requiring no touchup or repair) had stabilized in the 70% range with production volume at 6000 units per month. In early 1999, the product was selected for a major equipment expansion in
8. Asia. In order to meet the increased production demand, the company either needed to purchase additional test and repair equipment at a cost of $2.5 million, or the first test yield had to increase to above 90%. This follows because the rework needed to fix the failing units involved substantial labor content and production resources reducing throughput.
9. The improvement to the yields was the preferred situation due to the substantial savings in capital and production labor cost, and, thus, the problem was how to increase the yield in a cost-effective manner.
Example(PCB Project Planning)
A. Convene experts and perform one-factor-at-a-time (OFAT) experiments because the project is not important enough for a three to six month scope.
B. $2.5M is a substantial enough potential payback to apply six sigma using a define, measure, analyze, design, and verify (DMADV) process.
C. $2.5M is a substantial enough potential payback to apply six sigma using a define, measure, analyze, improve, and control (DMAIC) process.
Answer: Convening experts is often useful and could conceivably result in quick resolution of problems without need for formalism. However,
(A)Is probably not the best choice because: (1) if OFAT were all that was needed, the yield would
Likely have already been improved by process engineers; and (2) a potential $2.5M payoff could pay off as many as 25 person years. Therefore, the formalism of a six sigma project could be cost justified.
The problem involves improving an existing system, not designing a new one.
