A product is made of many parts that have been made from different materials. Many electronic products use parts made from plastics, sheet metal, machined parts, rubber, castings, etc. Parts made from these processes have unique properties, and manufacturing dictates the tolerances that can be specified with these parts. Sheet metal parts require a much wider tolerance band compared to machined parts, Plastic molded parts, made from mature molds and processes, have low variation in dimensions from batch to batch. Machined parts will vary from batch to batch, and many operators tend to make parts I on the high side of the tolerance, so they can be able to reduce some dimensions in the future if the parts fail to pass inspection tests.
Knowledge of manufacturing processes, and proper use of six sigma design for mechanical parts will reduce the need for conservative design thereby decreasing the costs of the product as well as providing for high-quality parts.
In mechanical design, statistical design analysis can be substituted for worst-case tolerance analysis. A case study in mechanical design tolerance analysis is that of a typical vibrating probe that is used for angioplasty medical operations. As shown in Figure 8.8, it consists of a vibrating element of wires wound around a magnetic barrel, and a cover to enclose the assembly. The vibrating barrel has an outside diameter 0.0075 士 0.0002 inches，and the winding coil around the barrel has an outside diameter equal to 0.0027 ± 0.0002 inches. The wires and vibrating barrel were purchased from outside suppliers, and therefore had fixed tolerances. The designer is faced with a dilemma: If the cover specifications are too loose, then the mechanical assembly gap between the cover, barrel, and the wires is too large, causing the assembly to come apart. If the cover specifications are too tight, then the mechanical PCB assembly design has interference. The statistical analysis allows for the best design to meet this contrasting set of conditions.
Using statistical design analysis, based on the RSS values ofσ, the design quality prior to manufacturing can be calculated as follows (Table 8.13):
Cover nominal = barrel nominal + 2 wire nominal + gap (6σsystem)
Using this RSS technique, it can be seen from Table 8.13 that the gap should be set to 0.0004 inches, regardless of whether one assumes three or six sigma incoming parts. The nominal of the cover will be equal to the nominal of the components of the assembly plus the gap, or 0.0133 inches. If the cover is given a similar tolerance of ±0.0002 inches as the purchased parts of barrel and wires, then the minimum of the cover (0.0131 inches) is in interference with the maximum of the assembly by 0.0004 inches (maximum barrel + maximum 2 wires =0.0077 + 0.0029 • 2 = 0.0135 inches). The expected defect rate is half the normal, since defects only occur on one side of the gap distribution.