Specifications

Cleco
Electric Tool Control TME-100 Series
58 PL12-1300 12/00 en04d141.fm, 30.01.2001
from the average increases. Since the shape resembles a bell, it is sometimes referred to as a
bell curve.
The normal curve is defined by two conditions - the average of all items produced, and the
amount of variation from the average. We can think of these as the center and width of the bell,
respectively. The center is the arithmetic average of all items produced. The width is expressed
in terms of the standard deviation which is a statistical calculation for the amount of variation
from the average. The standard deviation, represented by the Greek letter sigma (
σ), holds a
fixed relationship to the normal curve, as follows (see graph 4-2b):
• 68% of all items produced will be within +1
σ of the average (two sigma spread).
• 95% will be within +2
σ of the average (four sigma spread).
• 99.7% will be within +3
σ of the average (six sigma spread).
These two characteristics - the average and the standard deviation - provide the foundation for
statistical process control. By taking sample measurements during production, we can predict
the average value and the amount of variation for all items produced.
Fig. 4-2: a) The Normal Curve Fig. 4-2: b) Areas Under The Normal Curve
4.1.3 The Procedure
The procedure for establishing a statistical process control program consists of three phases.
The first is to obtain statistical control - a state of random and stable variation. The second is to
establish process capability. A state of statistical control in itself does not assure that the process
is capable of meeting the specification. The limits of variation must be equal to or less than the
total specification tolerance. The third is to monitor the process throughout production using the
control chart, to detect and correct conditions that upset the stable pattern of variation.
Use of the -R Control Chart includes the following steps:
1. Select the Characteristic
Since a separate chart is required for each characteristic, practical considerations limit use to
selected requirements. A good candidate is a characteristic that is important to the function of
the item. Others include those that are causing a high cost loss because of scrap or rework, and
those that require evaluation by destructive testing.
2. Designate the Sample Size
It is usually desirable to keep the sample size small so that variation within the sample is at a
minimum, but not so small that statistical validity is lost. The control chart described in this docu-
ment is structured for a sample size of five units. If a sample other than five is selected, the man-
ufacturing or quality engineer can provide alternate factors for use in calculating the control
limits and standard deviation (Steps 8 and 10).
68%
95%
99.7%
1σ1σ2σ 2σ 3σ3σ
Avg.
Avg.