engineers use statistical process control (SPC) techniques to
help keep a process within control limits. Defects such as a missing component
or an upside-down connector are easy to identify. Measuring an electrical
quantity provides a wider range of results and involves instruments that
require calibration. Most articles about SPC seem to either assume perfect measurements or assume someone else will discuss
In my opinion, information gathered by SPC tools can offer clues
that help determine when an instrument might need recalibration prior to its
scheduled trip to a cal lab. No one wants to reject good products only because
a DVM went out of calibration.
When you measure, say, a
voltage at a test point on many of the same product and plot voltage versus
number of units, you will see a graph similar to the one shown. The graph
includes the X-bar, or average, value for a good product. You can use this type
of SPC graph to determine how closely the measured voltages match the expected
result. I purposely exaggerated the 35 voltage points to make their value
distinct and easy to see.
This graph also includes
information about the distribution of the voltage values around the 20V
average. On its own, the distribution shows nothing abnormal for this small
sample. I artificially set an upper-control-limit (UCL) and a lower-control
limit (LCL). SPC software would calculate the UCL and LCL from gathered data.
In this type of graph, signs of process problems can include:
1. Eight or more sequential
points above or below the average. 2. Seven or more points in an
up or down "ramp." 3. Seven or more sequential
points with the same value. 4. Fourteen or more
sequential points that alternate above or below the average. 5. Any point above or below the
UCL or LCL lines. 6. Two sequential points
above or below Âą2s (standard deviation). 7. Four out of five
sequential points that fall outside Âąs on the same side of the average.
(Ref. 1, 2)
All these characteristics can arise from production variables,
depending on what you measure. For electrical measurements, recurrence of the
first three situations indicates a need to examine instrument measurement
accuracy and consider the need for recalibration.
In a move that strengthens its 3D design business, Stratasys continued a 15-month buying spree this week by announcing its plan to acquire GrabCAD, a provider of a cloud-based collaboration environment for engineers.
Many diverse markets take advantage of semiconductor IP; so many that no one can recite the entire list without leaving off several. So why do we track all the vertical markets? They all have a unique set of requirements and value attributes differently. One major vertical market segment is automotive.
Focus on Fundamentals consists of 45-minute on-line classes that cover a host of technologies. You learn without leaving the comfort of your desk. All classes are taught by subject-matter experts and all are archived. So if you can't attend live, attend at your convenience.