Sometimes there is no choice. If you
must get product out the door and all you have in stock is
rejected or defective
material, then you must sort through it and try to find the
good ones. (I would also look for another supplier or two)
This is very costly. (I would be trying to get the
supplier to pick up your cost of doing their job) But,
what choice is there? But you MUST meet customer's
requirements.
But first a little more on 100% inspection.
I am a firm believer that YOU
CAN NOT INSPECT QUALITY INTO A PRODUCT.
100% Inspection has an effectiveness of 40-65%. And that
doesn't count the 5-10% breakage. It
is a waste of people power. (Formally known as man-power)
I remember reading, I think it was in one of Juran's
books that 100% inspection is only 60% effective. But I have
found it to be a little less than that.
According to Dr. Jim Stewart of Northern
Illinois University in DeKalb, IL
While
working on my dissertation, I was reviewing some trade
magazines from the 50's. There were a number of case studies
showing 50-75% efficiency and a breakage rate (visual
inspections of wire wraps with pics) of 10-15%. Giving an
effectiveness of 40-65%
It was just about a year before the 'Six
Sigma' era. Motorola had just came out with a couple of
new radios. One was a KDT, a data terminal using 900Mhz, IBM
was the pilot customer. The other was the STX 800Mhz two-way
trunking radio and it was being sold as a pilot
project to the City Of Miami's Police Department.
The quality
of the KDT's was so bad that rumor had it there were two IBM
engineers testing each unit themselves before authorizing
shipments. The STX radios had received several complains that
the dispatchers could not hear the police officers in the
field. So the development and process
engineers got to work. But everything seemed to be fine on
that end. Ergo, it MUST be the labor force.
Motorola had released two new products,
both from what at that time was the 'Communication Sector' now
called 'Land Mobile' and before they got into full production
there were a gross amount of customer complaints.
I had already gone through, what at that
time was the ASQC
courses, where, as I recall, they said that 100% inspection is
only 60% effective, the quality manager decided that we would
100% inspect
the radios until we got 'x' good ones, then the sample
number would decrease. Well I had the proverbial cow. I
started my shift, being the only engineering support person on
2nd shift, I thought, if I have to look at every radio, then
I'm going to collect data so that it won't be a waste of time.
The radio factory only collected data on the prototypes and
maybe the first run. The Quality Engineering gurus were glad
to see some data finally being taken on actual product being
shipped.
At that time, Motorola tested for a test
called 'Dev' by modulating the radio with a 1khz tone. I
thought to myself, this is silly, who talks at 1khz. (Common
Sense) Since we were just whistling in to the radio and
then checking the Motorola service monitor anyway, I
started whistling not a single tone, but a sweeping tone
and taking my data. If a reading was a little low, I would try
around that frequency to see if I could produce a failure,
usually caused be an external
failure. At the end of the shift, I went to the office,
and on a Compaq 'portable', (you know, that white Comapq
with the green screen that looked like a sewing machine when
closed up and someone had to guts to call a 'portable'),
and with SQCpak entered my data. I then printed distribution,
X-Bar and Range
control charts
and left them on my boss's desk. The next day he went to the
daily product meeting and presented the Development
Engineering Manager with my charts.
Fortunately the Development Engineering
Manager actually looked at the charts and noticed that
consistently, my 'Dev' numbers where better than productions
and what they saw in the prototype. Production had an
automatic testing machine to test the radios, and the quality
department tested the radios by hand. The day after that, the
Development Engineering Manager waited for me to come in and
gave me some data the development team had collected to put
into the SPC,
(statistical
process control), software. The results were exactly
what they had gotten before. The very next day development
engineering watched me to find out what I was doing different.
It was of course the sweeping frequencies. (i.e. more
'real world' like)
This better 'Dev' number allowed them to
change a part in the radio that fixed the problem with the STX
radio. Now, or at least as of my last day there, Motorola was
testing all new radios with this sweeping frequency. What's my
point, we could have 100% inspected radios for years and never
found the problem. We were looking in the wrong place. So if
you find yourself needing to 100% inspect, at least collect
data and do the control charts, even just X Bar, Range and
a Distribution charts can tell you a lot. You never know what
it may reveal until you do.
Shortly after that I was promoted to
Component Engineering where I first heard of Six Sigma. I
wrote software that allowed Motorola and their suppliers
to use a standard SPC package to aid in starting their Six
Sigma program. This allowed suppliers to submit sample data
and control charts with every shipment coming into VQA.
Without this software, it would have been a mess trying to
train the inspectors what to look for with the different SPC
charts and without training the inspectors, the Quality
Engineers would have to look at each shipment and that would
have just made them over priced inspectors.
The concepts of Six Sigma are much
larger than just measurements of course. It's also about
getting a Return On Net Investments, (RONI), and
Return On Net Assets, (RONA).
