■ Also Eliminating Operator Differences
When manual processes remain,
Operator A.
Very skilled.
B.
Slightly off.
This is not stable.
So
jigs,
positioning,
pressing methods,
and work sequence
are standardized.
In other words, it is necessary to absorb differences in human skill through the process.
■ Converting the Feel of Skilled Workers into Numbers
“Press about this much.”
“Pull about this much.”
These cannot be reproduced.
They are converted into
pressure ○ N,
speed ○ mm/s,
and time ○ seconds.
In other words, volume production is also converting tacit knowledge into control values.
■ Process Control Is Needed Even After Automation
Automatic machines do not vary.
That is not the case.
Suction pressure,
application pressure,
camera position,
and servo position.
The equipment itself also has variation.
That is why even after automation,
equipment conditions are monitored.
In other words, replacing people with machines does not make quality control unnecessary.
■ Also Consider Differences Between Machines
Equipment unit 1.
Unit 2.
The same manufacturer.
The same specifications.
But they are not exactly the same.
When volume production quantities increase and multiple lines are introduced, differences between machines can appear.
In other words, conditions that work on one machine and conditions that work across multiple machines are different.
■ Transfer Between Factories Creates Even More Differences
It was established at the Japan factory.
It is transferred to an overseas factory.
Material procurement,
equipment,
temperature and humidity,
operators,
and transport
change.
Even if the drawings are the same, the process is not the same.
In other words, in manufacturing transfers, it is necessary to transfer not the drawings but even the manufacturing conditions.
■ The Value of Process Condition Documents
Material part number,
dies,
speed,
tension,
pressure,
temperature,
and inspection conditions.
These are recorded.
Do not rely on the memory of the person in charge.
In other words, documenting volume production know-how increases reproducibility.
■ Record Not Only Good Conditions but Also Failed Conditions
Condition A.
Success.
Condition B.
Air bubbles occurred.
Condition C.
Warping increased.
Such failure data is also important.
For similar materials next time, conditions to avoid will be known.
In other words, processing know-how is also the accumulation of failure data, not just success stories.
■ Inspection Is Not Only Pass/Fail Judgment
Dimensional measurement
and image inspection
are performed.
Only pass
or fail
is looked at.
But looking at the data over time,
trends become visible, such as
dimensions gradually getting larger
and positions gradually shifting.
In other words, inspection data can be used as a sensor for early detection of process abnormalities.
■ Trend Management Is Important
Within specification.
So no problem.
Rather than that, look at whether
the center value is drifting
or the variation is widening.
No defects have appeared yet.
But they may appear in the future.
So adjust early.
In other words, find process changes before defects occur.
■ Look at the Distribution, Not Just the Average
Average dimension 10.00 mm.
Good.
But widely distributed
from 9.80 to 10.20 mm.
In another lot,
from 9.98 to 10.02 mm.
Both average 10.00 mm.
But their stability is completely different.
In other words, in volume production, it is important to look at the range of variation, not the average.
■ The Concept of Process Capability
Does process variation fit within the drawing tolerance with sufficient margin?
This is looked at.
If process capability is high,
it is easier to maintain quality even if materials or environment change somewhat.
In other words, in stabilizing volume production, it is important not only to look at the resulting yield, but
to look at the margin of the process itself.
■ Do Not Hide Process Defects with 100% Inspection
The process is unstable.
But defects are removed by 100% inspection.
Only good parts reach the customer.
At first glance, no problem.
However,
inspection costs,
scrap,
and equipment stoppages
increase.
Fundamentally, rather than finding defects,
making a process that does not create defects is stronger.
■ Inspection Is the Last Line of Defense
Stabilize as much as possible within the process.
Then inspect.
This order is important.
Improving inspection technology is necessary, but
quality cannot be created by inspection alone.
In other words, quality is created in the manufacturing process and confirmed by inspection.
■ Using Image Inspection for Process Improvement
Image inspection detects
positional deviation,
chips,
and foreign matter.
Furthermore, data is accumulated.
Under which die condition did chips increase?
With which material lot did positional deviation increase?
These are looked at.
In other words, image inspection becomes not only a pass/fail device, but a data source for process learning.
■ The Same Thinking Applies to AI Inspection
AI is introduced.
It finds defects.
This alone is not enough.
Which defects are increasing?
Which process conditions are they related to?
These are analyzed.
In other words, value increases by progressing from inspection AI to process improvement AI.
■ Not Everything Needs to Be AI
For dimensions,
simple appearance,
and position,
conventional image processing or rule-based methods may be more suitable.
For complex abnormalities
and appearance requiring human judgment,
AI may be effective.
In other words, what matters is not using AI, but choosing the optimal inspection algorithm.
■ Traceability Is Important in Stabilizing Volume Production
A defect occurs.
If
material lot,
processing date and time,
equipment,
die,
operator,
and inspection results
are known, the cause is easier to trace.
In other words,
“when, what, and under which conditions it was made”
is recorded.
This makes volume production quality strong.
■ Management by Reel Is Also Effective
Reel number,
material lot,
processing conditions,
and inspection results
are linked.
Even if a problem appears at the customer, the scope can be narrowed.
In other words, supply format and quality information can be linked.
■ Being Able to Limit the Scope of Defects Is Also Valuable
No traceability.
One problem.
Recall of all lots.
On the other hand, with history,
it can be narrowed down to
only the target reels
and only the target time period.
In other words, traceability leads not only to quality assurance, but
to limiting losses when quality problems occur.
■ Change Control Protects Volume Production Quality
Change of material manufacturer.
Die change.
Equipment change.
Processing speed change.
Something is changed.
Then performance may also change.
That is why before and after the change is confirmed.
In other words, in volume production,
not treating small changes as small changes
is important.
■ Even "Equivalent Products" Are Not Necessarily the Same
Thermal conductivity
and thickness
are the same.
So the material is changed.
However,
hardness,
surface,
and peel force
differ.
Processing results change.
In other words, material changes require not only comparison of material specifications but processing evaluation.
■ Supply Chain Changes Also Become Process Changes
The material manufacturer changes its manufacturing site.
Raw materials change.
Surface treatment changes slightly.
On the processing manufacturer’s side, it is the same part number.
But behavior may change.
In other words, in long-term volume production, information sharing with suppliers is also important.
■ Volume Production Quality Includes Transport
Good right after processing.
Warping after arriving at the customer.
Reel pitch changes.
Peel force changes.
Then it is meaningless.
High temperature,
humidity,
vibration,
and transport time
are also considered.
In other words, it is necessary to look at quality at the time of use, not quality at shipment.
■ Packaging Is Also Part of the Process
Moisture-proof packaging.
Light shielding.
Vacuum packaging.
Trays.
Reel fixing.
These are chosen according to material properties.
In other words, packaging is not just the logistics department’s job, but a process that protects product quality.
■ Also Evaluate Long-Term Storage
The customer does not necessarily use it on the same day.
After storage for
3 months
or 6 months,
it is used.
At that time, whether
adhesion,
warping,
and peeling
have not changed
is confirmed as needed.
In other words, volume production quality includes the time axis.
■ Initial Performance Alone Does Not Mean Volume Production Is Stable
Initial thermal performance.
Passed.
But delamination after thermal cycling.
Positional deviation after long periods at high temperature.
Then it cannot be used in long-life products.
In other words, in stabilizing volume production, not only manufacturing variation, but
performance stability during the period of use is also important.
■ Look at Changes Due to Thermal Cycling
High temperature.
Low temperature.
Repeated.
Expansion differences between dissimilar materials,
degradation of adhesives,
and interface changes
occur.
As a result,
warping,
lifting,
and increased thermal resistance
may occur.
In other words, it is necessary to think of component quality right after processing and long-term thermal performance as connected.
■ Summary of Part 2: Reproducibility Is Protected by "Systems and Data," Not "Human Memory"
Differences in human skill are absorbed with jigs and control values.
Even with automation, differences between machines and between sites remain.
That is why not only drawings but the manufacturing conditions themselves are documented, and both successful and failed conditions are recorded.
Inspection does not end with pass/fail judgment.
Look at trends and distributions, and find changes in the process before defects appear.
Quality is not created by inspection, but created in the process and confirmed by inspection.
Traceability speeds up the investigation of causes and limits the scope of impact when problems occur.
Changes to materials, equipment and the supply chain are verified by processing evaluation, even for “equivalent products.”
And quality does not end at shipment.
Transport.
Packaging.
Long-term storage.
Thermal cycling.
Evaluate by “quality at the time of use,” including all of these.
In other words, what we looked at in Part 2 are the systems for “maintaining, watching over and protecting” a good process.
In Part 3, we consider how to deal with volume production defects that still occur, and the value OTIS can provide in stabilizing volume production.
Column Supervisor: Koji Kakumoto (President, OTIS Co., Ltd.)
After studying abroad and working in planning and development at a trading company, he joined OTIS Co., Ltd. in 2011. While primarily working in the Corporate Planning Department, he has also served concurrently in manufacturing and technical divisions, and since 2018 has served as President, working to drive business growth and strengthen the organization.
This article is a general technical explanation intended for educational purposes and does not refer to any specific company, product, or technology.
