ASMPT – The AD809 and
Lessons from the Early Days
During the company’s 50th anniversary, when
the screen displayed a certain old photo, it brought back a flood of memories!
One of those memories is of the AD809, the
first IC (integrated circuit) die bonder I worked with, back in 1987–1989. We had spent a
lot of time focusing on this machine, especially as after we had established
the opto-LED business with Sanyo HK’ support in process and material — our first real success in that opto LED area.
The next phase was moving into the IC
market. Our first major target was a top-tier subcontractor Amkor — the number one
subcon in the world at the time. But truth be told, I didn’t know much about
subcons or IDMs (Integrated Device Manufacturers) back then. My focus remained
solely on fine tuned the machine that could beat our competitors. I spent two
month in our old ASM Assembly Automation office 12/fl Watson’s centre.
I knew our competitors were strong — even though I had
never seen their actual machines myself. l honestly as it was second year in
the company. The sales team assured me they were.
In 1988, we conducted a one-month
evaluation of the Amkor K2 Buchon A809. A team of K2 engineers including JS
Park later joined our company to become head of ASM Korea. They ran tests on
legacy SOIC products. The results were promising — good scores on Bond accuracy +/-3 mils ,
angle of rotation +/- 3 degrees, Yield is higher than 99.93%, UPH (units per
hour), MTBA (mean time between assists), and MTBF (mean time between failures).
Then, three months later, Amkor P1 invited
us to participate in another equipment evaluation at their factory — ATP1 (Amkor
Philippine Factory 1). This time, we were up against three other competitors:
Kulicke & Soffa (K&S), MAGNUM, and AMI 3100). There were four machines
in total, each representing a different manufacturer. Three from USA and one
from Hong Kong.
Here’s how I remember it:
🔹 MAGNUM
Had a wafer handler.
The Field Service Engineer (FSE) was the MD
of the company — American ,surprisingly humble and polite to talk with.
The machine footprint was compact.
The linear bond arm was lightweight
and fairly fast.
🔹
KULICKE & SOFFA (K&S) model 6150
The machine was a huge and massive, imposing
unit with the wafer placed on top, facing downward. This is a beauty design as
the dust particle will not fallen onto the die surfaces.
The work holder was not very flexible. You
have to change the whole work holder from one leaden device to another.
The bond arm moved somewhat like a
ballpoint pen swinging 180 degrees up to pick the die and then press down to
place it on the lead frame bond pad.
According to the FSE (from Hong Kong), this
machine was expected to achieve a UPH of around 8,000 for SOIC-L, which was
much higher than the ASM 809’s 5500
However, the flexibility was poor — it reportedly took
two full days to convert from one device to another. That is the weak point for
a Suncon to use as they have frequent device conversion.
🔹 ASMPT (AD809)
I personally installed this machine, and I
had practiced the conversion process many times as that was the main concern
from Amkor. I remember our conversion
time was just 16 minutes —
And here’s what had to be changed each
time:
Wafer, Collet, Epoxy nozzle / shower
head, Magazine, Reset elevator and indexer, we also had to adjust bond
force and test epoxy disperse volume and position.
Honestly, considering the primitiveness of
the machine at the time, a 16-minute conversion was almost Guinness World
Record-worthy! As our competitors took more than one hours to two days!
The Magnum machine once had a wafer crash
during transfer and needed fine-tuning. I also felt the Managing Director of
Magnum wasn’t too familiar with the machine or its conversion process.
🔹 AMI 3300,
this was the main machine supplier at Amkor back then, they were using 3100 and
3300 was a new model.
It was driven by a high-power D.C. motor on
the Y-axis, moving at high speed.
The X-axis indexing was done via a series
of rollers and sensor system to accurately stop the lead frame in place.
The monitor was in color, whereas the ASM
AD809 had a tiny 9-inch black-and-white monitor.
The PRS (Pattern Recognition System) on the
AMI was still in 256 grey levels, while the ASM AD809 used the PD1000 PRS
system with 64 grey levels — which was actually more advanced for its purpose.
⚖️ The
Evaluation Results
Before the evaluation began, many customers
— including equipment engineers like Obi Gracia and the QA team — believed ASM was
not going to win. But after several weeks of rigorous testing involving:
Different die sizes, various leadframe
types,
They took Measurements of all bonded sample
units with 100% full check. And later we found that ASM AD809 was the winner in
many key parameters.
K&S lost credibility when one of their
boards failed in the second week, delaying their recovery by nearly two weeks.
MAGNUM had bonding results that were not as
good as ASM’s.
AMI was a strong contender — fast and accurate — but it couldn’t beat the AD809 in
conversion time and overall UPH.
When the final report was published, we
were overjoyed. ASM won the benchmark, and we were subsequently awarded orders
from Amkor Korea.
This was a big deal, because most smaller
customers tend to follow the decisions of major subcons like Amkor. And yes, we
did get additional orders stemming from this evaluation.
✅ Aftermath
& Reflection
The AD809 turned out to be a simple,
reliable, manually assisted low cost die bonder, well-suited for the IC
industry — especially for subcons — during the years 1989 to 1993.
💡 The
Moral of the Story?
Never underestimate yourself.
Success is often the marriage of
preparation and opportunity.
Had we not prepared thoroughly, had we not
believed in our machine, and had we not seized the opportunity when it came — we might have
missed out on something big. The AD809 taught me that even a “simple” machine, backed by
skill, preparation, and confidence, can win the day.
The pictures showed the HK former
governor Sir Wilson presented the HK Governor’s Awards of Hong Kong Design
competition 1988 to ASM former chairman Author de Pardo.
Paul
