Pick and place machine 1
If you’ve worked on hardware long enough, you’ve probably asked (or complained about) at least one of these questions:
From the outside, SMT looks straightforward: place parts, solder them, done.
In reality, modern pick-and-place machines are precision motion systems, optical measurement platforms, and real-time control systems combined. The constraints of these systems explain most of the “annoying” answers you hear from manufacturers.
Let’s break it down — from an engineer-to-engineer perspective.
Setup cost is not about board complexity. It is about machine preparation.
Before the first component is placed, the SMT line must be configured for your specific product:
A pick-and-place machine does not “auto-understand” your design. Every component becomes a data object in the machine’s system. Even a board with 20 parts still requires the same engineering steps.
That is why setup cost exists — you are paying for engineering time and machine occupation, not solder paste or components.
Stencil checking 1
This surprises many engineers, especially those used to software timelines.
In SMT, time is not dominated by placement speed alone. It includes:
Most factories do not start a production run until the line is stable. Running faster before stability simply increases defect rates and rework time — which actually delays delivery.
For low-to-mid volume orders, engineering and stabilization time dominates, not raw placement speed.
“Scrap” does not mean negligence.
During SMT placement:
High-speed placement machines are designed to reject anything that looks abnormal. These parts are physically thrown into a reject bin to protect overall yield.
This is why manufacturers often request extra components (for example +2% or +5%). It is not wasteful — it is a buffer against automated quality control doing its job.
BGA is not difficult because of placement — it is difficult because you cannot see the result.
Key challenges include:
BGA chip on PCB 1
From the machine’s perspective, BGA alignment depends heavily on lighting, contrast, and image algorithms. From the factory’s perspective, one hidden defect can mean a full board failure later.
That’s why BGAs trigger:
At 0201 and below, physics starts to dominate manufacturing.
Problems include:
Placement heads must use:
This is why some factories quote differently — or even refuse — ultra-small packages unless volumes justify the tuning effort.
Machines do not think in Gerbers. They think in coordinate systems.
Tooling holes and fiducials allow:
Without proper references, even the best placement system is guessing. That guess turns into placement offsets, rotated parts, and cumulative errors — especially on dense boards.
When a manufacturer insists on these features, they are protecting yield, not adding inconvenience.
A good EMS partner does not just own pick-and-place machines — they understand their limitations.
When a supplier:
They are not being difficult. They are aligning design intent with manufacturing reality.
At Fumax, we work closely with hardware engineers during DFM and early production, especially for low-to-mid volume projects where engineering decisions matter more than raw speed.
If you’re preparing for your next PCBA or box-build project and want feedback grounded in real SMT constraints, we’re always open for a technical discussion or quotation.
