2.1 Machine Architectures

Chipshooter vs flexible, gantry vs turret, and when to split/load-level—so the line hits TAKT without drama.

Pick-and-place machines come in two main personalities: chipshooters, built for raw speed on tiny passives, and flexible mounters, designed to handle everything from BGAs to odd-forms with high accuracy. Many high-volume lines pair them—chips first, complex parts second—while high-mix shops often run two flexibles in tandem or parallel to cut changeovers. Motion style also matters: gantry heads excel at accuracy and part variety, while turrets dominate chip speed but have part-height limits. How you arrange them—single stream, tandem split, role-split, or dual-lane—shapes throughput, with balancing based on placement times keeping takt time steady. Permanent passive banks, consistent vision teaching, and small in-line buffers protect uptime, while avoiding unnecessary flips or feeder swaps keeps the flow calm. When the machine mix matches the board mix and the workload is leveled, the line stops firefighting and simply runs to plan.

2.1.1 Two kinds of pick-and-place (and why you’d mix them)

Type	What it’s built to do	Strengths	Trade-offs	Best use
Chipshooter	Blazing through small passives (01005–0603), some small ICs	Very high parts/hour; tiny-part vision; low cost per placement	Narrower part range; changeovers hurt if you swap feeders often	High-volume boards with lots of repeat passives
Flexible mounter	“Anything you throw at it” (BGAs/QFNs/QFPs, odd-form, tall parts)	Broad component range; accurate; easier NPI	Lower headline speed on tiny chips	Mixed-tech boards; NPI and variant builds

House rule: if volume justifies it, run a chipshooter → flexible pair. If you’re high-mix/low-volume, two flexibles in parallel beat one of each—fewer changeovers and simpler kitting. (We’ll tune programs in 8.2 and feeders/splicing in 8.3.)

2.1.2 Heads & motion: gantry vs turret

Architecture	How it moves	Why lines love it	Where it bites
Gantry (moving head on X-Y; multiple nozzles)	Head sweeps over a stationary board; cameras under/over	Flexible parts range, great accuracy on BGAs/QFNs; modular (add heads/gantries)	Peak CPH lower than a pure turret; throughput depends on path planning
Turret (rotary wheel with many nozzles)	Wheel spins parts past vision → drops at speed	Insane chip speed; shines on dense passive fields	Limited tall/odd-form; setup sensitivity; not the tool for large/heavy parts

Modern lines blur the line (hybrid heads, dual-gantry modules), but the mindset still helps you design the flow: feed tiny parts where rate matters, feed complex parts where range and accuracy matter.

2.1.3 Line topologies (and when each wins)

Topology	Sketch	When to pick it	Notes
Single stream: Printer → PnP1 → Reflow	One mounter	Small boards, light BOMs, pilot/NPI	Simple, cheap; limited peak rate
Tandem split: Printer → PnP A + PnP B → Reflow	Two mounters in series	Most common volume setup	Balance parts so cycle times match (±10%)
Role split: Chipshooter → Flexible	Fast chips first	Boards dominated by chips + a few ICs	Keep high-hit chip feeders permanent
Parallel lanes / dual-lane	Two lanes through same machine	Short boards, tight TAKT	Great for pack-and-stack, but needs disciplined feeder mirroring
Top/Bottom mirror	Two lines per side	Dense double-sided builds	Sync changeovers; share feeder banks where possible

Conveyors/buffers and handshakes live in 8.4; keep them smooth so mounters never starve.

2.1.4 When to split or load-level (the 10-minute math)

Measure each machine’s placement time per board from logs (exclude board travel).
Find the bottleneck—the slowest station sets line TAKT.
Move part families (usually passives) from the slow machine to the fast one until their times are within ±10%.
If you can’t balance by parts, consider duplicate programs in parallel (two similar mounters doing half the placements each).
Re-run a First Article after any big split to confirm offsets/rotations stayed sane (that’s 8.5).

Think like line balancing: shift work off the constraint and protect it with feeder discipline. (Part VI digs deeper into bottlenecks and sustained throughput.)

2.1.5 Practical rules that keep uptime high

Permanent banks: park high-runner passives on fixed feeder banks; variants “hot-swap” the odd parts. (Feeder care/splicing in 8.3.)
Teach for success: lock vision teaching & rotation sanity during program creation so either machine can place the part without edits. (8.2)
Don’t ping-pong boards: minimize unnecessary flips/returns; if you must split by side, keep symmetrical feeder layouts so changeovers feel the same on both lines.
Buffers beat stars: small in-line buffers before the bottleneck absorb micro-stops and keep TAKT steady. (8.4)

2.1.6 Quick chooser (print this)

Mostly chips, few ICs → Chipshooter + Flexible in series.
High-mix, many changeovers → Two Flexibles (parallel/tandem), permanent passive banks.
Short boards, aggressive TAKT → Dual-lane with mirrored feeders.
Fragile BGAs, many QFNs → Bias toward gantry heads for accuracy; keep chip load reasonable there.

Bottom line: pick the machine mix that fits your part family (rate vs range), then balance work so no station is the hero. Use permanent feeder banks, clean programs, and simple buffers to keep the constraint fed. Do that, and the line hums—quietly, predictably, and on time.