PCB Prototyping & Small Batch

How a Dongguan Songshan Lake Prototype House Scaled from 1 to 7 Compact Pick and Place Machines in 5 Years

Yes, a prototype house can scale from manual assembly to a multi-machine SMT operation — and the path is more predictable than most people think. This Dongguan Songshan Lake PCB prototyping company started with 1 compact pick and place machine in late 2021, replacing 3 manual assembly workers. By mid-2026, they operate 7 machines: 3–4 dedicated to rapid-turn prototyping and 3–4 forming a dedicated small-batch production line — all within a 120m² facility.

The scaling didn't happen overnight. It followed a clear pattern that any prototype house can replicate: replace manual bottlenecks first → add machines as BOM variety grows → separate prototyping from production when volume crosses a threshold. This article breaks down exactly how they did it, including their machine choices, feeder management strategy, floor plan evolution, and the ROI at each stage.

Prototype House SMT Scaling Formula:
Machines Needed = (Weekly BOMs × Avg Setup Hours per BOM) ÷ Available Machine Hours per Week + 1 Buffer Machine

Example: 25 BOMs/week × 0.5 hrs setup each = 12.5 hrs. With 40 available hours/week: 12.5 ÷ 40 = 0.31 → 1 machine handles the load. But at 80 BOMs/week: 80 × 0.5 ÷ 40 = 1.0 → you need 2 machines minimum + 1 buffer = 3 machines.

5-Year Scaling Timeline: 1 → 7 Machines

This customer is a dedicated PCB prototyping service provider based in Dongguan Songshan Lake — one of China's most concentrated electronics R&D hubs. Their business is pure high-mix, ultra-low-volume: most jobs are 5–50 boards, with BOMs ranging from 15 to 120 unique line items. Before 2021, everything was hand-assembled.

Phase Period Machines Weekly BOMs Floor Space Key Event
Phase 0: Manual Before 2021 0 ~10–15 30 m² 3 manual assemblers. 2–5% placement error rate. Frequent rework.
Phase 1: Trial Late 2021 1 × HW-T4-50F ~15–20 30 m² First machine after seeing competitor results. Replaced 2 manual workers.
Phase 2: Validation 2022–2023 3 × HW-T4-50F ~30–45 50 m² Added 2 more machines as order volume grew. All machines shared prototyping workload.
Phase 3: Separation 2024 4–5 machines ~50–65 80 m² Moved to larger facility. Split into 2 prototyping + 2 production machines.
Phase 4: Scale 2025–2026 7 machines ~70–90 120 m² Added HW-T6-64 for complex BOM jobs. 3–4 prototyping + 3–4 production line.

Key Insight: They never replaced machines — they added them. Starting with a compact desktop machine that handles 80% of prototype BOMs, then adding more of the same model, proved far more flexible than upgrading to a single large machine. This kept changeover simple, operator training consistent, and capacity scalable in small increments.

Before vs. After: Manual Assembly vs. Compact SMT

The customer's core pain point before automation was clear: manual assembly could not keep up with BOM variety and accuracy demands. Here is the direct comparison:

Metric Manual Assembly (Pre-2021) 1 Machine (Late 2021) 7 Machines (2026)
Placement speed ~200–400 CPH (per person) ~5,000–6,500 CPH ~35,000–45,000 CPH (combined)
Placement error rate 2–5% (wrong component, polarity, orientation) <0.1% <0.1%
Workers per shift 3 manual assemblers 1 operator 2–3 operators (for 7 machines)
Boards per day (avg) ~8–15 (depending on complexity) ~20–40 ~120–200
Component types supported Limited to hand-solderable (≥0603, SOIC) 0201–large, QFP, QFN, BGA 0201–large, QFP, QFN, BGA, fine-pitch
Rework cost per board ¥50–200 (material + labor) Near zero Near zero
Customer satisfaction Complaints about errors and delays Significant improvement Key competitive advantage

The customer reported that the single biggest improvement was not speed — it was eliminating component mix-up errors. In manual assembly, a tired operator at 4 PM might place a 10kΩ resistor where a 100kΩ belongs. That one error meant rework, delayed delivery, and sometimes a lost customer. With the pick and place machine's vision system verifying every placement, that problem disappeared.

The Prototype House Machine Count Formula

How many compact pick and place machines does your prototype house actually need? Here is the calculation method derived from this case:

Step-by-Step Machine Count Formula:

  1. Weekly BOMs = number of unique PCB designs processed per week
  2. Avg Setup Time per BOM = feeder loading + program loading + first-article inspection (typically 0.3–0.7 hrs for experienced operators with pre-organized feeders)
  3. Avg Run Time per BOM = (boards per job × components per board) ÷ machine CPH (typically 0.2–1.5 hrs for prototype volumes)
  4. Total Machine Hours Needed = Weekly BOMs × (Setup Time + Run Time)
  5. Available Machine Hours = operating hours per week × machine utilization rate (typically 40–60 hrs × 75–85%)
  6. Base Machine Count = Total Machine Hours ÷ Available Machine Hours (round up)
  7. Final Count = Base Machine Count + 1 Buffer Machine (for peak periods, maintenance, and growth)

Real Calculation: The Dongguan Customer at Each Stage

Stage Weekly BOMs Avg Setup (hrs) Avg Run (hrs) Total Hrs Needed Formula Says Actual Machines
Phase 1 (2021) 18 0.6 0.5 18 × 1.1 = 19.8 19.8 ÷ 35 ≈ 1 1
Phase 2 (2023) 40 0.5 0.4 40 × 0.9 = 36 36 ÷ 35 ≈ 2 + buffer = 3 3
Phase 3 (2024) 58 0.4 0.4 58 × 0.8 = 46.4 46.4 ÷ 35 ≈ 2 + 2 prod = 4–5 4–5
Phase 4 (2026) 80 0.35 0.35 80 × 0.7 = 56 56 ÷ 35 ≈ 2 + 2 prod + buffer = 5–7 7

The formula tracks reality closely. The buffer machine becomes increasingly important as volume grows — at 7 machines, one unit can be down for maintenance without disrupting delivery schedules.

Prototype House SMT Readiness Checklist

Before buying your first machine, answer these 10 questions. If you answer "yes" to 7 or more, you are ready:

  1. Do you process 10+ unique BOMs per week?
  2. Do your BOMs typically have 20+ unique component line items?
  3. Is your current manual placement error rate above 1%?
  4. Do you have at least 15–20 m² of available floor space?
  5. Can you dedicate one operator to machine operation (even part-time)?
  6. Do your customers request component packages smaller than 0603 (e.g., 0402, 0201)?
  7. Do you have BOMs with QFP, QFN, or BGA packages that are difficult to hand-place?
  8. Are you turning away orders because of capacity constraints?
  9. Do you have compressed air (0.5–0.6 MPa) and stable 220V power available?
  10. Are at least 30% of your components available on tape/reel (not loose)?

Key Parameters for Prototyping SMT Equipment

Not all pick and place machines are suited for prototype work. Here are the parameters that matter most for a high-mix, low-volume environment — and why the Dongguan customer chose what they chose:

Parameter Why It Matters for Prototyping Recommended Range Dongguan Customer's Choice
Feeder count Determines whether a BOM fits in a single setup. More feeders = fewer split runs. 44–64 positions 50F (50 feeders) for most; T6-64 (64) for complex BOMs
Placement speed (CPH) Less critical for prototyping than for production. Changeover speed matters more. 5,000–15,000 CPH ~6,500 CPH (T4) / ~13,000 CPH (T6)
Min component size Must match the smallest package in your customer BOMs. 0201 imperial (0.6×0.3mm) 0201 supported on all machines
Max PCB size Must accommodate the largest board customers send you. ≥300×400mm for general prototyping 400×600mm (T4) / 450×1200mm (T6)
Vision system Flying vision + bottom camera for fine-pitch IC alignment. Multi-camera with BGA support Flying vision + bottom camera standard
Feeder type Quick-change feeders reduce changeover time dramatically. Pneumatic or electric quick-change Pneumatic quick-change feeders
Tray capacity For QFP/QFN/BGA ICs that don't come on tape. 2–4 internal tray positions Internal tray support on T6-64
Programming method Fast job setup from CAD/BOM data. Essential for frequent changeovers. CAD import + vision teach CAD/BOM import, auto-feeder assignment
Machine footprint Critical when scaling multiple machines in limited space. ≤1.5m × 1.5m per machine ~1.2m × 1.2m (T4)
Component height Prototype boards often have tall connectors or electrolytic capacitors. ≥15mm max height 15mm (T4) / 20mm (T6)

Critical Parameter for Prototype Houses: Feeder count matters more than placement speed. A 6,500 CPH machine with 50 feeders that runs a full BOM in one setup will outperform a 20,000 CPH machine with 30 feeders that needs 2–3 passes per board. The Dongguan customer confirmed: they never had a speed problem — they had a changeover problem until they matched feeder count to BOM complexity.

Feeder Management for Frequent BOM Changes

This is the operational secret that made 7-machine scaling possible. In a prototype house, you might change BOMs 5–15 times per day per machine. Without a disciplined feeder strategy, changeover time eats your capacity.

The Fixed + Flexible Feeder Model

The Dongguan customer organized feeders into two categories:

Category Count (per machine) What Goes Here Change Frequency
Fixed (Common) 15–20 positions 100nF, 10kΩ, 1kΩ, 1μF, 10μF, 0Ω, common transistors (MMBT3904/3906), popular voltage regulators Rarely — only when a reel runs empty
Flexible (Job-Specific) 25–35 positions Everything unique to the current BOM: specific ICs, non-standard resistor values, connectors, specialty components Every job change

Feeder Storage and Pre-Loading System

The customer built a simple but effective system:

  • Labeled feeder racks organized by component value (not by job), stored within arm's reach of each machine
  • Pre-loading while running: operator loads the next job's unique feeders onto spare feeder bases while the current job is placing
  • Changeover sequence: unload previous job's unique feeders → load pre-prepared next-job feeders → verify with first-article inspection → start (target: under 15 minutes)
  • Shared feeder pool: all 7 machines use the same feeder type, so feeders can move between machines without compatibility issues

Result: Average changeover time dropped from 45 minutes (Phase 1, unorganized) to under 15 minutes (Phase 4, with pre-loading and labeled storage). At 80 BOMs/week, saving 30 minutes per changeover recovers 40 hours of machine time — the equivalent of one full additional machine.

Floor Plan Evolution: 30m² → 120m²

The physical layout evolved with each phase. Here is the progression:

Phase Floor Space Layout Key Constraint
Phase 1 (2021) 30 m² 1 machine + 1 stencil printer + 1 small reflow oven in a line. Feeder storage on shelves above the bench. No room for expansion. Any second machine would require moving.
Phase 2 (2023) 50 m² 3 machines in a U-shape around a central operator station. One printer shared across machines. One reflow oven at the end. Shared printer became a bottleneck during peak hours.
Phase 3 (2024) 80 m² Two distinct zones: Prototyping Zone (2–3 machines, fast-turn) and Production Zone (2 machines, multi-day batches). Separate printer for each zone. Feeder storage between zones needed clear labeling to avoid mix-ups.
Phase 4 (2026) 120 m² Prototyping Zone: 3–4 machines with dedicated printer and small reflow oven. Production Zone: 3–4 machines forming a compact SMT line with printer + pick-and-place + reflow in sequence. Central feeder storage with labeled racks. AOI inspection station at the end of the production line. Operator walking distance — optimized with U-shaped machine arrangement.

The customer noted that ceiling height and compressed air routing were the two infrastructure items they underestimated when moving to the 80m² and 120m² facilities. Plan these before signing a lease.

When to Separate Prototyping from Production

This is the most important strategic decision a prototype house faces. The Dongguan customer's rule of thumb:

The Separation Rule:

If a production order is ever delayed because a machine is tied up on a prototype, it is time to split. This typically happens at 4–5 machines or 50+ weekly BOMs.

Signs You Need Separate Lines:

  • Delivery date conflicts: production batches miss deadlines because prototype jobs take longer than expected
  • Operator context-switching fatigue: the same operator handles both 5-board prototype runs and 500-board production batches in the same shift
  • Feeder chaos: production feeders get unloaded to make room for prototype BOMs, then need to be reloaded — doubling setup time
  • Customer complaints: prototype customers expect 24–48 hour turnaround; production customers expect on-time delivery. When both groups are unhappy, you have a structural problem, not a scheduling problem

How the Dongguan Customer Split:

Prototyping Line Small-Batch Production Line
3–4 machines (HW-T4-50F) 3–4 machines (HW-T4-50F + HW-T6-64)
Dedicated stencil printer (ASE Automatic) Dedicated stencil printer (XSE High-Precision)
Small reflow oven (HW-R306) Larger reflow oven (HW-R408)
1 operator manages 3–4 machines 1 operator + 1 assistant for 3–4 machines
5–25 boards per job 50–500 boards per job
15–30 BOM changes per day 2–5 BOM changes per day
Turnaround target: same day or next day Turnaround target: 3–7 days per batch

Configuration Recommendations by Stage

Based on this case and similar prototype houses we have worked with, here are the recommended configurations at each growth stage:

Entry Level — First Machine (10–25 BOMs/week)

Best for: Prototype house currently doing manual assembly, processing 10–25 unique BOMs per week, typical BOMs of 15–45 unique line items

Feeder need: 25–50 positions (mostly 8mm tape)

Recommended machine: HW-T4-44F/50F Desktop Placer — 44–50 feeder positions, ~6,500 CPH, 0201–large components, 400×600mm PCB

Pair with: ASE Automatic Stencil Printer + HW-R306 Desktop Reflow Oven

Floor space: ~15–20 m² for the full line

What the Dongguan customer says: "Start with the 50F, not the 44F. The extra 6 feeder positions mean fewer split runs on complex BOMs. The price difference is small; the time savings are not."

Standard — Scaling Up (25–50 BOMs/week)

Best for: Growing prototype house adding small-batch production, 25–50 BOMs/week, starting to get repeat production orders

Machine count: 2–3 × HW-T4-50F (same model for feeder interchangeability)

Layout: U-shape with shared operator station. Consider adding a second stencil printer if printer wait time becomes a bottleneck.

Floor space: ~40–60 m²

Add: HW-R408 Compact Reflow Oven if batch sizes exceed the R306's throughput

Critical upgrade: Labeled feeder storage racks. At 3 machines and 30+ BOMs/week, feeder organization becomes the difference between 15-minute and 45-minute changeovers.

Advanced — Separate Lines (50–90+ BOMs/week)

Best for: Established prototype house with dedicated production customers, 50–90+ BOMs/week, clear separation between sampling and batch production

Prototyping zone: 3–4 × HW-T4-50F for rapid-turn jobs. One operator manages all machines.

Production zone: 2–3 × HW-T4-50F + 1 × HW-T6-64 (64 feeders for complex BOMs, ~13,000 CPH). Form a proper compact SMT line with printer → pick-and-place → reflow → inspection.

Pair with: XSE High-Precision Stencil Printer + HW-R612E Reflow Oven for the production line

Floor space: ~80–120 m²

Add: AOI inspection station at the end of the production line for quality control on batch orders

See also: Small Batch SMT Line Solutions for full line configuration details

ROI Breakdown at Each Stage

The customer's rough ROI calculation (based on their actual numbers):

Stage Investment Monthly Labor Savings Monthly Rework Savings Monthly Revenue Increase Payback Period
Phase 1 (1 machine) ~$8,000–12,000 ~$1,500 (2 fewer workers) ~$300–500 ~$1,000–2,000 (new orders) ~5–8 months
Phase 2 (+2 machines) ~$16,000–24,000 ~$1,500 (1 more worker reduced) ~$200–400 ~$2,000–4,000 ~6–10 months
Phase 3 (+2 machines + facility) ~$25,000–40,000 ~$1,000 ~$300–500 ~$4,000–8,000 ~6–9 months
Phase 4 (+2 machines) ~$20,000–35,000 ~$500 ~$200–300 ~$3,000–6,000 ~5–8 months

Note: The revenue increase from new orders consistently outpaced labor savings. The customer reported that after Phase 2, their reputation for fast, error-free prototyping became their primary source of new business — customers who had been burned by manual assembly errors at other prototype houses switched to them and stayed.

Conclusion

The Dongguan Songshan Lake prototype house's 5-year journey from manual assembly to 7 compact pick and place machines is not an outlier — it is a replicable pattern. The key decisions that made it work:

  1. Start with one machine that fits 80% of your BOMs — the HW-T4-50F's 50 feeder positions covered nearly all prototype BOMs in a single setup
  2. Add machines, don't replace them — same model, same feeders, same operator training = lowest friction scaling
  3. Invest in feeder organization early — labeled storage racks and pre-loading cut changeover time by 67%
  4. Separate prototyping from production at the right time — not too early (wastes capacity), not too late (angers customers)
  5. Use the buffer machine rule — always have one more machine than your formula says, for peak periods and maintenance

If you run a PCB prototyping service and are considering your first or next compact pick and place machine, the most important step is mapping your actual BOMs to feeder counts. Once you have that number, the machine choice becomes obvious.

Planning to Scale Your Prototype SMT Capacity?

Send us your typical BOM, weekly volume, and PCB sizes. Our team will map your requirements to a specific machine configuration and line layout — based on the same methodology used by prototype houses across the Pearl River Delta.

Talk to Our SMT Planning Team

Frequently Asked Questions

Q1: How many compact pick and place machines does a prototype house need?

It depends on weekly BOM volume. A prototype house handling 10–30 unique BOMs per week typically starts with 1–2 machines for sampling only, scales to 3–5 when adding small-batch production, and reaches 6–8 when separating sampling and production lines. The formula: Machines Needed = (Weekly BOMs × Average Setup Hours per BOM) ÷ Available Machine Hours per Week + 1 Buffer Machine. A Dongguan Songshan Lake customer started with 1 machine in 2021 and scaled to 7 by 2026 — 3–4 dedicated to prototyping and 3–4 running a dedicated small-batch production line.

Q2: What is the ROI of switching from manual assembly to compact pick and place for prototyping?

For a prototype house doing 20+ boards per week with 30+ components each, a compact pick and place machine typically pays back in 6–12 months. Savings come from three areas: labor reduction (1 operator replaces 2–3 manual assemblers), error elimination (near-zero placement errors vs. 2–5% manual error rate), and throughput gain (5,000–8,000 CPH machine vs. 200–400 CPH manual). A Dongguan customer reported that one HW-T4-50F replaced 3 manual workers while eliminating component-mix-up rework costs entirely.

Q3: How do you organize feeders for frequent BOM changes in a prototype house?

Keep 15–20 common components (100nF, 10kΩ, 1μF, common ICs) on fixed feeder positions. For job-specific components, use labeled feeder storage racks organized by component value, not by job. Pre-load the next job's unique feeders while the current job runs. This approach reduced changeover time from 45 minutes to under 15 minutes per job at the Dongguan case study site. A 44–50 feeder machine like the HW-T4-44F/50F handles typical prototype BOMs (20–40 unique lines); 64+ feeders (e.g., HW-T6-64) are needed if many jobs share few common parts.

Q4: Should a prototype house separate sampling and small-batch production lines?

Yes, once weekly volume exceeds 15–20 unique BOMs. Mixing sampling and production on the same machines causes constant changeovers, delivery delays on production orders, and operator fatigue. The Dongguan customer separated lines at 4 machines: 2 dedicated to rapid-turn prototyping, 2 forming a small-batch production line. At 7 machines, they run 3–4 for prototyping and 3–4 for production. The rule: if your production orders are ever delayed because a machine is tied up on a prototype, it is time to split.

Q5: What PCB sizes and component types can a compact pick and place machine handle for prototyping?

Most compact pick and place machines support PCB sizes from 50×50mm up to 400×600mm or larger. For prototyping, the key spec is minimum component size — the HW-T4-44F/50F handles 0201 imperial (0.6×0.3mm) and fine-pitch ICs down to 0.3mm pitch. Component types supported: chip resistors/capacitors (0201–1206+), SOT/SOD transistors, SOP/SSOP/TSSOP ICs, QFP (0.4mm+ pitch), QFN, BGA (0.5mm+ pitch), LEDs, inductors, and connectors. Check the machine's vision system specs for the smallest supported package in your BOMs.

Q6: How much floor space does a prototyping SMT line need?

A minimal prototyping SMT line (stencil printer + 1 pick and place + small reflow oven) fits in 15–20 m². A 7-machine setup like the Dongguan case study occupies roughly 60–80 m² for equipment plus feeder storage, inspection stations, and material staging. The HW-T4-44F/50F has a footprint of about 1.2m × 1.2m. The Dongguan customer expanded from a 30m² workshop in 2021 to a 120m² facility by 2026. Plan for at least 3–4 m² per machine including operator access space and feeder storage.

Q7: What is the most common mistake prototype houses make when buying their first pick and place machine?

Underestimating feeder count. Prototype houses handle diverse BOMs with 25–55 unique line items. A machine with only 30–35 feeder positions forces constant feeder swaps. The second mistake: buying based on speed (CPH) alone — for prototyping, changeover speed and feeder flexibility matter more than raw placement speed. The Dongguan customer started with a 50-feeder machine (HW-T4-50F) which handled most prototype BOMs in a single setup, then added more machines as volume grew rather than upgrading to a single faster machine.

Q8: Can one pick and place machine handle both prototyping and small-batch production?

Yes, one machine can handle both if your total weekly volume is under 10–15 unique BOMs. Reserve morning hours for prototyping (fast-turn jobs) and afternoon/evening for production runs. Keep common components on fixed feeders and swap only job-specific ones. Once prototyping demand grows beyond 15 BOMs per week, add a second machine. The Dongguan customer crossed this threshold at around 20 BOMs/week and added their second machine within 8 months of the first. See our Small Batch SMT Line Solutions page for mixed-use configuration options.

Related Resources

Previous: LED Display Manufacturer Shenzhen Case Study Next: Compact SMT Line — Prototyping vs. Production