The THT Automation Playbook: How EMS Factories Remove 10 Manual Insertion Stations Without Breaking the Line
Diagnosis
Here is what we see when we walk a THT line for the first time. Boards arrive from the SMT line. A queue of operators each inserts two or three part numbers, passes the board on, and repeats. Nobody can tell you the labour minutes per board, because the work is measured in headcount, not in seconds. Yield is measured at final test, not at insertion, so reversed polarities and bent leads surface three stations downstream where they are expensive to fix. And when volume doubles, the answer is always the same: hire ten more people, if you can find them. The real problem is not that your people are slow. It is that you have never done the component triage that tells you which 80% of your through-hole work is already machine-ready.
Step 1 — Run a staffing audit, not a headcount estimate
Take your top five board part numbers. Count every through-hole component, classify it by lead geometry (radial, axial, odd-form, PIN/terminal), and record manual insertion seconds per component. Multiply by daily volume. This gives you labour minutes per board, the only number that survives a business case. Rule of thumb: a manual radial insertion takes 2.0-2.5 seconds including reach and orient. A machine does it in 0.2. That 10x gap is your entire ROI.
Step 2 — Triage the BOM into four buckets
Bucket A: radial taped (2.5/5.0/7.5/10 mm pitch) — machine-ready today. Bucket B: axial taped and jumper wire — machine-ready with a second machine. Bucket C: odd-form large parts (transformers, connectors, heat sinks, tact switches, fuse holders) — machine-ready with custom grippers. Bucket D: genuinely manual (cable assemblies, hand-soldered wire, mechanical sub-assembly). Most factories discover Bucket D is under 8% of insertion seconds, not the 40% they feared.
Step 3 — Sequence the line, not the machine
Auto-insertion only pays when boards flow without an operator. The correct order is: magazine loader to radial inserter to axial inserter to odd-form inserter to inline lead cutter to wave soldering to unloader. Buy the machines in the sequence that removes the largest labour bucket first, and install the loader and unloader from day one — they are 6% of the capital and they are the difference between a cell and a line.
Step 4 — Solve feeding before you sign the PO
Feeding is where automation projects die. Every component that does not arrive in tape, tube, tray or reel needs a feeding decision before the machine ships. Send 20 samples of each hard part and require a video feeding trial as an acceptance condition. If your supplier cannot show you the parts running, they are selling you a machine, not a solution.
Step 5 — Design the board for the machine you will own
Auto-insertion needs three things from your PCB: consistent lead hole tolerances, a clean insertion side free of tall SMD parts, and fiducial marks for vision correction. Adding a rule to your DFM checklist costs nothing. Retrofitting board designs after the machine arrives costs quarters.
Step 6 — Validate with a golden-board acceptance test
Acceptance is not ‘the machine turns on’. Acceptance is: run 500 of your actual boards, at your target rate, for one full shift, and measure insertion defects per million. Write that number into the contract. Anything less is a demo.
Our vehicle
This is exactly the sequence we run with customers, and it is built into the equipment we supply. The S3000 radial inserter covers Bucket A at up to 18,000 CPH. The S4000 axial inserter covers Bucket B, including jumper wires formed on the fly. The S7900 odd-form machine and our in-house gripper service cover Bucket C. SLD250 loaders and SUL250 unloaders make the cell unattended. And the auto-insertion readiness checklist turns Steps 1 and 2 into a spreadsheet your team can fill in this afternoon.
The compounding loop
Automation is not a one-time purchase, it is a compounding loop. Each board family you convert gives you real insertion-seconds data that makes the next business case easier to approve. Each new component you automate grows your gripper and feeder library, so the next project starts from a library instead of from zero. Customers who run this loop for 24 months typically find their marginal automation cost per new board family drops by half, because the hard engineering is already paid for. We publish every new feeder design, gripper trial and commissioning lesson on our YouTube channel so that library keeps growing for everyone.
Send us one board drawing and the BOM for your highest-volume THT assembly. We will run the staffing audit, triage your components into the four buckets, and send back a machine configuration with a payback calculation — free, and without a sales call attached to it.
How many operators does one radial insertion machine replace?
On a radial-heavy board, one S3000 with loader and unloader typically replaces 6 to 10 manual insertion operators per shift. The precise figure depends on radial components per board and daily volume, which is why we start every project with a staffing audit rather than a quote.
What is a realistic payback period for THT automation?
Most EMS customers in Southeast Asia, India and Latin America see 9 to 18 months payback on labour cost alone. Factories running two or three shifts, or those in markets with high labour turnover, frequently see payback inside 12 months.
Can we automate if our boards are high-mix low-volume?
Yes, but the configuration changes. For HMLV you prioritise fast changeover: quick-release feeder cassettes, recipe-based program recall, and offline programming. Radial and axial machines with recipe storage change over in under 15 minutes, which makes automation viable from roughly 300 boards per day.
What happens to the operators we displace?
In practice most factories do not reduce headcount, they reassign it. Automated insertion shifts people into inspection, testing, packing and second-shift coverage — work that was previously not getting done because everyone was on the insertion line.
