-
Who this checklist is for
-
Step 1: Lock your production profile before you talk to a vendor
-
Step 2: Draw your floor layout with real clearances, not catalog clearances
-
Step 3: Choose equipment against your 3-year board mix, not today's
-
Step 4: Ask for line-installed references, not catalog references
-
Step 5: Define what "setup" includes — in writing
-
Step 6: Treat material flow like a kitchen, not a warehouse
-
Step 7: Insist on a real acceptance run before you sign off
-
Common mistakes (from my own files)
Who this checklist is for
If you're a plant manager, sourcing lead, or process engineer about to buy your first or second SMT production line — whether you're starting from scratch in Vietnam or Mexico, or converting a through-hole shop into something that can actually run surface-mount — this is for you. My experience is based on roughly 40 line builds in the 6-to-30 placement-head range. If you're spec'ing a 100+ head inline with full traceability requirements, some of these steps will need more depth than I can honestly give you.
Back in 2018, I was the guy who ranked line quotes by sticker price and picked the cheapest one. Six weeks after install, we found out the supplier's 8mm feeder spec was only "compatible" if you also bought their proprietary feeder baseplates — which added six weeks to the timeline we couldn't afford. We'd promised a customer a delivery date before the line was even running. Total hit: about $18,000 in wasted budget and a renegotiated delivery penalty. That's when I started keeping this list.
Seven steps. Order matters — skip step 3 and jump to step 5, and you'll spend $20K+ on equipment you don't actually need.
Step 1: Lock your production profile before you talk to a vendor
Before you email a single supplier, spend one afternoon and write down these numbers. Just these:
- Your most complex board over the next 12 months — smallest component package, layer count, board dimensions, single-sided or double-sided
- Your batch sizes — 50 boards a run? 500? 5,000?
- Your real changeover frequency (once per shift or twice a week — be honest)
- Maximum board weight and warp — this hurts conveyors and vacuum nozzles more than any spec sheet admits
This is the step most people skip. They dive into quote requests, get numbers from three vendors, and start comparing things that aren't comparable. A line optimized for 5,000-piece singles runs looks nothing like one optimized for high-mix, low-volume — but both quotes arrive looking like "an SMT line" and cost roughly the same.
Checkpoint: If you can't answer all four questions on one page, you're letting the vendor define your project for you (they'll define it around whatever they have in stock).
Step 2: Draw your floor layout with real clearances, not catalog clearances
Catalog clearance for a placement machine is 800mm between units. Real clearance — when you're changing reels, when an operator needs to squeeze past, when a maintenance cart needs to roll through — is 1,200mm minimum.
Here's what to map out on your factory drawing:
- Mark the feeder/reel side separately. That's the side with the largest footprint — it's where the operator actually stands. Not the head side.
- Draw the reflow oven's in-feed and out-feed buffer zones. That oven needs clearance for heat dissipation, and that heat radiates onto anything within a meter.
- Reserve a spot for inspection. Don't tuck AOI into a corner — the moment it becomes a bottleneck, the whole line stops.
- Mark air, power, and exhaust drop points. If your compressor is 40 meters away, the pressure at the nozzle may not be what you spec'd.
Here's the detail most people miss: column spacing and fire-escape corridors. I've watched an entire line layout get rejected by a fire marshal before the first board was made (mental note: always send your factory drawing to the vendor *before* they propose a layout).
Step 3: Choose equipment against your 3-year board mix, not today's
The board you're running today isn't the board you should be spec'ing against. The board you'll run in 2027 is.
Concretely:
- If you might ever drop from 0201 down to 01005, buy placement precision that can get there — adding accuracy later costs more than buying it up front
- If you expect bottom-terminated components or PoP assemblies, spec the flux-transfer or dip option now. Retrofits are machine replacements.
- Size feeder capacity for your maximum, then add 30%. Everyone under-buys feeders by 20%.
- Verify your reflow profile headroom against your future thickest board, not today's.
When I put two recent line builds side by side — same rated placement speed, same price bracket — the difference in throughput wasn't placement speed at all. It was feeder count and changeover software. The line with 30% more feeder slots ran 40% more effective capacity per shift. Buying for today is the most expensive mistake on an SMT line — not the most obvious, but the most expensive.
Step 4: Ask for line-installed references, not catalog references
A catalog reference is someone the vendor has talked to on the phone, who uses maybe half the equipment you're buying. A line-installed reference is a factory that actually bought this line and actually runs it — one you can visit, walk up to the line supervisor, and ask real questions.
What to request:
- Three live installations, ideally same country, ideally 6+ months of runtime
- Contact name and phone number (not "our sales rep will arrange it")
- Actual OEE or utilization numbers — ask the customer, not the vendor
Then go. Drive out there. Buy the line supervisor a coffee. Ask what their real downtime is. If a supplier can't hand over three visitable references, that's a red flag — either unhappy customers or so few installs they don't have a base yet.
For pricing context: a mid-tier new in-line setup (printer + 2 placement machines + reflow + AOI) sits somewhere in the $180,000–$450,000 range depending on placement speed, feeder count, and software licensing level. Used equipment — relatively recent, working condition — lands around 40–60% of that. Those figures come from publicly listed distributor catalogs and used-market listings I reviewed in early 2025; verify current pricing before you commit.
Step 5: Define what "setup" includes — in writing
"Setup" is the most loaded word in SMT quotations. Make the supplier spell it out:
- Uncrating and mechanical positioning on arrival
- Electrical tie-in (often excluded — you'll pay your own electrician separately)
- Programming with your actual boards, not a demo PCB
- Operator training — how many hours, on-machine or classroom? Be specific.
- First article report and process window verification
- Yield or scrap commitment (get a number, not "industry standard")
If the contract says "installation included" and then has three lines of text, the cost is being shifted back to you. I've watched a quote pad on $8,000 for a line-item labeled "setup support" that turned out to be two technicians for four days, no training included (training was another $4,000). Get the scope in writing — before you sign.
Step 6: Treat material flow like a kitchen, not a warehouse
This is where most SMT production line layouts actually fail — not on machine selection, on material movement.
Draw the physical path your material takes from receiving dock to finished output. Then ask three questions:
- How far does a reel travel from stores to the line-side buffer? If it's over 20 meters, you're losing time on every changeover.
- Where does feeder prep and replenishment happen? If feeders are prepped at the machine, you've just added 15–25 minutes to every changeover.
- Does the line-side buffer hold the reel/coil sizes you actually use (many don't, and that's how cardboard boxes start appearing on the floor)?
Every good SMT shop I've walked through has a dedicated feeder prep bench within 5 meters of the line. Every bad one has feeders being prepped at the tail of the machine.
Step 7: Insist on a real acceptance run before you sign off
Acceptance runs frequently turn into demos. A demo is when the vendor's own engineer runs 50 boards with your best program and everyone claps. That's not acceptance.
Real acceptance looks like this:
- 8-hour continuous run with your actual program, including at least two changeovers
- Placement yield at or above the contracted threshold — use IPC-A-610 Class 2 as a baseline unless you're doing automotive, in which case Class 3 applies
- Operators (not your engineers — your real operators) running the line under the trainer's watch
- Signed first-article report and reflow profile records on the acceptance date
If the supplier gets hesitant about an 8-hour run, that itself tells you something.
Common mistakes (from my own files)
These are the ones that I've either made myself or watched others make, and most are avoidable with the checklist above in hand:
- Comparing quotes as if they're apples to apples when the equipment lists are completely different. Line up feeder count, head count, camera configuration, and software licenses side by side.
- Forgetting electrical tie-in, exhaust, and compressor sizing. Those three alone can run $15,000–$40,000 and are rarely in the main vendor quote.
- Paying in full before the first production board comes off. Stage payments against acceptance milestones.
- Leaving operator training out of the contract and out of the schedule. Training that happens "during ramp-up" is training that never happens.
- Comparing placement speed only. A 20% speed difference usually matters less than a 10% difference in changeover time or feeder capacity.
One more: if a supplier claims they do everything — sourcing, layout, setup, consumables, chemicals, after-sales — ask what they're genuinely best at. The best suppliers I've worked with will tell you what they *don't* do. The one I trust most told me on day one: "We don't do your factory electrical modifications — get a local electrician, contract them separately." That honesty is why I gave them everything else.
This checklist won't build the line for you. But it'll keep you from paying twice for the same lesson.