I've been buying capital equipment for a 180-person contract electronics manufacturer for the past six years. Our equipment budget runs about $1.2M annually, and I've negotiated with more than 30 vendors in that time. The checklist below is the one I wish I'd had in 2022, when our first line-sourcing project ran three weeks late and about $19,000 over because of things nobody flagged early enough.

This is for you if you're standing up a new SMT line, adding capacity, or moving from outsourced assembly to in-house. It works whether you're spending $300k on a used line or $2M on new equipment. It does not work if you're buying a single machine to bolt onto an existing line — that's a different, simpler problem.

Seven steps. Each one has a checkpoint you can actually verify. If you can't pass the checkpoint, don't move to the next step.

Step 1: Write down your board mix before you talk to anyone

Most sourcing projects start with 'we need an SMT line.' That sentence is useless to a vendor and dangerous for your budget. Before you request a single quote, document your top five boards: dimensions, layer count, component count per side, smallest component (0402? 01005?), double-sided or single-sided, annual volume per SKU, and how often you change over.

Here's the part that flipped my thinking. Everything I'd read about line selection led with throughput — chips per hour, boards per hour, that whole conversation. In practice, for a high-mix shop running 12 to 18 changeovers a day, changeover time dominates. In our 2023 audit, a 60k CPH machine with a two-hour changeover lost to a 35k CPH machine with a 15-minute changeover by roughly 40% in effective output. Same board mix, same shift pattern.

Checkpoint: you can fill in a one-page table for your top five boards without looking anything up. If you can't, you're not ready to talk to suppliers — you're ready to talk to your own engineering team.

Step 2: Build a total cost model before you compare a single quote

The machine price is maybe 60% of what you'll actually spend in year one. In our 2024 project, a line quoted at $86,000 per placement machine landed at $121,400 per machine once everything was counted. Nobody lied to us. The gap was all in line items that weren't on the quote because we hadn't asked.

What goes in the model:

  • Feeders — priced per unit, and the count is always higher than you think
  • Nitrogen generation, compressor capacity, and dry-air plumbing
  • Electrical: panel upgrades, voltage stabilization, UPS for the placement section
  • Floor reinforcement and vibration isolation
  • Installation, commissioning, and line balancing services
  • Operator and programmer training (days, headcount, on-site or factory)
  • Software licenses — perpetual or subscription, and the annual renewal number
  • Spare parts package for the first 12 months
  • Consumables: nozzles, belts, filters, solder paste

Checkpoint: feeder quantity and unit price appear as their own line in your spreadsheet. If a quote lists 'line configuration' as a lump sum, ask for the feeder breakdown in writing.

Step 3: Verify the building before you verify the machine

This is the step most teams skip, and it's the one that has cost me the most money. SMT equipment has brutal facility requirements: floor loading, three-phase power with clean grounding, compressed air at a specific dew point, exhaust for reflow, ambient temperature stability, and ESD-safe flooring. None of that is negotiable, and all of it has a lead time.

In Q2 2024 we were two weeks from signing on a reflow oven when our facilities manager mentioned the mezzanine floor was rated at 500 kg/m². The oven plus loader needed roughly double that. Reinforcement took three weeks and $14,000 we hadn't budgeted. We caught it. Barely.

Checkpoint: a marked-up floor plan showing your electrical panel location, remaining compressor capacity, column positions, and ceiling height at the reflow exhaust point — signed off by whoever runs your facility, not by you.

Step 4: Standardize the RFQ so the numbers mean the same thing

If you send five vendors five slightly different descriptions of your needs, you'll get five quotes you cannot compare. Build one specification sheet and send the identical document to everyone. Include: board dimensions, annual volume, required placement accuracy (and the Cpk you're holding them to), feeder counts by type, required changeover time, acceptance standard, training days, spare parts list, warranty terms, and response-time commitment.

One thing worth knowing when you read vendor throughput claims: the IPC-9850 standard defines how placement equipment performance is measured and reported. Vendors don't all test under identical conditions, so a 'CPH' number from one supplier is not directly comparable to another's. Ask what board, what component mix, and what percentage of the theoretical maximum they're quoting.

If a vendor can't tell you the test conditions behind their throughput number, treat the number as marketing, not engineering.

Checkpoint: at least three vendors return the same form. Blank fields come back with an explanation, not silence. Silence on a line item is usually where the surprise lives.

Step 5: Test the line in the demo room with your own board

Demo rooms are tuned. The boards are simple, the feeders are perfect, and the operator has run that exact job a hundred times. When I compared the demo-room cycle time against our installed line's first three weeks side by side, I finally understood why the gap was 18%. It wasn't the machine. It was feeder setup, program optimization, and operator familiarity — none of which shows up in a demo.

So bring your own board. Bring your awkward parts — the tall connector, the fine-pitch QFN, the part that always tombstone. Then watch three things:

  • Actual cycle time on your board, not the theoretical figure
  • First-pass yield on a 50-board run
  • Changeover time — ask them to demonstrate a full product change while you hold a stopwatch

Also ask for two reference customers with volume and mix close to yours. Not their biggest account. Yours. Ask those references about unplanned downtime and how fast spare parts actually arrived.

Checkpoint: you have a timed changeover on video and two reference calls completed. Not scheduled. Completed.

Step 6: Put acceptance criteria and spares into the contract

A purchase order that says 'one SMT line, per quotation' is not a contract. It's a hope. What you want in writing:

  • A factory acceptance test and a site acceptance test with pass/fail numbers — yield, cycle time, and placement accuracy measured on your board
  • Payment milestones tied to those tests, including a retention amount
  • The first 12 months of spare parts, itemized with prices, plus a stock commitment on critical components
  • Training scope: headcount, days, and whether programming is included
  • Software licensing terms and the renewal cost after year one
  • Response-time commitment for service calls, with a remedy if it's missed

Honestly, I'm not sure why some suppliers resist writing a cycle-time guarantee into the contract. My best guess is that they don't want to own the program optimization work. But that's exactly the risk you're paying them to take.

Checkpoint: there is a written answer to 'what happens if the line fails acceptance testing.' If the answer is a phone call, you don't have an answer.

Step 7: Plan the first 30 days before the equipment arrives

The line landing is the easy part. The hard part is the four weeks after. Before delivery, have these ready: golden boards for each product, a program library structure, a feeder calibration schedule, written work instructions for operators, and a maintenance calendar. Line the installation date up against your facility work — if the machine arrives in eight weeks and your electrical upgrade takes twelve, you're paying for equipment that's sitting in a crate.

Plan the ramp, too. Don't load the line to capacity on day one. Run your five hardest boards first as a stress test, while the vendor's engineer is still on site and still contractually obligated to help.

Checkpoint: a dated schedule that puts facility work, installation, training, first article, and ramp in one calendar — with the critical path highlighted.

Where this checklist breaks down

A few honest limits. This worked for us, but we run high-mix, low-to-medium volume with fairly stable demand. If you're a single-product shop running one board 24/7, changeover time barely matters and throughput should lead everything else. The calculus is different.

Common mistakes I've watched people make anyway:

  • Comparing machine prices instead of total cost. Feeder cost and software renewals are where budgets quietly die.
  • Ignoring facility lead time. Electrical upgrades and floor work routinely outlast the equipment delivery.
  • Forgetting the learning curve. Expect lower yield for the first few weeks. Build it into your delivery commitments to customers, not just your internal plan.
  • Skipping the spare parts list. A six-week wait on a $200 nozzle can idle a $300,000 line.
  • Assuming IPC-A-610 Class 2 and Class 3 acceptance are interchangeable. They aren't, and the difference changes your inspection labor and your rework rate.

Bottom line: five minutes of verification at the sourcing stage beats five days of correction after install. The checklist isn't exciting, and running it properly added about three weeks to our 2024 project. It also saved us an estimated $30,000 in change orders and unplanned facility work. That's a trade I'd make every time.