The short answer

When you're sourcing an SMT pick and place machine, the deciding number isn't the price on the quote and it isn't the placement speed on the brochure. It's repeatability in month twelve — how well that machine still places 0402s on your board after a year of two-shift operation with whoever you hired last spring.

The sticker price is maybe 40% of what the machine costs you over five years. The rest is feeders, nozzles, spares, service calls, changeover time, and the rework you eat because a head drifted out of calibration and nobody caught it for two weeks.

So the rule I actually work by: don't release a purchase order until the seller has run your board — your Gerbers, your component mix — and you've timed a changeover from one product to the next with a stopwatch in your hand. That part isn't negotiable. Price is.

Everything below is how I got to that rule, and where it stops being true.

Why you should take this from me (and where I'm not qualified)

I'm the office administrator for a 180-person contract electronics manufacturer. "Office administrator" is the job title; what I actually do is procurement. I handle equipment, MRO, and consumables — roughly $2.1M a year across 14 vendors — and I report to both operations and finance, which means I get squeezed from two directions when something goes sideways.

I took over equipment purchasing in 2021. Before that, our process engineer spec'd the machines and I just cut the checks. In our 2024 line expansion, I sat in on the full evaluation for the first time, and it changed how I buy.

Here's the story that did it. Early 2023, I found a used machine about $18,000 cheaper than anything our regular dealer quoted. Looked clean in the photos. The seller was friendly and fast. Then finance asked for the invoice and what I got was a handwritten receipt and a screenshot of a chat thread. No calibration certificate, no service history, no CE documentation. Finance wouldn't release payment, our compliance lead burned a week on it, and the deal died. We lost six weeks and I had to explain to my VP why the line still wasn't up.

The machine might have been perfectly fine. That's not the point. The point is that I couldn't prove it was fine, and in a purchasing role, unprovable is the same as unusable.

That's the admin buyer's real contribution to an SMT equipment decision — not the technical call, you need your process engineer for that, but the paper trail. Certificates, serial numbers, service records, spare parts lists, feeder inventory. Get those before you get excited about the price.

What the spec sheet doesn't tell you

The CPH number is a lab result, not your throughput

Placement rate gets measured under a defined standard. IPC-9850 governs how manufacturers characterize placement equipment performance, and that's genuinely useful — it means two machines' numbers are at least comparable. But it also means the number comes off a standardized test board under controlled conditions.

Your production rate is a different animal. It depends on how many feeder positions you have, how long it takes to swap feeder banks between products, how fast the vision system aligns a flex that isn't sitting perfectly flat, and how often an operator stops to clear a pickup error. I've watched a machine rated at 30,000 CPH produce less than one rated at 22,000 on a high-mix floor, because the second one changed over in 9 minutes instead of 25.

Most buyers focus on placement speed and completely miss changeover time. Speed only pays you back on long runs. Changeover pays you back every day, on every product.

The question to ask instead

The question everyone asks is "how many components per hour?" The question they should ask is "how many minutes, and how many people, to go from product A to product B — and can I watch it happen?"

The second question I'd add: "how many of these machines are running within 300 miles of me, and who services them?" Not because I want to buy local. Because spare parts lead time is where a cheap machine turns expensive. A $900 nozzle that takes five weeks to arrive is a five-week production problem.

Your customer never sees the machine

This is the part that took me longest to internalize. Nobody at your customer's company grades your pick and place machine. They grade the assembly you shipped them.

A slightly rotated part, a tombstoned 0603, a joint that reads cold because placement was a hair off center — that's what lands on the receiving inspection report, and that report is your reputation. IPC-A-610 splits acceptability into classes: Class 2 is the default for most commercial work, Class 3 for higher reliability. Whether you can hold Class 3 consistently is a function of placement accuracy, repeatability, calibration discipline, and whether anyone is checking.

The money you save on a machine shows up later as a note in someone else's audit. It's a slow, quiet way to lose a customer, and by the time it hits your numbers, you've been losing them for a year.

New, used, or through a sourcing partner

I have mixed feelings about used equipment. On one hand, a well-maintained machine from a shop that's upgrading is often the best value in this market — you get capability you couldn't otherwise afford. On the other, the risk isn't spread evenly. It lands all at once, on a Friday, when a board handler fails and there's no replacement in the country.

What I've settled on: used is fine when you can verify the maintenance history and buy the critical spares up front along with the machine. If you can't do either, the discount isn't really a discount.

Sourcing partners are the same story, just with different risk. If your requirement is narrow — one specific model, one specific year, factory warranty — go direct or through the manufacturer's channel. If your requirement is broad — you need options across several manufacturers at different price points and you don't have six weeks to call 20 dealers — a sourcing partner earns their margin. What I won't do anymore is let "we can match any price" be the entire basis of a relationship. Price matching is a service. It isn't proof of anything.

(Ballpark for context: entry-level automatic pick and place machines for low-volume work commonly quote in the $15,000–$40,000 range; mid-range machines with real feeder banks and vision alignment land somewhere between $60,000 and $150,000. Based on quotes we collected in Q1 2026 — verify current pricing, because this moves constantly and configuration swings the number by 50%.)

The checklist I actually use

Not exhaustive, but these are the items I won't skip before a PO goes out:

  • Demo on your actual board, your actual component mix, your actual panel size
  • Changeover timed in person or on camera, by your own engineer
  • Calibration certificates plus the most recent service records
  • CE or equivalent documentation, and a serial number you can trace
  • Spare parts list with lead times for the ten most-replaced items
  • Feeder count and feeder compatibility — feeders are where budgets go to die
  • A named service contact, written response time, and the nearest technician
  • Training included, and who pays for round two when round one doesn't stick

When this advice doesn't apply

If you're running one low-mix line making a handful of simple boards, most of this is overkill. A used machine with a good local technician and a drawer of spares will serve you for years, and paying a premium for repeatability you'll never measure is money down the drain.

Same if your boards are single-sided, 0805-and-up, low volume. Placement accuracy beyond what your component mix requires is a number you're buying for the spec sheet, not for the customer.

And I'll be honest about my own limits here: I'm not the person who should be judging head accuracy or vision resolution. I'm the person who makes sure the demo happens, the documents are real, and the payment doesn't leave until they are. Let your process engineer own the technical call. My job is making sure that call rests on something you can prove six months later.

Bottom line: buy the machine that still places correctly in year three, from someone who can prove they'll answer the phone in year two. If you can't get both, at least get the second one.