The Rejected Batch That Changed How I Read Laser Quotes

2026-11-17· Henrik Solberg

The Batch That Made Me Read the Fine Print

In March 2024, I stood in receiving with a 10× loupe in one hand and a rejection form in the other.

The delivery was 1,200 stainless steel brackets. Each one needed a laser-marked serial number and our logo on the underside—nothing exotic. I checked the first ten. Fine. Then I kept going, and by piece forty I could feel the engraving catch a fingernail on roughly one in five.

Contrast swung from dark grey to barely-there. Depth, measured on our profilometer, ranged from 0.012 mm to 0.061 mm against a 0.05 mm minimum spec.

The vendor's response arrived the same afternoon: "This is within industry standard for fiber marking."

Maybe. But "industry standard" isn't the standard we wrote into the purchase order, and it definitely isn't the standard printed on the drawing that ships with the part. We rejected the batch. They redid it at their own cost, and our assembly line sat idle for nine days.

(Note to self: nobody writes idle-line cost on a rejection form. Nine days across two product families came out to just over $38,000 in delayed shipments. That's the number that got management's attention—not the $2,900 redo.)

Why We Were Marking Parts At All

Some backstory, because laser marking wasn't the original plan.

We started in February 2023 with printed labels. A local digital press shop ran durable polyester labels for us—clean output, sensible pricing at our volumes, no capital expense. For eighteen months, that was the right answer.

Then a customer in outdoor equipment added a line to their spec: permanent identification that survives a solvent wipe and 500 hours of salt spray. Our labels didn't. Not even close. The adhesive failed around hour 260 in the test chamber, and the print itself started lifting at the edges.

That's how the digital printer vs laser printer question landed on my desk. Not as a preference—as a constraint. Digital printing gives you speed, color, and almost no setup cost on short runs. Laser marking gives you a mark that becomes part of the metal instead of sitting on top of it. Neither one is "better." They solve different problems, and the moment your spec says "survives 500 hours of salt spray," the choice makes itself.

So from October 2023 we outsourced laser marking to a job shop. The per-piece price was reasonable. Everything around the per-piece price was not.

The Quote That Grew 28% After We Signed

Here's the invoice structure I should have read more carefully:

  • $0.62 per piece for marking
  • $85 artwork/setup fee, charged every time we revised a part number
  • 250-piece minimum per part number
  • $45 per-box handling charge on outbound freight
  • 35% rush fee for anything inside ten business days

Our first annualized forecast said $19,400. The actual twelve-month spend was $24,900. That's 28% above the number I put on the budget slide.

Nobody lied to me. Every one of those fees was disclosed. I just didn't ask the right question, which isn't "what's the price"—it's "what's not included."

That's what I've come to believe after four years of reviewing supplier quotes: the vendor who lists every fee up front, even when the total looks higher, usually costs less by the time the year closes. The one with the cleanest-looking number is often just the one with the most practice at hiding the rest.

Bringing It In-House, and the Footprint Problem

By January 2024 we were quoting an in-house system. Capex estimate of $14,000–$18,000 against roughly $25,000 a year in outsourced spend. Payback inside a year, if nothing went wrong.

Things go wrong. More on that in a minute.

The first fork in the road was a diy laser engraver build versus a turnkey enclosed machine. I looked hard at the DIY route—open frame, off-the-shelf motion, source bought separately. It's genuinely viable if you have a mechanical engineer with spare hours and a room nobody else needs. We had neither. Diode-based hobby units were also out for a reason that has nothing to do with quality: bare stainless needs a wavelength and power density that a diode module on a hobby gantry won't hold consistently at production duty cycles. On coated metals and plastics, diode systems are perfectly respectable. On raw 304 stainless with a depth spec, they weren't the right tool.

The second fork was the source. We spec'd a 50 watt jpt fiber laser in MOPA configuration, because pulse-width control gave us the contrast range we needed on stainless without dumping heat into thin sections. A jpt laser source in that power class is a sensible middle—enough headroom for a 220 mm field at production speed, not so much that the enclosure and interlock requirements balloon. The same family runs up through 60W, 80W and 100W if we ever need to trade cycle time for power, and it covers UV heads if we move deeper into plastics.

Quotes for a 50 W MOPA-class source in early 2025 ran from roughly $2,800 to $5,500 depending on beam quality, interface, and warranty terms (based on vendor quotes we collected in Q1 2025; verify current pricing—this market moves fast).

Then the constraint that killed three options: floor space. I asked every integrator the same question—what's the smallest laser printer configuration that still gives me a 220 × 220 mm marking field without moving the part? Two of them answered with a footprint I couldn't fit next to our CMM. The third cleared it by 90 mm.

Safety wasn't negotiable. A 50 W fiber source in an open frame is a Class 4 laser product under IEC 60825-1, and that is not something you park on a production floor next to people. We required a Class 1 enclosed system with an interlock on every access panel. In the US, laser products also have to comply with FDA's 21 CFR 1040—verify current requirements at fda.gov before you sign anything.

The Mistake I'd Been Warned About

Our integrator told me, in writing, to run a five-piece first-article inspection at the start of every shift and to lock the parameter recipe behind a password.

I said we'd do it weekly. Weekly felt proportionate.

Six weeks in, we ran 400 aluminum tags on a Friday afternoon, then switched back to stainless brackets on Monday morning without a first-article check. The fixture had been nudged 1.4 mm during a tooling change. Nobody noticed. The lens had picked up dust from the aluminum run. Nobody cleaned it.

Sixty-two of the first 180 brackets failed depth or contrast. Scrap value: $4,180. Three days of recovery. And a customer-facing delay on an order I'd promised—the one part of this job I actually lose sleep over.

Everyone warned me. I didn't listen, because a weekly check felt like enough diligence for a process I'd already validated once. That's the trap, by the way. Validation isn't an event. It's a schedule.

What the Protocol Looks Like Now

Since June 2024, the line runs on a written checklist, and the checklist isn't optional:

  • Five-piece first article at every shift start and after any material or fixture change—depth measured three times per piece, contrast compared to a physical master under fixed lighting
  • Lens inspection and clean every 40 operating hours, logged by initials and date
  • Parameter recipes locked; changes require my sign-off and a fresh first article
  • Monthly depth audit on retained samples, kept for twelve months
  • Any out-of-tolerance result stops the run. No "we'll catch it downstream."

Reject rate went from 9.4% in May 2024 to 0.6% by August. Since then, zero customer escapes on marked features across eleven months of production.

What I'd Tell the Next Quality Manager

It took me about eighteen months and something like forty supplier quotes to understand that the expensive part of laser marking isn't the machine. It's the undocumented assumptions—the fixture height, the focal distance, the cleaning interval, the three fees nested in the terms.

Price the whole thing. Ask what's not included before you ask what it costs. Then write the answer into the contract, because the vendor who says "that's within industry standard" will say it again.

The other half of it: bring the standard with you. Whether you outsource to a job shop or put a 50 W JPT source on your own floor, the machine isn't what keeps quality consistent. The checklist is.

That's the part nobody puts on the quote.