Late October 2024. Our biggest client of the year — a 60-room boutique hotel — wanted 400 brass key tags, room number and serial code on each, mounted on rosewood blanks. Five weeks out.
By day job I'm the quality and brand compliance manager here. Everything ships past my desk before it reaches a customer — roughly 300 pieces a month through 2024, and I sign off twice per project: once at sample, once at final. We rejected about 12% of inbound supplier samples in 2024 on first pass. Not a flattering number, but an honest one.
The wood part wasn't the problem. Our diode wood engraver had been running walnut and maple for two years without drama. Slow, but predictable.
Brass was the problem.
Week one, we ran the existing etch process on a test batch of 40 tags. Inconsistent depth. Fuzzy edges. Two of them had a room number you could barely read under direct light. Not "a customer won't notice" inconsistent. "A hotel GM will notice at check-in" inconsistent. I rejected the batch — all 40 of them — and that meant a restart with a deadline already breathing on us.
The core issue was that we were asking our equipment to do something it was never built to do.
"Fiber is stronger" is the wrong explanation. Wavelength is the real one
I spent about six hours reading before I understood what the sales rep kept circling around: 1064 nanometers. I assumed it was marketing shorthand. It isn't. It's physics.
A diode laser — what most people buy for wood, leather, acrylic, anodized aluminum — typically runs around 450nm, which is blue, right in the visible range. Point that at brass, copper, raw silver, or uncoated steel and a large share of the energy bounces straight back off the surface. The energy never gets into the material. It reflects. What you get is a scorch mark that looks like engraving until someone rubs it with a thumb.
A fiber laser runs at 1064nm, in the near-infrared. Metals like brass and copper reflect substantially less of that wavelength, so the energy disperses into the metal instead of into the ceiling. That is the real difference between deep engraving and a surface smudge.
The spec that kept coming up, and that I confirmed on xTool's own product page and in a dealer call (November 2024): the xTool F1 Ultra is a dual-laser unit — 20W diode at 450nm plus 20W fiber at 1064nm, on a 220×220mm work area. Two physics in one box. That's the whole pitch, and it's a legitimate one.
Here's the part nobody puts in the brochure. Cutting metal depends entirely on which metal and how thick. On our test material, 20W fiber in the F1 Ultra needed 2–3 passes through 0.5mm brass for a clean cut. Roughly 3–4 passes on 0.8mm aluminum. Stainless above about 1.5mm, and you're running a heat-distortion experiment, not a production job. "Cuts metal" is technically true and practically conditional. Any honest answer about xTool F1 Ultra cutting metal has to include the numbers.
The three-week back-and-forth
I went back and forth for three weeks.
Option A: an enclosed fiber laser engraver, somewhere around $11,000, bench included, local service. Option B: the xTool F1 Ultra, which we saw at $3,999 on a January 2025 dealer quote sheet — check the current site yourself, because I've seen promos dip under $3,600.
On paper, Option A made more sense. Bigger, more dedicated to metal work, a real warranty relationship. Every spreadsheet column pointed at A: cost per piece, cycle time, service coverage.
My gut said B. Our floor is already tight and a second footprint machine wasn't going to fit. We still had the rosewood blanks to cut, and if one machine could handle both without splitting the job across two stations, that mattered more than the spec sheet admitted. And I've watched too many small shops buy a heavy machine that sits at 20% utilization for three years.
The thing that finally decided it was mundane. The contract was one order. Four hundred pieces. If no brass work came through next year, I'd be explaining an $11,000 ornament.
The $1,800 I spent to save $80
This is the part I'd take back.
Before ordering, the dealer offered a paid sample — same brass alloy as the job, $80, setup and material included, about three days.
I skipped it. I'd watched a handful of demo videos, the results looked clean, and $80 felt like a line item I could trim. We ran a test on brass coasters we already had in the scrap bin. Looked close enough. First pass was dark and crisp. I signed off and we started the run.
Two hundred tags in, it surfaced. The "dark" had 0.2–0.3mm of penetration into the plating, not the base brass. When we went to finish-clean the surface, sections went lighter in a splotchy, uneven pattern. Against our standard spec, visible without magnification.
200 blanks scrapped at about $6 each, roughly 40 hours of machining and cleanup, plus expedited re-buy on the material. About $1,800, all in.
Saved $80 on the sample. Spent $1,800 on the redo. That kind of arithmetic stays with you.
What we did on the second pass
The second time around, we ran a proper blind test on the actual key tag alloy. Three depth settings, same material, samples stripped of labeling. Then I pulled in three colleagues and asked each one to pick the version that looked most expensive.
All three chose the same one — 0.15mm, not the deepest. The deepest had a faint bright rim at the edge that read as a defect on polished brass. The shallowest looked printed.
That's now a house rule: on polished metal, deeper engraving can actually make the product look cheaper. It isn't in any spec document I've found, but it took our final first-pass acceptance rate from about 89% to 97%.
The wood side held up too. Rosewood bases kept running through the 450nm diode at the same parameters as before, and honestly the throughput was better than our old wood engraver. Sharing one machine across two material families is exactly what we bought it for.
On laser engraving on gold — slow down
We took a small run of gold-plated plaques and I tested the fiber side on them. Marking on plated gold came out clean, dark, repeatable.
Real gold is a different conversation. Gold conducts heat aggressively, and a few tenths of a millimeter of thickness change moves the result a lot. Thin sheets under 0.5mm on a 20W fiber unit are easy to burn through or distort. If someone brings us solid gold, we quote a sample first, then quote the job, then accept. We don't answer thickness or purity questions we haven't physically tested.
Where it stops — say this part out loud
The honest limitation is that this machine is not universal, and pitching it that way would be irresponsible. If you're shopping for an engraving machine for sale and any of these describe you, look elsewhere:
- Batch volume. A 220×220mm bed means constant loading and unloading. Past 300 pieces, the bottleneck isn't cutting speed, it's your hands.
- Thick stainless. Above 1.5mm, expecting a single-pass cut gets you distortion and blue heat marks.
- A full-shift factory floor. No enclosed cabinet, no industrial-grade extraction as a standard. It belongs in a workshop or small production cell, not a high-volume plant.
- Anyone who won't tune parameters. Dual laser means two sets of parameter logic. Carrying wood settings straight into brass is how you get the exact plating-scorch problem that cost us $1,800.
Where I'd recommend it without hesitation: a small shop running 50–500 mixed-material pieces a month that needs deep marking on brass, aluminum, and stainless, still cuts wood or leather to keep other orders alive, and has no room for a second machine. That was our situation. It fit.
What I'd do before you buy
Pay for the sample. That's it, that's the first item. Eighty dollars of setup against two hundred pieces of exposure is the cheapest insurance in the shop.
Match wavelength to material before you compare machines. 450nm and 1064nm aren't a premium tier and a budget tier. They're different tools for different substrates. Wood and leather belong on the diode side. If you want genuine depth in brass or copper, that's fiber territory, and no amount of marketing changes the reflectance curve.
Then blind-test your own samples before you commit to production parameters. Ours reversed my instinct completely.
One last number. Online (as of January 2025, at least) the F1 Ultra lists around $4,000, and for what it covers that's a fair price. But it isn't a universal engraver and it was never designed to be. It's a machine that runs two lasers to bridge two workflows — one that took us about $1,800 to learn how to feed correctly. That lesson is already paid for. You don't have to repeat it.
Leave a Reply