Factory-direct micro welding equipment
Dental frameworks, eyewear hinges, watch cases, thermocouple wire, sensor housings. Four benchtop models from the plant that builds them — each one right for a different kind of work, and wrong for something.
We will also tell you which parts we won't take.
Distributor or OEM buyer? Ask for a model recommendation and quoteThe weld is rarely what's at risk. Everything already finished around it is.
The parts list
A parts list, not an industry list. If yours isn't on it, send one and we'll find out.
Partial denture frameworks and clasp arms, implant bars, orthodontic attachments and retainer wire. Cobalt-chrome and titanium, almost always joined after casting, where a re-cast costs more than the joint ever did.
Denture and framework welding
Endpieces, hinges, bridges, pad arms, snapped temples. Titanium fronts, stainless hardware, plated brass components.
Eyeglass frame welding
Cases, bracelet links, clasp springbars, lugs. Stainless and precious metal, worked next to surfaces that are already finished.
Watch case and bracelet welding
Terminals, lead wire, thermocouple junctions, small sealed housings, battery tabs among them.
Chipped edges, worn detail, small inserts on tools that are already hardened and ground.
Hollow and thin-wall components, chain and findings assembly, settings joined as a production step.
Jewelry welding on the production sideRoutine here: cobalt-chrome, titanium, stainless, precious metals. High-reflectivity pure metals, zinc-bearing alloys and platings, and stone-set work all go through the limits check below first — we don't claim everything welds. Not on the list? That's what the weld sample evaluation is for.
Heat input
What decides how this job gets done usually isn't the machine. It's what's sitting next to the joint.
| What's next to the weld | What heat does to it | What we'd do |
|---|---|---|
| Acrylic denture base and resin | It whitens, blisters and lets go of the metal attachment long before the metal is bothered. | Weld the attachment before the resin goes on. On a finished case, plan a rebase rather than a rescue, and say the resin is already there when you send the part. |
| Plating and cemented lenses on frames | Plating discolours, and it doesn't discolour locally — the frame goes back for a full re-plate. Lens cement releases. | Weld at bare-metal stage. If the frame is already plated, price the re-plate into the job before you start, don't discover it afterwards. |
| Polished faces, brushed flanks, case gaskets | Heat tint on a polished surface doesn't buff back to the same finish, and a deformed gasket means the seal is gone even though the case looks fine. | Pull the gasket, mask the finished face, and plan to refinish that one facet — not the whole case. |
| Insulation, potting compound, adjacent components and magnets | The polymer gives up before the joint does, and a potted assembly holds the heat in instead of shedding it. | Weld before potting. If it's already potted, the fix is a design or sequence change, not a parameter change. |
| Stones set on the production side | Same energy, completely different outcome depending on the stone: filled, treated, organic, water-bearing. The treatment history usually isn't visible. | Judged piece by piece against a test weld. We don't publish a safe-distance rule, because there isn't one that survives contact with a treated stone. |
Four of those five end in the same place: change the order of operations, not the machine. That's the answer most equipment suppliers have no reason to give you.
Process limits
We'd rather lose the order than sell you a machine that can't do your job. Two things we turn down, four we test first — and a good deal of ordinary work we can set up without a test weld at all.
All four of these laser micro welding machines are pulsed benchtop units, and that depth band is where they live. It's a ceiling set by how the energy arrives, not a setting you can climb past — turning the power up on heavier section gets you a wider, hotter, still-shallow bead. If your joint needs full penetration through thicker material, this is the wrong process and we'll say so rather than sell you a machine that disappoints on the first job.
Stainless to titanium, copper to stainless are the two we see most. The bead can look immaculate and still crack straight through the interface the first time the part is loaded, because the failure is in the metallurgy of the fusion zone, not in the weld quality. We recommend a mechanical joint or a transition layer instead. This one lands hardest in dental (cobalt-chrome to titanium, cobalt-chrome to stainless on clasp work) and in eyewear (titanium front to stainless hinge), which is exactly where we get asked to "just weld it on" most often.
At 1080 nm the initial absorption is very low, so the energy either reflects — which can come back up the optical path — or conducts straight out into the body of the part. The signature failure is bistable: nothing, nothing, nothing, then a blow-through, all within a small change in setting. A test weld fixes the parameter window and tells us what reflection protection the job needs.
Zinc boils far below the melting point of the base metal, so the first thing the beam does is vaporise it — porosity, spatter, a spongy bead that looks solid from outside. Brass turns up constantly on frame components and on production hardware. Beyond the test weld, this one often needs the joint reshaped to give the vapour somewhere to escape.
Emerald is routinely oil- or resin-filled, some material is irradiated or heat-treated, opal carries water, pearl and amber are organic. The same heat gives you completely different outcomes, and treated stones frequently don't look treated. There is no rule we can publish that would be safe to follow — so we evaluate the piece. Manufacturing-side work only.
Here the limit isn't the laser at all: if the part can't be held rigid, the weld can't be repeated, and if shielding gas can't reach the pool, the joint oxidises whatever the parameters say. Assembled housings, joints buried behind other components, anything already sealed. We ask for the part in its pre-assembly state, or we quote it as a per-piece evaluation rather than as a repeatable process.
If it is already clear that yours is one we test first, send one real part: we weld it and write down the parameters we used, and the sending and the return are arranged in our first reply. If it is not yet clear, that is settled on the part before anything is posted.
Start a free weld sample evaluationThe range
Not good, better, best. Four different trade-offs — and each one is the wrong machine for something.
The category name still carries where it came from: bench welders built for jewelry workshops, and the product names kept it. What the machine does hasn't changed with the room it's in — a small spot, low heat input, a weld you can repeat — whether that room is a dental lab, a frame workshop or an electronics line.
The one that installs anywhere.
60% wall-plug efficiency, 2 kW maximum input, 110–240 V / 50–60 Hz — of the three models that state a supply figure, this one publishes a window rather than a single figure, so a standard single-phase circuit already sits inside it. Air-cooled, so there's no water circuit to plumb or maintain. Ytterbium fiber at 1080 nm, four average-power levels, pulse width 0.2–15 ms at 1–20 Hz, beam quality 1–2 mm·mrad, diode life over 100,000 hours. 610×290×430 mm, 32 kg net.
Cost Its average power is not the highest among the three models that give one.
QCW specificationsThe one that fits on a bench that's already full.
Making, cutting and welding on one machine — the only multi-function unit here — in 300×340×320 mm, the smallest body of the four. 1080 nm, several average-power configurations, air-cooled, diode life over 100,000 hours.
Cost Its average-power ceiling is the lowest among the three models that give one — where its range stops, the QCW's has only just started. Light work only — don't buy it for section it can't reach.
3-in-1 specificationsThe one that puts energy in and stops.
A pulse-width ceiling of 5.0 ms, which means it physically cannot sit on the part and keep heating it. Spot 0.1–2.0 mm, 0–20 Hz, desktop built-in water tank.
Cost Needs 220 V / 50 Hz / 60 A, so the supply usually has to be arranged first. At 840×450×500 mm and 60 kg with the tank inside the machine, moving it means moving the water too.
EU Style specificationsBuilt around seeing the joint, and keeping a rate while you weld it.
0–50 Hz where QCW and EU Style top out at 20. Spot 0.1–3.0 mm, pulse width 0.1–20 ms. Weld depth 0.1–3.0 mm. YAG, several average-power configurations, high-resolution microscope with a 7-inch HD CCD touchscreen, built-in air-cooled water tank. 765×462×570 mm, 38 kg.
Cost 5 kW total draw at 220 V / 50 Hz / 3 KVA.
High Value Type specificationsSide by side on one page: compare all four laser micro welding machines
Repeatability
One good weld is skill. Ten identical welds is parameters.
Across these machines the range runs from 0.1 ms at the short end to 20 ms at the long end, and the two ends behave nothing alike. A short pulse puts energy in and stops before conduction has time to carry it outward; a long one keeps feeding the part while the heat spreads. On an attachment backed by acrylic, or a joint a few millimetres from plating, that difference is the difference between a finished job and a re-do. It is also why "more power" is the wrong first adjustment — the neighbour usually cares about duration, not peak.
Peak is what couples you into a reflective surface in the first place; heat input is what the part has to absorb once you're in. They pull in opposite directions, which is why high peak with short dwell is a single setting rather than two independent knobs. Set peak too low on copper or silver and nothing happens until it suddenly does — the bistable failure. Too high on thin wall and you're through it. The window between those is what a test weld is for.
Spot diameter goes down to 0.1 mm at the fine end; open it up and the same energy spreads into a shallower, broader bead. Aim is the other half of the same problem: if you can't resolve the joint at the magnification the work needs, you're welding from memory, and the bead lands beside the crack instead of in it. That doesn't show up as a weld defect — it shows up as extra time on polish.
The settings behind a sound weld are a record, not a knack — pulse width, peak, frequency, spot, and the joint they were established on. Written down, they are the baseline the next run adjusts from instead of a blank page; undocumented, they leave with whoever happened to tune them that morning. That is the whole distance between a weld you got once and a weld the process can hold. Every sample evaluation we run ends with that set written out, because a setting you cannot reproduce was never a parameter.
Parameter starting points by material How pulsed laser welding actually works Choosing between the four
Our standing offer
The only page on this site where the whole thing is written out. Everywhere else just links here.
One or two real parts, in the state they'll actually be in when you weld them: same alloy, same section, same neighbours still attached. Tell us what sits next to the joint and what's already been done to the part — plating, heat treatment, resin, stones, potting. A part that's been cleaned up for us is a part we can't judge, and a coupon cut from bar stock tells us nothing about your job.
Our engineer welds it on the model we think fits the work and records the parameter set. Where it's useful, we weld the same joint at two settings so you can see the trade-off in your hand instead of taking our word for it — a shorter pulse against a wider bead, for instance, on the same part.
The parameter set we used, in writing. A model recommendation tied to the work and to your electrical supply, not to the size of the order. And the part itself, back to you if you want it — the sending and the return are arranged in our first email back to you.
If your part lands in the A-group limits above, we say so, name the mechanism that kills it — depth beyond what pulsed welding reaches, or an interface that will crack under load — and point at the change that would work instead: a mechanical joint, a transition layer, a different order of operations. Either way the part is yours — back to you or kept here, you decide. The alternative costs you a machine.
The four B-group cases above are decided here. That's what the test is for.
Send a part for evaluationBy industry
Same machine, five different rooms. The order below is our order of focus.
Nothing here is a batch. Every piece is different, the technician at the bench is often the person who owns the lab, and there's no scrap rate to hide a mistake in — getting it wrong means a full remake and a conversation with the practice.
What can't take the heat: the acrylic.
A frame workshop runs small repeat batches against a brand's spec; a watch bench runs one piece at a time and can't order a replacement for it. Opposite volumes, identical constraint.
What can't take the heat: the finish — plating, polish, the seal.
The person choosing the machine isn't the person welding on it. This is a process step that gets validated before it runs and signed off by someone who won't be there when it fails, so it has to survive vibration and thermal cycling months later.
What can't take the heat: the polymer around it.
The work arrives as a rescue. Something with a lot of hours already in it has failed, and a decision about saving or scrapping it is being made today, not next week.
What can't take the heat: the temper already in the base metal.
Production side only: repeat runs, more than one operator, and consistency across the batch mattering more than any single joint.
What can't take the heat: whatever was set or finished in the operation before this one.
Bench repair for retail jewelers and goldsmiths is a different trade with different economics, and we don't cover it here — that work lives on our sister site, jewelrywelderpro.com.
Before a purchase order
The things that have to be settled before a purchase order, written down instead of left to email.
How the plant is set up and what gets checked Who you're buying from
Private label, spare parts and the technical backup your customer will ask you for.
Talk to us about distribution and private labelAfter delivery
Delivery day is the start of it, not the end.
Remote video sessions with an engineer, plus a written parameter manual. The sessions run on your work rather than on a demo coupon — you weld your own parts while someone watches and corrects. The manual matters more than the sessions in the long run: it's what your second and third operator learn from after the calls are over.
The starting set from your sample evaluation stays with the machine. When the alloy changes, the supplier changes, or a new part comes in, send us the joint and we work out the adjustment with you remotely rather than leaving you to find it by trial.
Consumables and wear parts stay available for the working life of the machine, supplied direct. This is where factory-direct earns its keep — it matters more three years after the sale than it did on the day of it.
A commitment to keep it running rather than a clause to be read later. The same engineers who answer parameter questions handle warranty diagnosis, which usually happens over a call before anything is shipped anywhere.
What support covers, and how to reach it
Before you order
One or two is enough, as long as they're representative: same alloy, same section, same neighbours attached. Back comes the welded part, the parameter set in writing, and a model recommendation. If the part falls inside the limits we turn down, you get that answer instead, with the mechanism explained and an alternative suggested.
Argon for most of this work. Flow, and on some alloys the gas itself, follow the material and the joint geometry rather than the machine, which is why we don't publish one figure and call it universal. We establish both with you during the sample evaluation and write them into the parameter set that comes back with your part. Consumables and wear parts are supplied direct, long-term.
It depends on the model, and it's the item most often discovered too late. One of the four runs on 110–240 V / 50–60 Hz at a maximum 2 kW draw and is air-cooled, so there's no water circuit to plan for. The EU Style needs 220 V at 60 A, and the High Value Type 5 kW at 220 V / 3 KVA. Footprints run from 300×340×320 mm to 840×450×500 mm. Settle the supply before the machine ships rather than after it lands.
Remote video with an engineer plus a written parameter manual, included with the machine. We don't publish a number of days for this, because it depends on the operator and on the work — straightforward joints on familiar material come quickly, and the part that takes longer is judgement: knowing when a joint needs a fixture change rather than a parameter change. That's why the manual matters — it's what the next operator starts from.
Cobalt-chrome, titanium, stainless and precious metals are routine. Copper, silver and aluminium, zinc-bearing alloys and platings, and stone-set work go through a test weld first, for the reasons set out in the limits section. Dissimilar pairs that form brittle intermetallics we turn down rather than take. For a real answer about your part, send one.
Which of the four should I buy? The one that matches your supply, your bench and your class of work. Compare all four laser micro welders — and every sample evaluation ends with a recommendation.
Two routes
Two routes. Take the one that matches where you actually are.
Send one real part. We weld it, write down the parameters, and tell you which model fits — or tell you we won't take it and why. Free.
A model, a quote, private-label terms, spare parts for a machine you're reselling. Say which and we'll answer it directly.
Request a quote ·
WhatsApp +86 15327155363
Mon–Sat 9:00–18:00 GMT+8 · Elowen, Micro Welding Sales Engineer
Elowen is our micro welding sales engineer. You'll have a reply within 24 hours on business days (Mon–Sat, 9:00–18:00 GMT+8).
Email Request a quote or message +86 15327155363 and we'll pick it up from there.