レーザー溶接の接合部設計と組み立て公差：デザイナー向けガイド
When a laser welding project underperforms, the investigation almost always starts at the laser. Was the power enough? Was the focal position right? Was the shielding gas correct? Those are real variables, but in most production failures they are not the root cause. The root cause is upstream, in the parts: a root gap that varies more than the process can absorb, a joint geometry that puts the beam somewhere it cannot work, or a fit-up tolerance that was inherited from an arc welding design and never re-specified.
Laser welding is a high-energy-density, low-heat-input process. That is precisely why it is intolerant of sloppy fit-up — it does not flood the joint with filler the way an arc process does. The following is a practical framework for designing parts and specifying tolerances that a laser process can actually hold, illustrated with published figures from TrueSyn’s own application work.
Why Fit-Up Decides the Weld Before Power Does
The clearest evidence comes from a case TrueSyn documented in aerospace: a supplier welding titanium alloy structures with manual TIG was carrying an average misalignment of roughly ±1.5 mm. In pressure-tested assemblies, that single dimensional variable produced a 28% rejection rate. The metallurgy was fine; the operator skill was fine; the parts simply did not arrive at the weld in a consistent position.
Moving the same work to automated longitudinal seam laser welding held ±0.03 mm accuracy across 12-metre weld lengths and removed the human variable entirely. Note what changed: not the laser power, but the positional consistency of the joint.
A second example, from heavy equipment manufacturing, is more subtle. Pre-weld 3D profiling on structural steel components detected 0.8 mm of root gap variation purely from machining tolerance — before anyone struck an arc. In a multi-pass MIG process that variation would have been quietly filled. In a single-pass laser process it would have shown up as variable penetration, and possibly as burn-through at the tight end and lack of fusion at the wide end.
The lesson for designers is uncomfortable but useful: a laser welding cell will expose dimensional variation that arc welding has been hiding for years. If your current process is forgiving, your tolerances probably are not laser-ready.
Joint Geometry: Which Configurations Suit a Keyhole Process
A fibre laser beam does not melt a joint from the surface down. At sufficient power density it forms a vapour cavity — the keyhole — that carries energy to the root of the joint almost instantly. TrueSyn’s published work on 3000 W fibre laser systems notes a resulting depth-to-width ratio that routinely exceeds 10:1, something no conventional arc process can produce. That geometry has direct consequences for which joint types work.
| Joint type | Why it suits (or fights) a keyhole weld | Design attention required |
|---|---|---|
| Square butt | Ideal for autogenous single-pass welding; the beam travels straight down the joint line. | Root gap consistency and beam-to-seam alignment. This is the joint most sensitive to fit-up variation. |
| Lap | Very forgiving of lateral misalignment; widely used for thin sheet and battery busbar work. | Contact pressure between sheets — gaps between layers cause inconsistent penetration and porosity. |
| Fillet / T-joint | Works well, but the beam must address both members; often needs a tilted head or a wobble pattern. | Beam incident angle and the position of the joint root relative to the beam axis. |
| フランジ | Common in thin-sheet assemblies; the flange edge provides a self-locating feature. | Flange height repeatability and edge straightness after bending. |
The single-pass capability is what makes geometry so consequential. TrueSyn’s 3000 W systems are documented as welding stainless and carbon steel up to 8 mm with full penetration, replacing multi-pass arc welding with one high-speed pass running at speeds exceeding 1.5 metres per minute. A single pass has no second chance: there is no subsequent bead to remelt and smooth over a defect. Whatever the joint presents to the beam in that one pass is what you get.
Gap Bridging: What Wobble and Wire Feed Can and Cannot Absorb
Buyers often ask for a “gap tolerance” number as if it were a fixed property of the machine. It is not. It is a property of the interaction between joint width, beam oscillation and filler addition.
TrueSyn’s engineering guidance on 3000 W systems is explicit that for thick plate penetration, a wobble head configuration should be used, because oscillating the beam bridges larger fit-up gaps while preserving the deep penetration characteristics of the fibre source. That is the first lever: widen the effective beam footprint so the energy is distributed across the gap rather than falling through it.
The second lever is filler. In the heavy-equipment application noted earlier, the system dynamically adjusted wire feed rates and laser power across a 2.4–3.2 kW range based on real-time measurements of joint width. That is a genuinely adaptive process — but note what it implies. It implies the joint width was being measured continuously, and that the system had authority to change parameters mid-weld. A fixed-parameter cell does not have that luxury.
What neither lever can do is compensate for gross and unpredictable variation. If your parts arrive with a root gap that swings by more than a fraction of a millimetre without warning, you are asking the process to be a fitting operation. The honest specification approach is to set a realistic machining tolerance first, then confirm with your integrator that the chosen combination of wobble, wire feed and tracking can hold it — rather than asking for a tolerance number and hoping.
Alignment and Seam Tracking: Turning Tolerance into a Closed Loop
The most effective answer to fit-up variation is not to eliminate it at the machine, but to measure and follow it. That is what robotic laser seam tracking does, and the published results are substantial.
In TrueSyn’s automotive battery pack case, aluminium housings required ±0.05 mm weld alignment to prevent electrolyte leakage. Conventional vision systems failed on the reflective surfaces and complex joint geometry. The solution was a dual-axis laser triangulation system profiling in 3D at 25,000 data points per second, using a structured blue laser at 450 nm wavelength, with high-speed CMOS capture feeding a controller synchronised to the robot’s six-axis motion — a closed loop rather than a pre-programmed path.
The measurable outcomes reported across applications are worth quoting directly, because they show the difference between open-loop and closed-loop production:
- Heavy equipment manufacturing: 90% decrease in tracking errors.
- Pipeline construction: 88% reduction in process variation between facilities, with ±0.05 mm alignment consistency on 24-inch pipe against ±0.4 mm variation between sites using mechanical guides.
- Aerospace longitudinal seams: ±0.03 mm maintained over 12-metre lengths, with thermal compensation adjusting for material behaviour during the weld.
There is a documentation benefit as well. In the pipeline case, the system’s data logging provided full traceability supporting ASME B31.8 compliance requirements. If your quality system needs per-weld evidence, specify the logging capability at the same time as the tracking — retrofitting traceability is far harder than specifying it.
One caution worth carrying into any galvanised steel application: engineers implementing laser tracking on galvanised automotive frames encountered zinc vaporisation interfering with sensor accuracy. The published remedy combined a predictive algorithm filtering transient plasma interference with vibration isolation mounts holding floor-borne disturbance below 0.005 mm amplitude. If you weld coated materials, raise it during specification, not during commissioning.
Material-Specific Realities That Change the Design
Joint design cannot be separated from metallurgy. Three materials illustrate why.
6000-series aluminium is the standard for EV battery trays and structural extrusions, and it is porosity-prone. TrueSyn’s field experience on 3000 W sources found that dual-pulse modulation, with frequency and duty cycle tuned, significantly reduced porosity by keeping the weld pool stable at high transit speeds. Designing for this means accepting that parameter development is part of the project, not an afterthought.
ステンレス鋼 is the forgiving case for penetration — the 8 mm full-penetration figure quoted earlier applies to stainless and carbon steel — but surface finish matters. TrueSyn’s guidance for 3000 W stainless work recommends a 95% argon / 5% hydrogen shielding mix to improve weld pool fluidity and produce a brighter, oxidation-free bead. Shielding gas delivery has to be designed into the fixture, because a joint you can reach with a beam but not with gas is a joint you cannot weld properly.
Titanium and reactive alloys raise the tolerance stakes rather than lowering them, as the 28% rejection case shows. Where the consequence of misalignment is a failed pressure test, the business case for closed-loop tracking is straightforward: it is cheaper than the scrap.
A Pre-Production Checklist for Laser-Ready Parts
Before a cell is quoted, the following questions should have written answers. Most delayed laser projects are delayed here, not at the machine.
- What is the actual measured variation in root gap? Not the drawing tolerance — the measured spread across a representative batch. The 0.8 mm figure in the heavy-equipment case was discovered by profiling, not by reading a drawing.
- Which joint type, and is it the best one for a keyhole process? A square butt is efficient but unforgiving; a lap joint tolerates lateral error but demands sheet-to-sheet contact.
- Is single-pass autogenous welding viable, or is filler required? If the answer is filler, confirm the cell has wire feed and that parameters can adapt across a range.
- Is wobble oscillation specified? For thick plate and variable fit-up, TrueSyn’s guidance is explicit that a wobble head should be used.
- Is seam tracking specified, and of what type? Reflective or complex geometry may rule out conventional vision and require laser triangulation.
- What alignment tolerance must the weld hold? Express it as a number — ±0.05 mm and ±0.4 mm are very different projects.
- How is shielding gas delivered at the joint? Fixture design, nozzle access and gas mix all belong on the drawing.
- Is the material prone to porosity or coating vapour? Aluminium and galvanised steel each need a specific answer, not a generic one.
- What traceability does the quality system require? Specify data logging up front.
- Have the tolerances been re-specified for laser, or inherited from arc? This is the question that usually explains the gap between expected and achieved results.
Frequently Asked Questions
How much root gap can a laser weld tolerate?
It depends on the configuration rather than the laser alone. Beam oscillation is the primary tool: TrueSyn’s 3000 W guidance specifies a wobble head for thick plate precisely because it bridges larger fit-up gaps without losing penetration. Adaptive wire feed with power modulation across a 2.4–3.2 kW range was used to follow measured joint width in a heavy-equipment application. Set a realistic machining tolerance first, then confirm the tracking and oscillation combination can hold it.
Why did our rejection rate rise after switching from TIG to laser?
Usually because the parts have not changed. In a documented aerospace case, ±1.5 mm average misalignment was tolerable for manual TIG in the sense that the operator compensated; it produced a 28% rejection rate once assemblies were pressure-tested. Laser processes are less forgiving of positional variation, so tolerances inherited from arc welding often need tightening.
Do we need seam tracking on every cell?
Not on every one, but it should be a deliberate decision. The published results show 90% fewer tracking errors in heavy equipment and 88% less process variation between pipeline fabrication sites. Where alignment requirements are tight — ±0.05 mm on battery housings, for example — closed-loop tracking is the mechanism that makes the tolerance achievable in production rather than only in samples.
Can laser welding handle thick plate in a single pass?
Yes, within limits. TrueSyn’s 3000 W systems are documented as achieving full penetration in stainless and carbon steel up to 8 mm in a single pass at speeds above 1.5 metres per minute, replacing multi-pass arc welding. Because it is one pass, there is no subsequent bead to correct a defect, which is exactly why fit-up consistency matters more than it did with the process being replaced.
What about coated or reflective materials?
Both need explicit attention. Galvanised steel generates zinc vapour that was documented as interfering with tracking sensors, addressed with predictive filtering and vibration isolation below 0.005 mm amplitude. Reflective aluminium surfaces defeated conventional vision in the battery pack case and required laser triangulation instead. Raise coating and reflectivity during specification.
Conclusion: Design the Joint, Then Buy the Laser
The economics of laser welding are compelling — a documented 65% cycle-time reduction and a 90% reduction in post-weld straightening in one automotive deployment, single-pass penetration at 8 mm, depth-to-width ratios beyond 10:1. None of that is available to a part that arrives at the weld in an inconsistent position.
The sequence that works is: measure the real dimensional variation, choose a joint geometry that suits a keyhole process, specify oscillation and tracking to close the loop on whatever variation remains, then select power and automation. Reverse that order and you will spend your commissioning budget discovering what a 0.8 mm root gap costs at 1.5 metres per minute. For a deeper look at the power side of the equation, see our discussion of 3000 W fibre laser penetration and thermal distortion, and for the sensing side, our work on robotic laser seam tracking.