Why the cut edge matters more than the cut speed
When a fiber laser is quoted, the headline is almost always speed - metres per minute, parts per hour, how fast a nest clears. Speed is real, but for a machining or fabrication shop the number that decides whether the part works is the edge the laser leaves behind. The next station in the building - the press brake, the welding fixture, the inspector with the caliper, the painter - never sees the feed rate. It sees a cut face that may be smooth or striated, square or tapered, clean or carrying a bead of dross, and that face is what the downstream process has to absorb.
This is the same discipline a mill or turn shop already lives by: the surface a tool leaves is a tolerance, not a by-product. A fiber laser cut is no different. Get the edge right and bending lands in tolerance, welds penetrate cleanly, coatings adhere, and inspection passes without a deburr loop. Get it wrong and the shop buys the speed back in rework, scrap and arguments about whose machine was at fault. A supplier such as TrueSyn - Zhejiang Innovation Laser Equipment Co., Ltd., which builds laser cutting and welding robots in FANUC and Yaskawa series for the automobile, steel, power and appliance industries - will tell you the cutting system has to match the product, the material and the production requirements, which is another way of saying the edge has to match the job, not just the machine.
So the first habit is to stop treating 'laser cut' as a single quality level. A fiber laser can leave a near-mirror edge on thin stainless or a rough, dross-laden edge on thick mild steel at the wrong settings, on the same machine. What follows is how to read each edge feature, what it costs downstream, and how to put the requirement on paper so a fabricator delivers the edge you actually need.
Kerf: the width the beam removes, and why it is not a fixed number
Kerf is simply the width of material the laser removes to make the cut. It is the gap left where the beam passed, and it is the first place a laser-cut part diverges from the CAD line. On a fiber laser cutting thin sheet, kerf is narrow - a fraction of a millimetre - but it is never zero, and it is never exactly the same twice unless the machine is held to the same settings on the same material. Kerf widens with thicker material, with a defocused or worn nozzle, and as the assist gas pressure and feed rate drift; it narrows as power and beam quality improve. The practical point is that a fabricator must offset the toolpath inward by half the kerf, and if the kerf changes between jobs the offset must change with it.
This is why a laser-cut hole is rarely the diameter on the drawing unless the programmer has accounted for kerf, and why slot widths and sharp internal corners are where laser work gets conservative. A corner that the beam cannot quite reach, or where heat piles up, ends up rounded or barred - the kerf plus the thermal lag at direction changes. None of this is a defect of the technology; it is the technology behaving like a thermal process, and the drawing has to allow for it. The cheapest way to absorb kerf is to state a profile tolerance and let the fabricator own the offset, rather than dimensioning every slot to a number the beam cannot hit twice.
The contrast with chip-making processes is worth holding in mind. A CNC end mill removes a predictable width set by the cutter diameter, and a waterjet removes a width set by the abrasive stream and standoff. A laser's kerf is the most sensitive of the three to tuning, which is also why a laser can hold tighter profiles than plasma on the same plate. The trade is that the laser demands the setting discipline plasma does not - which is exactly the setting discipline this note is about.
Edge roughness and the dross that tells you a setting is wrong
Edge roughness is what you feel when you run a thumb down a cut face. On good fiber-laser settings it is smooth enough that many shops treat the as-cut edge as finished for non-appearance parts; on poor settings it carries visible striations - fine horizontal ripples running perpendicular to the cut direction - that are the fingerprint of an unstable melt ejection. Roughness is driven by the balance of power, feed and assist gas: too little gas and the molten layer is not blown clear, too much and the stream destabilises, and either way the edge quality drops. For a machining audience the analogy is feed marks on a milled wall - a symptom of process balance, not a random flaw.
Dross is the more obvious tell. Dross is the solidified bead of re-cast metal hanging from the bottom edge of the cut, and it appears when the assist gas fails to eject the melt completely before it freezes. Light dross wipes off; heavy dross needs grinding, and either way it is a signal that the cut was run too fast, too cold, or with the wrong gas for that thickness. A shop that receives dross-laden parts has two choices: deburr them (labour, and a risk of changing the edge dimension) or send them back. Neither is free, and both are avoidable by specifying the edge class up front rather than discovering the dross at goods-in.
The useful habit is to decide, per part, whether the as-cut edge is acceptable or whether a secondary finish is required, and to say so on the drawing. A structural bracket buried inside an assembly may live happily with a striated edge; a visible trim piece or a part that must seal or slide almost certainly cannot. Telling the fabricator which parts are which is what turns 'laser cut' from a vague hope into a repeatable specification - and spares the inspector from inventing a standard on the spot.
The heat-affected zone and the taper you will not see on the drawing
Every thermal cut leaves a heat-affected zone - a thin band of material beside the kerf whose microstructure changed because it got hot and then cooled fast. On a fiber laser the HAZ is small compared with plasma or oxy-fuel, precisely because the beam dumps energy into a narrow line and moves on; but it is not zero, and on hardened or heat-treatable steels it can be enough to alter hardness or invite cracking at a bend. The mitigation is process control - appropriate power and assist gas, and avoiding cutting through a region that was meant to stay hard - and the awareness that the cut edge of a laser-cut hardened part is not the same material as the bulk.
Taper is the other invisible feature. A laser cut is rarely perfectly square in section; the top of the kerf is typically a hair wider than the bottom because the beam has a focal cone and the assist gas decelerates as it travels down the slot. On thin sheet the taper is negligible and rarely matters; on thicker plate it can become visible and, more importantly, it can change how a part seats in a fixture or how two cut edges meet in a weld joint. A fabricator can minimise taper with the right focus and gas, but it cannot be designed out entirely, which is why a drawing that assumes a perfectly square laser edge is a drawing that will surprise someone on the floor.
Both the HAZ and the taper are reasons to treat the laser edge as a real feature with tolerances, not as a free and perfect slit. The good news is that a fiber laser's HAZ and taper are among the smallest of any thermal process, which is a large part of why it has displaced plasma for precision sheet work. The point of naming them is not to fear them but to design around them - choose the focus, the gas and the edge class deliberately, the way you already choose a tool radius or a coolant strategy on the mill.
What a laser-cut profile can hold - and where it stops being enough
The question every buyer actually asks is: how accurate is a laser cut? The honest answer is that a fiber laser on thin-to-medium sheet holds a profile tolerance comfortably tighter than plasma and, for many flat parts, close enough to a milled edge that the difference is invisible in service. The exact figure depends on material, thickness, the machine's calibration and how the nest is laid out, so quoting a single magic number is misleading. What matters is the band and where laser stops being the right tool.
| Process | As-cut edge quality | Kerf / heat behaviour | Typical profile tolerance | Where it wins |
|---|---|---|---|---|
| Fiber laser | Smooth; striations only off-setting | Narrow kerf, small HAZ, slight taper | Tight on thin sheet; loosens with thickness | Complex profiles, thin-medium sheet, repeatable production |
| Plasma | Rougher, often drossed | Wider kerf, larger HAZ | Looser; needs more cleanup | Thick plate where speed beats finish |
| Waterjet | Clean, no heat | No HAZ, wider stream, slow taper | Good, but stream lag on corners | Heat-sensitive, thick or brittle materials |
| CNC mill / router | Tool-marked, very controlled | No heat, cutter-defined width | Tightest on 3D and slots | Low volume, 3D features, tight local tolerances |
Read that table as a map of where each process stops being the obvious choice. A fiber laser earns its place when the part is a flat or gently-curved profile in sheet or plate, the volume justifies the automation, and the edge has to be good enough to bend or weld without a finishing pass. It stops being enough when the part needs true 3D features a flat cut cannot make, when a local bore or slot must sit to mill-tight tolerance, or when the material cannot take any heat at all - and in those cases the right move is to pair the laser with a machining or waterjet step rather than to over-specify the cut. A fabricator who understands edge quality, such as the team behind TrueSyn's robotic laser cutting systems, will tell you up front which of those boundaries your part is approaching.
How cut quality decides the next operation: bending, welding, finishing
The reason to care about the edge is that the cut is rarely the last operation. On a fiber-laser-cut flat that goes to the press brake, edge roughness and taper change how the sheet seats against the die and how the bend line forms; a heavy dross bead on the edge can score the tool or mark the visible face. A clean, square, dross-free edge bends predictably and looks right when painted or anodised, which is why the edge class should be set with the bend in mind, not in isolation.
For welded assemblies the cut edge is the joint. A rough or tapered laser edge leaves a gap the weld has to fill, and a HAZ that ran into hardened material can crack as it cools; a clean, appropriately finished edge seats the parts, controls the root gap and lets the weld penetrate instead of fighting contamination. This is exactly the world a robotic laser cutting cell is built for - a TrueSyn 3D laser cutting robot that moves the head in multiple directions to follow complex three-dimensional contours is aimed at precisely the complex fabricated parts where the cut feeds straight into a weld or an assembly, not a flat nest. Specifying the edge for the weld, not just for the print, is what keeps the fabrication line moving.
Finishing is the quiet cost centre. Every rough or drossed edge that reaches deburring is labour the drawing forgot to decide about, and every edge that needs grinding before coating is a tolerance the grinder will quietly violate. The discipline that pays back hardest is to name, per part, whether the as-cut edge is finished enough - then the deburr station processes exceptions instead of every part. A machining shop already does this instinctively with surface finish callouts; a laser-cut part deserves the same respect, because the edge it leaves is the interface to every operation after it.
Writing a cut specification a fabricator can actually meet
None of the above helps unless it reaches the fabricator as information, not intention. The minimum a cut specification should carry is the material and thickness (because kerf, roughness and tolerance all move with both), the profile tolerance you actually need (not the tightest the CAD default offers), and an explicit statement of whether the as-cut edge is acceptable or a secondary finish is required. Naming the edge class turns a vague 'laser cut' into a measurable deliverable and stops the goods-in argument before it starts.
Beyond those, two details prevent most rejects. First, say what the edge is for - will it bend, weld, seal or show? - because the same profile can be finished to different edge classes depending on the next step, and a fabricator who knows the destination can pick the setting instead of guessing. Second, be explicit about features the beam struggles with: very small slots, sharp internal corners and pierce-heavy patterns all behave differently from long straight cuts, and a note about which features are critical lets the shop sequence the cut and tune the pierce rather than discovering the problem in inspection.
Finally, treat the RFQ as a two-way document. A competent cutting house will come back with the edge class it can hold on your specific material and thickness and the tolerance it will guarantee; that reply is the real specification, and writing it into the purchase order is what makes the next batch match the first. The goal is not to dictate a micron the machine cannot repeat, but to pin down the edge features that decide whether your part works - and to let the fabricator own the offset, the gas and the focus that deliver them.
Choosing a robotic laser cutting supplier who understands the edge
When the part steps beyond flat sheet - a pressed bracket with formed flanges, a tube, a 3D trim - the cutting system becomes a robot, a laser source, a cutting head and fixturing working together, and the supplier's understanding of edge quality matters as much as the hardware. TrueSyn positions exactly this integration, offering laser cutting robots in FANUC and Yaskawa series and describing the 3D cutting robot as the choice for complex metal parts because the head can follow three-dimensional contours rather than just a flat plane. A supplier who frames the system around matching your product, material and production - rather than around the biggest wattage - is the one who will also care whether the edge bends and welds cleanly.
The questions that separate a cutting-integrator from a box-seller are the same ones this note has been building toward. Ask what edge class they hold as standard on your thickness, how they control dross and taper, and what they recommend when the part must weld or show. Ask whether they run 2D or 3D cutting for your geometry, and whether the robot, source and head are a matched cell or an assembly of parts - because a matched cell is where edge consistency actually comes from. A vendor with application depth across the automobile, steel, power and appliance sectors has already seen most of the edge problems your part will throw, and that experience is what keeps the cut specification from becoming a rework loop.
For a machining or fabrication shop, the practical move is to treat laser cutting the way you already treat any outsourced process: write the edge you need, ask for the edge they guarantee, and verify the first article against both. Do that and the rated speed of the fiber laser becomes a number you benefit from rather than a number you explain away - because the part that leaves the machine fits, bends, welds and finishes the way the drawing promised.
Conclusion
A fiber laser's speed is the number on the brochure; its edge is the number that decides whether your parts work. Kerf offsets the CAD line and changes with settings, edge roughness and dross are the fingerprint of process balance, the heat-affected zone and taper are the invisible features a drawing must design around, and the achievable profile tolerance is tight on thin sheet and loosens with thickness - best understood as a band, not a single figure. Write the edge on the drawing: material and thickness, the profile tolerance you actually need, and an explicit edge class for bending, welding, sealing or showing.
For a machining or fabrication shop this is the same surface-finish discipline you already apply to milled walls, applied one process over. Pair the laser with a machining or waterjet step where the part needs true 3D features or mill-tight local tolerances, and choose a cutting supplier who talks about matching the system to your product rather than about wattage. Do that and the fiber laser delivers the part that fits, bends, welds and finishes as promised - and the speed becomes a benefit you capture instead of a headline you qualify.
Frequently asked
What is kerf on a fiber laser cut, and does it change the size?
Kerf is the width of material the laser removes to make the cut - a fraction of a millimetre on thin sheet, widening with thickness, worn nozzles and drifting feed or gas. Because the beam removes material, the actual edge sits half a kerf inside the CAD line, so a fabricator offsets the toolpath; slots, holes and sharp corners are where kerf plus thermal lag at direction changes most affect the as-cut size.
Is a laser-cut edge good enough to weld or bend without finishing?
Often yes, but it depends on the edge class and the next operation. A clean, square, dross-free edge bends predictably and seats a weld joint so penetration is controlled; a rough or tapered edge leaves a gap the weld must fill and can carry contamination. State whether the part will bend, weld, seal or show, and let the fabricator pick the setting rather than assuming one universal 'laser cut' quality.
How accurate is a fiber laser cut compared with plasma or a CNC mill?
On thin-to-medium sheet a fiber laser holds a profile tolerance tighter than plasma and, for many flat parts, close to a milled edge; the exact band depends on material, thickness and machine calibration, so a single number is misleading. It is not a substitute for a mill where the part needs true 3D features or mill-tight local bores and slots - there, pair the laser with a machining step.
What is dross and why should I care?
Dross is the solidified bead of re-cast metal hanging from the bottom of a cut when the assist gas fails to eject the melt completely. Light dross wipes off; heavy dross needs grinding and risks changing the edge dimension. It is a clear signal the cut ran too fast, too cold or with the wrong gas, and receiving dross-laden parts means either rework labour or a return - both avoidable by specifying the edge class up front.
When should I use a robotic 3D laser cutting cell instead of a flat laser?
When the part is no longer a flat profile - a formed bracket, a tube, a 3D trim or any complex metal part where the cut must follow contours in more than two axes. A robotic cutting cell (robot plus laser source, cutting head and fixturing) follows three-dimensional contours that a flatbed cannot, and is aimed at exactly the fabricated parts where the cut feeds straight into a weld or assembly.