When a Light Source Developer Launches Its Own Laser for SLS Printing

• 1
  What changes when a developer launches its own laser
• 2
  Spot size, hatch spacing and the energy window
• 3
  Where powder-bed printing still falls short
• 4
  Joining a printed body to machined interfaces
• 5
  How to qualify a laser-sintered part before you commit
• 6
  Powder-bed printing versus CNC machining

An SLS machine sinters powder with a scanned infrared beam. For years most builders bought that laser from a third party and tuned the optics around whatever the vendor shipped. When a light source developer launches its own laser, it controls the emitter, the diode stack and the drive electronics together. That matters because the beam is not a point. It has a diameter, a divergence angle and an energy distribution across the spot.

The practical gain is not raw power. It is repeatability. A 100 W fiber laser with a clean beam profile can sinter the same way a 200 W unit with a ragged profile does, because the energy lands where the galvo mirrors aim it. Beam quality is measured as M². Values near 1.0 mean the beam stays tight through the focusing lens. Values of 1.5 or higher mean the spot smears as it travels across the build plate.

In-house laser design also lets the builder match wavelength to the powder. Most polymer SLS runs at 10.6 μm with a CO₂ source, while metal platforms use 1.06 μm fiber or 1.07 μm disk lasers. You cannot swap those on a whim. The absorber in the powder is chosen for one band. A laser built for nylon will not sinter 316L, no matter how the software is tuned.

So the launch of a proprietary laser is really a claim about consistency across the build envelope. Corners of a 400 mm build plate are the hard part. If the spot grows or the power droops at the edge, the outer parts come out porous while the center parts look fine. Closing that gap is the whole engineering problem.

• 1
  Beam qualityM² near 1.0 keeps the focused spot tight across the plate.
• 2
  Wavelength match10.6 μm for polymers, 1.06 μm for most metals.
• 3
  Edge consistencyPower and spot size must hold to the corners of the build.

The focused spot sets the smallest feature you can sinter. A 50 μm spot can resolve thin walls and small holes. A 400 μm spot cannot, no matter how fine the STL mesh is. Hatch spacing, the distance between adjacent scan lines, should be roughly 70 to 80 percent of the spot diameter. Push it wider and you leave unsintered powder between lines.

Every powder has an energy window. Too little energy and the particles only tack together, giving low density and poor strength. Too much and the polymer degrades or the melt pool boils, which leaves voids and a rough top surface. The window is narrow for polymers and wider for metals, but it always exists.

Layer thickness shifts that window. Thin layers of 0.06 to 0.08 mm need less power per pass and give finer Z resolution. Thick layers of 0.12 to 0.15 mm build faster but show stair-stepping on slopes and need more energy to bond through. A stable laser lets you sit near the middle of the window instead of riding its edge.

Scan strategy matters as much as the beam. Most systems alternate the scan direction by 90° between layers, and some re-melt the border contour after the fill. This reduces residual stress and stops the part from curling at the edges. A laser with fast modulation can follow that contour without leaving a heat trail.

• 1
  Hatch spacingAbout 70–80 percent of the spot diameter.
• 2
  Layer range0.06–0.08 mm fine, 0.12–0.15 mm fast.
• 3
  Contour passRe-melt the border to cut curl and stress.

SLS gives you geometry that milling cannot reach: internal channels, lattice cores, merged assemblies. It does not give you a machined surface. As-built polymer parts land around Ra 8–15 μm. Metal parts are rougher and often need support removal plus a finish pass. If a bore must fit a bearing, or a face must seal against an O-ring, the printed surface is the starting point, not the finish.

Tolerances are the second limit. Powder-bed machines typically hold ±0.1 to ±0.3 mm on small features, and thin walls can warp during cooling. That is two orders of magnitude looser than what CNC turning and milling deliver. On our own 5-axis centers we hold ±0.005 mm and finishes of Ra 0.2–0.8 μm when a drawing calls for it.

Porosity is the third. Even a well-tuned laser leaves some internal voids, and those voids are where cracks start under fatigue. For a display model or a duct, that is fine. For a load-bearing bracket that sees 10⁶ cycles, it is not.

The honest split is this: print the shape, machine the interface. Printing gets you the complex form in days. Machining gets you the critical 20 percent of surfaces that actually touch something else. Trying to force one process to do both jobs usually costs more than using each where it is strong.

• 1
  As-built finishRa 8–15 μm on polymer, rougher on metal.
• 2
  Printed tolerance±0.1 to ±0.3 mm on small features.
• 3
  CNC tolerance±0.005 mm with Ra 0.2–0.8 μm available.

A common production route is hybrid. Print the organic body, then machine the mating faces, bores and threads. The two halves have to locate against each other, so the print needs datum features that survive the build. Add flat pads, bosses or a machined ring where the chuck or vise will grip.

Shrinkage is the trap. Polymer parts cool and pull in by roughly 1 to 3 percent depending on the material and the wall section. If you print a 100 mm bore and expect a 100 mm bearing seat, it will not fit. Print undersize and let the boring bar bring it home, or design a clearance and use a shim.

Threads are better cut than printed. Printed threads are weak in shear and their pitch drifts with layer height. A printed boss with a drilled pilot hole, then tapped on a mill, holds far more torque. The same logic applies to bearing bores, seal grooves and anything with a flatness callout.

Plan the sequence before the first build. Decide which surfaces will be machined, leave 0.3 to 0.5 mm of stock on them, and note the datums on the drawing. That single step prevents most of the rework we see on hybrid parts, whether they come from our own printers or a customer's machine.

• 1
  Leave stock0.3–0.5 mm on any surface that will be cut.
• 2
  Add datumsFlat pads or bosses give the vise something to hold.
• 3
  Cut threadsDrill and tap after printing for real torque capacity.

Do not qualify a powder-bed process on a decorative part and then trust it on a functional one. Build a test coupon with the same wall thickness, the same orientation and the same powder lot as the real part. Measure density, dimensions and surface finish on that coupon first.

Density is the number to watch. Archimedes testing or micro-CT gives you a real figure. Below roughly 99 percent density, fatigue life drops fast. If the coupon comes in low, the fix is usually energy density, not more laser power. Check hatch spacing, scan speed and layer thickness in that order.

For anything regulated, keep the paper trail. We work to ISO 9001:2015, IATF 16949:2016, ISO 13485:2016 and ISO 27001:2022, and we inspect 100 percent of parts before shipment. Material certificates, in-process checks and final reports are available on request, so a printed-then-machined part can ship with the same documentation as a fully machined one.

One more check: repeat the build. A single good coupon proves the machine can do it once. Three consecutive builds with the same settings prove the process is stable. That is the difference between a prototype and a production route.

• 1
  Coupon firstSame wall, same orientation, same powder lot.
• 2
  Density targetAbout 99 percent or better for fatigue parts.
• 3
  Repeat three timesStable settings across builds, not one lucky run.