Understand VFA in 3D Printing: How Vertical Fine Artifacts Form

• 1
  What VFA in 3D Printing Actually Is
• 2
  Why the Motion System Prints a Vertical Pattern
• 3
  The Engineering Cost of Ignoring Vertical Banding
• 4
  How to Confirm VFA Before Changing the Process
• 5
  Process Levers That Actually Reduce Banding
• 6
  Material and Geometry Effects You Should Expect
• 7
  When VFA Is Not the Real Problem
• 8
  Step by Step: A VFA Assessment Routine
• 9
  Where VFA Shows Up and How Strong It Runs
• 10
  Which Fix Fits Which Wall

Vertical fine artifacts are repeating vertical bands or ripples on the near-vertical outer walls of a printed part. They are not layer lines. Layer lines run horizontally and match your layer height, so a 0.05 mm layer gives a 0.05 mm step. VFA run the other way, up and down the wall, and their spacing has nothing to do with layer height.

The spacing is usually set by the motion system, not the slicing. On a belt-driven machine the belt tooth pitch can print itself into the wall. On a ball-screw machine the screw lead and the motor pole count can leave a footprint. A wall that should read Ra 1.6 μm can sit at Ra 3.2 μm or worse, with a visible 1–3 mm rhythm across it.

The reason engineers care is simple. VFA survives most of the standard fixes. You can dry the filament, lower the layer height, and slow the outer wall, and the banding stays. It is a mechanical signature, so it has to be treated as one.

One more boundary. VFA is a geometry-dependent defect. Curved or sloped walls often show it clearly. A flat top surface and a horizontal floor usually do not, because the toolpath there does not excite the same vibration.

Every axis has a loop. The controller reads position, compares it with the commanded path, and pushes current into the motor. When the commanded direction reverses, the mechanical play in the drive train has to be taken up before the tool actually moves. That take-up is not perfectly repeatable, and the error repeats at a fixed interval along the wall.

Direction reversal is the trigger. On a square pocket, the outer wall reverses direction at each corner, so the signature lands at the corners. On a round boss, the direction changes continuously, so the signature spreads into a fine vertical ripple that follows the toolpath.

The mass being moved matters. A heavy gantry or a large heated bed has more inertia, so the take-up and settling take longer. This is why the same file can look clean on a small machine and banded on a large one.

Belt tension, pulley runout, and screw preload all change the amplitude. So does the belt tooth profile. A worn belt with uneven tension produces a stronger, more irregular band than a fresh one at correct tension.

The practical point: the pattern is periodic, and periodicity is a clue. If you measure the band spacing with a caliper and it matches a belt pitch or a screw lead, you have found the source.

On a cosmetic part, VFA is a finishing problem. On a functional part it is a fatigue and fit problem. A vertical ripple is a stress riser. Under cyclic load, a rippled wall can start a crack earlier than a smooth one of the same nominal thickness.

Fatigue is the expensive case. In aerospace and medical work, a wall that carries load in tension and compression is the last place you want a periodic notch. The band spacing is small, so the notch is shallow, but it repeats hundreds of times around a boss.

Sealing faces are the second case. An O-ring groove or a gasket land printed with vertical ripple will not seat evenly. Leak testing then fails on a part that measures within tolerance on a CMM, because the CMM touches discrete points and misses the ripple between them.

The third case is tolerance stack-up on a mating pair. Two printed walls with opposite phase can interfere by the peak-to-valley amplitude, which is often 20–60 μm on an as-printed metal wall.

None of these are reasons to avoid additive. They are reasons to specify the wall condition, not just the nominal dimension.

• 1
  FatiguePeriodic ripple acts as a notch on load-bearing walls.
• 2
  SealingO-ring and gasket lands need a smooth, flat land.
• 3
  FitPeak-to-valley amplitude adds to the stack on mating pairs.
• 4
  Finishing costBanded walls need more stock removal to clean up.

Measure the spacing first. Put the part under a toolmaker's microscope or a low-power optical comparator and record the distance between band peaks. If the number matches a belt pitch, a screw lead, or a known motor pole interval, you have a mechanical source.

Then check the amplitude. A contact profilometer across the wall gives Ra and Rz. Ra tells you the average roughness. Rz tells you the peak-to-valley, which is what actually drives fatigue and fit. Report both.

Compare against a witness coupon. Print a plain vertical cylinder in the same material and orientation as the production part. If the cylinder bands and the production part bands at the same spacing, the machine is the source. If the cylinder is clean, the part geometry is exciting the motion system.

Look at the direction of the banding relative to the toolpath. Bands that line up with corners point at reversal take-up. Bands that wrap smoothly around a round feature point at continuous direction change.

Only after this should you touch parameters. Changing layer height or temperature without knowing the spacing wastes a build.

Mechanical first. Re-tension belts to the machine maker's spec, replace worn belts, and check pulley runout with a dial indicator. On screw-driven axes, verify preload and look for axial play. A loose axis will band no matter what the slicer says.

Then motion limits. Lower the outer wall acceleration, not just the feed rate. Most banding comes from the acceleration phase, where the loop is working hardest. Cutting acceleration by 30–50 percent on the outer wall often removes visible banding at a small time cost.

Then the input shaping or resonance compensation feature in the firmware. This is the single most effective software lever. It measures the machine's ringing frequency and cancels it. It has to be calibrated per machine, and it drifts when you change tool head mass.

Then the toolpath. A smooth continuous path around a feature excites less than a path with many short segments. Where the geometry allows, a slightly larger arc tolerance produces a smoother wall.

Last, post-processing. If the wall is a sealing or fatigue surface, a light CNC skim or abrasive flow can remove the ripple. That is a cost you should plan for at quoting, not discover at inspection.

Metal powder bed is not immune. The laser galvo has its own dynamics, and the recoater stroke adds a periodic disturbance. Banding on a metal wall tends to be finer, in the 0.1–1 mm range, but it can still reach 20–40 μm peak-to-valley on a vertical surface.

Titanium and nickel alloys are the harder cases. Both hold a sharp as-built surface, and both are used in load-bearing parts. A titanium bracket and an Inconel duct are exactly the geometries where a periodic notch matters.

Sintering changes the picture. Binder jet and metal extrusion both shrink during sinter, and the sinter smooths some of the banding. But the shrinkage is not uniform, so a banded green part can warp into a banded and distorted finished part.

Geometry is the other half. Tall, thin walls ring more than short, thick ones. Small round bosses show more ripple than large flats. A part with many direction reversals on the same wall will always be harder than a part with one continuous sweep.

If your design has a sealing groove on a tall thin wall, plan for a post-machining pass. There is no print setting that reliably fixes that combination.

Not every vertical mark is a motion artifact. Extrusion width variation from inconsistent flow can leave a vertical seam-like pattern. Wet filament causes popping and surface pits that can look rhythmic. On a metal machine, a worn recoater blade leaves a repeating streak that follows the blade, not the toolpath.

A quick test separates them. Change one variable at a time. Dry the material and reprint. Swap the nozzle or recoater. Rotate the part 90 degrees on the build plate. If the pattern rotates with the part, it is the part geometry. If it stays with the machine axis, it is mechanical.

There is also a point of diminishing return. On a non-critical cosmetic wall, chasing the last 10 μm of Rz can double the build time. On a fatigue-critical wall, it is worth every minute.

Decide which wall is which before you tune. That decision belongs in the drawing and the inspection plan, not in the operator's head at 2 a.m.

Once the wall condition is specified, quotation and process planning get much simpler for everyone.