OrcaSlicer Z-Wobble & Z-Banding: Diagnose and Fix (2026)

The first Z-wobble print I ever diagnosed was a 200 mm vase in matte PLA on an Ender 3 V2. I’d just upgraded the hotend, leveled the bed twice, and run a fresh PID tune. The first layer was glass. The seam was clean. And running my thumb up the side of the print, I could feel a soft, regular ripple every eight millimetres. Like corduroy stretched over a glass bottle. I spent the next four hours convinced it was my PETG slicer profile. It wasn’t. It was the leadscrew. The cheap, slightly bent leadscrew Creality had shipped with the printer.

I’m telling you this because almost every Z-defect article I can find online treats Z-wobble and Z-banding as the same problem. They aren’t. They look similar at a glance, they live in the same neighbourhood of your print, and the fixes for one will do absolutely nothing for the other. If you swap your leadscrew because someone on a forum told you to and the artifact was actually a PID tune issue, you’ve spent thirty quid and a Saturday on the wrong fix. Worse, you’ll start to believe nothing fixes this, which is what most “I gave up” forum posts boil down to.

This guide is the long version of the conversation I wish I’d had with myself five years ago. I’ll separate Z-wobble from Z-banding from VFA from layer shift with a proper disambiguation table, then I’ll walk through mechanical causes, thermal causes, the OrcaSlicer settings that actually matter, the math behind the so-called “magic layer heights,” and a diagnostic workflow that gets you to the root cause cheaply. There are per-printer notes for Bambu, Voron, Prusa, Creality, Sovol, and Anycubic at the end, and an FAQ for the questions that come up over and over on the orcaslicer.net inbox. If this is your first stop, my OrcaSlicer troubleshooting guide is the broader hub.

Z-wobble vs Z-banding vs VFA vs layer shift

Spend the first cup of coffee on this section. If you walk away from this article with one thing, it should be the ability to look at a print and immediately know which of these four defects you have. The fixes diverge from this point.

Z-wobble is a smooth, periodic horizontal ripple on a vertical wall, where the vertical period of the ripple matches the leadscrew lead. “Lead” is the axial distance the nut travels in one full rotation of the screw. A TR8x8 leadscrew, the most common in low-cost FDM printers, has an 8 mm lead. A TR8x4 has a 4 mm lead. A TR8x2 has a 2 mm lead. If you can stick a digital caliper against the side of your print and measure that the bands repeat every 2, 4, or 8 mm consistently, you’re looking at Z-wobble. The cause is mechanical and lives in the Z drivetrain.

Z-banding is a horizontal stripe or set of stripes whose spacing does NOT match the leadscrew lead. It’s irregular, or it correlates with feature transitions (a wall count change, the start of an infill region, a tall thin part), or it tracks the bed heater duty cycle. The cause lives in thermal control or extrusion: PID instability, filament diameter swing, fan ramping, or layer-time effects. Most “Z-wobble nothing fixes it” forum threads I’ve read are actually Z-banding cases where the poster never measured the period.

VFA stands for Vertical Fine Artifacts. These are sub-millimetre ridges, often spaced around 2 mm but really tied to belt pitch (one GT2 tooth = 2 mm of belt travel) or motor full-step distance, not the Z leadscrew. VFAs are an XY problem, not a Z one. I’ve covered VFAs at length in the OrcaSlicer VFA test guide; the short version is that input shaper, pressure advance, and belt tension are the levers, and none of those touch Z-wobble.

Layer shift is a catastrophic XY displacement where one layer (or a stack of layers) sits offset sideways from the rest of the print. It is not a Z defect at all. It’s caused by lost steps on X or Y, belt slip, or a mechanical strike. There’s a full OrcaSlicer layer shifts article for that one because the diagnosis is entirely different.

Here’s the cross-reference table I keep open on a second monitor when I’m troubleshooting a stranger’s print over Discord:

The single best diagnostic move you can make is to measure the band period with a caliper. If you can’t measure it (matte filament, very subtle defect), print a thin tall tower in glossy filament and try again under raking light. Without that measurement, every other troubleshooting step is a guess.

One more clarification before we move on. Some online guides use “Z-banding” as an umbrella for both defects, then prescribe leadscrew swaps to people whose problem is a 4 °C temperature swing. That’s the gap this guide is trying to close. From here on, I’ll use “Z-wobble” only for the mechanical, period-matched ripple, and “Z-banding” only for the thermal or extrusion-driven stripe.

Root causes of Z-wobble (mechanical only)

Z-wobble is a mechanical artifact. Every cause in this section lives in the Z drivetrain or the frame that supports it. None of them can be fixed by changing a setting in OrcaSlicer, although some of them can be masked enough to look fixed if you’re not paying close attention.

Bent leadscrew

Far and away the single most common physical cause, especially on printers under five hundred quid. Cheap TR8 leadscrews ship bent from the factory more often than the manufacturers will admit. The diagnosis is free and definitive: remove the leadscrew, lay it on a flat surface (a kitchen worktop or a piece of float glass), and roll it. If you see a visible gap appear and disappear as it rolls, or if it rocks rather than rolls smoothly, the screw is bent. Straightening it almost never works, and even if it appears to work, the internal stress shows up as wobble again a few months later. Replacement with a quality TR8x8 (or whatever lead your printer uses) is the answer.

Rigid coupler transmitting screw runout

The coupler between your Z motor shaft and the leadscrew matters more than most people realise. If it’s a rigid aluminum sleeve with grub screws, any wobble in the screw and any misalignment between motor mount and screw axis transfers directly to the nut, and from the nut to the bed or gantry. Rigid couplers are the worst-case for Z-wobble. A flexible (helical) coupler is better. An Oldham coupler is best, but it asks more of your installation.

Coupler over-tightened or screw bottomed against motor shaft

If you cranked the grub screws hard against a screw that’s seated against the end of the motor shaft, you’ve effectively pinned the screw against the motor and lost the small axial play that lets the screw self-center on the nut. The cure is to loosen the grub screws, lift the screw a millimetre or two so it isn’t bottomed out, and re-tighten with gentle torque. Free fix. Often missed.

Loose Z-motor mount

The Z motor sits in a mount at the bottom of the column on most bedslingers, or at the top in some CoreXY designs. If those bolts have loosened (and they do, vibration loosens everything), the motor can shift in place during a Z move. The leadscrew walks. Check during a slow manual Z move and watch the bottom of the screw for any lateral movement.

Frame flex

Most relevant to bedslingers and to any printer with a tall, unsupported gantry. The Ender 3 V2 has a single Z motor and a single leadscrew on the left side; the right side of the X gantry is free-floating, supported only by the rollers on the V-slot. As the gantry rises, gravity and acceleration loads cause the right side to lag behind the left by a few hundredths of a millimetre. That lag becomes periodic if anything in the Z drivetrain has a periodic component. Dual-Z upgrade kits fix this category outright.

Anti-backlash nut wear or wrong nut

POM (delrin) anti-backlash nuts wear over time, and worn nuts develop play that shows up as inconsistent Z position. A worn nut alone doesn’t usually produce classic period-matched Z-wobble, but it can produce a near-Z-wobble appearance with a slightly variable period. Do not, repeat, do not buy a “universal anti-backlash nut” from a generic listing and drop it in. Anti-backlash nuts are printer-specific. The threads, the mounting holes, and the spring pre-load all have to match your Z assembly. Buy the one your printer’s community knows works.

Bedslinger geometry vs CoreXY

Bedslingers (Ender 3, Sovol SV06 line, Bambu A1, Prusa MK4) move the bed on the Y axis under a fixed-Z gantry. Any wobble at the Z interface gets amplified at the print surface because the print is being built on the moving bed. CoreXY machines (Bambu X1C, P1S, Voron) move only the toolhead in XY; the bed itself only moves in Z. CoreXY tends to show less visible Z-wobble for an equivalent mechanical defect because the bed is more stable. They still have leadscrews and can still suffer the same mechanical issues. The CoreXY isn’t immune; it’s just better at hiding things.

Belt-driven Z (Bambu A1)

The Bambu A1 is the odd one out among popular machines because its Z motion uses a belt to synchronise two leadscrews from a single motor. If that belt loses tension, one leadscrew lags the other, and the artifact you get isn’t a periodic ripple. It’s a layer-by-layer tilt that looks like the front of the bed is lifting unevenly compared to the back. Don’t try to diagnose A1 Z issues with the same workflow as an Ender. Read the Bambu Lab wiki page for the A1 Z-axis leadscrew kit replacement first.

Root causes of Z-banding (thermal and extrusion)

This is the section that distinguishes this guide from the dozens of others online. Most articles I’ve read mention “filament” and “PID” in a single sentence and move on. They’re the entire ballgame for Z-banding, so I’m going to break them open.

Hotend PID untuned or tuned with fan off

If your hotend’s PID was tuned with the part-cooling fan off, the moment that fan ramps up during a real print, the cooling profile of the heat block changes. The PID can’t hold the setpoint as tightly. Temperature oscillates ±2 to ±5 °C around the target. Viscosity changes with temperature. Extrusion rate fluctuates. You get bands. The fix is to re-run PID autotune with the part-cooling fan running at the speed you’ll actually print at. On Marlin, that means M106 S255 before M303 E0 S210 C8. On Klipper, run PID_CALIBRATE HEATER=extruder TARGET=210 with the fan started first via M106 S255. Once you’ve got new values, store them. The temperature tower guide covers this in more depth on the slicer side.

Filament diameter variation

Cheap spools can vary by ±0.05 mm or worse along a single roll, and a few unlucky budget brands swing wider than that. The math here is unforgiving. Volumetric flow rate depends on the cross-sectional area of the filament, and area scales with the square of diameter. A 5% diameter swing translates to roughly a 10% volumetric flow swing. That’s plenty to produce a visible band. The diagnostic is to take ten caliper measurements along the length of a spool, rotating the filament a few degrees between each measurement (round-out-of-round is also a thing). If the variance is bigger than ±0.02 mm, your filament is the variable.

Bed heater cycling

Less of an issue on modern Marlin and Klipper builds that PID-control the bed, but on older Marlin with bang-bang bed control, the bed temperature swung wide enough to visibly affect print quality. The bed cycles on, the chamber warms, the part contracts a little less; the bed cycles off, the chamber cools, the part contracts a little more. Layers laid down during those cycle transitions look different from layers laid down at peak or trough. If you’re on Marlin and you’ve never enabled PIDTEMPBED in firmware, this is worth checking.

Fan curve cycling

A fan curve that ramps abruptly is one of the easiest banding causes to overlook. If you’ve got OrcaSlicer set to keep the part-cooling fan off for the first three layers and then jump straight to 100%, the moment the fan engages, the hotend block sees an aggressive cooling step. The hotend PID plays catch-up. Temperature dips. You get a band right at that Z height. The fix is to smooth the ramp. Set the fan to come on gradually over five to ten layers instead of stepping straight to full.

Layer height not a multiple of the full-step Z distance

This is the “magic numbers” thing you’ve heard about, and it’s real. I’ve given it its own section below, but the short version is: for a standard 8 mm lead screw with a 1.8° stepper (200 steps per revolution), one full step of Z motion is 8 mm divided by 200, which is 0.04 mm. Layer heights that are clean multiples of 0.04 mm keep the Z motor mostly on full steps, which produces more consistent step-to-step thickness than odd values. 0.20 mm is a multiple of 0.04, so it works. 0.25 mm is not, so the Z motor has to microstep between full steps, which introduces a small but visible thickness variation. Sometimes.

Adaptive layer height set too aggressively

OrcaSlicer’s variable layer height feature is useful when used carefully and a banding factory when used carelessly. The feature lets the slicer choose different layer heights for different Z regions of the print, thicker where geometry is simple, thinner where curves need detail. If you set a wide min-to-max range with low smoothing, the print shows visible transitions between thickness regions. It’s not a bug exactly; it’s the feature doing what you told it to. Several OrcaSlicer GitHub issues (#1787, #5709, #9810 among others) document edge cases where it produces banding even with reasonable settings. For a first round of banding diagnosis, turn adaptive layer height off and reprint the same model. The full adaptive layer height writeup goes deeper into the parameter math.

Minimum layer time and small-feature dwell

If your slicer is set to enforce a minimum layer time (say, 5 seconds per layer to give cooling time for tiny features), the hotend dwells over the print at full temperature for longer than usual on small layers. When the next larger layer follows, the first pass is hotter than the rest of the print and you get a band right at the transition. The lever is OrcaSlicer’s “Slow down for layers shorter than” setting in the Cooling tab, plus the related minimum layer time field. Crank both down and reprint; if the band moves or fades, you’ve found it.

Pressure advance or flow ratio drift

Extrusion-rate inconsistency from a poorly calibrated pressure advance or a flow ratio that’s a few percent off can produce mid-layer banding, especially on prints with lots of corners. The fix is in the under-extrusion troubleshooting guide and the over-extrusion companion on blobs and zits. Calibrate both before assuming the problem is mechanical.

OrcaSlicer settings worth investigating

I want to give you a concrete list of settings to open and check before you blame the hardware. All the field names below are taken from OrcaSlicer 2.3.x; older builds may put things in slightly different tabs but the labels haven’t moved much in two years.

• Quality → Layer height. Set to a multiple of 0.04 mm for any printer with an 8 mm lead leadscrew. See the magic numbers section.
• Variable layer height (toolbar button on the right of the plate view). Turn off for diagnosis. The smoothing slider and the min-to-max range both affect banding when on.
• Cooling → Close fan the first X layers. If this is set to a high number and your fan jumps straight to print speed afterwards, that transition is a banding source.
• Cooling → Full fan speed at layer. Set this to ramp the fan in over several layers rather than slamming to full instantly. A 5-10 layer ramp is gentle enough for most filaments and removes the abrupt cooling step.
• Cooling → Slow down for overhangs and Overhang fan speed. Sudden fan or speed shifts for overhang regions can produce a band that tracks the overhang boundary. If you see a band that lines up with the start of a curve, this is the suspect.
• Cooling → Slow down for layers shorter than (minimum layer time). Drives hotend dwell on small layers. If banding correlates with feature size, here’s where to look.
• Speed → Outer wall speed and Overhang speed. Mid-layer speed changes for overhang or wall transitions can produce periodic-looking banding tied to feature geometry.
• Filament → Pressure advance and Flow ratio. Rule out extrusion-driven banding with proper calibration before touching hardware.

One thing worth flagging because it confuses people: OrcaSlicer’s “Slow down for overhangs” feature changes outer wall speed when the slicer detects an overhang based on the angle threshold you set. If your model has features that trigger this on some layers and not others, you’ll see speed-driven banding that looks like Z-banding but is actually slicer behaviour. The diagnostic is to disable overhang slowdown for one test print and see if the banding tracks.

The layer height magic numbers

I get one variation of this question every week: “what layer heights should I use to avoid Z-banding on my Ender 3?” The honest answer is more nuanced than the meme suggests, but the meme is rooted in real physics. Here’s the math.

A standard NEMA 17 stepper motor has 200 full steps per revolution (1.8° per step). Most low-cost FDM printers use 1/16 microstepping on the Z driver, giving 3200 steps per revolution at the controller. A typical Z leadscrew on consumer printers is TR8x8, which means 8 mm of axial travel per full rotation.

So one full step of Z motion = 8 mm / 200 = 0.04 mm.

One microstep at 1/16 = 8 mm / 3200 = 0.0025 mm.

Stepper motors are most positionally consistent at full-step positions. Microsteps in between are real, but the torque profile across a microstep cycle isn’t perfectly uniform, especially under load. The practical implication is that layer heights which land exactly on full-step positions are slightly more consistent in actual Z position than layer heights that don’t.

For an 8 mm lead leadscrew with 200-step motor, the “magic” layer heights are multiples of 0.04 mm:

• 0.04 mm (rarely used, too slow)
• 0.08 mm
• 0.12 mm
• 0.16 mm
• 0.20 mm (the classic default)
• 0.24 mm
• 0.28 mm

Layer heights like 0.15 mm, 0.25 mm, and 0.30 mm aren’t multiples of 0.04 and live between full-step positions, so the Z motor has to microstep to reach them. In practice, with a healthy printer and reasonable filament, the difference is small. On a printer with marginal Z mechanics or with a load-heavy gantry, it can show up as faint banding.

For TR8x2 (2 mm lead): one full step = 2 / 200 = 0.01 mm. Almost any “normal” layer height (0.10, 0.15, 0.20, 0.25) is a clean multiple of 0.01, so the magic number argument essentially doesn’t apply to TR8x2 machines.

For TR8x4 (4 mm lead): one full step = 4 / 200 = 0.02 mm. The same logic applies. Most common layer heights fall on full-step positions.

The argument is loudest for TR8x8 because its full-step distance (0.04 mm) is large enough that ordinary layer heights can miss it. If your printer has dual leadscrews and a sturdy gantry, you can usually ignore the magic numbers and use 0.20 mm without thinking about it. If your printer is a budget bedslinger and you’re chasing banding, switch to 0.20 mm or 0.16 mm explicitly and see what happens.

Mechanical fixes in order of cost and effort

Going cheapest and easiest to most expensive and most invasive. Work down the list; don’t skip ahead.

1. Lubricate and clean the leadscrew. White lithium grease, Super Lube, or any PTFE-based grease. Wipe the screw clean first with a paper towel and a bit of isopropyl alcohol, then apply a thin layer of grease and run the Z up and down its full range ten times. Half the “wobble” cases I’ve helped diagnose were actually binding from a dry or contaminated screw. The Bambu Lab wiki has a clean reference page for lead screw lubrication that applies to most printers, not just Bambu’s. Free if you already have grease, two minutes of work.
2. Loosen coupler grub screws and re-seat. Loosen both grub screws on the rigid coupler, lift the leadscrew a couple of millimetres so it isn’t bottomed against the motor shaft, then re-tighten gently. Lets the screw self-align with the nut. Free.
3. Roll the leadscrew on a flat surface to diagnose. Remove it, lay it on glass, roll it. Confirms or rules out a bent screw before you spend on a replacement. Free.
4. Replace rigid coupler with a flexible (helical) coupler. Five to ten quid. Compensates for minor misalignment between motor and screw, and absorbs some of the runout from a slightly bent screw. Won’t fix a badly bent screw, but it’ll mask a marginal one.
5. Install an Oldham coupler. Ten to twenty-five quid for a kit. An Oldham coupler is a three-plate design that decouples lateral leadscrew motion from the nut entirely while still transmitting rotation. It’s the gold-standard fix for stubborn Z-wobble on Ender-class printers. TH3D’s Creality leadscrew Oldham coupler kit is the most widely cited.
6. Replace the leadscrew. Ten to thirty quid for a quality TR8x8 from ReliaBot, FYSETC, or a known machinist supplier. Cheap eBay screws are often bent out of the box; this is not the place to save five pounds. If the diagnostic roll test showed a bent screw, this is the actual fix, not the coupler.
7. Replace the POM nut with a printer-specific anti-backlash nut. Only if your current nut shows wear, and only with a nut your printer’s community has confirmed fits. Anti-backlash nuts pre-load the screw threads with a small spring or split design to eliminate axial play. They do not fix bent leadscrews. They do reduce micro-play that can show up as inconsistent Z position.
8. Reduce gantry weight (advanced). Mostly applies to heavily modded bedslingers with aftermarket toolheads. Lighter gantry equals less torque load on the leadscrew, which means any wobble in the screw produces less lateral force at the nozzle. Most owners shouldn’t worry about this.
9. Upgrade to dual Z. For printers that shipped with a single Z motor and single leadscrew (older Ender 3, Anycubic Kobra with a single screw). A dual-Z kit adds a second leadscrew on the opposite side of the gantry plus a Y-splitter or independent second driver. Reduces unsupported gantry tilt dramatically. Forty to seventy quid for a quality kit.

Per-printer notes

Every printer family has its own Z personality, and the diagnostic and fix routes differ enough that a generic guide misses the point. Here are the printers that come up most often in our inbox.

Bambu X1C

Rigid CoreXY frame with two leadscrews driven by a single Z motor through a belt-and-pulley arrangement. The X1C is mechanically stable; if you’re seeing horizontal banding on an X1C, the cause is almost always thermal or filament, not the Z drivetrain. Run a known-good spool first. Then check that your part-cooling fan curve isn’t stepping abruptly. If it is mechanical, Bambu’s X1 Z-axis leadscrew replacement guide is the official path. Lead-screw silicone caps stop the screws from shaking at the top of long travels and are worth installing if yours came without them.

Bambu P1S

Mechanically close to the X1C in the Z assembly. Same diagnostic approach. The Bambu community forum has multiple threads with titles like “P1S Z banding nothing fixes it,” and the resolution in most of those threads is either PID tune with fan running or a bad spool. Don’t pull the Z apart on a P1S until you’ve ruled out filament and PID.

Bambu A1 and A1 mini

Belt-driven Z. Two leadscrews synchronised by a belt and a single motor. Failure modes differ from leadscrew-only printers. If one side of your A1 is lifting differently from the other, the belt has lost tension or one of the pulleys has loosened. Bambu’s A1 wiki has the official Z-axis kit replacement procedure linked above. A bent leadscrew on an A1 is rare but not impossible; the same diagnosis (remove and roll on glass) applies.

Voron 2.4

Four independent Z motors, one at each corner, no leadscrew coupler. The Voron’s Z calibration relies on QUAD_GANTRY_LEVEL (QGL), a Klipper macro that probes the bed at four points and adjusts each motor’s offset to level the moving gantry. If you’re seeing Z banding on a Voron 2.4, run QGL three to five times before assuming any mechanical fault; it sometimes takes multiple passes to fully level. If banding persists, the most likely cause is one motor binding (check belt tension on its tower) or a worn lead nut. Voron documentation on software configuration including QGL is the canonical reference. Voron 0 and Trident use Z_TILT_ADJUST instead, but the principle is the same.

Prusa MK4 and MK4S

Dual Z motors driving dual TR8 leadscrews with smooth-rod linear guides. The MK4 is reasonably forgiving of Z mechanical issues thanks to the dual setup and decent factory leadscrews. If you’re getting banding on an MK4, thermal causes outweigh mechanical by a wide margin. The Prusa forum thread “Z-banding or Z-wobble, do we have a solution yet?” has hundreds of MK4 cases and the dominant resolutions are PID tune and filament swap.

Ender 3 and Ender 3 V2

Single Z motor, single leadscrew, V-slot guide on the opposite side. The most Z-wobble-prone common printer in the wild. If you bought an Ender 3 in the last five years and you’re seeing Z-wobble, your three options in order are: replace the leadscrew, add an Oldham coupler, and upgrade to dual-Z. All three are well-trodden upgrade paths with mature kits.

Sovol SV06, SV06 Plus, SV07

Dual leadscrews from the factory. Better than the Ender 3 baseline. SV06 had a documented run of slightly bent factory leadscrews; if you bought yours in the first year of release and see Z-wobble, that’s the prime suspect. SV07 ships with anti-backlash spring nuts on both leadscrews and is mostly free of stock Z issues. The full troubleshooting guide has a Sovol-specific section if you want more depth.

Anycubic Kobra 2 and Kobra 3

Single leadscrew on most variants. Bedslinger. Diagnostic and fix path is essentially identical to the Ender 3 line; the components are different brands but the architecture is the same.

Creality K1 and K1 Max

CoreXY with dual leadscrews. Stiffer frame than the Ender line. Z-wobble on a K1 is uncommon; if you have it, check the leadscrew couplers and run QGL-equivalent leveling first.

A diagnostic workflow that doesn’t waste filament

This is the section I’d hand someone who just walked up to my bench and said “my prints have lines on them.” Run this end to end before you buy a single replacement part.

1. Measure the band period with a caliper. Squeeze the caliper jaws against a clean section of the wall and measure peak to peak between adjacent bands. Write the number down. If it’s 2 mm, 4 mm, or 8 mm consistent across multiple measurements, you almost certainly have Z-wobble. If it’s irregular or doesn’t match a leadscrew lead, you have Z-banding.
2. Print TeachingTech’s Z banding test tower. The simple Z banding/wobble test tower on Printables is the de facto community test. Four millimetres wide, eighty millimetres tall, single perimeter, flat front face. Print one in a glossy filament and you’ll have a clean read.
3. Check whether the pattern follows Z height. Look closely at the tower. Is the banding worst at the bottom, worst at the top, or uniform? Worst-high suggests frame flex (more unsupported gantry above). Worst-at-a-specific-Z suggests a bent leadscrew in that section. Uniform suggests thermal or filament.
4. Re-run PID autotune with the part-cooling fan running. This is the most-skipped step in the entire pipeline. Get it done before you touch hardware. The Marlin M303 documentation covers the syntax for Marlin firmware; Klipper users run PID_CALIBRATE.
5. Reprint the tower at a different layer height. If you were at 0.20 mm, try 0.16 mm. If the band period changes proportionally to layer count (same screw rotations, fewer layers per band), you’ve confirmed it’s tied to the Z drivetrain. If the bands disappear, you were on a non-magic layer height for your leadscrew lead.
6. Reprint the tower at half speed. If banding fades or disappears, you’re chasing an extrusion-rate or resonance issue, not a leadscrew. If it’s identical at half speed, mechanical.
7. Swap filament spools. Try a known-good roll. If banding vanishes with the new filament, the old roll had diameter variation. Cheap insurance.
8. Roll the leadscrew(s) on flat glass. Power off, remove the screw, lay it on glass, roll it. Bent or not bent, you’ll know in thirty seconds.
9. Manually rotate Z with the printer powered down. Spin the leadscrew by hand and feel for binding or tight spots. Smooth motion the whole way is the goal. If you feel resistance at one Z position, either the screw or the nut or the linear guide is dirty or damaged at that height.

What input shaper, anti-backlash nuts, and Oldham couplers each actually do

Three pieces of advice come up in every forum thread on Z defects, and at least one of them is regularly given to people for whom it won’t do anything useful. Let me unpack each.

Input shaper does not fix Z-wobble. Klipper has Z-axis input shaping (shaper_freq_z, shaper_type_z), and it does a real thing: it smooths out the acceleration profile of Z moves to reduce ringing on Z-hops and quiet the machine down. It targets resonances in the 20-80 Hz range typical of XY axes. A Z leadscrew rotating once every couple of seconds during a normal print is operating far below that frequency band; the wobble it produces is not a resonance phenomenon, it’s a geometric one. Input shaper has nothing to grab onto. The Klipper resonance compensation documentation is clear on what shaping addresses, and a bent leadscrew isn’t on the list. My full OrcaSlicer input shaper guide covers what it does fix.

Anti-backlash nuts fix axial play, not bent screws. An anti-backlash nut is a two-part nut with a spring or split design that pre-loads the threads against the leadscrew, eliminating the small axial play that builds up as a regular POM nut wears. That play can show up as inconsistent Z height between layers, but it does not produce the smooth, periodic, lead-matched ripple of classic Z-wobble. If your leadscrew is bent, no nut in the world will compensate for the lateral motion of the screw. Anti-backlash nuts are also printer-specific; don’t drop a “universal” one in without checking it fits your assembly.

Oldham couplers do fix Z-wobble from a marginal leadscrew or motor misalignment. The Oldham coupler is genuinely the most effective single mechanical fix for Z-wobble that doesn’t involve a leadscrew replacement. It works because of its three-plate sliding design, which transmits rotation but allows lateral movement between the motor shaft and the leadscrew. Any wobble in the screw stays in the screw; it doesn’t get transmitted to the nut as a lateral force. If your leadscrew is mildly bent and you can’t or don’t want to replace it, an Oldham coupler is the most cost-effective workaround.

FAQ

How do I tell Z-wobble from Z-banding in thirty seconds?

Caliper. Measure peak-to-peak distance between bands. If it’s a clean multiple of your leadscrew lead (usually 2, 4, or 8 mm), it’s Z-wobble. If it’s irregular or doesn’t match any lead, it’s Z-banding. The eye alone isn’t reliable.

I’ve replaced the leadscrew and I still have banding. What now?

You almost certainly had Z-banding, not Z-wobble, and the leadscrew was the wrong fix. Run the diagnostic workflow above, starting with caliper measurement and PID autotune with the fan running. The vast majority of “I replaced everything and it didn’t help” cases are thermal in origin and resolve with a proper PID tune and a known-good filament spool.

Does layer height really matter, or is it a meme?

It matters more on printers with sloppy Z mechanics and less on printers with tight ones. The math is real: for an 8 mm lead screw, the full-step distance is 0.04 mm, and layer heights that aren’t multiples of 0.04 force the motor to microstep. The visible effect of that is small on a healthy printer with a good filament. It can be significant on a budget printer with marginal Z motion. If you’re already chasing banding, stick to 0.20 mm or 0.16 mm and remove one variable.

My Bambu X1C has banding and the leadscrew is fine. What’s the cause?

Almost always filament or PID. Bambu firmware tunes PID at the factory, but it can drift with time and especially with hotend or nozzle changes. Run a manual PID calibration with the fan on. Try a different filament brand. Inspect the heater cartridge and thermistor seating.

Can I run my Voron 2.4’s QGL more than once per print?

QGL is normally part of the START_PRINT macro, so it runs once per print. You can call it multiple times in succession when troubleshooting; some Voron owners run it three to five times consecutively when first commissioning a printer or after any mechanical change. For routine prints, once is enough on a healthy machine.

Is Z-banding worse with PETG than PLA?

It can be, because PETG has a wider sweet spot for temperature and a more aggressive viscosity curve in the print temperature range. A PID swing of ±3 °C produces a more visible band in PETG than in PLA. The fix isn’t to switch filament; it’s to tighten your PID tune with the fan running at the actual print speed.

Does my anti-backlash spring nut need maintenance?

If it’s POM (delrin), check the spring tension every six to twelve months of heavy use. The plastic compresses slightly over time. Most printer communities have a recommended replacement interval; for the Sovol SV06 line, two years of regular use is a reasonable rule of thumb. The full OrcaSlicer recalibration schedule has a maintenance interval table.

Why does my banding only show up on tall thin parts and not on wide flat ones?

Two reasons. First, tall thin parts have very short layer times per layer, which can trigger minimum-layer-time enforcement and hotend dwell. Second, tall thin parts make the gantry and frame work hardest at their position of maximum unsupported height, which amplifies any frame flex. Try printing the same tall part in the middle of the bed versus near a corner; if banding intensity changes, frame flex is contributing.

Closing thoughts and what to do next

Z-wobble and Z-banding share a visual family resemblance but live on opposite sides of the printer. One is metal: a screw, a coupler, a nut, a frame. The other is heat and pressure: PID, filament, fan, extrusion. The single most valuable habit you can build for diagnosing them is the caliper measurement, before any forum post, before any part order, before any setting tweak. Measure. Write the number down. Then decide.

If you’ve worked through this guide and you’re still stuck, the most useful next steps are usually one of three things. Print the TeachingTech test tower in a glossy filament and re-measure with a sharper read. Cross-reference what you’re seeing against the main troubleshooting hub, which links into every related defect article on the site. Or, if you’ve ruled out mechanical entirely and you’re stuck on thermal, walk through the temperature tower, under-extrusion, and over-extrusion guides in order to pin down the exact extrusion variable.

OrcaSlicer itself isn’t the cause of either defect, but it’s the lens you’re going to use to see them. Make sure you’re on a current build before you do real diagnostic work; older builds had a handful of known variable layer height edge cases that produced banding in their own right. Grab the current release straight from the OrcaSlicer GitHub releases page if you haven’t updated in a while. We don’t host the installer ourselves and we never will; the official source is the only one worth trusting.

And a closing favour. If you go through this whole process, fix your Z, and end up with a print that’s mirror-smooth where it used to look like corduroy, drop a note in the comments with what your final cause was. The pattern data from real readers is what makes the next version of this guide better than the last one. The articles that aged best on this site are the ones where the comments turned into a second guide underneath the first.

Related OrcaSlicer guides

• OrcaSlicer Troubleshooting: Master Index of Every Common Fix
• OrcaSlicer Print Time Estimate Wrong: Why and How to Fix
• Elephant Foot in OrcaSlicer: Compensation Done Right (2026 Guide)
• OrcaSlicer Settings Not Saving: Diagnose the Profile Reset Bug
• OrcaSlicer Tolerance & Fit Test: Functional Parts (2026 Guide)