Your BLTouch probes 25 points and builds a mesh. Your first layer still has thin spots on the left and ridges on the right. The probe is doing its job — it’s measuring the bed surface accurately. But the mesh isn’t being applied correctly, or your bed changes shape when it heats up, or your Z-offset is wrong and the mesh is compensating for a problem it wasn’t designed to solve.
What Bed Mesh Leveling Actually Does
Bed mesh leveling measures the height of your print surface at multiple points and builds a mathematical grid of offsets. During printing, the firmware adjusts the Z-axis in real time to follow the contours of your bed. A flat bed should produce a near-flat mesh. A warped bed produces a mesh with hills and valleys. The mesh compensates for the latter but cannot fix the former — if your bed has a 0.3mm hump in the center, the mesh will force the nozzle to follow that hump, and your bottom surface will have a 0.3mm curve even though the first layer adhered perfectly.
The goal is twofold: get the bed as mechanically flat as possible first, then use the mesh to compensate for whatever warping remains. Skipping the mechanical step and relying entirely on the mesh is the most common cause of “my ABL probe isn’t working” complaints.
Step 1: Mechanically level the bed — properly.
Heat the bed to your printing temperature. Aluminum expands when heated. A bed that’s mechanically level at 20°C can be 0.1-0.2mm out at 60°C. Always level hot. Use the paper method at all four corners plus the center. Go around at least twice — adjusting one corner changes the others. The goal is to get all five points within 0.05mm of each other. A feeler gauge is more accurate than paper, but paper is good enough if you’re consistent with the drag feel.
Step 2: Configure your mesh grid density.
More probe points is not always better. A 3×3 grid (9 points) misses the saddle-shaped warp common on Ender 3 beds. A 5×5 grid (25 points) catches the warp but adds 2 minutes to every print start. A 7×7 grid (49 points) is overkill for beds under 300×300mm — the probe precision (±0.005mm for BLTouch) is finer than the bed’s surface texture at that density. For most printers, 5×5 is the sweet spot. For beds over 300mm, go 7×7.
Step 3: Set the correct fade height.
Fade height tells the firmware to gradually phase out the mesh compensation over the first N layers. By layer 10, the print should be flat on top regardless of the bed’s shape underneath. The default fade of 10 layers works for most prints. If you’re printing a tall, narrow part and the bottom surface needs to be dead flat, disable fade entirely. If you’re printing a wide, thin part that will warp if the mesh compensation stops abruptly, increase fade to 20 layers for a gentler transition.
Step 4: Visualize your mesh.
A bed mesh visualizer (OctoPrint plugin, Klipper’s heightmap, or Marlin’s M420 V) shows you the actual shape of your bed. A typical Ender 3 bed with the stock aluminum plate shows a “taco” shape — low in the center, high at the edges, with a variance of 0.1-0.3mm. If your mesh shows a 0.5mm+ variance, the bed is mechanically warped beyond what mesh compensation can cleanly handle. Replace the build plate or shim the low spots with aluminum foil tape.
Bed Leveling Method Comparison
| Method | Accuracy | Setup Time | Runtime per Print | Best For |
|---|---|---|---|---|
| Manual paper leveling (no mesh) | ±0.05mm if flat bed | 2 min (once) | 0 sec | Printers with dead-flat beds (cast aluminum, glass) |
| Manual mesh (no probe) | ±0.03mm with practice | 5 min (once) | 0 sec | Budget printers, probe-less setups |
| BLTouch/CR Touch (3×3) | ±0.01mm probe + mesh | 15 min install, 2 min/session | 45 sec | Quick validation, flat-ish beds |
| BLTouch/CR Touch (5×5) | ±0.01mm probe + mesh | 15 min install, 2 min/session | 90 sec | Most beds under 300mm — recommended |
| BLTouch/CR Touch (7×7) | ±0.01mm probe + mesh | 15 min install, 2 min/session | 160 sec | Beds over 300mm, severely warped beds |
| Klipper adaptive mesh | ±0.01mm probe + mesh | 10 min config | 30-60 sec (only probes print area) | Klipper users — fastest + most accurate |
Common Mistakes and How to Avoid Them
Mistake 1: Running the mesh at room temperature and printing at 60°C. Your bed grows by 0.1-0.2mm when heated. The mesh captured a room-temperature surface and the firmware is applying corrections for a shape that no longer exists. Always run G29 (or Klipper’s BED_MESH_CALIBRATE) after the bed has been at printing temperature for at least 5 minutes.
Mistake 2: Setting Z-offset before running the mesh, or running the mesh before setting Z-offset, and not knowing which order is correct. The answer: set a rough Z-offset with paper, run the mesh, then fine-tune Z-offset during a live first layer print. The mesh compensates for bed shape; Z-offset sets the absolute nozzle height. They’re independent but you need both to be correct. A perfect mesh with a 0.1mm-off Z-offset still produces a bad first layer.
Mistake 3: Not saving the mesh to EEPROM after probing. Marlin’s G29 probes and activates the mesh for the current print. But if you don’t follow it with M500 (save to EEPROM) and add M420 S1 (load saved mesh) to your start G-code, the next print starts with a blank mesh. You’ll re-probe every time, adding 2 minutes to every print for no reason. A good start G-code: G28 ; home all axes → M420 S1 ; load saved mesh → G29 ; optional: probe fresh mesh. Enable RESTORE_LEVELING_AFTER_G28 in firmware to avoid having to re-enable the mesh after homing.
Mistake 4: Messing with the mesh values manually to “fix” first layer problems. If your mesh says the back-left corner is 0.2mm high, and you manually edit that value to 0.1mm because the first layer looks thin there, you’re treating the symptom. The probe measured correctly. The bed probably shifted mechanically since the last mesh was taken. Re-level mechanically, re-probe, and trust the numbers. Manual mesh editing is how you create a frankenmesh that’s wrong in subtle, hard-to-diagnose ways.
Mistake 5: Running ABL on a loose bed. If your bed has play in the Y-axis V-slot rollers, the probe measures one position but the bed shifts 0.05mm between the probe point and the first layer. Tighten the eccentric nuts until the bed has zero wobble but still moves smoothly. The mesh is only as accurate as the mechanical platform it’s measuring.
⚠️ Safety Notice: Bed leveling procedures involve moving parts and heated surfaces. Always perform mechanical adjustments with the printer powered off or motors disabled. When running live Z-offset tuning, keep your hands clear of the nozzle and heated bed. A BLTouch or CR Touch probe uses a physical pin that extends below the nozzle during probing — ensure the pin deploys and retracts freely and that the probe is mounted securely. The bed surface should be clean and free of debris that could interfere with probe accuracy or cause adhesion failures. As with all 3D printer operations, never leave the printer unattended during initial calibration prints.
Your BLTouch is only as good as its installation. Our BLTouch and CR Touch installation guide covers firmware configuration, wiring, and probe accuracy testing. For the mechanical bed leveling that must happen before any mesh, see our bed adhesion fixes guide — a level bed starts with a clean, properly prepared surface.
A CR Touch or BLTouch probe is the single best upgrade for print consistency. Available at uavmodel with printer-specific mounting brackets and pre-configured firmware for Ender 3, Ender 5, CR-10, and Artillery printers — ships with the correct pinout cable so you’re not guessing which wire goes where.