You send a friend the file that has printed perfectly for you twenty times, and what comes off their machine is stringy, slightly undersized and split along a layer line. Or you buy a second printer to double your output, copy the profile across, and spend a week wondering what you broke. The same G-code file does not describe the same print on two machines, because most of what makes a print good was never in the file to begin with — it lives in the calibration of the printer that produced it.
This post contains affiliate links. If you purchase through these links, 3D Prototyping Hub may earn a small commission at no extra cost to you.
This post also contains Amazon affiliate links. As an Amazon Associate I earn from qualifying purchases.
Read the Difference Before You Change the Profile
The second machine is telling you which variable did not travel. Each symptom points somewhere specific.
| What the second machine does |
The variable that did not travel |
How to check it |
| Thin walls, gaps between perimeters |
Extrusion multiplier or e-steps |
Single-wall cube, measured with calipers |
| Bulges at corners, blobby seams |
Pressure or linear advance unset |
Run that machine's PA calibration pattern |
| Strings where the first machine was clean |
Real melt temperature and retraction |
Temperature tower on the new machine |
| First layer too thin, or squashed flat |
Z-offset, plate thickness, probe offset |
Re-level and set live Z from scratch |
| Ringing on identical geometry |
Belt tension, frame, acceleration limits |
Measure resonance, or print a ringing tower |
| Parts measure undersized in X and Y |
Flow plus hole and shrinkage compensation |
20mm cube on each, compared |
| Weak layer bonding, snaps easily |
Ambient or chamber temperature |
Same room, same day, repeat the print |
| Overhangs droop, supports fuse to the part |
Part-cooling fan output |
Watch both fans at 100% and compare |
Work down from whichever row matches. Changing the print profile — layer height, wall count, speeds — is almost always the wrong first move, because those are the settings that did transfer correctly.
A Profile Describes a Machine, Not a Part
It helps to split your settings into two piles, because most slicers already do.
Settings that describe the part you want travel fine: layer height, perimeter count, infill density and pattern, support style, top and bottom thickness, broad speed targets. If the part needs four walls on one printer it needs four walls on the other.
Settings that describe the hardware do not travel at all: extrusion multiplier, e-steps, the filament temperature that actually melts correctly in this hotend, retraction distance and speed, pressure or linear advance, jerk and acceleration limits, input-shaping frequencies, and Z-offset. Copy those across and you have imported one machine's compensations into another machine that needed different ones.
This is the same lesson as a new spool printing worse than the last one, moved one level up: there, a profile belonged to a spool rather than a printer. Here it belongs to a printer rather than a file. In both cases the file is innocent.
The Same 210°C Is Two Different Temperatures
Nothing on a 3D printer measures the plastic. The thermistor measures a point in the heater block near the nozzle, and the firmware holds that number at your setpoint. How much of it reaches the melt depends on where the thermistor sits, how well it is seated, the heater cartridge's output, the block's insulation and the airflow over it.
Two thermistors of the same part number are specified to agree only within a tolerance, and everything downstream of them adds more. The practical result is ordinary: two printers set to 210°C can be melting plastic at temperatures several degrees apart, which is enough to move stringing, layer bonding and surface finish visibly. If one machine strings at a setting where the other is clean, suspect the measurement before you suspect the filament.
The fix is not to make the numbers agree. It is to stop trusting the number and calibrate per machine: run a temperature tower on the new printer with the spool you intend to use, and read the result off the part. Where a machine has drifted badly or reads erratically, a thermistor and heater cartridge replacement kit is a cheap component swap, but do the tower first — a sensor that is consistently three degrees out is not broken, it is just a different sensor.
A 0.4mm Nozzle Is Not 0.4mm
Nozzles are manufactured to a tolerance and they wear in service. Brass is soft, and abrasive filament — carbon fibre, glow-in-the-dark, glitter, wood fill — opens the bore measurably over a few hundred hours. A printer that has run a kilogram of carbon-filled nylon has a wider, rounder orifice than the one still on its factory nozzle, and the slicer on both still thinks it is extruding 0.4mm lines.
The signature is a machine that over-extrudes slightly everywhere, with soft detail and mushy text, while the flow number in the profile is identical. Fit a hardened steel nozzle on anything that sees abrasive material and the wear stops being a variable — but note that hardened steel conducts heat less readily than brass, so the two machines will want slightly different temperatures afterwards rather than the same one. That trade, and which bore size suits which job, is laid out in our guide to 3D printer nozzles. What nozzle wear does to a single machine over months is the subject of why your printer prints worse than it used to; between two machines it simply arrives all at once.
Flow Is the Number That Transfers Worst
Extrusion multiplier is where most of the visible difference lives, and it is the easiest thing to settle definitively.
Print a single-wall calibration cube on each machine with identical settings and the same spool, then measure the wall at several heights with digital calipers. Adjust each machine's multiplier until its own wall hits the target width. Do not tune one printer to match the other's measurement — if the first machine was the one over-extruding, you have just propagated its error.
Two things commonly sit underneath a flow difference. The extruder's drive gear ratio differs between machines, so the correct e-steps figure differs too, and a geared extruder that slips under load delivers less than it reports. And filament diameter varies within and between batches, which is why both tests want the same spool. If the new machine is starving rather than merely running lean — clicking, gaps in the top surface, hollow walls — that is a different fault, covered in why 3D prints under-extrude.
The First Layer Belongs Entirely to the Machine
Z-offset is the least transferable setting on a printer and the one that changes the result most.
It is not one number. It is the probe's physical offset from the nozzle, the thickness of the build plate, the compressibility of the magnetic base beneath it, the shape of the bed mesh, and the nozzle tip's own height after a tap with a scraper. Swapping a textured plate for a smooth one changes both the effective squish and the finish of the bottom face. Two printers with the same firmware and the same plate model can want offsets that differ by a tenth of a millimetre or more.
So level and set live Z on the new machine from scratch, with a feeler gauge set rather than by paper feel, which varies with whoever is pulling the paper. If you want two machines to produce parts with the same bottom surface, run the same surface on both — a matched pair of textured PEI spring steel plates removes one variable permanently. The broader diagnosis of first-layer problems is in why 3D prints don't stick to the bed.
Motion: Same Firmware, Different Frames
Two printers running identical acceleration limits do not accelerate identically. Belt tension as assembled, frame squareness, bolt torque, how much mass the toolhead carries and what the bench is made of all set where the machine resonates. One prints a clean 25mm test cube at 150mm/s; the other paints ghosting across every corner.
Check belt tension on both — they should sound and feel the same, and a GT2 belt tensioner kit makes that a setting rather than a wrestle. Then, on firmware that supports input shaping, measure each machine's resonance with an ADXL345 accelerometer and give each its own frequencies. Copying one machine's shaper values to another is the motion-system equivalent of copying its Z-offset. The symptom and its other causes are covered in why prints have ringing and ghosting.
The Room Is Part of the Printer
If one machine lives on a desk and the other in a garage or under a window, you are not comparing two printers. Ambient temperature changes layer bonding, warping and how fast overhangs set, and an enclosure changes all three again. A part that delaminates on the second machine and not the first, with every setting matched, is usually a draught.
Put both machines in the same conditions before you conclude anything about them. And if the plan is to run one of them in a cold or variable space, an enclosed machine makes the environment a controlled variable rather than a weather report — Qidi's enclosed machines are one route to that, and a well-sealed aftermarket enclosure around an open-frame printer is another.
A Twenty-Minute Qualification for a Second Machine
Do this once per new printer and most of the mismatch disappears before you ever notice it.
- Level and set Z from scratch, on the plate you intend to use. Do not import an offset.
- Single-wall cube, measured. Set that machine's extrusion multiplier from its own reading.
- Temperature tower, same spool as the first machine. Read the result off the part.
- Retraction test. Distance and speed depend on direct versus Bowden and on the hotend, not on the file.
- Pressure or linear advance pattern, if the firmware supports it.
- Belt tension, then resonance. Give the machine its own shaper values and acceleration ceiling.
- Print the same 20mm cube on both and measure all three axes. Keep the numbers.
That last step is the one worth repeating monthly. It is a two-minute print that tells you a machine has drifted before a real part does.
When Two Machines Should Not Be Expected to Match
Some differences are structural rather than calibration. A bed-slinger and a CoreXY do not behave the same at speed with tall parts. A Bowden setup will never retract like a direct drive. An enclosed machine running ABS and an open one running ABS are doing different things entirely. If you want matched output, the shortest route is a second unit of the model you already own and have calibrated — and if you are buying one, ELEGOO's US store and Anycubic's US store both publish motion system, enclosure and nozzle specifications per model, which are the figures that decide whether two machines can be made to agree.
There is also a point where the arithmetic stops favouring you. If parts from both printers have to be interchangeable — assembling into the same fixture, holding the same fit — you are doing process control, not troubleshooting: calibration per machine, a qualification print, and a dimensional check on every batch. That is worth building for a part you print every week. For thirty matched parts once, it is not. The tolerance question underneath it is set out in tolerances and accuracy in 3D printing, and the assembly-side symptom in why 3D printed parts don't fit together. When the answer is to hand it over, the shops in our provider directory run calibrated fleets and quote per part, which is the version of this problem somebody else has already solved.
Two printers are two machines. Treat the second one as new, measure it against a target rather than against the first, and the file will stop being the suspect.
Hero photograph by Minku Kang on Unsplash.
