Phase 2 · The filament · L10

The temperature tower

One print, seven temperatures, one number. The first measurement of Phase 2, and the one every other measurement is standing on.

The method is a temperature tower — a single tall print whose nozzle temperature drops by 5 °C every 10 mm. You get every candidate temperature on one object, printed with one spool, on one machine, in one session. That is the only way to compare them honestly: seven separate prints would differ by seven other things too.

1What you are actually looking for

Not "the best-looking band". The goal is the lowest temperature that still bonds layers properly, and those are different questions. Hotter almost always looks glossier and smoother — and strings more, sags more on bridges, curls more on overhangs, and takes longer to freeze, which caps how fast you can print later.

So you are searching downward from "obviously hot enough" until bonding fails, and then stepping back one band. The failure you are hunting for is invisible from the outside, which is why the tower gets destroyed afterwards.

The snap test is the whole lesson in one action. Break a band across the layer lines. A granular, ragged fracture that tears through the plastic means the layers fused into one solid. A clean, flat separation along a layer line means they didn't — they are stacked, not welded, and that part will fail in service no matter how good it looks.

2Build the tower

Any smooth-walled tall model with an overhang and a bridge will do. What matters is the G-code you wrap around it. With a layer height of 0.2 mm, 50 layers is exactly 10 mm — so a band change every 50 layers gives you 10 mm of each temperature.

Set the filament profile's nozzle temperature to the highest value in your sequence, then step down. Stepping down means the machine is never asked to heat up mid-print and wait, which would stall in place and leave a blob.

;AFTER_LAYER_CHANGE
;[layer_z]
{if layer_num == 50}M104 S245{endif}
{if layer_num == 100}M104 S240{endif}
{if layer_num == 150}M104 S235{endif}
{if layer_num == 200}M104 S230{endif}
{if layer_num == 250}M104 S225{endif}

M104 sets the target and returns immediately — it does not wait for the nozzle to get there.1 That is what you want (a wait would pause the print and scar the wall), but it has a consequence: the first two or three layers of each band are still on their way to the new temperature. Judge the middle of each band, never the boundary.

Open your machine. The command is the same; where you put it and what else is already there is not.

3Read it, then destroy it

• Photograph it first Raking light from one side, ruler alongside. You are about to break it, and the photo is the only thing that survives into the log.
• Mark the bands Pencil the temperature on each 10 mm section while you still know which is which. Every mis-recorded tower in history was mis-recorded at this step.
• Rank on bridges and overhangs Look at the underside of each bridge and the edge of each overhang. Sag and curl both get worse as temperature rises — find where they become acceptable.
• Snap every band Across the layers, with pliers if needed. Granular tear = bonded. Clean split along a layer = not bonded. Work from the coldest band upward and find the first one that tears rather than splits.
• Take that band, not the one above it The lowest temperature that passes the snap test is your answer. Then set the first layer 5 °C above it, because the first layer is fighting a cold bed and a cold sheet.

Write the winner into Filament Settings → Filament → Nozzle, both fields. Then log it — with the machine, the spool, the colour and the date, per Lesson 19. Colour matters: pigment loads change flow and thermal behaviour within the same brand and material, so "eSun PETG" is not a spool identity, and "eSun PETG, black, batch on the label" is.

4Find the fault

Click each. Every one of these produces a tower you can read confidently and wrongly.

5Check yourself

From memory. Question 3 reaches back to Phase 0.

§References

1. Prusa Research. Buddy firmware specific G-code commands. M104 is listed as "Set Extruder Temperature" with standard Marlin behaviour and no Buddy-specific parameters — it sets the target and returns without waiting. Link.
2. Prusa Research. Stringing and oozing. Stringing is caused by "very high printing temperatures and/or using incorrect retraction settings"; the recommendation to lower the nozzle by 5–10 °C; and the note that filament which "might have gathered moisture" will cause a lot of stringing. Link.
3. Prusa Research. Max volumetric speed. "With larger nozzles, consider increasing the nozzle temperature to allow for even bigger MVS" — the documented link between heat and achievable flow that explains why the MK4S ships hotter profiles than the MK3S. Link.
4. This course. Master tuning guide, §2.1 and §4 — the after-layer-change tower method, the judging priority order, and the per-material starting temperatures. Link.

Lesson L10 of the prusa-profile-tuning course · all lessons · The Calibration Card · glossary