PET Bottle Recycler – Filament Extruder

The aim is to build a PET filament extruder that can process plastic from discarded bottles and double as a test bench for a RAMPS 1.4–based, multi-hotend setup. Bottles are cut into a continuous strip, guided into a two-stage extrusion section built from off-the-shelf 3D printer hotends, and pushed through a 1.75 mm die. A passive spool on the output side handles filament storage, with tension set mechanically. Control is handled by a RAMPS 1.4 board and its stock LCD, running a stripped-down firmware configured for extrusion-only operation instead of full 3D printing.

The hardware layout follows an existing open-source design rather than being fully original, and this build is intentionally treated as an experiment: a way to see what works, what doesn’t, and what to change before designing my own recycler from scratch. All original model creators are credited in the project files and documentation.

Mechanically, the extruder is kept as simple as possible: a small aluminum frame holds the bottle cutter, feed screw, extrusion section, and spool in a straight line. The main challenge is keeping the feed path smooth so the PET strip doesn’t snag or twist before it reaches the hotends.

Electronics are based on a RAMPS 1.4 stack, which provides heater control, temperature sensing, and stepper driving. The role of a normal 3D printer is reduced to a single axis of motion: the feed screw, plus heater and fan control.

Once this setup is properly understood, there are a few clear hardware upgrades worth trying. On the process side, the first step is to split cutting and extrusion into two separate stages: process full bottles into consistent PET strips first, store them, and only then feed those strips through the extruder. This should make the system more predictable than going directly from bottle to filament. On the hardware side, I want to experiment with different heater blocks and cartridges, better thermal isolation around the hotends, and a redesigned spooling system that keeps tension steady without too much tweaking.

PET is much fussier to work with than typical 3D printer filaments. It runs hotter, absorbs moisture easily, and reacts differently depending on the bottle it came from. Wall thickness turned out to be a real problem: some bottles are thin and soft, others are thick and stiff, and mixing them in one batch causes the required torque and flow rate to jump around. The closer you want to get to a clean 1.75 mm filament, the more important it is to stick to one or two bottle types with similar wall thickness.

One experiment that helped was heating the bottles with a heat gun before cutting and blowing compressed air inside them. That expands the bottle and thins the walls a bit, making the resulting strip more uniform and easier to feed. It also makes it possible to play with different wall thicknesses and see how each behaves through the same hotend setup without immediately choking the system.

Getting consistent filament diameter is still all about balancing temperature, feed rate, and spool tension. Too much heat or pull force and the filament necks down; too little and it comes out under-melted or lumpy. The second heating zone and a dedicated output cooling fan helped smooth out the melt, but only after a lot of small tweaks. I also started experimenting with insulating the heater blocks to see whether a more stable temperature profile along the hotend stack will improve melt behaviour and diameter stability.

Mechanically, even small misalignments in the bottle cutter or feed path showed up later as diameter ripple, so keeping the strip guide simple and straight mattered more than any fancy mechanics. One other lesson is that planning the wiring and control on something like RAMPS is worth the time, even for a small build. Having clear heater channels, thermistor mapping, and fan outputs documented made it easier to experiment without losing track of what was connected where.

The build reuses a lot of standard 3D printer hardware rather than starting from raw stock. The frame is based on 2020/2040 aluminum extrusions with V-wheels riding in the slots, held together with a mix of M3, M4, and M5 screws and 2020-series T-nuts. A generic screw kit (Hilitchi-style button-head assortment) covers most fasteners, with extra M4×8 mm screws and T-nuts for mounting brackets and guides.

The hot section uses two complete 3D printer hotend assemblies as separate heating zones. RAMPS 1.4 drives the first hotend on the main hotend output and the second on the heated-bed channel, each with its own 40 W cartridge heater and 100 K thermistor. Two 12 V fans cool the hotends, and a third fan is aimed at the filament right after it leaves the die.

A single NEMA 17 stepper motor drives the feed screw that pushes the PET strip through the extrusion section. The filament then winds onto a simple free-spinning spool with printed guides and adjustable friction. Most custom parts—the bottle-cutting jig, motor mounts, guides, and small brackets—are 3D printed. Off-the-shelf POM V-wheels can be used on the extrusion frame, but they can also be printed in TPU using the supplied files if needed.

This extruder started life as a remix of an existing PET pultrusion design for the Ender 3 platform and then turned into a dedicated test rig. The mechanical layout is heavily based on the Recreator 3D MK5Kit Ender3 Pultrusion Unit, with small changes to fit my parts, goals, and the RAMPS 1.4 electronics. All original model creators for the printed and metal parts are credited in the project documentation, and any downloadable files for this build clearly reference their work.

On the electronics and firmware side, RAMPS 1.4 documentation and the Marlin source code were the main references for reusing the hotend and heated-bed outputs as two independent heater channels. General community resources on PET filament recycling, drying, and extrusion helped with picking starting temperatures and feed rates.

Most of the custom parts were modeled in CAD and printed. Fusion 360 was used to edit metal parts, lay out brackets, and design the 3D printed components, while simple online STL box and panel generators helped with quick enclosures. For planning the wiring and connectors around the RAMPS board, a small schematic in an EDA tool like EasyEDA was enough to block out power rails, heater channels, thermistors, fans, and motor wiring.