CentrifugalDustSeparator: print files for a workshop dust separator

A repository of print files, not of code

The first thing to understand is what kind of project this is. The repository description is three dimensional files for printing, the primary language is listed as unknown because there is no code, and the top level of the repository is a set of directories, one per build, plus a folder for an anemometer jig and a folder for T-Loc parts. What you get are CAD models and slicer files, not an application. The README describes the Centrifugal Dust Separator as the second version of the design, marked as a beta release, and it points to the author's YouTube channel for build videos and a test video, with the first and second parts of the build video linked separately. There is also an original build section, which credits a Pilson Guitars build project and links its GitHub repository, so the lineage is a community design that was iterated by this author rather than something designed from nothing. For a reader used to software repositories, the interesting part is that the engineering record lives in the changelog rather than in commit messages, because the changelog here is written in a first person narrative about what broke during printing and what changed as a result.

Why there are four build folders

The four directories exist for a simple reason that the README states up front: tube suppliers may vary based on your geographic location, and so do the tube dimensions, so the files are adjusted accordingly. The base MK2 version is designed for six inch clear PVC, and the parts and supplies section links to a six inch clear PVC tube. The European build, MK2_EU, is designed around an acrylic tube of 150 by 3 millimetres, where the real size is 151 by 3.5 millimetres, and the United Kingdom build, MK2_UK, is designed around a 150 by 3 millimetre acrylic tube whose real size is 150.05 by 3.1 millimetres. Those last two decimals are the whole point, because a scaled model for a tube that is three and a half millimetres thick where the model assumed three will not seal. A fourth folder, MK2_EU_200mm, holds the step files for a single 200 millimetre tube version with third stage integration, added in the changelog entry for the sixteenth of August 2025. The README also says that for parts which are not listed in the subfolders, such as the Y-pieces, you should use the normal MK2 version, so the folders are not duplicates of the same set.

Print the scaling ring first, not the impeller

The single most useful piece of advice in the README is also the most unusual. It says to print the scaling ring first, to check your tube dimensions, and links a short video showing the check. That single part turns the whole regional folder question from a guess into a measurement, because a ring printed to the dimensions of one build tells you immediately whether the tube you can actually buy matches that build or whether you need a different folder. The rest of the printing constraints are stated just as plainly. All parts are designed to print on the Bambu Lab build volume of 256 by 256 by 256 millimetres cubed, and only the Y-pieces are not compatible with the Prusa build plate, which is an important exception if your printer is a Prusa, since the Y-piece is a structural part of the assembly rather than a decorative one. The README also notes that the scaling ring needs to be checked before committing to the rest of the print, and the assembly adds hardware, since the parts and supplies list includes o-rings sized for each build, a collection container, compression tape and a printed T-Loc.

Slicer files, a mirrored impeller and a wall loop direction

The repository carries a wide spread of file formats, which tells you what the author wanted people to be able to do. The step files are the master geometry, and alongside them there are STL files, 3mf files, DXF and SVG files for the flat parts, and a Fusion archive in f3d form for the baseplate and the cleat, so that the design can be opened and edited in a parametric modeller rather than only printed. There are Bambu Lab and OrcaSlicer files, which is a deliberate inclusion rather than an accident, because the slicer configuration is part of the design. The changelog shows why. A mirrored impeller was added, and a community member pointed out that the print imperfections on it were a wall loop direction problem, so the recommendation is to use the open source OrcaSlicer and set the wall loop direction to clockwise. A later entry records that the impeller and the mirrored impeller had their non-manifold edges error fixed. A separate entry records that the REV2 Y-pieces were redesigned to need less filament by removing the need for support, and a later one adds Prusa printable Y-pieces, which is what removed the Prusa exception for that part.

The bill of materials is a Festool container and some tubing

The parts and supplies section is where the design stops being a set of files and becomes a shopping list, and it is worth reading in full before printing anything. The base MK2 version is designed for six inch clear PVC tube, with an EU alternative in 150 by 3 millimetre acrylic and a UK alternative in the same nominal size. The o-rings differ by build, with a 140 by 3 for the MK2 version and a 120 by 3 for the EU version, and the changelog explains the reason for that switch, since the EU inlet and outlet parts were adjusted for a better o-ring seal and fit with the smaller ring. The collection container is a Festool VAB-20/1, part number 204294, with the note that the author received 204295 because his was a pack of three, and the US price and a free shipping source are listed. The seal is made from compression tape, 50 millimetres by 30 metres by 3 millimetres, and the Sytstem box connection is a 3D printable T-Loc, with the original model linked from a printing site. There is also a folder for a BT-100 anemometer jig, and the changelog notes that the latch and latch cover for the T-Loc were not printed by the author, who provides the step files used for the design and invites people to try printing them.

The changelog is where the real engineering is

The changelog runs from the twenty first of July 2024 to the sixteenth of August 2025 and reads like a build diary, and a few entries are worth quoting for the engineering content they contain. The twenty eighth of September 2024 entry records a failure mode that would be expensive to rediscover: a normal nut will not lock the impeller assembly for the long run, because PLA will shrink and compress over time and the nut loosens, so the fix is a spring washer and a lock nut. The twenty ninth of July 2024 entry records the printing fix for the mirrored impeller with the OrcaSlicer wall loop direction, and the first of August records that all the 3mf and STL files became available and that the non-manifold edges error on the impeller and mirrored impeller was fixed. The twenty fourth of August 2024 entry records the EU inlet and outlet being adjusted for a better o-ring fit, and the nineteenth of August records the switch to 120 by 3 millimetre o-rings with an honest note that the parts still needed changing for a better fit and that test prints would follow. Two entries carry a warning that matters for anyone choosing a build, since the UK step files and the EU files were both added with the note that they were not printed or tested yet. That kind of candour is more useful to a builder than a polished claim of readiness, and the whole set is described as a beta release.

Where this fits, and what to check before you start

This is a design for a maker with a printer, a shop vacuum and a tolerance for assembly work, and it is worth being clear about that. You are not buying a product. You are downloading a set of models, printing them, buying a tube, an o-ring set, a Festool collection container and compression tape, and building something that has to seal well enough to separate dust from the airflow. The realistic risks are geometric and print related rather than software related. Clear tubes have real wall thickness that differs from the nominal dimension, which is why the scaling ring exists and why there are three regional builds. Printed parts shrink and creep, which is what the nut and washer note is about. And a mirrored part needs a specific slicer setting to print cleanly. The repository's own Beta label and the untested notes on two of the regional builds are the honest way to read the maturity of the set. The MIT licence covers the files, so you can print, modify and redistribute the parts freely, and the credit chain runs through the Pilson Guitars original project and the author's YouTube build videos, which are the place to see the assembly order rather than inferring it from the folders.

Editorial conclusion

Adopt these files if you build your own cyclone separator from printed parts and a shop vacuum and you want a design that has already been printed, iterated and fixed by someone else, and if you are willing to work from the four build folders rather than one universal set. It does not fit you if you expect a turnkey product, because the parts list includes tubes, o-rings, a Festool collection container and a Sytstem box, and the assembly depends on prints you make yourself. Before you start, print the scaling ring named in the README to check your tube dimensions, and pick the folder that matches your supplier, since the EU and UK builds use 150 millimetre acrylic with slightly different real dimensions and the USA build is designed around six inch clear PVC.

Frequently asked questions

Official sources

1. Issues
2. License: MIT
3. Makerr-Studio/CentrifugalDustSeparator- on GitHub
4. README