I had bounced off CAD twice before. Fusion 360 wanted a cloud account I did not want to give. SolidWorks wanted a Windows VM and a budget. OnShape wanted everything I made to live on someone else's server. Every time I tried to learn CAD as an abstract skill I gave up after a week because nothing I drew had a reason to exist.

Two things changed it. The first was a project: I wanted to build my own hobby rockets, the kind that fly on motors I cast myself, and that meant designing parts that nobody sells. The second was a YouTube channel that finally taught FreeCAD without making it look harder than it is.

This is the post about that. The CAD half. The rocketry half is a hobby with its own established communities, certification, and safety practices. Anyone interested in the propellant chemistry side should read the standard amateur references rather than a blog like this one.

The channel that made FreeCAD click#

@deltahedra3D is a small YouTube channel that teaches modern FreeCAD on the 1.x branch. The 1.x release is the one that finally cleans up the parts of FreeCAD that everyone complained about: the topological naming problem is mostly solved by the Realthunder branch landing upstream, the Part Design workflow is consistent, the sketcher is genuinely good.

What the channel does right is teach you the workflow, not the buttons. Every tutorial is a real part. You watch someone build a nosecone or a coupling or a fin can, you see how they make a sketch, attach a feature, name a reference geometry, and you absorb the mental model rather than memorising menu paths.

The thing I had been missing in every other tutorial: there is one correct way to model in Part Design, and once you have it, every other operation falls into place. Sketches go on planes. Planes go on bodies. Features attach to features. References are named. Parametric updates work when you do this. They explode when you do not.

After three of those videos I could model parts I had been failing to model for months in other tools. The mental model is the whole game.

What I was actually drawing#

The rocketry side gave me a parts list that no off-the-shelf vendor sold for what I wanted to fly. Specifically:

  • Nosecones: ogive and elliptical, sized to the body diameter I was using, with shoulder lengths that matched my tube wall thickness.
  • Body tubes and centring rings: I bought tubes, I drew the rings. The motor mount tube sits concentric to the body tube, the rings hold them in place, the fins slot through the body tube and bond to the motor mount.
  • Fin sections: trapezoidal fins with a specific airfoil cross-section. Calculated the stability margin in OpenRocket, then drew the actual cuts in FreeCAD to match.
  • Recovery bays: bulkheads, eyebolts, shock-cord anchors. The bulkhead is the boring part of the rocket and the part that holds the whole thing together when the parachute deploys.
  • Motor casings and end closures: this is where the CAD work paid off the most, because the motor is the part nobody sells in the diameters I was experimenting with.

Each of these is the same FreeCAD pattern: a Body, a starting sketch, a Pad, attached features for cuts and fillets, references named cleanly so the part updates when I change a parameter. The first nosecone took me four hours. The fifth one took twenty minutes. The geometry was the easy half. The discipline of building a tree that updates correctly was the hard half, and learning it on a rocket was the right call because every part has a clear purpose.

What rocketry the hobby actually looks like#

Amateur rocketry in 2026 has a serious infrastructure around it. Three organisations are worth knowing about:

  • NAR (National Association of Rocketry) certifies hobby fliers for High Power Rocketry in three tiers. L1 lets you fly motors up to a certain impulse class, L2 doubles that, L3 lets you fly the biggest stuff hobby flyers fly.
  • Tripoli has its own certification track and is the body more associated with experimental motors.
  • Local clubs run launch sites with FAA waivers, range safety officers, and a culture that takes safety extremely seriously. Showing up at a club launch with no clue is the right way to start. The community is friendly and the safety conversations are mandatory before you launch anything substantial.

I had a stack of FreeCAD-designed parts before I had a single certification. That is the wrong order. I worked on the certification side after the CAD side caught up, and the safety mentality you absorb from being at launches changes how you design from then on. Centring rings get fillets so the model rocket motor cannot bind. Bulkheads get tested to load before they go into a rocket that flies. The hobby is not about being clever. It is about being thorough.

The propellant side of amateur rocketry is a topic with established literature and decades of community practice. People mix their own motors for serious technical reasons: scaling beyond what commercial vendors sell, matching impulse profiles to specific airframes, learning the discipline. It is a legal hobby in most jurisdictions when done through the established channels (NAR L2 or Tripoli, range with appropriate clearances, mentor relationship). I will not write about the specific compositions or procedures here. The Richard Nakka site, Sutton's "Rocket Propulsion Elements", and the propellant resources at Apogee Components are the canonical starting points for anyone who wants that knowledge. Earning it through reading those sources is part of the safety culture. Skipping it by reading a blog post would defeat the point.

What FreeCAD taught me that the rockets did not#

Designing parts that have to fit together is a different skill than drawing pretty things. The fin slots in the body tube have to match the fin section's root chord, with a fit that allows a thin glue layer. The motor mount tube has to fit inside the body tube with the centring rings tolerancing the gap. The threaded end closures of the motor casing have to actually screw together.

That is tolerance design, and FreeCAD's parametric tree is the right tool for it. Every fit is a parameter. The body tube ID is one value. The fin root chord is body_tube_id - 2 * wall_thickness - fin_thickness. Change the body tube and everything updates. Get the formula wrong once, see the failure, fix the parameter, the whole assembly stays consistent.

I had no idea how badly I needed this until I had it. A real parametric tree is what makes "iterate" mean something other than "redraw the whole thing". Every other CAD package I had touched promised parametric and delivered "you can change values but only if you remember in what order you made every feature". FreeCAD 1.x on the Realthunder-derived branch actually delivers.

Where the projects live now#

I have four flying rockets in various states of preparation, two motor casings drawn and ready to print in PA-CF, and a stack of design files I keep iterating on. The launches happen at the local club site under their range protocols.

The FreeCAD muscle is permanent now. I use the same parametric discipline for printer parts, for the camera mount that holds the K2 Plus webcam, for the cable channels I design for projects unrelated to rockets. The project chose the tool. The tool stayed for everything else.

If you have bounced off CAD too, find a project where every part has a reason to exist, watch @deltahedra3D for an afternoon, and try modelling that project from scratch in FreeCAD 1.x. The third part you draw will be the one where it clicks. The first two are tuition.