Core A Design

Core A was developed in 2021 as the first removable Core subassembly. It uses a high-performance "launch tube" configuration, where a PVC pipe inserted into the bottle provides a performance boost. It requires specialized tools and use of PVC cement. Core A was deprecated in 2023 in favor of Core C, which is far simpler to build.

Design Goals

This discussion highlights initial challenges from the early adaptation of 3D-printed parts for pressurized systems and high-stress functional components. 

For context, the starting point of Core A's design process were the initial Alpha and Beta prototypes, which used multiple PVC fittings and specialized brass valve adapters. These Cores were not removable from the Base, so when one or the other failed, the entire launcher had to be thrown out.

The problem of removability in Beta cores

Beta Cores (pictured) were permanently fixed by cemented PVC pipe fittings required to interface to brass air valves with NPT threads (pictured). It was definitely a kludge, and one that cost about $3-5 for the difficult to find parts. To make Core A removable, I first had to find a way to interface to air valves that did not protrude from the pipe profile.

Several existing designs use a PVC end cap with a hole drilled in it for an automobile tire valve, but this still required having a specialized fitting and access to a power drill. It would also protrude radially, even if using hard-to-find internal end caps.

A version of this idea could work, though; interfacing a tire valve with the inside of the PVC pipe and somehow holding it in with a 3D-printed part. The new assembly would have to hold pressure both in the macro sense (prevent the valve from popping out) and micro sense (an airtight interface).

The tire valve stem

Millions of tire valve stems are made each year. The pictured TR414 tire valve can be found in car and light truck tires everywhere and costs $0.25 USD in bulk. The brass end is a standard Schrader valve which is also found on mountain-style bicycles. Several sizes are available; all Cores use the longer TR414.

A happy serendipity allows for an airtight seal against the smooth pipe bore. The small bumped ring measuring 16.0mm fits perfectly inside ASTM Schedule 40 1/2" nominal pipes, which measures 15.6-8mm.

Core A components

A 3D-printed Valve Holder retains the tire valve stem in place, and a cylindrical backstop prevents it from sliding too far into the lower pipe. This all fits within the outer diameter of the PVC pipe, finally allowing removability.

Shown here are:

Core A Center geometry

With the interface to the air pump solved for now, we move onto designing the Core A Center. This part interfaces to the bottle and Base and transfers forces between them. It must have air passages for pressurization, be strong against leveraged shear forces from loading rockets on the upper tube, seat an O-ring, and transfer the force of the launch to the Base.

Pictured are CAD models with cutaways, showing air passages, interface areas for PVC pipes, and a shallow groove for the O-ring.

The O-ring groove was challenging to size. Too much O-ring protrusion caused binding at launch; not enough resulted in pressure leakage. Using a thicker O-ring helped with holding pressure but weakened the printed part. Requirements for standardization with later Core designs further constrained options.

Core A Center with compression bolt

Early testing showed that the Core Center was subjected to brutal forces beyond the reliable capability of 3D prints. The failure mode pictured was common: the top of the Core would shear off at the O-ring groove.

Inspiration came from reinforced concrete. Like PLA layers, concrete is strong to compressive forces and weak to bending and shear forces. Reinforced concrete uses steel rebar under tension to keep from cracking under bending and shearing. A stainless steel bolt provided this tension.

Shown is an early Core A Center prototype using a 10-24 bolt. Final versions use a larger 1/4″-20 bolt with a nylon-inlay locknut to prevent backing off.

This eliminated shearing failures entirely and increased options for O-ring sizes.

Use of PVC cement to bond PVC to PLA

The last issue was bonding the parts together. These joints needed to both stay together and not leak. Here, we run into an material property of common FDM filaments: they do not bond well with common adhesives.

My searches online and testing have not found any readily-available, not extremely toxic solvents that do so. Other alternatives like epoxy and cyanoacrylate glue (superglue) have their own limitations such as brittleness and it’s unclear if they’re bonding that well either. Printing parts in ABS opens up bonding options, but ABS printing releases toxic fumes and therefore requires special printer setups.

What I was already using, and was immediately available, was PVC cement.  For context, especially you are not a plumber or material scientist (I am neither), it is important to understand how PVC cement – and by extension plastic solvent welding – works. The cement actually chemically dissolves the PVC polymer at the molecular level. When you join two pieces of PVC pipe together with cement, the pipes are dissolved and then held in place until the solvent migrates away and the pipes fuse together at the molecular level in an airtight seal in what’s called a solvent weld.

Unfortunately, neither PLA nor PETG dissolve in PVC cement, and therefore do not solvent weld. They do not form an airtight seal because they do not bond on a molecular level.

However, what saves the design is that the very porosity of 3D prints that makes them frustratingly not air-tight. All those nooks and crannies wick in the PVC and PVC cement via capillary action and form a tight interference fit. In the Launch Tube assembly process, the primer and cement are intentionally laid on thick to take advantage of this, as well as seal the majority of those microscopic air passages. While it's possible for Core A to have pinhole leaks, they do not meaningfully affect performance.

This interference fit turns out to be very strong. In all my testing, I have not had a PLA/PVC or PETG/PVC joint come apart, despite intentional abuse and stress-testing past 100psi.

Refinement of Core A design

Having a removable Core design allowed rapid iteration to improve the design.

Over 20 were built with many destructively tested to find weaknesses. Design variables included:

Core A cut the cost of the launcher in half and eliminated custom plumbing parts, making it widely buildable in regions with Schedule 40 pipe.

Its dependability allowed me to shift my focus to the Base design, especially clamping and compatibility with a wide range of bottles. Core A remained the only Core option during the rapid Base evolution from Beta to Versions 1.0-1.2.

Moving on from Core A

During work on the Base subassembly, I considered next steps of Core design. Core A was limited by its dependence on Schedule 40 PVC pipe, which is easily found only in North America. I wanted the design to be globally accessible, which was only possible if the design did not require PVC pipe at all.

This crazy pipe dream became the basis for the design of Core B.