CNC Carbon Fiber Filament Winder

Design started with setting basic dimensions of the frame, as well as determining the maximum size tube that could be manufactured. In the end, the base frame was decided to be 1000mm x 500mm and can wind a mandrel of ~850mm length and 100mm diameter.

A dual-rail linear axis design was decided on early. This design allows for weight to be evenly distributed between the two rails and reduce belt-slip inducing wobble during gantry movement. It also allows for a second stepper motor to be retro-fitted to work in parallel with the original in case of insufficient torque or modifications that increase the mass of the gantry.

Updated gantry assembly in context of base frame and linear axis rails.

Cross-section of the gantry assembly. Carbon fiber tow unwinds from the spool, passes through a series of rollers to tension, through a resin bath to impregnate tow with resin, and finally out of the tow head. Gantry slides on linear axis rails with two V-slot gantry assemblies (purchased). Tow head rotates with a GT2 belt drive and stepper motor.

Resin bath detail. Designed for a disposable 100mL measuring dish to hold resin, meaning that the whole bath does not have to be reprinted with each use. Resin bath "cap" clamps measuring dish in place and has an internal channel for the tow to pass through, scraping off excess resin.

Updated Tube Winder assembly with rotational axis assemblies and all hardware.

Rotational axis motor mount assembly detail. Stepper motor drives a rotating shaft with a 3d printed chuck to hold mandrel via GT2 belt drive. 

Rotational axis motor mount cross-section. A M12 shaft passes through 2 press-fit bearings and whole shaft assembly rotates as one piece, driven by a stepper motor and GT2 belt drive.

Most custom-designed parts were designed to be 3D printed. Over the course of two weeks, myself and my co-lead printed all of the parts on our personal printers.

While parts were 3D printing, we calibrated and set up the stepper motors used for the winder. All 3 motors connect to a CNC shield and Arduino UNO running grbl firmware. With this setup, we can send GCODE through a computer program (We used UGS - Universal GCODE Sender) to control the winder.

Rollers coming into contact with the carbon tow were covered in teflon tape to minimize tow breakage.

The aluminum extrusion frame and linear axis rails were assembled and fit perfectly! Great effort was taken to ensure that the base frame was square and level. Effort was also spent to ensure that the linear axis rails were parallel so as to not restrict gantry motion.

After a few work sessions in the basement of Hesse Hall, the winder was starting to take shape! Next, out to our garage to perform final assembly and get this thing running.

One of the first steps was to run a dry tow path to make sure that the tow was routed as intended. 

After calibration, the first dry run was ready to go! To avoid wasting expensive carbon tow and resin, we ran the winder (with same GCODE as intended for carbon fiber) with curling ribbon to test it out. We did encounter some issues with ribbon slippage, but the rest of the winding ran incredibly smoothly. This gave us confidence to proceed with our first wet layup as slippage would be minimized with resin binding things together.

The first run went beautifully! The winder performed exactly as intended and laid a thick tow onto the mandrel. There are a few minor issues to resolve, such as slippage of the tow and GCODE errors leading to slightly imperfect wrapping. Nonetheless, these issues are process issues, not hardware issues and I am incredibly pleased by the winders' performance!

For the mandrel, PVC pipe with a layer of parchment paper was used. The parchment forms a barrier against resin and PVC provides structure.

With PVC pipes, we were limited to standard circular tubes at pre-defined diameters. However, after experimenting and dialing in a process, we are able to use PVA, a 3D printing material normally meant for dissolvable supports to create mandrels. This allowed us to rapidly prototype non-standard tube geometries that could be easily dissolved out after the composite was cured.

The resulting tubes had an extremely well controlled inner diameter which was perfect for the next phase of this project...

Using the tube winder and processes we dialed in along the way, we successfully designed and wound tubes for a Bike Builders of Berkeley special: the club's first lugged carbon fiber downhill mountain bike. With learnings about tensile and bending strength from our Instron testing, the frame designers were able to choose wall thicknesses and wrapping angles to optimize their design.

Rear triangle fit check at the club's garage.

A deconstructed bike: filament wound carbon fiber tubes, 3D printed aluminum lugs, and blood/sweat/tears.

Presenting the BBB DH 1.91

Of course, it's Oski approved