cct winding machine

during my internship at the lawrence berkeley lab, I designed and built a prototype three-axis machine that winds canted cosine theta (cct) coils, assisting in research in magnet design for future particle accelerators.

the curve of their geometry is defined by the following parametric equations:

where r is the radius of the curve, α specifies the tilt angle, and p specifies the spacing of the curve each revolution.

when a cct mandrel is machined, a groove is milled along the cct curve such that it remains normal to the surface of the cylinder throughout the curve.

when an even number of cct coils are nested within each other, they emit magnetic fields such that the solenoidal fields (horizontal) cancel out, while the transverse fields (vertical) sum. this results in a field that bends a particle beam that passes through the cct magnet. these bending magnets are essential in high-energy physics experiments to guide and redirect particle beams.

in order to maintain tangency to the cct curve, a minimum of three axes of motion is needed: two rotational axes, and one translational axis. to confirm this, we built a manual version of the machine using aluminum extrusion and low-friction bearings. the machine features a a rotating shaft for the cct mandrel, a linear carriage, and an arm mounted to the carriage that pivots directly under the mandrel.

once we confirmed our proposed axes of motion could maintain tangency as we envisioned, we then added stepper motors to drive the machine.

the carriage is driven by a gates gt3 belt...

while the mandrel and arm motors are coupled directly to drive shafts.

an esp32, a buck converter, and three driver boards, one for each stepper motor, are soldered to a permanent breadboard.

cables routed!

the toolpath for the machine is written in arduino ide. in a config.h file, the user needs to specify some parameters. these should be the same parameters that were used to generate the mandrel's cct curve.

after performing a homing function for all three axis, the machine will then transition into the regular toolpath.

• a value for t increases constantly.

• the mandrel's rotation is constant, directly relating to t.

• the carriage's position is defined by the z(t) function.

• the arm's rotation is driven by the tangent line to the derivative of the curve.

please refer to the project report's appendix for a more detailed explanation of the motion of the machine.

while the machine performed perfectly for mandrels with a larger bend radius (as shown above), the simple tension + tangent approach proved to be insufficient for some mandrel geometries that feature a tighter bend radius.

one possible solution we explored was to use a "shoe," that matches the contour of the mandrel's groove. this allows the machine to pre-bend the wire to the correct shape, to then be pressed into the groove. this idea showed promise, but will need further exploration to be confirmed as a viable solution.

up until now, the wires shows in the images are a regular, stranded, insulated copper wire that is significantly easier to wind than the real superconducting wires. for a future winding machine, it will be necessary to use significantly beefier motors and shafts. in addition, a tensioning mechanism should be implemented to ensure consistent winding.