The measurements for the aluminum sections can be taken directly off the table drawings. Aluminum is easy to work with, light, and does not try to fly towards magnets. The sections are joined together in a jig with Pop rivets, and then bolted to the upper and lower plywood magnet holders.
The steel rod is cheap and can be bought in 4 meter lengths from the local hardware store. It does try to attach itself to any magnets close by. Some types of stainless steel are non magnetic and may be more suitable. Aluminum threaded rod is not available at reasonable cost, but may be the best. Aluminum may be best for a double rotor machine. The threaded rod makes the frame components quite adjustable. Each part has its own nuts and washers, and can be freely moved, then locked tightly into place.
The plywood is cheap and easy to work with. I make up patterns as accurately as possible, and then use these to mark slightly oversize plywood pieces. These are cut with a hand jigsaw and bolted to the pattern. A router in a router table makes quick accurate pieces ready for glueing together. I use PVA glue. Not very space age materials, but cheap and readily available. The table is used to glue bits together on, and some of the precut frame alum sections are used as clamps.
The complete machine as it exists now is strong and stable while attached to the table. If I try to lift it from one side it is very flexible. To be removed from the table will require a complete triangular spaceframe built around and attached to it. For initial testing it appears to be quite adequate.
The frame is made in 14 sections. This is a decision I made based on the Smartdrive motor, which I copied, and has 3 sets of 14 coils . A few months after construction started I discovered that the Smartdrive has now been converted to 36 coils, or 3 sets of 12. Engineering wise, 12 frame sections makes more sense than 14, but then the distance between components is longer and may need to be stronger. I have reduced the coil count to 3 at present, but may need to go to 6. Each "coil"
is comprised of 4 individual coils. Either of these numbers work well with 12 sections, but 18 sections may be better. 18 would definitely be better if the overall machine was scaled up in size to say 3 meters across.
Everything is assembled on the table according to the drawings glued onto the table. Once the table and radius arm has been set up carefully, very little measuring needs to be done, and mistakes are rare.