I spent Monday trying to get a plastic model built around the wooden mold (in preparation for the copper), but ran into significant problems.
As shown, the mold has since crack significantly, making the circumference much larger then originally planned. Even with the cracks, it proved difficult to get my test plastic around the mold without creating a large number of seams. The seams are a problem because they are not smooth and would need to be brazed. I have, for the meantime, rejected using the mold to strap the copper around.
As I mentioned previously, paying someone else to do the metal spinning is too expensive, and I put more thought into doing it myself, but decided again that too, for a number of reasons:
- I don’t have the right tools (I would need pretty much everything on that page)
- I need a lathe with a larger space between the spindle (the part that spins on either side of the piece) and the bed (the part that holds both side together). The largest piece of copper I can fit onto the my current lathe is 11″ in diameter which is not enough copper to push over a 15″ long mandrel. Good metal spinning lathe are worth thousands of dollars.
- After talking to a supplier, it sounds like the mandrel I have created is not a hard enough wood, nor is it centered properly.
I have since been investigating using “3D Printing” technology which uses metal powders, and adhesives to first make a “green” object, which is then baked, and then infused with bronze. I have sent in a request for quote to Prometal (US based) and Shapeways (UK based) and I am hoping to get the cavity made for under $2K CND. The beauty of using 3D printing technology is that I can design the cavity in two parts including all the mounts necessary for the tuning plate and the two probes in Rhino3D (which I already own) and then have it printed with 0.1mm accuracy! I am also looking into getting a much cheaper plastic model “3D printed” and then electroplated with copper.
I spent the afternoon at the university running more simulations – I currently have a problem where if I run a simulation with the highest number of tetrahedrals possible (~600K), I get different results then if I run it with fewer tetrahedrals (~300K to ~445K). The results are different enough that it looks like the TE0,1 mode is resonating at a different frequency. With the ~600K model, I tried using the tuning plate at the back of the cavity to bring the TE0,1,n mode back to resonate at 2.45Ghz, but with ~600K tetrahedrals, it takes five to six hours just to run one simulation. I then make a small change to the position of the tuning plate and retry the simulation but the length of time for the iterations makes the method unfeasible. My analysis is that the higher number of tetrahedrals changes the results because the resonant frequency of the TE0,1,n mode is sensitive to the accuracy with which all the round parts are meshed at. Everything in this model is round, from the probes to the tuning plate to the entire cavity itself, which makes an accurate meshing very difficult. However, I am confident the model will work, but it will need a tuning mechanism to make the TE0,1 mode resonant at 2.45Ghz and I am going ahead with the design. (This could be my biggest mistake yet, if the cavity costs $2K to get built and it doesn’t resonate as expected)
I did get the new L gasket for the vacuum chamber and I was very happy to see that it fit:
Now I am waiting for:
- Vacuum gauge – I unfortunately lost my bid and will have to find another one.
- Drill bits for the acrylic – they should be here within a week.