G-Zero - Cavity Build
Possible Build Methods
Multiple ways were considered for building the cavity, but each rejected, typically because the size of the cavity resulted in large costs. The final solution was to have the cavity printed on a 3D Fused Deposition Modeling machine (FDM) and then lined with copper. Below are a brief analysis of each of the rejected methods. Keep in mind that the cavity is roughly 16" long, 11" at the wide end, 7" at the narrow end and 11" high, which if split in half would require a block 16"L x 11"W x 5.5" high.
This was looked at in two ways, first, as a single thick walled tube which could be put on a metal lathe then machined from the inside out. The second method was to create the cavity in two halves by machining each half from a solid block. The materials considered were aluminum and copper, however, the metal blocks were expensive, easily costing over $3 to $4K and wastage high considering the cavity was hollow.
This was seriously considered, especially because it could be applied to an inexpensive 3D printed plastic part. However it was rejected for the following reasons:
- Cost - upwards of $500, which is expensive considering a thick 4' by 8' (yes foot) sheet of pure copper goes for $360 (in 2008)
- Electroplated Thickness - typically only a couple microns, where a thousand microns is 1 mm. Unfortunately, because of the high power requirement, it isn't so much penetration of the microwave energy as it is momentary heating. With just a bit of heat bubbling the plastic and electroplated surface it could then arc something like this.
- Adhesion - getting the copper layer to stick, or plate evenly sounds very difficult. Part of the problem is that for electroplating, it needs about an amp per square inch, which results in a lot of amps given the inside volume of the part. Too many amps leads to unpredictable results.
Metal spinning has been around a long time and the material costs for the copper would have been inexpensive. However, getting the mold or "mandel" made, around which the copper could be spun was again prohibitively expensive, upwards of $3K. Wood lathes are relatively inexpensive and an attempt to see how hard it was to make the mandel and spin copper was attempted. It proved too difficult.
A relatively new technology, 3D printing had the benefit of useful tolerances at 0.2mm and relatively inexpensive at under a thousand dollars for the plastic cavity. It was also because of the willingness of a local provider (Motive Industries) to take on strange jobs at excellent prices, that 3D printing was the final pick. It turns out the slope of the cavity cone section was just the right angle to be printable from the base up in a free standing manner. Even though the part was inside the build envelope of the machine, a Statasys FDM Maxum, because of calibration problems, the part was printed in two sections.
Once the cavity case was printed in plastic (Figure ), 16oz copper sheeting was waterjet cut to fit exactly inside the cavity (figure ). The end caps and inside tuning plate, which would handle most of the current, were curt from 32oz copper, mostly for proper heat dissipation. The copper liner was then fit to the cavity, marked off (figure ), removed and soldered together (figure ). Once the copper liner was built it was refitted to the plastic.
The appropriate holes for the probes where then drilled through the plastic and copper liner. A 3D model (figure ) shows how the probes were mounted using thin copper sheeting wound around the probe's dielectric to provide grounding between the copper liner and the probe mount.
The initial design for the probe, although easy to implement and useful for low power testing, is not well designed for high power. Attempted improvements, for example, soldering the wrapped copper to the cavity and adding a cap meant working around the plastic as shown in the pictures below.
It was easier to solder on copper probe mounting blocks cut with a CNC machine:
Alas, after sourcing the copper block and contacting five different CNC shops, none of them were willing to cut the mount. In the end, the next best solution was to solder the copper plate onto the round wrapped copper cylinder. Once finished, the cavity was then tested .