The first step is to make a model of the roof using modelling clay. I decided to invent a cover for the folding roof that probably never existed for the Karmann Ghia, but is easier to make than a full folded-down roof with rods and levers and everything.
Here is the model after smoothing the surface with water. Not bad for a first attempt, but much too big, especially because the model will be used as a positive mold and the final part will be formed over it, so it will be even bigger. The squarish shape of the part also looks out of place on the rounded and flowing body.
The next version was much better. It can be seen here after drying for 24 hours and making final touches to the shape with fine sanding paper. The part is a little bit too small so that the thermoformed part will (hopefully) fit perfectly. There is only one attempt, because the process will destroy the clay model.
Now for that industrial equipment...
This is an Erkodent Erkoform-D vacuum thermoforming device that is normally used in dental labs to make retainers. It is a precision thermoforming device that uses very low pressure (0.2 bars) to wrap a sheet of plastic around a positive mold.
The device weighs around 10kg, is about 25 years old and if you are patient, you can get one for little money on Ebay. Note that, with the exception of the heat lamp on the top, the device is purely machanical. It uses Venturi tubes to create 0.2 bars of low pressure from 8 bars of high pressure. All that can break are the seals, and those were still fine on that old device.
Of course this also means that you need some means to generate high pressure first. I am using a 1400W, 8 bar "silent" compressor for this and, yes, it still sounds as brutal as it looks. Hearing protection is highly recommended when operating this thingy. Fortunately, you only have to fill the reservoir once in order to drive the Erkoform.
The clay shape is placed in the vacuum chamber of the device. The substance below the shape is granulated stainless steel, which allows for a homogenous airflow around the mold. Note that the container is filled insufficiently, which will cause blur on the edges of the plastic sheet. Fortunately, this did not matter in this case.
The raw material processed by the machine: a 100mm disc of polystyrene, 1mm thick. The apparatus can process thicker sheets, but this is what I use. Note that the device is specified for polymethyl acrylate (PMA), but it works equally well for PS and probably for other kinds of thermoformable plastics.
The disc is inserted into the frame and fixed with springs under the grey knobs on each side. The frame is then swung over the heat lamp and heated for a while. Modern devices use a sensor to tell you when the temperature is sufficient, but on this machine you will have to look at the sheet and use visual judgement. The sheet will start to "wobble" slightly when it reaches the glass transition point of the material.
The frame is then moved back to the vacuum chamber and pressed down. This deactivates the lamp and opens the valve of the Venturi tubes. The thermoforming process itself takes a fraction of a second (but I usually hold down the handle for a few more seconds just to be sure).
Here is the thermoformed sheet after removing it from the frame. You can see the structure of the granules at the center which blurs out toward the edges. This is because the chamber was insufficiently filled. The part itself looks fine, though.
The clay model is now stuck in the sheet. You normally cannot use molds that will be enclosed by the thermoformed part, because you cannot remove the mold afterward. In fact you need a "draft angle" of at least 3 degrees; i.e. the mold must be tilted toward its center by ≥ 3 degrees on all points of its surface, or it will most probably get stuck.
In order to extract the part from the sheet, it is first cut out roughly using a milling drill.
The shape is then extracted more closely with a cutter.
At this point, I am bending the part slightly to remove it from the clay model. This is considered cheating. You cannot rely on the final part to be bendable, but in this case it worked fine. Because the plastic part wraps around the clay model on all sides, a small part of the clay had to be removed, thereby destroying the clay model.
Here is the final plastic part after some further cutting and filing.
Fitting test. Looks good!