The machining was done in two operations. All the fancy 3d profiling was done with the part flipped over and held in soft jaws.
Front side:
Back side, before gluing in the magnets:
Here's the assembled driver. I'm using a cheap off-the-shelf pair of earpads.
This version sounded pretty terrible. The pcbway flex-pcb was way too thick, making it both too heavy and too stiff. The response was way to heavy at low frequencies, with a terrible ~100 hz resonance, and there were all sorts of weird notches in the frequency response. But it did make sound, so that was a good start.
I ordered some Pyralux samples from Dupont, and tried etching my own drivers. I first tried with the LF7012R variety, which has 1/2 oz copper, 13 micron kapton layer, and 13 micron adhesive. I measured the total thickness to be 45 microns where there was copper, and ~25 microns without copper, which is half the thickness of the pcbway material. At the time I thought that was their thinnest material, but I hadn't looked through their datasheets carefully enough. The AC 091200EV material has 1/4 oz copper, 12 micron kapton, and no adhesive layer, so it should be in total something like 20-25 microns thick with copper, and 12 without. I've got a sample of that on the way, and will report back when I've etched some.
After a few tries I figured out a pretty good process for etching the pyralux. I did the classic toner transfer and ferric chloride etching method. The etching part worked really well, but doing the toner transfer proved finicky. It took a few tries with a few different laser printers to get the process to work.
Etched:
Assembled into the driver:
This one sounds much, much better, although there's definitely still a little weirdness going on. Listening to it plugged into a function generator and sweeping frequencies, the low-frequency resonance is gone, and over all it's much flatter, but there are still some strange notches in the response. Unfortunately I don't have a good setup for quantifying the frequency response at the moment, but it's definitely a little off right now. But it's getting better. I'm hopeful that the even thinner flex pcb material will improve the situation further. Reducing tension in the membrane might help as well.
