Step 1: Tools and Materials
The material used in professional manufacturing of hearing aids and IEMs, called otoplastik, is made primary by two companies (from what I can tell), Egger and Dreve, both of which are German. This stuff is on the expensive side, partly because they don't sell small quantities, but the pro-rated price (price of material per set of shells made) is actually only a few dollars, if that's any consolation. Dreve makes an acrylic lacquer, that is supposed to be used to "build up" a finished shell, but it's the smallest quantity available, 20mL for ~$30. You could probably make 3-5 sets of shells from that, depending on thickness. The material that is actually made for shell making is significantly cheaper in terms of $/mL but the smallest quantity sold is 91mL and that costs ~$60.
In terms of the materials needed, again, I've discussed/referenced a lot of this in my other 'Ible, so I'm not going to go into too much detail.
Also, while I will be focusing on showing you how to make hollow shells with the UV acrylic, I did make some shells using urethane epoxy, just as proof of concept (anticipating that most people aren't going to want to go through the hassel and expense of the UV acrylic route). So keep that in mind when reviewing the materials listed below.
-Impression making kit
-Nailpolish (or lacquer paint)
-UV curable acrylic (available from Lightning Enterprises)
Balanced armature drivers (Mouser or Colsan)
Caps and resistors (if you want/need crossovers)
-Connecting wire (from IEMs to source)
-Hydrocolloid material (like gelatin)
-2oz disposable wax lined paper cups
-Dremel (with lots of bits!)
-A helping hands
-Some long wood screws
Step 2: Making the Impressions
Step 3: Casting the Impressions
Step 4: Shaping/building Up the Master Casts
When I made the solid earpiece IEMs, I did the shaping/build up with the final product, basically because I didn't know any better. After doing some further investigating, I learned that this is typically done early on, with the actual silicone impressions. You do the shaping by remove material with a sanding disk on a rotary tool, and the build up with a sticky wax. From this, the earmold is made. I dislike that conceptually because it means if you make a mistake modifying your impressions you don't have anything to go back to. But also, because of my next step (for the UV material), using wax wouldn't work anyway. So what I did was make an intermediary cast, which I'm calling the master, because it is going to be from this that the final earmold is made.
First, make a cast from your earmolds. You can use any kind of plastin resin, doesn't have to be clear, doesn't have to be any specific kind of plastic, but here I'm using a clear acrylic (from Electron Microscopy Sciences). Once they cure, you will first shape (using a rotary tool), then build up (using nailpolish or lacquer paint or something), testing each iteration for comfortable fit and proper seal. Build up is also necessary to smooth out the casts, which will have a rough, prickly texture at first. Once you've reached a point where you get a proper seal test them out for an extended period of time. This will ensure that you can maintain the seal while turning your head, opening your mouth, etc., and also that they will be confortable for extended use. If you build them up too much they will exert pressure on your ear canals which can be hard to detect in a short period of time, but will become evident after ten or 15 minutes (especially once you take them out).
Step 5: Making Shells, Part 1
I'll just mention that I used a piece of equipment that not everyone is going to have access to, which although it isn't necessary, definitely makes it easier, and that is a roller mixer. It basically looks like one of those hot dog grills in a 7/11; it has a bunch of rollers, and an RPM adjustment, you lay down cylindrical containers in between the rollers and it keeps the contents tumbling (kind of like a clothes dryer). It would be super easy to make something to rotate your molds -- you basically just need an electric motor, mount a plate on the shaft, put some double sided tape on it, and stick your molds on. Voilà. You *could* do it by hand, but I would recommend finding something that cures much faster than the Crystal Clear, which takes about 2 hours before it's viscous enough that it will no longer deform under gravity. But it needs to be an epoxy, something that cures by smoothly transitioning from a liquid to a solid. For example, the acrylic casting resin I bought from EMS isn't an epoxy, and goes through a weird gel phase before it turns into a solid, so you wouldn't be able to use it to do this.
With the Crystal Clear and the roller mixer, I was able to make very nice shells that would have been perfectly usable (and the Cystal Clear really is crystal clear). As I talked about in the other 'Ible, this stuff is allergenic, so they need to be coated in something. In order to do that, I stuck some putty on the end of a long wood screw, and stuck it to the bottom inner surface of the shell. Then I just dipped them in a cup of nail polish, twirled a bit to get an even coat, let the excess drain off to one side, and then hung them to dry on my helping hands. The rim is somewhat jagged, so to finish them I wet sanded with a piece of 220 grit wet/dry paper on a nice smooth surface.
One caveat is that it's next to impossible to do this without getting some air bubbles trapped in the plastic. Not a huge deal, purely asthetic, but in order to minimize this as much as possible, when mixing the two components, the resin and the hardener, it's best to do so via tumbling, not stirring. You can either gently pour it back and forth between two cups, or if you have a tallish container, tip it to as much of an angle as you can and just keep rotating it. Eventually you'll get a homogeneous mixture that has very few air bubbles.
Step 6: Making Shells, Part 2
The way this is done professionally is to make an earmold out of a clear hydrocolloid material (which you can buy at some of the places that sell the UV acrylics, but comes it huge bulk quantities and is expensive). Then, you fill the earmold with your UV compound, put it in a UV chamber, covering the top so that it doesn't cure, and zap it for a period of time. Because your exposing the outter surface through the earmold material, it cures from the outside it. The longer it gets UV light, the thicker the wall will be. Also, the UV chamber ensures that all surfaces are getting even exposure, giving you a uniform wall thickness. I don't have a UV chamber, but I made a little UV light with (7) 330mW longwave UV LEDs (some quick math tells me that's... 2.31W total!) that I wired up on a breadboard and powered with a hacked ATX power supply using the 12V output. The problem with this is that I have to apply the light in sections, which means it's basically impossible to get an even wall thickness throughout. But you can get pretty close. I made my earmolds with the cheapest, most common hydrocolloid there is: gelatin. The more gelatin you use the firmer the material will be, but the more opaque it will be as well. I found two packets of Knox pure gelatin per 4oz of water was a pretty good ratio. I put the plastic master casts I made upside down in metal measuring cups sitting in an ice water bath and then covered them with the hot liquid gelatin (this is why you couldn't use wax to do your buildup). Once it set, I used a butter knife to shimmy them out (obviously, using little wax lined paper cups is easier because you can just tear them away, and that's what I eventually used later on, but instead of an ice water bath I just put them in the fridge). After popping out the plastic casts from the gelatin molds, I filled them with the Fotoplast Lacquer, and exposed each of 4 "sides" for about 30 seconds, and then poured out the excess. If there are any real thin spots you can just add some more compound and give it another zap. I can almost guarantee that if you try this, you will mess up the first one you do, but I would recommend starting at a shorter interval, maybe 10 or 15 seconds, and going from there. You can always add more, but you can't take it away (easily) once it's there. And given the variables (the thickness of your molds, what you use for your hydrocolloid, how opaque it is, how strong your UV light source is, what kind of UV material you're using, etc.) what worked for me isn't necessarily going to work for you.
Now, although this produced a usable shell, I'll just mention that this material yellows quite a bit, which isn't noticable when your using it as a thin top layer, but when it's this thick it's very apparent. Also, this material dries to a rock hard plastic, like plexiglass. For what it's worth.
Step 7: Making Shells, Part 3
Now, the first set of shells I made with this material I used the same technique as with the Fotoplast Lacquer, filling the shells and shining the light for a period of time. The results was that the walls were really, really thick, and the tip was solid (which is a problem). The second set, the tip was hollow, but I had spots so thin in other places that they had holes. So what I ended up doing was sort of a hybrid borrowing from the epoxy technique. I put enough of the S/IO material in the molds to coat them evenly, and then hit them with the UV light. Again, where there were thin spots (which are much easier to see with the colored material, btw) I just added a few more drops and lit it immediately before it could run or disperse. This ended up working perfectly.
Using a Dremel I drilled small pilot holes for the acoustic tubing outlets, and then used a conical griding bit to ream out the holes to the right diameter. I used a small cylindrical grinding bit to make the holes for the connectors. I found that grinding is preferred to drilling because drill bits have a tendency to crack/splinter/shatter the material as you increase the hole size.
A note on the female connectors I used: It's hard to find really small connectors that are suitable for this purpose. If you end up buying supplies from Colsan Micro, do yourself a favor and be sure to buy some of the IEM connectors they sell. You'll be able to use any off the shelf IEM cable with them. Here I'm using DIN connector pins, and as you can see in one of the photos, the females are waaaay too long. Also, these have a slit in them, with the two halves crimped together, which puts tension on the male pin to hold it in. Because of how I'm mounting these, that wouldn't have worked anyway. So I cut them at the bottom of that slit, and drilled out the bottom of the connector (after the slit it's solid) with a 1mm drill bit (the male pin is 1mm diameter) until it was deep enough to accept the male. I also cut the other end a bit (which is also just a hollow tube, for inserting and soldering your wires).
Using the flat side of a drill bit, I installed the acoustic dampers into the acoustic tubing (greens for the tweeters, yellows for the woofers). I first dry fit the tubing so that I could cut it to the right length. After I got it how I liked it, I applied some of the Fotoplast lacquer to glue it in place. I also glued the drivers to the tubing the same way (I wired the drivers to the connector pins before installing). I positioned the tubing so that it was recessed in the tip, which I did so that I could go back and fill it with more material, that way the tip would be nice and smooth (if the tubing goes all the way through, it's very hard to smooth and can be uncomfortable when inserting; it also looks much better this way). The way I backfilled the tip was to use the same size drill bits I used to insert the acoustic damplers, one in each tube, then added more of the S/IO material, cured it, and twisted the bits out with a pair of pliers. Then I just sanded the tip and gave it a coat of lacquer.
I didn't talk much about drivers, it was discussed length in my other 'Ible, but here I'm using Sonion drivers, 2091i as the woofer and 2389 the tweeter. With the drivers mounted, I *gently* coaxed the adaptors through the holes in the shells. You have to be very careful; if you put too much tension one the solder pads of the drivers you can rip them off. Also, you have to be careful that none of you wires touch the cases of the drivers, otherwise they'll short. Using really small gauge magnet wire would be advisable as it would avoid this problem. I also went over crossovers in depth in my other 'Ible, so I'll just say that I went for a crossover point of about 1000hz, which meant using a 2uF cap (tantalum, not ceramic) and a 10ohm resistor.
With the drivers mounted, the tips finished, and the connector in place, I plugged them in to test them out. In one of them the wiring was much cleaner and I didn't have any problems. The other one I had a wire that was intermittently shorting against one of the cases so I had to go back and reposition. Once everything's how you like it, it's time to make the faceplates.
To do this, I first traced the outline of the tops of the IEMs on aluminum foil, then filled the outlines with my compound, and cured it. Turns out this material has enough shrinkage when curing that the ends curled up a bit. Instead of using more UV material, sanding, adding more, etc., I just put down some pools of the Smooth On, and put the blue chips in it, essentially building them up so that they'd be nice and smooth. After it cured, I roughly cut them out and dry fit them for positioning, then added Fotoplast to the rims of the IEMs, held down the face plates, and lit them up. After they cured, I sanded them down to the shape of the IEMs, applied more lacquer to smooth them out, cured it. And... done.