Note: If you're already familiar with casting parts, jump directly to the specifics about embedding capacitive touch in concrete.
We use a lot of fabrication methods when quickly prototyping one-off parts, but finding a way to scale up can be a major hurdle.
There are two immediate issues when going from prototype to production:
Casting parts is one common workaround for the scaling issue. To do so, first a model (called a pattern) is created to look like the part you want to make. Then a mold is created around it, the pattern is removed (either by prying it out, melting it, or splitting the mold), and the desired material can be poured into the mold for making the desired parts.
Making a good mold is an artform unto itself. But even with a basic silicone mold that you simply pour over a printed part, you’ll likely get pretty far without much trouble (see HomeMade Modern's quick video on silicone mold making). Of course there are lots of lessons to learn in making a reusable mold that lasts (minimizing air bubbles, selecting the best orientation, splitting the part/mold if necessary, and choosing the proper mold material), but even a “bad” mold can still tell you a lot and produce halfway decent castings.
As for durability, that comes down to the material you use in your casting — three main types are commonly used; metals, polymers, and concrete/plaster.
Casting metal parts is common on an industrial scale, but it brings with it some major downsides. First are the dangers. Molten metal is super hot and can release all kinds of noxious fumes depending on what type you’re working with. Then there’s the issue of mold durability, often requiring a new mold each time a casting is produced. And the expensive furnace needed to control the melting. And the post-processing of cast metal parts.
For small-run production in a modest makerspace, casting metal parts is just more overhead and risk than it’s worth (though we imagine it'll continue to become more accessible in the future).
Casting polymers (i.e. plastic parts) is pretty straightforward (just mix two liquids together and pour in the mold). But plastics are, well, plastic-y, and depending on how they’ll be used, that makes them potentially not desirable or durable enough.
Many plastics dry, crack, or decompose over time (especially if exposed to UV light, like say, the sun) and slowly bend or creep under stress (for example, a part that is supporting weight might bow in the days or weeks post-cast). Additionally, the resins, before being mixed and cured, often have some pretty nasty health/safety warnings, which is a bit concerning when you’re combining gallon buckets of the stuff on a daily basis.
Which brings us to casting concrete and plaster. When mixed properly, concrete sets up rock hard, holds up to stress, cures with minimal shrinkage, has a nice hefty feel, and lasts a lifetime. Plaster is similar but often much more brittle, so we'll focus on concrete here.
If you're just getting started, you may want to check out Bold Maker Studio's concrete casting tutorial showing how to mix CementAll for casting small concrete objects. It's cheap, simple, and has given us consistent results.
There are, however, two major design constraints to keep in mind beyond just getting the mix right: don't use thin edges or sharp corners (they will easily chip and crack); and limit forces to compression (not tension) as much as possible.
That second point is the tricky one — while concrete can withstand a massive amount of compressive force on top of it (think of tall concrete walls), it breaks rather easily under tension (it’s hard to visualize “pulling” on concrete, but this is what makes it feel brittle if you’ve ever snapped a thin piece of concrete with your hands).
Fortunately, there are three workarounds for concrete’s low tensile strength:
The other important factors in producing a good concrete casting are ensuring the proper ratio of mix/water (hint: use a gram scale!), reducing evaporation while curing (concrete doesn’t “dry,” it cures, and it needs water to do so; just cover with a plastic sheet so the exposed parts aren’t brittle), and removing as many air bubbles as possible (tapping on the sides while pouring makes a huge difference; you can also vibrate the mold with anything that vibrates).
Which brings us to the crux of this post — what about embedding interactive elements (like capacitive touch) into the concrete itself?
Much has been written and shared about casting one-off items in concrete (from countertops and planters to modern lamps and LED clocks), and clearly care must be taken when embedding various electronic elements within the concrete due to its watery, alkaline nature (wet concrete has a pH of around 13, enough to cause burns if you’re not careful). Generally electronics don't like being submersed in corrosive water, so they need to be encapsulated.
The good news is that concrete is pretty electrically inert (for example, a 1cm^2 contact area over a distance of 1cm away gives around 12k for typical concrete), so once it’s set there’s not much risk of shorting things out in some spectacular fashion. But it’s not nearly as inert as you might imagine, acting as a kind of time-varying semiconductor that slowly loses conductivity (resistance increases) over many weeks and can itself act as a capacitive plate alongside other materials.
We’re currently experimenting with casting a concrete base for LightNudge (our color-changing barometer with a capacitive touch input).
While we’re not planning to cast the PCB itself into the concrete (it will rest on a shelf inside when fully set), the 5V DC power jack and a metal capacitive touch point (a wire soldered to the back of a steel tack) will be cast directly into each unit.
To our surprise, many of our initial cast concrete parts (even days after being fully set) still show a fairly low resistance (around 100k-200k Ohms) between the capacitive touch point and the power jack. 200k Ohm might sound like a large enough value to dismiss, but it definitely throws a wrench into reliably sensing a user’s touch where the overall changes are often quite small.
To accurately sense capacitive inputs, the system should be relatively stable over time (so that the microcontroller itself can control the signal and accurately read the capacitance). In cases where significant parasitic resistance is unavoidable but fairly consistent, there are other workarounds to offset its contribution.
Unfortunately for us, concrete slowly changes over time and varies part-to-part, making it difficult to finalize the capacitive sensing code and feel confident that it will properly function across different batches of concrete and continue to reliably sense a user's touch days or weeks later as it cures.
While we could drastically change the design to no longer use a part with exposed metal, that brings with it its own issues (since the capacitive input sensing area does in fact need to be conductive). We ended up deciding to coat/paint the metal with an insulator (also known as "potting") such that the resistance between the concrete body and the sensor could be drastically reduced.
We tried metal-adhering spray paint, which looked great and offered some reduced conductivity (by about a factor of 2 with a single coat), but didn’t quite cut it for us.
We also tried dipping the metal element in acrylic enamel (shown in the image above), but the consistency of the paint seemed to leave exposed gaps in the finish where the metal could still come into contact with the concrete.
In the end, we settled on using an oil-based enamel designed for use on metal (easy to find at our local hardware store for about $5 a can; shown above). It takes a full 24 hours to dry, but the finish is consistent and doesn’t leave regions exposed like the acrylic enamel seemed to do.
One coat (just dipping in the can, swirling it around a bit, and letting the excess drip off) seemed adequate, but we doubled down with a second coat (the element on the right in the picture above) just in case.
Now the measured resistance between the touch point and the power jack is greater than the limits of our multimeter (more than 40,000k Ohm to use the same units as before; basically infinite resistance for our purposes). That means coating the exposed metal with oil-based enamel did the trick, and we’re back in business using our metal wire and tack as a capacitor for touch sensing again. Success!
While not as visually warm as the original wooden base, the cast concrete parts feel much more substantial and won't warp over time (as wood is prone to do when the humidity changes).
There's still more to iron out before we can actually ship a concrete product (sanding, finishing, improved structural packaging for a heavier item, user testing, etc.), but sorting out the capacitive touch brings us a big step closer.
So despite this being very much a work in progress, we were pleasantly surprised at how easily we could directly embed an interactive element in concrete and wanted to share our excitement along the way.
The ability to include a wider range of electronics in concrete opens up new possibilities in often overlooked spaces. From stone-like electronic gadgets to massive interactive concrete walls (entire buildings?), we look forward to seeing a lot more concrete tech in the near future.
All hacks posted by Makefast Workshop are open source and shared without any strings attached for your amusement, use, and continued experimentation.