[Alan Yates] is a hacker’s engineer. His job at Valve has been to help them figure out the hardware that makes virtual reality (VR) a real reality. And he invented a device that’s clever enough that it really should work, but difficult enough that it wasn’t straightforward how to make it work.
In his presentation at the Hackaday Supercon 2016, he walked us through all of the design and engineering challenges that were eventually conquered in getting the Lighthouse to market. We’re still a bit overwhelmed by the conceptual elegance of the device, so it’s nice to have the behind-the-scenes details as well.
Every once in a while, we stumble on an amazing resource that’s not exactly new, but it’s new to us. This is the case, in spades, with The Engines of Our Ingenuity, a radio show that’s been running since 1988!
Each episode covers an invention or engineering marvel, and tells the story of how it came to be, and puts each device into its historical and cultural context. Want to get the lowdown on how we safely bring fire into the kitchen? Or the largest land transport vehicle, NASA’s crawler? And what’s up with lobsters anyway?
[CNLohr] needs no introduction around these parts. He’s pulled off a few really epic hacks. Recently, he’s set his sights on writing a simple, easy to extend library to work with the HTC Vive VR controller equipment, and in particular the Watchman controller.
There’s been a lot of previous work on the device, so [Charles] wasn’t starting from scratch, and he live-streamed his work, allowing others to play along. In the process, two engineers who actually worked on the hardware in question, [Alan Yates] and [Ben Jackson], stopped by and gave some oblique hints and “warmer-cooler” guidance. A much-condensed version is up on YouTube (and embedded below). In the links, you’ll find code and the live streams in their original glory, if you want to see what went down blow by blow. Code and more docs are in this Gist.
The NASA workmanship standards are absolutely beautiful. I mean that in the fullest extent of the word. If I had any say in the art that goes up in the Louvre, I’d put them up right beside Mona. They’re a model of what a standard should be. A clear instruction for construction, design, and inspection all at once. They’re written in clear language and contain all the vernacular one needs to interpret them. They’re unassuming. The illustrations are perfectly communicative. It’s a monument to the engineer’s art.
Around five years ago I had a problem to solve. Every time a device went into the field happily transmitting magic through its myriad connectors, it would inevitably come back red tagged, dusty, and sad. It needed to stop. I dutifully traced the problem to a connector, and I found the problem. A previous engineer had informed everyone that it was perfectly okay to solder a connector after crimping. This instruction was added because, previously, the crimps were performed with a regular pair of needle nose pliers and they came undone… a lot. Needless to say, the solder also interfered with their reliable operation, though less obviously. Stress failures and intermittent contact was common.
Say you have a team of French engineers, a lake in the summer, a wizened old machinist, and some gigantic bungee cords. What would you build? The answer is clear, a human-launching crossbow. (Video, and making-of embedded below.)
You can start out watching the promo video because it looks like a lot of fun, but don’t leave without watching the engineering video. What looks like a redneck contraption turns out to be painstakingly built, and probably not entirely a death trap. The [Rad Cow] team even went so far as to purchase metal cart wheels.
Everyone else on the Intertubes would tell you not to do this at home. We say go for it. That is, draw up reasonable plans, work with an obviously competent machinist, and make something silly. It’s not going to be more dangerous than the stuff that [Furze] pulls off.
Today’s engineers are just as good as the ones that came before, but that should not be the case and there is massive room for improvement. Improvement that can be realized by looking for the best of the world to come and the one we left behind.
Survivorship bias is real. When we look at the accomplishments of the engineers that came before us we are forced to only look at the best examples. It first really occurred to me that this was real when I saw what I still consider to be the most atrocious piece of consumer oriented engineering the world has yet seen: the Campbell’s soup warmer.
This soup warmer is a poor combination of aluminum and Bakelite forged into the lowest tier of value engineering during its age. Yet it comes from the same time that put us on the moon: we still remember and celebrate Apollo. It’s possible that the soup warmer is forgotten because those who owned it perished from home fires, electrocution, or a diet of Campbell’s soup, but it’s likely that it just wasn’t worth remembering. It was bad engineering.
In fact, there’s mountains of objects. Coffee pots whose handles fell off. Switches that burned or shocked us. Cars that were ugly and barely worked. Literal mountains of pure refuse that never should have seen the light of day. Now we are here.
The world of engineering has changed. My girlfriend and I once snuck into an old factory in Louisville, Kentucky. The place was a foundry and the only building that survived the fire that ended the business. It happened to be where they stored their professional correspondence and sand casting patterns. It was moldy, dangerous, and a little frightening but I saw something amazing when we cracked open one of the file cabinets. It was folders and folders of all the communication that went into a single product. It was an old enough factory that some of it was before the widespread adoption of telephony and all documents had to be mailed from place to place.
We’ve all taken apart a small toy and pulled out one of those little can motors. “With this! I can do anything!” we proclaim as we hold it aloft. Ten minutes later, after we’ve made it spin a few times, it goes into the drawer never to be seen again.
It always seems like they are in everything but getting them to function usefully in a project is a fool’s errand. What the heck are they for? Where do people learn the black magic needed to make them function? It’s easy enough to pull out the specification sheet for them. Most of them are made by or are made to imitate motors from the Mabuchi Motor Corporation of Japan. That company alone is responsible for over 1.5 billion tiny motors a year.
More than Just the Specs
In the specs, you’ll find things like running speed, voltage, stall current, and stall torque. But they offer anything but a convincing application guide, or a basic set of assumptions an engineer should make before using one. This is by no means a complete list, and a skip over the electrics nearly completely as that aspect of DC motors in unreasonably well documented.
The first thing to note is that they really aren’t meant to drive anything directly. They are meant to be isolated from the actual driving by a gear train. This is for a lot of reasons. The first is that they typically spin very fast, 6,000 – 15,000 rpm is not atypical for even the tiniest motor. So even though the datasheet may throw out something impressive like it being a 3 watt motor, it’s not exactly true. Rather, it’s 3 N*m/s per 15,000 rotations per minute motor. Or a mere 1.2 milliwatt per rotation, which is an odd sort of unit that I’m just using for demonstration, but it gives you the feeling that there’s not a ton of “oomph” available. However, if you start to combine lots of rotations together using a gear train, you can start to get some real power out of it, even with the friction losses.
The only consumer items I can think of that regularly break this rule are very cheap children’s toys, which aren’t designed to last long anyway, and those powered erasers and coffee stirrers. Both of these are taking for granted that their torque needs are low and their speed needs are high, or that the motor burning out is no real loss for the world (at least in the short term).
This is because the motors derate nearly instantly. Most of these motors are hundreds of loops of very thin enameled wire wrapped around some silicon steel plates spot welded or otherwise coerced together. This means that even a small heat event of a few milliseconds could be enough to burn through the 10 micrometer thick coating insulating the coils from each other. Practically speaking, if you stall a little motor a few times in a row you might as well throw it away, because there’s no guessing what its actual performance rating is anymore. Likewise, consistently difficult start-ups, over voltage, over current, and other abuse can quickly ruin the motor. Because the energy it produces is meant to spread over lots of rotations, the motor is simply not designed (nor could it be reasonably built) to produce it all in one dramatic push.
This brings me to another small note about these tiny motors. Most of them don’t have the carbon brushes one begins to expect from the more powerful motors. Mostly they have a strip of copper that’s been stamped to have a few fingers pressing against the commutator. There’s lots of pros to these metal contacts and it’s not all cost cutting, but unless you have managed to read “Electrical Contacts” by Ragnar Holm and actually understood it, they’re hard to explain. There’s all sorts of magic. For example, just forming the right kind of oxide film on the surface of the commutator is a battle all on its own.
It’s a weird trade off. You can make the motor cheaper with the metal contacts, for one. Metal contacts also have much lower friction than carbon or graphite brushes. They’re quieter, and they also transfer less current, which may seem like a bad thing, but if you have a stalled motor with hairlike strands transferring the pixies around the last thing you’d want to do is transfer as much current as possible through them. However, a paper thin sheet of copper is not going to last very long either.
So it comes down to this, at least as I understand it: if bursts of very fast, low energy, high efficiency motion is all that’s required of the motor over its operational life then the metal strip brushes are perfect. If you need to run the motor for a long stretches at a time and noise isn’t an issue then the carbon brush version will work, just don’t stall it. It will cost a little bit more.
Take Care of Your Tiny Motors
To touch one other small mechanical consideration. They are not designed to take any axial load at all, or really even any radial load either. Most of them have a plastic or aluminum bronze bushing, press-fit into a simple stamped steel body. So if you design a gearbox for one of these be sure to put as little force as possible on the bearing surfaces. If you’ve ever taken apart a small toy you’ve likely noticed that the motor can slide back and forth a bit in its mounting. This is why.
Lastly, because most of these motors are just not intended to run anywhere near their written maximum specifications it is best to assume that their specifications are a well intentioned but complete lie. Most designs work with the bottom 25% of the max number written on the spreadsheet. Running the motor anywhere near the top is usually guaranteed to brick it over time.
These are useful and ubiquitous motors, but unlike their more powerful cousins they have their own set of challenges to work with. However, considering you can buy them by the pound for cheaper than candy, there’s a good reason to get familiar with them.