What’s Inside? — Vol 1: Leap Motion
Growing up as a kid, I loved taking things apart to see if I could figure out how they worked. It seemed like a mystery waiting to be solved. Toys, radios, electronic clocks, car parts, you name it.
It was fascinating to see beyond just what’s outside, sometimes even having to figure out how to open the enclosures. A whole new world of mechanisms, parts, and design revealed as you peeled the “onion”.
Putting them back together successfully? I had a decent track record. Probably around 75–80% much to the dismay of my parents. But I think they saw how excited I’d get and how much fun it was for me.
I’m sure this habit is a big part of why I ended up becoming an engineer that designs and builds hardware. Yes, I know, classic…
Developing electronics for a living has given me a new appreciation for elegant design and understanding of the challenges involved. It’s also a great way to learn how other engineers and companies think.
To this day, I haven’t stopped taking gadgets apart. Recently, I’ve been wondering if it would be fun to document these teardowns. The now famous ifixit.com has done a great job with theirs. I decided to give it a shot.
So why not just go over to ifixit.com and browse their work?
Yes, they are the pros at this. But it’s never the same as getting my own hands dirty and trying to understand how a product was designed and put together. It’s a lot more satisfying.
I’m hoping to add my personal thoughts along the way as well: There are questions and back stories that websites like ifixit.com leave untouched. Which CM produced this product? Why might the engineers have made certain design decisions? How does a particular section of circuitry work?
Lots of potential detective work. Let’s get started!
Gesture Recognition — Leap Motion
We are now living in the sci-fi future of being able to interact with computers just by using our hands. This type of consumer technology was pioneered by the original Kinect. The diminutive Leap Motion takes gesture recognition to a whole new level with its size, accuracy, and speed.
You may already know by now that the Leap Motion works by capturing stereographic video illuminated by near-IR LEDs at high frame rates. This information is sent over to the host PC where the mathematical magic happens to calculate the position of your hands and fingers. It’s quite ingenious actually if you dive into the patents behind it.
It’s been over a year since I got my Leap Motion. The device does what it promises really well in my opinion. But it’s been collecting a little bit of dust due to the lack of practical applications for me. With the resurgence of Virtual Reality headsets, I think the Leap Motion will have renewed interest from developers and users alike.
Taking The Covers Off
Did I say the Leap Motion is tiny? It’s a cool minimal design with a curved aluminum chassis that holds everything tightly packed together. The dark plastic cover — transparent to near IR wavelengths — is seamlessly glued onto the chassis. (more on that later)
The micro USB 3.0 port and the indicator LED round out the accessible features on the chassis itself. It is surprisingly difficult to get inside the device. The rubber bottom cover and the screws underneath come off easily but don’t get you any closer to the guts.
Unfortunately removing the top cover proved to be a little destructive. I had to put in a couple of notches in the chassis so that I could pry the cover open. You can see there is a carefully designed rubber cover over the top PCB. Note the bars on each side of the center LED and the partial cover on the side LEDs. I’m sure these are constructed to create the correct illumination coverage and leakage.
You can also see the wide-angle lenses that help create the sensing area in space for the device. Don’t worry I dig into the lenses and the image sensors in a little bit.
The one IC that’s visible through the rubber cover is a serial FLASH chip, the Macronix 25L320E, with 32 Mbits of space used for storing the firmware for the USB controller that we’ll see shortly.
The rubber cover comes off easily (small amount of glue) revealing the top PCB. The silkscreen has a very good clue about who’s manufacturing the Leap Motion! — “Ver06 SUNNY”.
Sunny Optical is a well known manufacturer of optical systems. It looks like Leap Motion decided to use them for the complete manufacturing of the device. This leads me to believe Sunny Optical also put the camera modules (image sensor + lens assembly) together as well.
The top PCB reveals another curious design detail. The left and the right IR LEDs have small riser PCBs to increase their z-stackup height in relationship to the center LED! This must be an important part of the optical design since they went through with the trouble of making these extra PCBs. I’m guessing the raised LEDs work together with the features on the rubber cover that partially cover these LEDs.
Let’s talk about some of the circuitry next.
Once the screws on the four corners are unscrewed, it’s easy to take off the top PCB which is attached to the bottom camera module PCB with a board-to-board connector. Each LED seems to have individual on/off circuitry that include a current limiting resistor (R305, R307, R309).
I’m guessing the crystals Y3 (32KHz) and Y2 (high frequency) are for the USB controller on the other side of the PCB. The top right group with the small inductor looks like a switching power supply, probably generating one of the rails from the USB power.
On the bottom side of the PCB you can see the USB controller IC — U5. This is the brains of the operation. It’s the CYUSB3014-BZX made by Cypress Semiconductor part of their FX3 SuperSpeed USB 3.0 peripheral controller family. Check out the press release from Cypress for details on how the whole system works at a high level.
The other major part on the bottom side is U301 which is the FDD6685 P-Channel MOSFET by Fairchild. My best guess is that this part of the circuit is used for adjusting the intensity of the IR LEDs. I’ll talk about the USB connector later in the bonus section.
The bottom PCB is attached to the plastic enclosure that’s screwed to the chassis. (the screws were underneath the outside bottom rubber cover). With a little careful prying, this plastic cradle comes off.
Of interest is the conductive mesh foam cube tucked near the USB port. This helps ground the USB 3.0 connector to the chassis for EMC/EMI purposes. And there is additional mesh tape on the bottom side of the camera PCB that helps ground this PCB to the chassis as well.
I’m thinking that the mesh also helps make good thermal connection to the outside chassis for heat sinking the cameras which will produce a decent amount of heat especially at high frame rates.
Not a lot of circuitry on this board other than the camera modules and supporting passive components. The lenses are threaded to the base of the camera modules and hot glued to set the focus. Do not unscrew these if you want to preserve the factory set focus and therefore proper operation of the Leap Motion.
I unscrewed the lens assemblies since I’m feeling adventurous. I’ll attempt to re-set the focus later on. You can’t yet see the image sensors themselves. I believe the rectangular pieces of glass are filters for only accepting light at the IR wavelengths of interest. It’s hard to tell how many elements the lens assembly is made of.
Looking at electronics under a high magnification microscope usually reveals many more details. Now that we’re done with disassembly, let’s take a look at several exciting magnified sections of the device.
You can see the USB 2.0 signal pair routed from the connector to L1 which is most likely a common-mode filter. You can also see the USB 3.0 signal pairs routed to U10 which is probably an ESD protection IC. Next, these signals are typically routed very carefully using impedance and length matching techniques into the controller IC.
The final treat is a close look at one of the image sensors! The square glass part is most likely the filter I mentioned earlier. The purplish rectangular part is the image sensor itself. I may speculate on the part number of this image sensor in the future. The aspect ratio is a big clue.
I can say with high confidence that this is a CMOS image sensor given the total price of the Leap Motion itself. The microscope I’m using doesn’t have high enough magnification to see the individual pixels on the sensor, sadly.
The parting shot of the Leap Motion completely disassembled.
I think the Leap Motion team came up with a very elegant and simple design. It’s always educational to see the tough assembly decisions and challenges fellow engineers worked on. I’m impressed with how tightly packed this little guy is.
I’ll report back how the reassembly goes. I’d like to improve my success rate.
Resurrection — (Updated on 12/01/2014)
This teardown has a happy ending: Yesterday morning I carefully reassembled the Leap Motion, cleaned the lenses, and ran the recalibration on the Control Panel.