Even before Nintendo Switch started shipping at the beginning of March, the left Joy-con was experiencing intermittent connectivity issues. Antenna connectivity is always a tricky problem because there’s so many different things (mechanical, electrical, and software) that can affect antenna performance. Nintendo has come out in the meantime and said that the issue was caused by “manufacturing variation” and has now been corrected. Sam and I love a good failure analysis hunt, so we acquired a set of Joy-cons with this issue from a friend, and set about our investigation.
Intermittent behavior usually requires that something is changing throughout use or over time with changing conditions (such as heat or humidity). For antennas, there is also an element of range and sensitivity to interference, which could play a role. Following our methodology for failure analysis, we brainstormed a list of hypotheses that could cause intermittent performance in Nintendo Switch, and then tore the units down layer by layer to either support or refute the hypothesis. Sam used the Instrumental System to capture detailed images that a Nintendo engineer might have been able to use to debug the problem.
Potential Root Cause: Hands detune the antenna.
Something we noticed right away during the teardown, is that the antennas for the left Joy-con and right Joy-con are of completely different design and are in different locations. Based on what we assume to be a pretty standard grip, the left Joy-con antenna is right underneath where Sam’s hand mashes up next to the enclosure, where the right Joy-con antenna seems to have more room to breathe. While something like this could definitely cause an intermittent issue (and is a bit reminiscent of the famous antennagate), it’s a fundamental design issue in the product and wouldn’t be a “manufacturing variation”.
Potential Root Cause: Bad connection between PCB and antenna
The left Joy-con has a “board antenna” -- the antenna is embedded directly into the printed circuit board. While there could still be a manufacturing defect in the PCB itself, or variation in how it's de-panelized, we don’t see any evidence of anything out of the ordinary here.
Potential Root Cause: Poor solder for a critical bluetooth component
Poor solder is most apparent on passives -- and those in the vicinity look good. There is also a solder “jumper” (likely used for antenna testing) that looks OK, too.
Potential Root Cause: Other components or the antenna are moving with respect to each other, changing the “load” on the antenna
Antennas need room to breathe, especially with respect to metal components in the device. It seems reasonable that with impassioned play, it might be possible to squeeze the device and move components closer or further away from the antenna. Alternatively, maybe components inside the joystick (which moves) could be at fault.
This is where our failure analysis hunt got interesting. We noticed that there is a flex circuit that runs through the bluetooth antenna area. Flexes have metal traces, which you’d want to keep a controlled distance away from the sensitive tip of the antenna. The flex has been flat-patterned for maximal efficiency (it’s straight with no bend in the design), even though it routes through a bit of a tortuous path. As a result, depending on assembly, that flex could end up in multiple places: along the side wall, right on top of the antenna, or high up above the antenna. This seems like the most likely cause of the “manufacturing variation” that Nintendo mentions.
Based on a quick teardown of a left Joy-con with the reported intermittent signal issue, the flex seems to be the most likely culprit. We did see another post that said Nintendo is refurbishing the early Joy-cons by adding a piece of foam on the antenna. While the poster believes the foam is conductive -- that doesn’t add up for us (putting metal right next to an antenna is usually a bad thing). It’s more likely that the foam is acting as a spacer that is keeping the flex from getting too close to the antenna, and enforcing a consistent air-gap.