Oct 22, 2016, hellekin
There is software. There is hardware. There is nightmare. Hellekin's venture into the scary world of Tombstone.
To many, EE looks like Lego: you put colorful pieces together and they all fit well. As you contemplate masters at work, their movement look natural, their painting obvious, their sculpture seamless.
You may think it's an easy task for hardened engineers and OpenMoko veterans to fit the pieces of the puzzle together. Well, not quite. As colorful as they are, the pieces tend to give headaches from sourcing to placement (and replacement). Electronics engineers not only have to deal with the inherent complexity of their field, they also need to care for the economics part (are these pieces available? Will this supplier be shipping in time? Etc.)
Trying to follow the daily conversations between Joerg and Werner regarding the details of the Neo900 design, I often feel like a trainee who knows not enough to wonder in awe, and just enough to understand this is definitely not as easy as they make it look. They start throwing TLAs all around and then obscure deep-links that get you to PDFs that cannot shy in comparison to EU law and bureaucratic jargon: in my library, they'd all go to the same 'encrypted books' where I store Chinese and Farsi presents I might read when someone visits and knows the language. Yet these two seem like ballet dancers at the Bolchoi, catching meaning in mid-air, responding to each other with references I can't fathom, yet alone catch a causal link between one line and the next.
In theory, there's no difference between theory and practice. In practice, there is.
-- Jan L. A. van de Snepscheut
For example: how do you connect the UPPER board of the Neo900 with the LOWER board?
Easy I'd say: align the UPPER and LOWER boards, place the components according to the specs, and let the reflow machine do its work: done. Right?
Que nenni! How would you lay butter in a closed sandwich? Each board has its components applied separately, so the boards need to match exactly. We're talking precision here. A tenth of a millimeter or two. Well, I'd say, with Werner 'littlefinger' Almesberger on our team, we can do magic. Indeed, you'd reply, but magic doesn't exist.
In theory, apply the map (schematics) to the territory (the SMD) works. But in practice, there are many details to look out for that may ruin the show if not considered carefully. Without going to the extremes of mass-manufacturing mishaps like the recent exploding Samsung Galaxy Note 7, which seems to have been a terrible disaster, but when put into perspective, only represents a mere 0.01% of devices produced, and a total loss of 1% of their annual revenue, electronics has its non-deterministic magic, like the Tombstone effect: this happens when the solder paste applied to each component's pads don't melt evenly, creating torque that will lift the tiny critter on its side.
See, when the solder paste changes phase, it can happen that a component moves from its original position, it drifts slightly from where it's supposed to be, and that "slightly" can make all the difference between a working device and a wasted circuit. No matter how precise you are, nor what soldering technique you use, here you enter the fantastic realm of chemical complexity, fluid mechanics, and homegeneity.
In the prototype phase, you would resolder the faulty component and that would take a few hours at most. But in production, this type of event can ruin the batch. Samsung didn't fall for this, but for pressing a bit too much the Lithium Ion battery, leading to a fragile battery: when they heat up, a damaged cell can ignite the volatile and flamable chemicals filling the pressurized battery. This happens all the time to many products: from hoverboards to Tesla cars, and even laptops and iPhones. Cases abound over the years. Of course the larger the production, the higher the probability of something going wrong.
Fortunaly it's not as scary as it sounds: production lines are aware of these issues and engineers making machines take a lot of precautions to lower the probabilities to extreme rarity. But still, it requires a lot of attention.
I learned about the tombstone effect as an extreme (and fun) illustration of drifting components: that using SMD technology, sometimes a small component can float in the soldering paste as it liquefies, eventually leading to this funny effect (also called Manhattan effect for its vague resemblance to the skyline in New York City--it would take a serious amount of unluck to 'grow' several capacitors at once).
Joerg wanted to explain to me the problematic of using B2B connectors between UPPER and LOWER boards. The case is different from either the exploding batteries and the tombstone effect, as the connectors are rather large components. Yet, their relative alignment on the board require a precision of about 0.2 mm so they can clip in easily.
In the course of my understanding I went from awe to boredom, since I discovered that kind of considerations can look scary (especially when it concerns mass-production and the consequent recalls like the case of Samsung Galaxy Note 7) but are in fact one in a myriad of elements experienced engineers need to take into account on a daily basis. And that's what makes the difference between looking easy and being easy.
Here, a small insight into the complexity of research & development, and how it goes way beyond simple puzzle solving: sometimes, venturing into puzzle design can be puzzling. Bottom-line: at any scale details matter!
The Neo900 project aims to provide a successor of N900 Nokia Internet Tablet™ device, with faster CPU, more RAM and LTE modem, basing efforts on an already existing, mature and stable free platform - the OpenPhoenux GTA04, following the spirit of freedom known from Openmoko devices.