Surface mount devices can take some adjusting to for hackers primarily used to working with through-hole components. Despite this, the lure of the hottest new parts has enticed even the most reticent to learn to work with the technology. Of course, time rolls on and BGA parts bring further difficulties. [Nate] from SparkFun worked on the development of the RedBoard Artemis, and broke down the challenges involved.
The RedBoard Artemis is an Arduino-compatible devboard built around the Ambiq Apollo3 chip. In addition to packing Bluetooth and 1 MB of Flash, it’s also capable of running TensorFlow models and using tiny amounts of power. The chip comes in an 81-Ball Grid Array at 0.5mm pitch, which meant SparkFun’s usual PCB fabrication methods weren’t going to cut it.
An initial run of prototype boards was run using 4 layers, blind and buried vias, and other fancy tricks to break out all the necessary signals. While this worked well, it was expensive and inefficient. The only part of the board that needed such fabrication was around the chip itself; the rest of the board could be produced with cheaper 2-layer methods. To improve this for mass production, instead, an SMD module was created to house the Apollo3, which could then be dropped into new designs on cheaper boards as necessary.
[Nate] does a great job of explaining the engineering involved, as well as sharing useful tips for others going down a similar path. So far, this is just part 1, with future posts promising to cover the RF shield design and FCC certification process. [Nate] has always been keen to share his wisdom, and we can’t wait to see what comes next!
Hackaday Editors Mike Szczys and Elliot Williams are back after last week’s holiday break to track down all of the hacks you missed. There are some doozies; a selfie-drone controlled by your body position, a Theremin that sings better than you can, how about a BGA hand-soldering project whose creator can’t even believe he pulled it off. Kristina wrote a spectacular article on the life and career of Mary Sherman Morgan, and Tom tears down a payment terminal he picked up in an abandoned Toys R Us, plus much more!
Take a look at the links below if you want to follow along, and as always tell us what you think about this episode in the comments!
Most Hackaday readers will be a pretty dab hand with a soldering iron. We can assemble surface-mount boards, SOICs and TSSOPs are a doddle, 0402s we take in our stride, and we laugh in the face of 0201s. But a Twitter thread from [Greg Davill] will probably leave all but the most hardcore proponents of the art floundering, as he hand-wires a tiny FPGA in a BGA package to the back of a miniature dot-matrix LED display module.
As far as we can see the module must once have had its own microcontroller which has been removed. We’d guess it was under an epoxy blob but can’t be sure, meanwhile its pads are left exposed. The Lattice LP1k49 fits neatly into the space, but a web of tiny wires are required to connect it to those pads. First, [Greg] populates the pads with a set of “tombstoned” tiny (we’re guessing 0R) resistors, then wires them to the pads with 30μm wire. He describes a moment of confusion as he attempts to tin a stray hair, which burns rather than accepting the solder.
The result is a working display with a new brain, which surprises even him. We’ve seen more than one BGA wiring over the years, but rarely anything at this scale.
You may have heard the phrase “flip-chip” before: it’s a broad term referring to several integrated circuit packaging methods, the common thread being that the semiconductor die is flipped upside down so the active surface is closest to the PCB. As opposed to the more traditional method in which the IC is face-up and connected to the packaging with bond wires, this allows for ultimate packaging efficiency and impressive performance gains. We hear a lot about advances in the integrated circuits themselves, but the packages that carry them and the issues they solve — and sometimes create — get less exposure.
Let’s have a look at why semiconductor manufacturers decided to turn things on their head, and see how radioactive solder and ancient Roman shipwrecks fit in.
It’s said that the electronic devices we use on a daily basis, particularly cell phones, could be so much smaller than they are if only the humans they’re designed for weren’t so darn big and clumsy. That’s only part of the story — battery technology has a lot to do with overall device size — but it’s true that chips can be made a whole lot smaller than they are currently, and are starting to bump into the limit of being able to handle them without mechanical assistance.
Or perhaps not, if [mitxela]’s hand-soldering of a tiny ball-grid array chip is any guide. While soldering wires directly to a chip is certainly a practical skill and an impressive one at that, this at least dips its toe into the “just showing off” category. And we heartily endorse that. The chip is an ATtiny20 in a WLCSP (wafer-level chip-scale package) that’s a mere 1.5 mm by 1.4 mm. The underside of the chip has twelve tiny solder balls in a staggered 4×6 array with 0.4 mm pitch. [mitxela] tackled the job of soldering this chip to a 2.54-mm pitch breakout board using individual strands from #30 AWG stranded wire and a regular soldering iron, with a little Kapton tape to hold the chip down. Through the microscope, the iron tip looks enormous, and while we know the drop of solder on the tip was probably minuscule we still found ourselves mentally wiping it off as he worked his way across the array. In the end, all twelve connections were brought out to the protoboard, and the chip powers up successfully.
If you’re an Android fan, there’s a good chance you’ve heard of the Nexus 5X. The last entry in Google’s line of low-cost Nexus development phones should have closed the program on a high note, or at the very least maintained the same standards of quality and reliability as its predecessor. But unfortunately, a well known design flaw in the Nexus 5X means that the hardware is essentially a time-bomb. There are far too many reports of these phones entering into an endless bootloop right around the one year mark to say it’s just a coincidence.
The general consensus seems to be that faulty BGA chip soldering on the CPU works lose after about a year or so of thermal stress. Whatever the reason, [hillbillysam] recently found himself the proud owner of a dead Nexus 5X. Resigned to the fact that he would need to get a new phone, he at least wanted to get some of his data off the device before it went to that big landfill in the sky.
As it turns out these bootlooped phones can temporarily be revived by cooling them down, say by putting them in the freezer for a few hours. There’s plenty of debate as to why this works, but even our own [Lewin Day] can testify that it does seem to get the phone booting again; though only until it comes back up to operating temperature. With this in mind, [hillbillysam] reasoned that if he kept the phone as cold as possible while it was running, it may stay operational long enough for him to pull his files off of it over USB.
He couldn’t exactly freeze the phone in a block of ice, but remembering his high school chemistry, he came up with something pretty close. By adding salt to water, you can significantly lower temperature at which it freezes. Putting the phone into a watertight bag and submerging it in this supercooled solution is an easy and non-destructive way of keeping it very cold while still being accessible over USB.
His Nexus 5X was able to keep kicking the whole time it was luxuriating in its below-freezing saltwater bath, giving him plenty of time to copy everything he needed. It doesn’t sound like the kind of spa day we’d like to have personally, but to each their own.
It’s amazing how hackers are nowadays building increasingly complex hardware with SMD parts as small as grains of sand. Getting multilayer PCB’s and soldering stencils in small quantities for prototyping is easier than ever before. But Pick-and-Place — the process of taking parts and stuffing them on the PCB in preparation for soldering — is elusive, for several reasons. For one, it makes sense only if you plan to do volume production as the cost and time for just setting up the PnP machine for a small run is prohibitive. And a desktop PnP machine isn’t yet as ubiquitous as a 3D printer. Placing parts on the board is one process that still needs to be done manually. Just make sure you don’t sneeze when you’re doing it.
Of course the human is the slow part of this process. [Colin O’Flynn] wrote a python script that he calls MeatBagPnP to ease this bottleneck. It’s designed to look at a row in a parts position file generated from your EDA program and highlight on a render of the board where that part needs to be placed. The human then does what a robotic PnP would have done.
A bar code scanner is not necessary, but using one does make the process a bit quicker. When you scan a code on the part bag, the script highlights the row on the spreadsheet and puts a marker on the first instance of it on the board. After you’ve placed the part, pressing the space bar puts a marker on the next instance of the same value. The script shows it’s done after all parts of the same value are populated and you can then move on to the next part. If you don’t have a bar code scanner handy, you can highlight a row manually and it’ll tell you where to put that part. Check it out in the video below.
Of course, before you use this tool you need some prior preparation. You need a good PNG image of the board (both sides if it is double-sided) scaled so that it is the same dimensions as the target board. The parts position file generated from your EDA tool must use the lower left corner of the board as the origin. You then tell the tool the board dimensions and it scales up everything so that it can put the red markers at the designated XY positions. The script works for single and double-sided boards. For a board with just a few parts, it may not be worth the trouble of doing this, but if you are trying to manually populate a complex board with a lot of parts, using a script like this could make the process a lot less painful.
The project is still fresh and rough around the edges, so if you have comments or feedback to offer, [Colin] is listening.
[Colin]’s name ought to ring a bell — he’s the hacker who built ChipWhisperer which took 2nd Prize at The Hackaday Prize in 2014. The MeatBagPnP project is a result of having worked at building increasingly complex boards manually and trying to make the process easier. In addition to the walk-through of how the script works after the break we’ve embedded his other video from three years back when he was stuffing parts — including BGA’s — the hard way and then reflowing them in a Chinese oven with hacked firmware.