We’ve become sadly accustomed to consumer devices that seem to give up the ghost right after the warranty period expires. And even when we get “lucky” and the device fails while it’s still covered, chances are that there will be no attempt to repair it; the unit will be replaced with a new one, and the failed one will get pitched in the e-waste bin.
Not every manufacturer takes this approach, however. When premium quality is the keystone of your brand, you need to take field failures seriously. [Dalibor Farný], maker of high-end Nixie tubes and the sleek, sophisticated clocks they plug into, realizes this, and a new video goes into depth about the process he uses to diagnose issues and prevent them in the future.
One clock with a digit stuck off was traced to via failure by barrel fatigue, or the board material cracking inside the via hole and breaking the plated-through copper. This prompted a board redesign to increase the diameter of all the vias, eliminating that failure mode. Another clock had a digit stuck on, which ended up being a short to ground caused by pin misalignment; when the tube was plugged in, the pins slipped and scraped some solder off the socket and onto the ground plane of the board. That resulted in another redesign that not only fixed the problem by eliminating the ground plane on the upper side of the board, but also improved the aesthetics of the board dramatically.
As with all things [Dalibor], the video is a feast for the eyes with the warm orange glow in the polished glass and chrome tubes contrasting with the bead-blasted aluminum chassis. If you haven’t watched the “making of” video yet, you’ve got to check that out too.
Blindsided by missing pin allocations? Perhaps you’re working on a piece of hardware and you notice that the documentation is entirely wrong. How can you get your device to work?
[Dani Eichhorn]’s troubles began when running an IoT workshop using a camera module. Prior to the work, no one had through to check if all of the camera modules ordered for the participants were the same. As it turns out, the TTGO T-CAM module had a number of revisions, with some even receiving a temperature/pressure sensor fixed on top of the normal board.
While the boards may have looked the same, their pin allocations were completely different.Changing the pin numbers wouldn’t have been difficult if they were simply numbered differently, but because the configurations were different, errors started to abound: Could not initialize the camera
As it turns out, even the LillyGo engineers – the manufacturers of the board – may have gotten a bit lost while working on the pin allocations, as [Eichhorn] was able to find some of the pins printed right onto the PCB, hidden behind the camera component.
To find information not printed on the board, a little more digging was required. To find the addresses of the devices connected to the I2C bus, running a program to find peripherals listening on the bus did the trick. This was able to print out the addresses of the SSD1306 OLED display driver and the microphone for the board at hand.
To find the pins of peripherals not printed on the PCB or hidden on the silkscreen, a GPIO scanner did the trick. This in particular worked for finding the PIR (passive infrared) motion sensor.
We picked up a few tips and tricks from this endeavor, but also learned that reverse-engineering anything is hard, and that there isn’t any one method for finding pin allocations when the documentation’s missing.
One of the step changes in electronic construction at our level over the last ten or fifteen years has been the availability of cheap high-quality printed circuit boards. What used to cost hundreds of dollars is now essentially an impulse buy, allowing the most intricate of devices to be easily worked with. Many of us have put away our etching baths for good, often with a sigh of relief.
We’re pleased that [Riyas] hasn’t though, because they’ve etched an STM32 dev board that if we didn’t know otherwise we’d swear had been produced professionally. It sports a 176-pin variant of an STM32F4 on a single-sided board, seemingly without the annoying extra copper or lack-of-copper that we remember from home etching. We applaud the etching skill that went into it, and we’ll ignore the one or two boards that didn’t go entirely to plan. A coat of green solder mask and some tinning, and it looks for all the world as though it might have emerged from a commercial plant. All the board files are available to download along with firmware samples should you wish to try making one yourself, though we won’t blame you for ordering it from a board house instead.
It’s always nice to see that single board computers are not the sole preserve of manufacturers. If the RC2014 Micro doesn’t isn’t quite your style, there’s always the Blueberry Pi which features a considerably higher penguin quotient.
[Mastro Gippo] found himself in a pickle recently, with the development of an enclosure for the Prism electric vehicle charger. The body had been sorted out, but the front cover needed work. It had to be visually appealing, and ideally should provide the user feedback on the charging process. After some thought, [Mastro] decided to explore the possibilities of using a PCB as a part of a commerical product enclosure.
For a variety of reasons, using a specially designed PCB was an attractive solution for the team’s enclosure. Wanting something cost effective, easily customizable, and something that would help with emissions compliance, a PCB seemed like a great idea. With this in mind, [Mastro] prepared a series of prototypes. These feature see-through sections for LEDs to shine through, as well as a capacitive button and gold-plated logo. The fact that the front cover is a PCB makes the integration of the electronic components a cinch.
Before heading into full production, [Mastro] began to question why this technique isn’t used more often. Deciding to research further, [Bunnie Huang] was tapped to provide some advice on the concept. Noting that there can be issues with lead content, as well as the fact that PCBs aren’t often produced with proper regard to aesthetics, there were some pitfalls to the idea. Additionally, ESD testing can be difficult, while the in-built capacitive button would face issues in wet conditions.
None of these are showstoppers however, and [Mastro] has persevered, combining the front cover PCB with an adhesive plastic sheet for added protection. We fully expect that if more manufacturers explore this route, it may be a more viable technique in future. It’s also a great example of knowing when to ask others for help – it’s not the first time we’ve learned from [Bunnie’s] broad experience!
There are a few common lessons that get repeated by anyone who takes on the task of assembling a few hundred PCBs, but there are also unique insights to be had. [DominoTree] shared his takeaways after making a couple hundred electronic badges for DEFCON 26 (that’s the one before the one that just wrapped up, if anyone’s keeping track.) [DominoTree] assembled over 200 Telephreak badges and by the end of it he had quite a list of improvements he wished he had made during the design phase.
Some tips are clearly sensible, such as adding proper debug and programming interfaces, or baking an efficient test cycle into the firmware. Others are not quite so obvious, for example “add a few holes to your board.” Holes can be useful in unexpected ways and cost essentially zero. Even if the board isn’t going to be mounted to anything, a few holes can provide a way to attach jigs or other hardware like test fixtures.
Other advice is more generic but no less important, as with “eliminate as many steps as possible.” Almost anything adds up to a significant chunk of time when repeated hundreds of times. To the basement hacker, something such as pre-cut and pre-tinned wires might seem like a shameful indulgence. But cutting, stripping, tinning, then hand-soldering a wire adds up to significant time and effort by iteration number four hundred (that’s two power wires per badge) even if one isn’t staring down a looming deadline.
Even if you don’t work in a nuclear power plant, you might still want to use a Geiger counter simply out of curiosity. It turns out that there are a lot of things around which emit ionizing radiation naturally, for example granite, the sun, or bananas. If you’ve ever wondered about any of these objects, or just the space you live in, it turns out that putting together a simple Geiger counter is pretty straightforward as [Alex] shows us.
The core of the Geiger counter is the tube that detects the radiation. That’s not something you’ll be able to make on your own (probably) but once you have it the rest of the build comes together quickly. A few circuit boards to provide the tube with the high voltage it needs, a power source, and a 3D printed case make this Geiger counter look like it was ordered from a Fluke catalog.
The project isn’t quite finished ([Alex] is still waiting on a BNC connector to arrive) but seems to work great and isn’t too complicated to put together, as far as Geiger counters go. He did use a lathe for some parts which not everyone will have on hand, but a quick trip to a makerspace or machinist will get you that part too. We’ve seen some other parts bin Geiger counters too, so there’s always a way around things like this.
One of the more popular ways of rolling out your own custom PCB is to simply create the model in your CAD program of choice and send it off to a board manufacturer who will take care of the dirty work for you. This way there is no need to deal with things like chemicals, copper dust, or maintaining expensive tools. A middle ground between the board manufacturer and a home etching system though might be what [igorfonseca83] has been doing: using an inexpensive laser engraver to make PCBs for him.
A laser engraver is basically a low-power laser CNC machine that’s just slightly too weak to cut most things that would typically go in a laser cutter. It turns out that the 10W system is the perfect amount of energy to remove a mask from a standard PCB blank, though. This in effect takes the place of the printer in the old toner transfer method, and the copper still has to be dissolved in a chemical solution, but the results are a lot more robust than trying to modify a printer for this task.