It’s the most basic of functions for a camera, that when you point it at a scene, it produces a photograph of what it sees. [Jasper van Loenen] has created a camera that does just that, but not perhaps in the way we might expect. Instead of committing pixels to memory it takes a picture, uses AI to generate a text description of what is in the picture, and then uses another AI to generate an image from that picture. It’s a curiously beautiful artwork as well as an ultimate expression of the current obsession with the technology, and we rather like it.
The camera itself is a black box with a simple twin-lens reflex viewfinder. Inside is a Raspberry Pi that takes the photo and sends it through the various AI services, and a Fuji Instax Mini printer. Of particular interest is the connection to the printer which we think may be of interest to quite a few others, he’s reverse engineered the Bluetooth protocols it uses and created Python code allowing easy printing. The images it produces are like so many such AI-generated pieces of content, pretty to look at but otherworldly, and weird parallels of the scenes they represent.
It’s inevitable that consumer cameras will before long offer AI augmentation features for less-competent photographers, meanwhile we’re pleased to see Jasper getting there first.
Before video games, there were pinball machines. Not that they don’t exist today, but a modern pinball machine will likely have microprocessors and other fancy things that traditional pinball machine designers could never dream of. [Eli] had one of these mechanical machines from 1974 as a kid and, later, encountered a more modern machine with a rudimentary microprocessor and other integrated circuits onboard. One thing this enabled is the ability to remember high scores. But you have to physically look at the machine, and you can only see the top four scores. [Eli] decided to adapt the machine to upload high score data to the Internet, and it is a fun project.
[Eli]’s design goals were to make it automatic and robust. That is, if the network is down or the machine loses power, you shouldn’t lose high score data. In addition, he didn’t want to change the appearance or damage the 40-year-old machine. You can see a video of how it all turned out below.
The Laser Cue machine is one of many built around the “Williams System 7” platform. A 6808 CPU, along with some I/O chips to manage all the lights, sensors, and bells. The game has only 1K of RAM, 12K or ROM, and 128 bytes (no prefix, just bytes) of RAM with battery backup. There was even a common “operating system” called Flipper ROM, and that’s actually documented over on GitHub.
The ESP32 version of the WiFi interface board
Since the memory for the machine is all in external chips, it was a reasonable idea to replace the CPU with a board that monitored signals on the board. The CPU would plug into this new board, and then a newer microcontroller with an Internet connection could eavesdrop on bus traffic. However, removing the old CPU and jamming pins into the ancient socket was worrisome, so instead, [Eli] elected to tap into a test connector that was already on the board but not plugged into anything.
An ESP32 is more than capable of the speeds, although connecting to 5 V logic was a bit of a problem. The CPU has 5 V tolerant pins, but some of the 25 available pins on the development board either set items on boot or may briefly be outputs and were thus unusable. To reduce the necessary pins, [Eli] decided to do some of the decoding in separate logic. Instead of using TTL chips, he elected to use a programmable logic array.
After that, it seemed it would be straightforward, but there was something preventing the ESP32 from reading each bus cycle. [Eli] never got to the bottom of it but instead switched to the Raspberry Pi Pico W. Using the chip’s special I/O processors made the job easy, and it worked perfectly. The rest of the project was just fit and finish. Be sure to read to the end to find out the lessons learned which might help you on your next similar project.
A modern DIY machine might even have an FPGA inside. Don’t have room for a big full-sized pinball machine? No problem.
[Shelby] at Tech Tangents recently wrapped a project / obsession to obtain an old HP ScanJet 4C, get it running on a PC and put it through its paces. After after nearly five years, three scanners, and untold SCSI cards and drivers later, he finally succeeded. The first big problem was getting a working scanner. These don’t stand up well to shipping, and one arrived with broken mirrors. And when he finally got one that worked, sorting out SCSI controller and driver issues was surprisingly complicated, though ultimately successful.
The HP ScanJet 4C was introduced in 1995, and was notable for its scanning quality, its resolution ( 2400 DPI interpolated / 600 DPI optical ), and selling for under $1000. Except for replacement parts concerns, particularly the customized triphosphor fluorescent bulb assembly, it would still be a very competent scanner today. For this reason, [Shelby] will not be using it as his daily use scanner. Continue reading “Deep Dive Into The HP ScantJet 4C”→
Moore’s law isn’t strictly holding anymore, but it is still true that most computing systems are at least trending towards lower cost over time, if not also slightly smaller size. This means wider access to less expensive hardware, even if that hardware is still an 8-bit microcontroller. While some move on to more powerful platforms as a result of this trend, there are others still fighting to push these platforms to the edge. [lcamtuf] has been working to this end, stretching a small AVR microcontroller to not only play a classic video game, but to display it on a color display. Continue reading “Pushing Crates In 8-bit Color”→
A regular pair of pliers is fine most of the time, but for delicate work with squarish objects you can’t go wrong with a pair of parallel pliers. [Neil Paskin] decided to make his own pair from scratch. (YouTube)
The jaws were machined down from round stock in [Paskin]’s mill before heat treating and tempering. The steel portions of the handles were cut from 16 gauge plate steel and half of them were stamped on a fly press to make the bridging section around the pivot bolt. The filler for the handles is copper on one side and brass on the other as [Paskin] didn’t have enough brass of the correct size to do both.
The steel and filler were joined with epoxy and copper pins before beveling the edges and sanding to give a comfortable contour to the handles. The bolts for the pliers started as ordinary hex bolts before being machined down on the lathe to a more aesthetically-pleasing shape and size. The final touches included electrolytically etching a logo into the bridge and then spraying down the pliers with a combination lubricant and corrosion preventative spray. This is surely a pair of pliers worth handing down through the generations.
Over the past two decades, e-paper has evolved from an exotic and expensive display technology to something cheap enough to be used for supermarket price tags. While such electronic shelf labels are now easy to find, actually re-using them is often tricky due to a lack of documentation. Luckily, [Aaron Christophel] has managed to reverse engineer many types of shelf labels, and he’s demonstrated the results by turning one into an ultra-low-power clock called Triink. It’s based on a 128×296 pixel e-ink display paired with an nRF52832 BlueTooth Low-Energy SoC and uses just 65 micro-amperes on average: low enough to keep it running for more than a year on a single battery charge.
Power on the left, e-ink on the right: the custom PCB is clever and compact, too
The clock is housed in an enclosure that’s simple but effective: a 3D-printed triangular prism with a slot for the screen and space for the 18650 lithium battery. One side can be opened to access the internal components, although that’s really only needed to charge the battery. You can see how cleverly everything snaps together in the video embedded below. Continue reading “Low Power Challenge: E-Paper Shelf Label Becomes Ultra-Frugal Clock”→
HF radios often use toroidal transformers and winding them is a rite of passage for many RF hackers. [David Casler, KE0OG] received a question about how they work and answered it in a recent video that you can see below.
Understanding how a conventional transformer works is reasonably simple, but toroids often seem mysterious because the thing that makes them beneficial is also what makes them confusing. The magnetic field for such a transformer is almost totally inside the “doughnut,” which means there is little interaction with the rest of the circuit, and the transformer can be very efficient.
The toroid itself is made of special material. They are usually formed from powdered iron oxide mixed with other metals such as cobalt, copper, nickel, manganese, and zinc bound with some sort of non-conducting binder like an epoxy. Ferrite cores have relatively low permeability, low saturation flux density, and low Curie temperature. The powder also reduces the generation of eddy currents, a source of loss in transformers. Their biggest advantage is their high electrical resistivity, which helps reduce the generation of eddy currents.
If you haven’t worked through how these common little transformers work, [David]’s talk should help you get a grip on them. These aren’t just for RF. You sometimes see them in power supplies that need to be efficient, too. If you are too lazy to wind your own, there’s always help.
