We know, you’ve already got a USB to serial adapter. Probably several of them, in fact. But that doesn’t mean you couldn’t use one more — especially when it’s as as cleverly designed as this one from [Anders Nielsen].

The first thing you notice about this adapter, and the big departure from the ones that are likely littering your parts bin, is that it terminates in a full-size male DSUB9 connector. With the ability to be directly plugged into a RS-232 port, this adapter will certainly catch the eye of retrocomputer enthusiasts. With a clever arrangement of jumpers, you can even reconfigure the RX and TX lines to be straight-through or cross over as needed.

But if you’re working with something that doesn’t have a literal serial port, no worries. All of the lines coming from the CH340G chip are broken out to a header so you can connect it up to whatever device you’re working with via jumpers.

In fact, if you’re really sure you’ll never need that RS232 feature, the PCB is even designed in such a way that you can simply snap it off. Admittedly it might seem a little odd to get a device like this if you didn’t want that capability. But once broken off, it’s not like the components go to waste. [Anders] has designed the board in such a way that if you flip it over and install a right-angle header, you can use the RS232 segment on a breadboard.

But the list of features doesn’t stop there. There’s also a 3.3 V regulator on board that you can use to power external circuits, as well as breakouts for the data lines in the USB-C connector. In keeping with the theme of the device, that part of the PCB can also be snapped off if you want to use it elsewhere.

Most folks probably’ won’t need all the capabilities offered by this particular serial adapter, and that’s fine. We’re still happy that it’s out in the wild and available for the community to use and adapt as an open source project.

You wouldn’t typically associate graphing calculators with artificial intelligence, but hacker [KermMartian] recently made it happen. The innovative project involved running a neural network directly on a TI-84 Plus CE to recognize handwritten digits. By using the MNIST dataset, a well-known collection of handwritten numbers, the calculator could identify digits in just 18 seconds. If you want to learn how, check out his full video on it here.

The project began with a proof of concept: running a convolutional neural network (CNN) on the calculator’s limited hardware, a TI-84 Plus CE with only 256 KB of memory and a 48 MHz processor. Despite these constraints, the neural network could train and make predictions. The key to success: optimizing the code, leveraging the calculator’s C programming tools, and offloading the heavy lifting to a computer for training. Once trained, the network could be transferred to the calculator for real-time inference. Not only did it run the digits from MNIST, but it also accepted input from a USB mouse, letting [KermMartian] draw digits directly on the screen.

While the calculator’s limited resources mean it can’t train the network in real-time, this project is a proof that, with enough ingenuity, even a small device can be used for something as complex as AI. It’s not just about power; it’s about resourcefulness. If you’re into unconventional projects, this is one for the books.

Most people know about the numerical constant pi (or π, if you prefer). But did you know that pi has an evil twin represented by the symbol ϖ? As [John Carlos Baez] explains, it and its related functions are related to the lemniscate as pi relates to circles. What’s a lemniscate? That’s the proper name for the infinity sign (∞).

[John] shows how many of the same formulas for pi also work for the lemniscate constant (the name for ϖ). Some  (as John calls them) “mutant” trig functions use the pi-like constant.

Mathematically, a circle is a point (the center) with a curve that describes x2+y2=r2. The lemniscate is a particular instance of a Cassini oval where r2=cos2θ. We all know the circumference of a circle—basically, the perimeter—is 2π; the perimeter of the lemniscate is 2ϖ.

Why does any of this matter? Well, [John] shows how it connects to elliptic curves and the Gauss constant.

Like pi, the lemniscate constant probably never ends, but it is roughly 2.622057. Will this be useful in your next project? Probably not. Will it help you win some bar bets? Maybe.

Then again, if you are bored calculating more digits of pi, here’s something new to try. Not that you need that many digits.

If we’ve learned anything over the years, it’s that the only thing hardware hackers love more than a device festooned with buttons is one that’s covered in LEDs — so it’s no surprise that this “Mr Christmas” jukebox caught the eye of [Roberts Retro]. But while the holiday gadget might have been mildly entertaining in its stock configuration, he quickly realized that what it really needed was an ESP32 retrofit. After all, what good are all those buttons and LEDs if you can’t bend them to your will?

For the first half of the video, [Robert] treats us to a detailed teardown of the device, which as you might imagine, is largely hollow inside. This gave him plenty of room to graft in new hardware, which is really the best gift any of us could hope to find under the tree. In addition to the ESP32 development board, the jukebox also received a number of WS2812B addressable RGB LEDs, and a DFPlayer module to handle music playback.

With all the buttons wired up to inputs on the ESP32, [Robert] can reconfigure the jukebox to do pretty much whatever he wants with just changes to the software. In the video, he demonstrates how the buttons can be used to trigger the playback of individual songs stored on the DFPlayer’s SD card, which essentially replicates it’s stock functionality. A few lines of changed code later, those same buttons can be used to control devices via Home Assistant.

To get into the holiday spirit, [Roberts Retro] shows off the completed jukebox controlling his ESP-enabled LEGO train set — another of his festive upgrades that we covered back in 2022.

We all know and love the humble seven-segment display, right? And if you want to make characters as well as numbers, you can do an okay job with sixteen segments off the shelf. But if you want something more art-deco, you’ll probably want to roll your own. Or at least, [Ben] did, and you can find his designs up on GitHub.

Taking inspiration from [Posy]’s epic investigation of segmented displays, [Ben] sat down with a sketchpad and created his own 20-segment font that displays numbers and letters with some strange, but frankly lovely, segment shapes. There is no center line, so letters like “T” and numbers like “1” are a little skewed, but we think it’s charming.

We’ve seen about a bazillion takes on the seven-segment idea over the years here. Most recently, we fell in love with this 21-segment beauty, but honestly the original eight(!) segment patent version is charming as well. Anyway, picking a favorite segmented display at Hackaday is like picking your favorite child, if you have a few hundred children. We love them all.

Thanks [Aaron] for the tip!

An unfortunate reality of pre-1990s computer systems is that any manuals and documentation that came with them likely only existed on paper. That’s not to say there aren’t scanned-in (PDF) copies of those documents floating around, but with few of these scans being indexable by search engines like Google and Duck Duck Go, they can be rather tricky to find. That’s where the Manx catalog website seeks to make life easier. According to its stats, it knows about 22,060 manuals (9,992 online) across 61 websites, with a focus on minicomputers and mainframes.

The code behind Manx is GPL 2.0 licensed and available on GitHub, which is where any issues can be filed too. While not a new project by any stretch of the imagination, it’s yet another useful tool to find a non-OCR-ed scan of the programming or user manual for an obscure system. As noted in a recent Hacker News thread, the ‘online’ part of the above listed statistics means that for manuals where no online copy is known, you get a placeholder message. Using the Bitsavers website along with Archive.org may still be the most pertinent way to hunt down that elusive manual, with the Manx website recommending 1000bit for microcomputer manuals.

Have you used the Manx catalog, or any of the other archiving websites? What have been your experiences with them? Let us know in the comments.

[Filip] got his hands on a sweet old Hammond X5 organ, but it had one crucial problem: only half of the keys worked. Each and every C#, D, D#, E, F, and F# would not play, up and down the keyboard, although the other notes in between sounded just fine.

Those of you with an esoteric knowledge of older electric organs will be saying “it’s a busted top-octave generator chip”, and you’re right. One of the TOGs worked, and the other didn’t. [Filip] rolled his own top-octave generator with a Pico, in Python no less, and the old beauty roared to life once more.

But what is a top-octave generator, you may ask? For a brief period of time in the early 70s, there were organs that ran on square waves. Because a musical octave is a doubling or halving of frequency, you can create a pitch for every key on the organ if you simply create one octave’s worth of pitches, and divide them all down using something as simple as a binary counter IC. But nobody makes top-octave chips any more.

Back in 2018, [DC Darsen] wrote in asking us if we knew about any DIY top-octave designs, and we put out an Ask Hackaday to see if you all could make a top-octave generator out of a microcontroller. We got a super-optimized code hack in response, and that’s worth checking out in its own right, but we always had the nagging suspicion that a hardware solution was the best solution.

We love how [Filip]’s design leans heavily on the Pico’s programmable input/output hardware modules to get the job done with essentially zero CPU load, allowing him to write in Python and entirely bypassing the cycle-counting and assembly language trickery. The voltage shifters and the switchable jumpers to swap between different top-octave chip types are a nice touch as well. If you have an organ that needs a top-octave chip in 2024, this is the way we’d do it. (And it sounds fantastic.)