An AI By Any Other Name

November 16, 2025 by 5 Comments

While there are many AI programs these days, they don’t all work in the same way. Most large language model “chatbots” generate text by taking input tokens and predicting the next token of the sequence. However, image generators like Stable Diffusion use a different approach. The method is, unsurprisingly, called diffusion. How does it work? [Nathan Barry] wants to show you, using a tiny demo called tiny-diffusion you can try yourself. It generates — sort of — Shakespeare.

For Stable Diffusion, training begins with an image and an associated prompt. Then the training system repeatedly adds noise and learns how the image degenerates step-by-step to noise. At generation time, the model starts with noise and reverses the process, and an image comes out. This is a bit simplified, but since something like Stable Diffusion deals with millions of pixels and huge data sets, it can be hard to train and visualize its operation.

The beauty of tiny-diffusion is that it works on characters, so you can actually see what the denoising process is doing. It is small enough to run locally, if you consider 10.7 million parameters small. It is pretrained on Tiny Shakespeare, so what comes out is somewhat Shakespearean.

Continue reading “An AI By Any Other Name” →

A photo of the LEGO sorter

Making A Machine To Sort One Million Pounds Of LEGO

November 16, 2025 by 19 Comments

You know what’s not fun? Sorting LEGO. You know what is fun? Making a machine to sort LEGO! That’s what [LegoSpencer] did, and you can watch the machine do its thing in the video below.

[Spencer] runs us through the process: first, quit your day job so you can get a job playing with LEGO; then research what previous work has been done in this area (plenty, it turns out); and then commit to making your own version both reproducible and extensible.

A sorting machine needs three main features: a feeder to dispense one piece at a time, a classifier to decide the type of piece, and a distributor to route the piece to a bin. Of course, the devil is in the details.

Continue reading “Making A Machine To Sort One Million Pounds Of LEGO” →

The Fastest (68k) Macintosh Might Not Be An Amiga Anymore

November 15, 2025 by 13 Comments

Amiga and Atari fans used to lord over their Apple-eating brethren the fact that Cupertino never moved to the most advanced 68k processors — so for a while, thanks to 68060 accelerator cards, the fastest thing running Macintosh software was an Amiga (or Atari). After all these years, the Macintosh community is finally getting the last laugh, as [zigzagjoe] demonstrates an actual Macintosh booting with a 68060 CPU for the first time in a thread on 68KMLA. Video or it didn’t happen? Check it out below.

The Mac in question is a Quadra 650, which is a good choice since it was about the last thing Apple sold before switching to PowerPC, and ran the 68040 processor. [Reinauer] had already produced a 68040-to-68060 socket adapter (the two chips not being pinout compatible), so the hardware part of the battle was already set. Software, however? That was a different story, and where [zigzagjoe] put in the effort.

Continue reading “The Fastest (68k) Macintosh Might Not Be An Amiga Anymore” →

A baby blue hatchback with red accents drives down a road with blurry trees and a blue sky in the background.

Hyundai Paywalls Brake Pad Changes

November 15, 2025 by 61 Comments

Changing the pads on your car’s brakes is a pretty straightforward and inexpensive process on most vehicles. However, many modern vehicles having electronic parking brakes giving manufacturers a new avenue to paywall simple DIY repairs.

Most EVs will rarely, if ever, need to replace their mechanical brake pads as in most driving situations the car will be predominantly relying on regenerative braking to slow down. A hot hatch like the Ioniq 5N, however, might go through brakes a lot faster if it spends a lot of time at the track, which is what happened with Reddit user [SoultronicPear].

Much to their chagrin, despite buying the required $60/wk subscription to the National Automotive Service Task Force (NASTF) and the $2,000 interface tool, [SoultronicPear]’s account was suspended because it was not intended for use for anyone but “service professionals.” Not exactly a Right-to-Repair friendly move on Hyundai’s part. After trying a number of different third party tools, they finally found a Harbor Freight T7 bidirectional scan tool was able to issue the parking brake retract command to perform the pad swap, albeit not without throwing some error codes in the process.

Electrification of vehicles should simplify repairs, but manufacturers are using it to do the opposite. Perhaps they should read our Minimal Motoring Manifesto? There is a glimmer of hope in the promises of Slate and Telo, but we’ll have to see if they make it to production first.

A circuit board in the shape of a business card is shown. The circuitry is confined to the left side of the board, and the rest is used for text.

(Neural) Networking With A Business Card

November 15, 2025 by 7 Comments

A PCB business card is a great way for electrical engineers to impress employers with their design skills, but the software they run can be just as impressive as the card itself. As a programmer with an interest in embedded machine learning, [Dave McKinnon] wanted a card that showcased his skills, so he designed one that runs voice recognition.

[Dave] specifically wanted to run a neural network on his card, but needed to make it small enough to run on a microcontroller. Voice recognition looked like a good fit for this, since audio can be represented with relatively little data, a microphone is cheap and easy to add to a circuit board, and there was already an example of someone running such a voice recognition network on an Arduino. To fit the neural network into 46 kB, it only distinguishes the words “one” through “nine,” and displays its guess on an LED seven-segment display. [Dave] first prototyped the system with an Arduino, then designed the circuit board around an RP2040.

Continue reading “(Neural) Networking With A Business Card” →

Using The Pyroelectric Effect To Identify Broken MLCC Capacitors

November 15, 2025 by 5 Comments

Vintage computer hardware can fail in a variety of fascinating ways, with [Bits und Bolts] dealing with an interesting failure mode, in the form of degraded MLCC capacitors on Voodoo 2 graphics cards. These little marvels of miniaturized surface-mount technology enable the placement of ceramic capacitors with very little space required, but as they degrade over time or due to physical damage, they can cause big issues in a circuit.

In the case of the two Voodoo 2 GPUs that [Bits und Bolts] was trying to fix, the clue that something was wrong was graphical glitches, which seemed to be related to something dragging down the 5V rail. Using the standard ‘inject voltage and see what gets hot’ method, he discovered a couple of dead MLCCs and replaced them. But something was still dragging the rail down. Unfortunately, whatever it was wasn’t enough to heat up the part in question, and no sane person wants to desolder hundreds or even thousands of MLCCs on a PCB and see whether it makes a difference.

Ultimately, the pyroelectric effect was used to hunt down the culprit, saving countless hours of work. This is a property of certain naturally electrically polarized crystals, in which the material generates a voltage when heated or cooled. Materials like that used in MLCCs, for example.

Continue reading “Using The Pyroelectric Effect To Identify Broken MLCC Capacitors” →

A closeup of a transparent-bodied example of the new Steam Frame VR headset

The Engineering Behind Valve’s New VR Headset

November 15, 2025 by 11 Comments

Valve’s new Steam Frame is what all the well-connected YouTubers are talking about, but most of them are talking about what it’s like to game on it. That’s great content if you’re into it, but not exactly fodder for Hackaday — with one exception. [Gamers Nexus] gives us a half hour of relatively-unedited footage of them just chatting with the engineers behind the hardware.

It’s great stuff right from the get-go: they start with how thermal management drove the PCB design, and put the SoC on the “back” of the chip, sandwiched betwixt heat pipes. We don’t usually think of taking heat through the PCB when building a board, so it’s a neat detail to learn about before these things get into the hands of the usual suspects who will doubtless give us teardown videos in a few months.

From there wanders to power delivery — getting the voltage regulators packaged properly was a challenge, since impedance requirements meant a very tight layout. Anyone who has worked on this kind of SBC might be familiar with that issue, but for those looking in from the outside, it’s a fascinating glimpse at electrical sausage being made. That’s just the first half.

The heat-regulation conversation is partially repeated the next conversation (which seems to have happened first) where they get into the cooling requirements of the LCD screens. This requires less than you might think, as they like to run warm for fast refresh. It’s really more about keeping your face cool. They also they discuss acoustic vibration — you don’t want your integrated audio shaking your IMUs apart — and why the prototype was being blasted with freakin’ laser beams to monitor it.

If you haven’t seen or read any other coverage on the Steam Frame, you’re going to miss some context here, but if you’ve not hid under a rock for that announcement, this is amazing detail to have. We’re hugely impressed that Valve let their engineers out of their cubicle-cave to talk to media.

Sure, it’s not an open-source VR headset, but compared to the deafening silence coming from the likes of Meta, this level of information is still awesome to have.

Continue reading “The Engineering Behind Valve’s New VR Headset” →