2026-10-05 04:00:49
When [Thomas] showed his 1960s-era Tektronix 545A oscilloscope in a recent video, it made us both nostalgic and happy. Nostalgic because we miss the days when our oscilloscopes were “mainframes” that could take plug-in modules. Happy because we noticed the two handles on top to manage hauling the almost 70 pounds of tubes, transformers, and glass around if you didn’t have the requisite cart.
Not only did old scopes have plugins so you could reconfigure them, but there were also handy plugins and racks that could power them so you could build different test setups easily. This scope was an early version of that idea, but it only accepted a vertical section plugin. The mainframe part of the scope has 75 tubes, and the plugin, a type D, has six tubes, too. The power consumption was about 500 watts!
If you’ve used a scope like this before, you’ll probably remember a lot of the oddities mentioned in the video. If not, you’ll gain a new appreciation for your modern, sophisticated, and lightweight scope. Of course, the old boat anchor wasn’t working. Apparently, these scopes had ceramic points that didn’t like regular solder, so there was a spool of silver solder inside the unit for repairs.
[Thomas] made a lot of progress, although it still needs work to get it working perfectly. We have no doubt he’ll get it working in time. Even the graticule on these old scopes was an issue.
2026-10-05 01:00:31
If you saw a device that had a laptop form factor playing sound and video, alt-tabbing over to play a game or take some notes, then back to the video — well, that’s normal; that’s what laptops do. If you learned the heart of this machine was an Arduino-compatible AVR microcontroller board, you’d probably do the same spit-take we did. Well, put down your coffee, because [Vladimir]’s AVR Laptop project really is that impressive.
The micro at the heart of this amazing laptop is an ATmega2560. That’s a pretty powerful chip by 8-bit microcontroller standards, but it’s still an 8-bit microcontroller, so getting the FreeRTOS-based system and all of the functions he wants onboard took some careful coding. The ST7920 128×64 monochrome LCD isn’t being driven by any standard library, for example — [Vladimir] built his own driver that does only what he wants. Even in 1-bit color, getting 24 fps video playback was a challenge, especially with sound, and the write-up gets into how that was accomplished.
The code is pending review before release, but we do look forward to seeing how he pulled it all off. In the meantime, we can at least watch it in action in the demo video embedded below.
The vibe of making an 8-bit machine do things you don’t expect to fit in 8 bits reminds us of SymbOS for the Z80. For the sysadmins out there, we’ve also seen an AVR web server.
Thanks to [Vladimir] for the tip!
2026-10-04 22:00:45
[Tolko] wanted to encourage kids to have some physical activity, but noted that they only wanted to play games. That led to BlobJump, a Raspberry Pi-based game sort of like the famous dinosaur runner game. The difference? Your body — or more precisely, a belt worn on your body — is the controller. Your jumping and ducking make your on-screen avatar do the same thing. Watch the video below to see the game in action.
Each belt contains an ESP32-C3, an MPU-6050, and enough rechargeable battery to make it all work. Everything is wireless, of course. We didn’t see any video of the actual kids, so we don’t know how active they were or whether they figured out they could just have a seat and move the belts in their hands. Even so, it looks like a fun game.
We couldn’t help but think that, beyond encouraging activity, this might be a great project to get a kid excited about building hardware and software. You could certainly do worse in that department.
This isn’t a totally new idea, but we liked the execution. If you are truly lazy, you can skip playing the game completely.
2026-10-04 19:00:40
If you want to experiment with LLMs, you typically have a choice of sending your requests to someone else’s computer or fielding a very large GPU and CPU setup to run models locally. However, a recent crop of projects aims to bring bigger models to much more modest hardware.
One example is Strata, a project from [Niko1221], which lets you run a 125-billion-parameter LLM on hardware you might already have for gaming. It won’t run on your old Pentium laptop, but it doesn’t require a supercomputer-like farm of graphics cards, either.
Strata can use several Qwen3.8 model variants, including different quantizations of the original model as well as coding and other specialized versions. Qwen3.8-Flash-Next is a mixture-of-experts model containing 24,576 small experts, of which only ten are needed for each token. The clever part is that Strata effectively treats VRAM as a cache for the much larger model. Frequently used experts stay on the GPU, while the complete collection normally remains in system RAM. The model also includes a roughly 29 GB lookup table that stays on the SSD and is accessed as needed.
The software also uses the model’s multi-token prediction machinery for speculative decoding, allowing several candidate tokens to be checked in a single pass. According to the project, an RTX 5070 with 12 GB of VRAM can produce roughly 50 to 90 tokens per second, depending on quantization. Tokens, of course, aren’t usually entire words, but it is still a respectable clip, once everything gets set up.
We did have some trouble setting everything up due to some incompatibility with the NVIDIA C compiler, our gcc version, and some headers, but your problems will surely be different. The setup.sh file asks you a few questions on the first run. After that, it just handles your selected startup options, which can take a few minutes while everything loads.
Once running, Strata lets you interact through a web browser. It also exposes OpenAI- and Anthropic-compatible APIs on localhost, so existing chat front ends, coding assistants, and other tools can use the local model without much special handling. Of course, you can’t expect its answers to compete with the big models out there for every task. When asking about Hackaday, for example, it got a lot of it right but also got confused about who founded the site and our authors (unless we forgot that [Tom Nardelli] once wrote some posts). Turning up the “thinking level” and turning down the temperature didn’t help much, although it did move its confusion to different facts. It did better when asked to identify some problem code or outline how to port a particular C compiler to a new target.
You’ll still want at least 32 GB of system RAM, 12 GB of VRAM, and around 80 GB of storage, so “modest” is relative. Still, it’s a neat demonstration of how mixture-of-experts models and some clever memory management can stretch ordinary PC hardware surprisingly far.
These economical LLMs can even run on older hardware, just slower.
2026-10-04 16:00:48
When you look into the night sky, you can’t help but wish to bring the outer worlds to your doorstep. One of the best ways to do that is to photograph the Moon’s surface to display in a frame. However, [Sebastian Lague] found a simple image too lackluster compared to something with a bit more style, such as a plotted image of the lunar terrain.
Why plotting? Well, the Moon is not defined in the same way with contrasting colors as Earth is. The Moon is peppered with craters and differing elevations that separate regions, so why not build an entire DIY plotter from scratch? [Sebastian Lague] did exactly this with a custom algorithm to take the elevation maps and create vector drawings which are plotted on his custom plotter.
The plotter’s design is simple on its face, using an Arduino and stepper motors, as many other plotters have. [Sebastian Lague] also shares insights into their process and iterative designs, moving from a level arm to raise a marker to a linear method. While the Moon plots are impressive on their own, why stop there? [Sebastian Lague] decided to plot Mars and then return home with Earth plots.
While the plotter itself is rather simple, all the work behind the contour maps and turning the ideas into reality is nothing to take for granted. However, if you want a more complicated side of plotting, make sure to check out this multi-colored delta plotter!
2026-10-04 13:00:40
It’s no secret that the size of webpages has been slowly — or not so slowly — creeping up over the years. Does a simple website need to span multiple floppy disks? [PortalRunner’s] doesn’t — or rather, he doesn’t want to burn that much of his limited mobile data. So he built a self-hosted HTTPS proxy to compress images and websites, block ads, and save his bandwidth.
The proxy itself is based on mitmproxy rather than reinventing the wheel. The biggest gains are, of course, from knocking images down to size — since he’s only going to be looking at them on a tiny screen, he can not only apply the better compression of .webp rather than the more common .PNG and .JPEG, but also toss unneeded information by reducing the resolution. Exactly how much quality to toss is left up to the individual user in a handy config.
Likewise, if you’re willing to wait a little longer for pages to load, you can force every webpage to get the highest possible level of compression — otherwise the server is only going to do that with ridiculous Java-heavy pages that absolutely need it. As a bonus, going through your home server means this acts like a private VPN to help protect your data on the go. If you want the details, check out the GitHub link above or the video embedded below.
This project is explicitly about saving data; it won’t help older hardware get online the way the publicly hosted FrogFind search engine or private-hosted MacProxy servers do. If you really want to escape the bloat of the modern internet, a proxy can also help you go way back.
