Corral Some Zippy Blue Flames Into 3D Printed Troughs

[Steve Mould] came across an interesting little phenomenon of blue flames zipping around a circular track. This led to diving down a bit of a rabbit hole about excitable mediums, ultimately leading him to optimize the shapes and come up with some pretty wild variations which he shows off in a video (also embedded below.)

After figuring out that the moving flame depended on combustion of fuel vapor in an environment that didn’t allow for the whole surface to stay lit at once, [Steve] tried to optimize the design of 3d-printed channels and raceways to encourage this effect, and he came up with some pretty novel ones. The 3D models are here if you’d like to try them for yourself (we especially like the “figure eight” and “rays” models.)

The video is an excellent show & tell of everything [Steve] dove into, complete with plenty of demonstrations of harnessing this effect to create some nifty running flames. Check it out in the video below, and if unintuitive physical effects are your thing, don’t miss [Steve]’s peeling apart of the turntable paradox.

Continue reading “Corral Some Zippy Blue Flames Into 3D Printed Troughs”

Train A GPT-2 LLM, Using Only Pure C Code

[Andrej Karpathy] recently released llm.c, a project that focuses on LLM training in pure C, once again showing that working with these tools isn’t necessarily reliant on sprawling development environments. GPT-2 may be older but is perfectly relevant, being the granddaddy of modern LLMs (large language models) with a clear heritage to more modern offerings.

LLMs are fantastically good at communicating despite not actually knowing what they are saying, and training them usually relies on PyTorch deep learning library, itself written in Python. llm.c takes a simpler approach by implementing the neural network training algorithm for GPT-2 directly. The result is highly focused and surprisingly short: about a thousand lines of C in a single file. It is a highly elegant process that does the same thing the bigger, clunkier methods accomplish. It can run entirely on a CPU, or it can take advantage of GPU acceleration, where available.

This isn’t the first timeĀ [Andrej Karpathy] has bent his considerable skills and understanding towards boiling down these sorts of concepts into bare-bones implementations. We previously covered a project of his that is the “hello world” of GPT, a tiny model that predicts the next bit in a given sequence and offers low-level insight into just how GPT (generative pre-trained transformer) models work.

Wine In Beverage Cans Had A Rotten Egg Problem, Until Now

Aluminum beverage cans are used for all kinds of drinks, but when it comes to wine there are some glitches. Chief among them is the fact that canned wine occasionally smelled like rotten eggs. Thankfully, researchers have figured out why that happens, and how to stop it. How was this determined? As the image above hints at, lots and lots of samples and testing.

What causes this, and why don’t other beverages have this problem? Testing revealed that the single most important factor was the presence of molecular sulfur dioxide (SO2), a compound commonly used in winemaking as an antioxidant and antimicrobial.

It turns out that the thin plastic lining on the inside of beverage cans doesn’t fully stop molecular SO2 from reacting with the surrounding aluminum, creating hydrogen sulfide (H2S) in the process. H2S has a very noticeable rotten egg smell, even in low concentrations.

Researchers discovered that if a canned beverage contained more than 0.5 ppm of molecular SO2, a noticeable increase in hydrogen sulfide was likely to be present within four to eight months. The problem is that since most wines aim for around 0.5 ppm of SO2, the average can on wine sitting on a shelf will have a problem sooner rather than later. The more SO2 in the wine (reds tend to contain less, whites more), the worse the problem.

Simply increasing the thickness of the plastic liner is an imperfect solution since it increases manufacturing costs as well as waste. So, researchers believe the right move is to use a more durable liner formulation combined with a lower SO2 concentration than winemakers are usually comfortable with. Unlike bottles, cans can be hermetically sealed which should offset the increased oxidation risk of using a lower concentration of SO2. The result should be wine as a canned beverage, with a shelf life of at least 8 months.

The research is published here and gives a great look at just how one approaches this kind of scientific problem, as well as highlighting just how interesting the humble aluminum beverage can really is.

Let The Solder Scroll Take Care Of Your Feed Needs

[Victor]’s nifty tool the Solder Scroll is a handheld device that lets one feed solder out simply by turning something a little like a scroll wheel. It looks like an intuitive and comfortable design that can adapt to a wide variety of solder thicknesses, and is entirely 3D printed.

One part we particularly like is the feed system. One rolls a wheel which feeds solder out using a mechanism a lot like extrusion gears in many 3D printer hot ends. Both wheels have ridged surfaces that grip and feed the solder; their gears mesh with one another so that moving one moves both in unison.

Solder feed tools like this have seen all kinds of interesting designs, because while the problem is the same for everyone, there are all kinds of different ways to go about addressing it. We love this one, and we have seen many other takes that range from a powered, glove-mounted unit to an extremely simple tool with no moving parts. We’ve even seen a method of hacking a mechanical pencil into a new role as a solder feeder.

DIY Quad-Motor Go-Kart Is A Thrilling Ride

[Peter Holderith] set out some time ago to build an electric go-kart. That by itself is not terribly unusual, but where his project diverts from the usual is in the fact that each of the four wheels has an integrated hub motor.

It might not look it, but each wheel has an integrated hub motor.

This kart project is a bit of a work in progress, with [Peter] previously building (then scrapping) a failed attempt at a cheap suspension system. But it’s completely operational with all four wheels able to deliver a monstrous amount of power despite being limited by the power supply (a battery pack salvaged from an Audi Q5 Hybrid).

The kart might not look it, but it weighs 177 pounds (80 kg) with the battery and motors accounting for nearly half of that. What is is like to drive? “Nothing short of thrilling,” says [Peter]. It’s got no suspension and is pretty bare bones, not to mention limited in power by the battery, but [Peter] finds it a satisfying drive that nevertheless delivers car-like cues in the driving experience. The build isn’t done, and [Peter] plans to see if more power is available by switching battery chemistries rather than add more battery weight.

Building and driving electric vehicles can be remarkably satisfying, and it’s an area in which hobbyists can meaningfully innovate. Self-balancing one-wheeled vehicles for example look like a ton of fun. Heck, researchers have discovered that even rats seem to enjoy driving just for the fun of it.

Unraveling The Secrets Of Apple’s Mysterious Fisheye Format

Apple has developed a proprietary — even mysterious — “fisheye” projection format used for their immersive videos, such as those played back by the Apple Vision Pro. What’s the mystery? The fact that they stream their immersive content in this format but have provided no elaboration, no details, and no method for anyone else to produce or play back this format. It’s a completely undocumented format and Apple’s silence is deafening when it comes to requests for, well, anything to do with it whatsoever.

Probably those details are eventually forthcoming, but [Mike Swanson] isn’t satisfied to wait. He’s done his own digging into the format and while he hasn’t figured it out completely, he has learned quite a bit and written it all up on a blog post. Apple’s immersive videos have a lot in common with VR180 type videos, but under the hood there is more going on. Apple’s stream is DRM-protected, but there’s an unencrypted intro clip with logo that is streamed in the clear, and that’s what [Mike] has been focusing on.

Most “fisheye” formats are mapped onto square frames in a way similar to what’s seen here, but this is not what Apple is doing.

[Mike] has been able to determine that the format definitely differs from existing fisheye formats recorded by immersive cameras. First of all, the content is rotated 45 degrees. This spreads the horizon of the video across the diagonal, maximizing the number of pixels available in that direction (a trick that calls to mind the heads in home video recorders being tilted to increase the area of tape it can “see” beyond the physical width of the tape itself.) Doing this also spreads the center-vertical axis of the content across the other diagonal, with the same effect.

There’s more to it than just a 45-degree rotation, however. The rest most closely resembles radial stretching, a form of disc-to-square mapping. It’s close, but [Mike] can’t quite find a complete match for what exactly Apple is doing. Probably we’ll all learn more soon, but for now Apple isn’t saying much.

Videos like VR180 videos and Apple’s immersive format display stereoscopic video that allow a user to look around naturally in a scene. But to really deliver a deeper sense of presence and depth takes light fields.

Fail Of The Week: Can An Ultrasonic Cleaner Remove Bubbles From Resin?

[Wendy] asked a very good question. Could putting liquid resin into an ultrasonic cleaner help degas it? Would it help remove bubbles, resulting in a cleaner pour and nicer end product? What we love is that she tried it out and shared her results. She purchased an ultrasonic cleaner and proceeded to mix two batches of clear resin, giving one an ultrasonic treatment and leaving the other untouched as a control.

Sadly, the test piece had considerably more surface bubbles than the untreated control, as well as a slight discoloration.

The results were interesting and unexpected. Initially, the resin in the ultrasonic bath showed visible bubbles rising to the surface which seemed promising. Unfortunately, this did not lead to fewer bubbles in the end product.

[Wendy]’s measurements suggest that the main result of putting resin in an ultrasonic bath was an increase in its temperature. Overheating the resin appears to have led to increased off-gassing and bubble formation prior to and during curing, which made for poor end results. The untreated resin by contrast cured with better color and much higher clarity. If you would like to skip directly to the results of the two batches, it’s right here at 9:15 in.

Does this mean it’s a total dead end? Maybe, but even if the initial results weren’t promising, it’s a pretty interesting experiment and we’re delighted to see [Wendy] walk through it. Do you think there’s any way to use the ultrasonic cleaner in a better or different way? If so, let us know in the comments.

This isn’t the first time people have tried to degas epoxy resin by thinking outside the box. We’ve covered a very cheap method that offered surprising results, as well as a way use a modified paint tank in lieu of purpose-made hardware.