Tell Time And Predict The Heavens With This Astronomical Timepiece

May 31, 2024 by Donald Papp 2 Comments

Looking for a new project, or just want to admire some serious mechanical intricacy? Check out [illusionmanager]’s Astronomical Clock which not only tells time, but shows the the positions of the planets in our solar system, the times of sunrise and sunset, the phases of the moon, and more — including solar and lunar eclipses.

One might assume that the inside of the Astronomical Clock is stuffed with a considerable number of custom gears, but this is not so. The clock’s workings rely on a series of tabs on movable rings that interact with each other to allow careful positioning of each element. After all, intricate results don’t necessarily require complex gearing. The astrolabe, for example, did its work with only a few moving parts.

The Astronomical Clock’s mechanical elements are driven by a single stepper motor, and the only gear is the one that interfaces the motor shaft to the rest of the device. An ESP32-C3 microcontroller takes care of everything else, and every day it updates the position of each element as well as displaying the correct time on the large dial on the base.

The video below shows the clock in operation. Curious its inner workings? You can see the entire construction process from beginning to end, too.

Continue reading “Tell Time And Predict The Heavens With This Astronomical Timepiece” →

The Secret Behind The Motion Of Microsoft’s Bendy Mouse

May 25, 2024 by Donald Papp 7 Comments

The Surface Arc is a designed-for-travel mouse that carries flat, but curves into shape for use. It even turns on when it’s bent and shuts itself off when it’s flat. The device isn’t particularly new, but [Mr Teardown] was a bit surprised at the lack of details about what’s inside so tears it down in a video to reveal just how the mechanism works.

The mechanism somewhat resembles a beaver’s tail, and locks into place thanks to a magnetic connector at the base that holds the device’s shape.

The snap-action of the bending is accomplished with the help of a magnetic connection near the bottom end of the mouse’s “tail”, locking it into place when flexed. Interestingly, the on and off functionality does not involve magnets at all. Power control is accomplished by a little tab that physically actuates a microswitch.

There are a few interesting design bits that we weren’t expecting. For example, there is no mechanical scroll wheel. The mouse delivers similar functionality with touch sensors and a haptic feedback motor to simulate the feel and operation of a mechanical scroll wheel.

[Mr Teardown] finds the design elegant and effective, but we can’t help but notice it also seems perhaps not as optimized as it could be. There are over 70 components in all, including 23 screws (eight different kinds!), and it took [Mr Teardown] the better part of 45 minutes to re-assemble it. You can watch the entire teardown in the video embedded just under the page break; it’s a neat piece of hardware for sure.

If you’re in the mood for another mouse teardown, we have a treat for you: an ancient optical mouse from the 80s that required a special surface to work.

[via Core77]

Continue reading “The Secret Behind The Motion Of Microsoft’s Bendy Mouse” →

Everything You Ever Wanted To Know About The ULN2003

May 25, 2024 by Donald Papp 27 Comments

The ULN2003 IC is an extremely versatile part, and with the help of [Hulk]’s deep dive, you might just get some new ideas about how to use this part in your own projects.

Each of the seven outputs works like this simplified diagram.

Inside the ULN2003 you’ll find seven high-voltage and high-current NPN Darlington pairs capable of switching inductive loads. But like most such devices there are a variety of roles it can fill. The part can be used to drive relays or motors (either brushed or stepper), it can drive LED lighting, or simply act as a signal buffer. [Hulk] provides some great examples, so be sure to check it out if you’re curious.

Each of the Darlington pairs (which act as single NPN transistors) is configured as open collector, and the usual way this is used is to switch some kind of load to ground. Since the inputs can be driven directly from 5 V digital logic, this part allows something like a microcontroller to drive a high current (or high voltage, or both) device it wouldn’t normally be able to interface with.

While the circuitry to implement each of the transistor arrays isn’t particularly complex and can be easily built by hand, a part like this is a real space saver due to how it packs everything needed in a handy package. Each output can handle 500 mA, but this can be increased by connecting in parallel.

There’s a video (embedded below) which steps through everything you’d like to know about the ULN2003. Should you find yourself wanting a much, much closer look at the inner secrets of this chip, how about a gander at the decapped die?

Continue reading “Everything You Ever Wanted To Know About The ULN2003” →

How To Find Replacement Parts When Model Numbers Don’t Match

May 22, 2024 by Donald Papp 24 Comments

[Sharad Shankar] repaired a broken TV by swapping out the cracked and malfunctioning image panel for a new one. Now, part-swapping is a great way to repair highly integrated modern electronics like televisions, but the real value here is something else. He documented his fix but the real useful part is his observations and guidance on how to effectively look for donor devices when the actual model of donor device can’t be found.

The usual approach to fixing a device by part swapping is to get one’s hands on two exact same models that are broken in different ways. But when it comes to consumer electronics with high turnovers — like televisions — it can be very difficult to actually locate any particular model once it’s no longer on shelves. [Sharad Shankar]’s broken TV was a 65″ TCL R646 purchased in 2021, and searching for a second 65″ TCL R646 was frankly like looking for a needle in a haystack. That’s when he got a visit from the good ideas fairy. Continue reading “How To Find Replacement Parts When Model Numbers Don’t Match” →

Quad-Motor Electric Kart Gets A Little Too Thrilling

May 21, 2024 by Donald Papp 39 Comments

[Peter Holderith] has been on a mission to unlock the full potential of a DIY quad-motor electric go-kart as a platform. This isn’t his first rodeo, either. His earlier vehicle designs were great educational fun, but were limited to about a kilowatt of power. His current platform is in theory capable of about twenty. The last big change he made was adding considerably more battery power, so that the under-used motors could stretch their legs a little, figuratively speaking.

How did that go? [Peter] puts it like this: “the result of [that] extra power, combined with other design flaws, is terror.” Don’t worry, no one’s been hurt or anything, but the kart did break in a few ways that highlighted some problems.

The keyed stainless steel bracket didn’t stay keyed for long.

One purpose of incremental prototyping is to bring problems to the surface, and it certainly did that. A number of design decisions that were fine on smaller karts showed themselves to be inadequate once the motors had more power.

For one thing, the increased torque meant the motors twisted themselves free from their mountings. The throttle revealed itself to be twitchy with a poor response, and steering didn’t feel very good. The steering got heavier as speed increased, but it also wanted to jerk all over the place. These are profoundly unwelcome feelings when driving a small and powerful vehicle that lurches into motion as soon as the accelerator is pressed.

Overall, one could say the experience populated the proverbial to-do list quite well. The earlier incarnation of [Peter]’s kart was a thrilling ride, but the challenge of maintaining adequate control over a moving platform serves as a reminder that design decisions that do the job under one circumstance might need revisiting in others.

How To Lace Cables Like It’s 1962

May 20, 2024 by Donald Papp 34 Comments

Cable harnesses made wire management a much more reliable and consistent affair in electronic equipment, and while things like printed circuit boards have done away with many wires, cable harnessing still has its place today. Here is a short how-to on how to lace cables from a 1962 document, thoughtfully made available on the web by [Gary Allsebrook] and [Jeff Dairiki].

It’s a short resource that is to the point in all the ways we love to see. The diagrams are very clear and the descriptions are concise, and everything is done for a reason. The knots are self-locking, ensuring that things stay put without being overly tight or constrictive.

According to the document, the ideal material for lacing cables is a ribbon-like nylon cord (which reduces the possibility of biting into wire insulation compared to a cord with a round profile) but the knots and techniques apply to whatever material one may wish to use.

Cable lacing can be done ad-hoc, but back in the day cable assemblies were made separately and electrically tested on jigs prior to installation. In a way, such assemblies served a similar purpose to traces on a circuit board today.

Neatly wrapping cables really has its place, and while doing so by hand can be satisfying, we’ve also seen custom-made tools for neatly wrapping cables with PTFE tape.

Try Image Classification Running In Your Browser, Thanks To WebGPU

May 20, 2024 by Donald Papp 3 Comments

When something does zero-shot image classification, that means it’s able to make judgments about the contents of an image without the user needing to train the system beforehand on what to look for. Watch it in action with this online demo, which uses WebGPU to implement CLIP (Contrastive Language–Image Pre-training) running in one’s browser, using the input from an attached camera.

By giving the program some natural language visual concept labels (such as ‘person’ or ‘cat’) that fit a hypothetical template for the image content, the system will output — in real-time — its judgement on the appropriateness of such labels to what the camera sees. Again, all of this runs locally.

It’s maybe a little bit unintuitive, but what’s happening in the demo is that the system is deciding which of the user-provided labels (“a photo of a cat” vs “a photo of a bald man”, for example) is most appropriate to what the camera sees. The more a particular label is judged a good fit for the image, the higher the number beside it.

This kind of process benefits greatly from shoveling the hard parts of the computation onto compatible graphics cards, which is exactly what WebGPU provides by allowing the browser access to a local GPU. WebGPU is relatively recent, but we’ve already seen it used to run LLMs (Large Language Models) directly in the browser.

Wondering what makes GPUs so very useful for AI-type applications? It’s all about their ability to work with enormous amounts of data very quickly.