There’s an old joke that the world’s greatest secret agent was Beethoven. Didn’t know Beethoven was a secret agent? That’s why he was the greatest one! While most people have some idea about the CIA, MI6, and the GRU, agencies like the NRO and GCHQ keep a much lower profile. GCHQ (Government Communications Headquarters) is the United Kingdom’s electronic listening center housed in a 180 meter round doughnut. From there they listen to… well… everything. They are also responsible for codebreaking and can trace their origin back to Bletchley Park as well as back to the Great War. So what’s inside the Doughnut? National Geographic managed to get a tour of GCHQ and if you have any interest in spies, radios, cybersecurity, or codebreaking, it is worth having a look at it.
Of course, only about half of the GCHQ’s employees work in the Doughnut. Others are scattered about the UK and — probably — some in other parts of the world, too. According to the article, GCHQ had a hand in foiling 19 terrorist attacks, arresting at least two sex offenders, and prevented about £1.5 billion of tax evasion.
With the Jetson Nano, NVIDIA has done a fantastic job of bringing GPU-accelerated machine learning to the masses. For less than the cost of a used graphics card, you get a turn-key Linux computer that’s ready and able to handle whatever AI code you throw at it. But if you’re trying to set up a lab for 30 students, the cost of even relatively affordable development boards can really add up.
Which is why [Tea Vui Huang] has developed jetson-emulator. This Python library provides a work-alike environment to NVIDIA’s own “Hello AI World” tutorials designed for the Jetson family of devices, with one big difference: you don’t need the actual hardware. In fact, it doesn’t matter what kind of computer you’ve got; with this library, anything that can run Python 3.7.9 or better can take you through NVIDIA’s getting started tutorial.
So what’s the trick? Well, if you haven’t guessed already, it’s all fake. Obviously it can’t actually run GPU-accelerated code without a GPU, so the library [Tea] has developed simply pretends. It provides virtual images and even “live” camera feeds to which randomly generated objects have been assigned.
The original NVIDIA functions have been rewritten to work with these feeds, so when you call something like net.Classify(img) against one of them you’ll get a report of what faux objects were detected. The output will look just like it would if you were running on a real Jetson, down to providing fictitious dimensions and positions for the bounding boxes.
If you’re a hacker looking to dive into machine learning and computer vision, you’d be better off getting a $59 Jetson Nano and a webcam. But if you’re putting together a workshop that shows a dozen people the basics of NVIDIA’s AI workflow, jetson-emulator will allow everyone in attendance to run code and get results back regardless of what they’ve got under the hood.
We’ve said it before, but we cast a wary eye at any superlative claims that come our way. “World’s fastest” or “world’s first” claims always seem to be quickly debunked, but when the claim of “World’s Smallest Benchy” is backed up by a tugboat that two dozen E. coli would have a hard time finding space on, we’re pretty comfortable with it.
Of course the diminutive benchmark was not printed just for the sake of it, but rather as part of a demonstration of what’s possible with “microswimmers”, synthetic particles which are designed to move about freely in microscopic regimes. As described in a paper by [Rachel P. Doherty] et al from the Soft Matter Physics lab at Leiden University, microswimmers with sizes on the order of 10 to 20 μm can be constructed repeatably, and can include a small area of platinum catalyst. The catalyst is the engine of the microswimmer; hydrogen peroxide in the environment decomposes on the catalyst surface and provides a propulsive force.
Artificial microswimmers have been around for a while, but most are made with chemical or evaporative methods which result in simple shapes like rods and spheres. The current work describes much more complex shapes — the Benchy was a bit of a flex, since the more useful microswimmers were simple helices, which essentially screw themselves into the surrounding fluid. The printing method was based on two-photon polymerization (2PP), a non-linear optical process that polymerizes a resin when two photons are simultaneously absorbed.
The idea that a powered machine so small could be designed and manufactured is pretty cool. We’d love to see how control mechanisms could be added to the prints — microfluidics, perhaps?
Machinists have a lot of neat shop tricks, but one especially interesting one is shrink-fitting tools. Shrink-fitting achieves an interference fit between tool and holder by creating a temperature difference between the two before assembly. Once everything returns to temperature, the two parts may as well be welded together.
The easiest way to shrink-fit machine tooling is with induction heating, and commercial rigs exist for doing the job. But [Roetz 4.0] decided to build his own shrink-fitting heater, and the results are pretty impressive. The induction heater itself is very simple — a 48 volt, 20 amp power supply, an off-the-shelf zero-voltage switching (ZVS) driver, and a heavy copper coil. When the coil is powered up, any metal within is quickly and evenly heated by virtue of the strong magnetic flux in the coil.
To use the shrinker, [Roetz 4.0] starts with a scrupulously clean tool holder, bored slightly undersized for the desired tool. Inside the coil, the steel tool holder quickly heats to a lovely deep brown color, meaning it has gotten up to the requisite 250-300°C. The tool is quickly dropped into the now-expanded bore, which quickly shrinks back around it. The advantage of this method over a collet or a chuck is clear in the video below: practically zero runout, and the tool is easily released after another run through the heater.
You say you’ve got no need for shrink-fitting tools? How about stuck bolts? Induction heaters work great there too.
It’s a fact of life for CNC router owners — swarf. Whether it’s the fine dust from a sheet of MDF or nice fat chips from a piece of aluminum, the debris your tool creates gets everywhere. You can try to control it at its source, but swarf always finds a way to escape and cause problems.
Unwilling to deal with the accumulation of chips in the expensive ball screws of his homemade CNC router, [Nikodem Bartnik] took matters into his own hands and created these DIY telescopic ball screw covers. Yes, commercial ball screw covers are available, but they are targeted at professional machines, and so are not only too large for a homebrew machine like his but also priced for pro budgets. So [Nikodem] recreated their basic design: strips of thin material wound into a tight spring that forms a tube that can extend and retract. The first prototypes were from paper, which worked but proved to have too much friction. Version 2 was made from sheets of polyester film, slippery enough to get the job done and as a bonus, transparent. They look pretty sharp, and as you can see in the video below, seem to perform well.
It’s nice to see a build progress to the point where details like this can be addressed. We’ve been following [Nikodem]’s CNC build for years now, and it really has come a long way.
Never underestimate the power of nostalgia. In an age when there are more megapixels stuffed in the sensor of a smartphone camera than the average computer display can even represent, why would jagged images from a 20-year-old grayscale camera with pixels numbering in the thousands still grab attention? Maybe what’s old is new again, and the coolness factor of novelty is something that can’t be quantified.
The surprise I had last Monday when I saw my Twitter notifications is maybe only second to the feeling I had when I was invited to become a Hackaday contributor. I’d made a very simple web app which mimics a Game Boy Camera using the camera from your phone or desktop, and it got picked up by people so much that I’m amazed my web host is still holding. Let’s look at why something seemingly so simple gained so much traction.
When they were invented in the 1950s, Nixie tubes were a huge leap forward in display technology. In the days before affordable LEDs made seven-segment displays a commodity, there were few alternatives to the charming glow of the clear and legible characters inside Nixies. Sturdy and reliable, the cold-cathode displays found their way into everything from scientific instruments to test equipment, and even some of the earliest computers and the equipment that formed the foundation of the Space Race sported the venerable tubes.
But time marches on, and a display that requires high voltage and special driver circuits isn’t long for a world where LEDs are cheap and easy to design with. Nixies fell from favor through the late 1960s and 1970s, to the point where new tubes were only being made by the Russians, until that supply dried up as well. Rediscovered by hobbyists for use in quirky clocks and other displays, any stock left over from the Nixie’s heyday are quickly being snapped up, putting the tubes on the fast track to unobtainium status.
That’s not to say that you can’t get brand new Nixie tubes, of course. Artisanal manufacturers like Dalibor Farný have taken the Nixie to a whole new level, with big, beautiful tubes that are handcrafted from the best materials. Reviving the somewhat lost art of Nixie manufacturing wasn’t easy, but the tubes that Dalibor makes in a castle in the Czech Republic now find their way into cool clocks and other builds around the world. He’ll join us on the Hack Chat to dive into the art and science of Nixies, and what’s going on with his mysterious “Project H”.
Click that speech bubble to the right, and you’ll be taken directly to the Hack Chat group on Hackaday.io. You don’t have to wait until Wednesday; join whenever you want and you can see what the community is talking about.