Amateur radio is a pursuit with many facets, some of which hold more attention than others for the hacker. Though there have been radio amateurs for a century it still has boundaries that are being tested, and sometimes they come in surprising places.
A recent first involved something you might consider a done deal, a transatlantic radio contact. On June 16th, for the first time ever a contact was made between the operators [D41CV] and [FG8OJ] at 3867 km across the Atlantic ocean from the Cape Verde Islands to Guadeloupe, on the 2 metre (144 MHz) band. If this means little to you it’s worth explaining that the 2 m band is a VHF band, with a range normally similar to that you’d expect from an FM broadcast station. Nobody has ever done this before, so it’s a significantly big deal.
Before you dismiss this as merely some radio amateur chasing grid squares and thus not particularly impressive, it’s worth talking about both the radio mode used and the unusual atmospheric conditions that were carefully sought for the achievement. The attempt was made to coincide with a prediction of transatlantic tropospheric ducting, and the mode employed was [Joe Taylor K1JT]’s FT8. This is a digital mode designed especially for weak-signal and long-distance work. It is theorised that the propagation was so-called surface ducting, in which the signal travels and is reflected between the surface of the sea and a relatively low-level reflective layer of atmosphere. The contact really pushed the limit of what is possible with radio, and while you wouldn’t use it for a voice conversation, proves that there are new tricks in an old hobby for the hardcore experimenter.
The project is built around a ticking four-digit display. The blue LEDs give it a modern touch, and it’s attached on top of an Arduino Pro Mini 3.3V. This enables the whole module to be powered by a coin cell, for an incredibly compact and tidy timer that is barely bigger than the display itself. There’s also a buzzer attached, which chirps each second, somewhat heightening the stress level in the immediate vicinity.
With a functioning timer, [deshipu] then went for comedy points, by hooking it up to a trio of bananas. This is widely considered more courteous than attaching it to a detonator circuit and actual dynamite, and is key to staying off government watchlists.
It’s a piece that would be amusing at a Halloween party or similar, and is easily completed by any beginner learning Arduino. It goes without saying that, while this is amusing, it’s a build that should very much not be bandied about in public or used for a prank. In this day and age, even touting a custom clock can draw unwelcome attention, so it’s important to be careful. Video after the break.
If you were a British kid at any time from the 1950s to the 1980s, the chances are that your toy shop had a train set in it. Not just any train set, but a full model railway layout in a glass case roughly the size of a pool table, with a button that when pressed started a timer and set a little tank engine off on a circuit with a pair of coaches. Magical for a generation raised on black-and-white TV, but probably not something that would cut it with today’s youth. A modern take on the glass-case layout comes from [Jack Flynn], who has created a coffee table with an automated and computerised N-gauge railway layout inside it. And this is definitely a railway rather than a railroad, the main locomotive is a Brush Type 4, a British Rail Class 47 diesel.
The modelling is a work of art, with a slightly idealised British street scene in an oval of double track against a backdrop of a rocky hillside. In the hill is an unexpected surprise which you can see on the video we’ve placed below the break, and beneath it lie the electronics. A Teensy handles the track switching and all the various LED lights around the board, a Sprog DCC controller takes care of the trains, and overseeing everything is a Raspberry Pi running some custom software in Python with a web interface for control. We probably wouldn’t be able to resist a bit of remote-control railway action if our coffee table had a layout like this one!
We will admit it: mostly when we see a homebrew CPU design on an FPGA, it is a simple design that wouldn’t raise any eyebrows in the 1970s or 1980s. Not so with [Henry Wong’s] design, though. His x86-like design does superscalar out-of-order execution, just like big commercial modern CPUs. Of course [Henry] designs CPU architectures for Intel, so that’s not surprising. You can see a very detailed talk on the design in the video, below. You can also read the entire thesis project.
[Henry] starts out with a description of FPGAs and soft processors. He also covers the use of multiple instruction issue to increase the virtual clock rate of a CPU. In other words, if a 100 MHz CPU can do one instruction at a time, it won’t be any faster — in theory — than a 50 MHz CPU that can do two instructions at once. Of course, trying to do two at once has some overhead, so that won’t be completely true.
When we recently discussed the skills that we might wish to impart upon a youngster, one of those discussed was the ability to speak more than one language. If any demonstration were required as to why that might be the case, it comes today in [Byfeel]’s Notif’Heure, an ESP8266-powered clock and display (French-language, Google Translate link). If we only watch for English-language projects, we miss much of the picture.
The project began life in April 2018 (Google Translate link) and has since speedily progressed through many software versions to the current v3.2. In hardware terms it’s pretty simple: an ESP8266 development board drives a set of LED matrix displays. In the software though it has the primary function of an NTP-synchronised clock, there is also support for notification display and integration with the Jeedom home automation package.
We’ve featured innumerable ESP8266 clocks over the years, but surprisingly this is the first one with Jeedom integration. With so many to choose from it’s difficult to pick examples to show you, so perhaps it’s time to go to the truly ridiculous with this twelve-ESP monster.
The core of the machine is a moving platform combined with a rolling pin, that can be set to a desired height to roll the dough into a set thickness. This is key to baking top-notch croissants, which [Alex] takes very seriously. His initial model used a table leg for a rolling pin, fitted with a threaded rod down the centre. This had significant issues with both runout, and uneven diameter across its length. Additionally, its frame had not held up after a recent move, and [Alex] was keen to start again.
The new model starts with attention paid to the basic engineering issues. The table leg is replaced with a professional-grade rolling pin, fitted with 3D-printed gears that accurately align the axis of rotation to the centre of the pin. A rack and pinion drive is also added to move the dough platform. Finally, a locking pin system is used to set the desired height of the dough.
It’s a useful project for the keen baker, and one that leans heavily on additive manufacturing methods. Producing such a tool in the years before 3D printers would have required significant effort to produce the required gears and mating components, so it’s impressive to see how easily something like this can come together these days. A hacker mindset can always be handy for baking – don’t forget, you can improve your bread crusts with steam! Video after the break.
In 2019, using AI to evaluate artwork is finally more productive than foolish. We all hope that someday soon our Roomba will judge our living habits and give unsolicited advice on how we could spruce things up with a few pictures and some natural light. There is already an extensive amount of Deep Learning dedicated to photo recognition but a team in Croatia is adapting them for use on fine art. It makes sense that everything is geared toward cameras since most of us have a vast photographic portfolio but fine art takes longer to render. Even so, the collection on Wikiart.org is vast and already a hotbed for computer classification work, so they set to work there.
As they modify existing convolutional neural networks, they check themselves by comparing results with human ratings to keep what works and discard what flops. Fortunately, fine art has a lot of existing studies and commentary, whereas the majority of photographs in the public domain have nothing more than a file name and maybe some EXIF data. The difference here is that photograph-parsing AI can say, “That is a STOP sign,” while the fine art AI can say, “That is a memorable painting of a sign.” Continue reading “AI And Art Appreciation”→