One of the delights of our tips line is that from time to time it brings us retrocomputing hardware that, despite years of reporting, we were not aware existed. [Hitmanmcc] has just such a machine, an NEC PC Engine LT. It’s a PC engine in a laptop form factor, and like many of this super-rare console, it has succumbed to capacitor failure. We’re treated to the process of bringing it back to life.
Replacing capacitors was only part of the story for this repair, as the electrolyte had caused damage elsewhere on the board. In particular there is a small transformer that forms part of an inverter to generate an LCD bias voltage, and this had been destroyed. Fortunately the art of switching power conversion has advanced in the decades since the console was produced, and a small module was procured to do the same job.
The result of all this surgery is another rare console rescued from e-waste, and an opportunity for the rest of us to take a look too. The PC engine is a relative rarity here, but we’ve had a few hacks over the years. This converter for its American cousin is one.
Whenever the topic of fusion power comes up, someone will say it’s only 10 years away from commercialization in an excited tone, and someone older or more cynical will point out that it’s been 10 years away since Eisenhower was president. So it’s with a certain-sized crystal of sodium chloride that we share the news here that the US-based Commonwealth Fusion Systems is applying to feed 400MWe into the grid there by the early 2030s.
The early 2030s is, notably, less than ten years from now.
Commonwealth Fusion Systems isn’t a bunch of nobodies out to suck up venture capital; they’re a talented group of researchers from MIT’s well-known plasma laboratory out to suck up lots of venture capital and hopefully build reactors along the way. So far, the second part is going better than the first: they’ve raised a couple billion dollars, which has let them make great strides in building their SPARC reactor– like crafting the big magnet we told you about in 2021. As that article describes, SPARC is the precursor to the later, larger ARC reactor they hope to hook to the grid in slightly under a decade. Alas, SPARC remains under construction as of this writing. ARC is evidently in the final planning stages, with a physical location determined and grid-tie applied for at the “Fall Line Fusion Power Station” in Virginia.
CFS’s reactors are of the Tokamak type that has been favoured at universities since the 1970s. From China to Europe’s ITER who are also planning to produce power before another decade passes— though not, notably, into a power grid. While promising, Tokamaks aren’t the only game in town, either– steampunk startup General Fusion started making plasma last year, though while if it works it has some big advantages, that one is probably the traditional “ten years away” still.
What do you think? Will fusion power be in the grid before humans make it back to the moon? Add the flying-car potential of eVTOL and we might finally get close to the future we were promised.
COBOL is not the first language anyone would ever think of when writing a First Person Shooter– after all , it’s the Common Business Oriented Language, not the Common Game Oriented Language. For Youtube-based hacker [icitry] though, that’s the point. The only way to determine if COBOL would be enough to write an FPS game was to do it.
Sure, you could rest on your laurels knowing that the language is Turing complete and therefore capable by definition, but what’s the fun in that? Now the pipeline for this game is as hacky as anything– COBOL doesn’t exactly have a robust graphics stack or a lot of libraries for pushing pixles, so he’s outputting each frame of the game as raw bitmap to STDOUT, and letting ffplay assemble the images. Control enters the same way, with the terminal set to raw input and the COBOL program reading STDIN.
As for what the images consist of, he’s going for a standard Wolfenstien-inspired raycasting shooter. [icitr] provides a decent explanation of the raycasting algorithm, along with why implementing in COBOL is a silly thing to try. That’s a theme here; he’s able to implement sprites and the logic to move and attack enemies, while constantly complaining about COBOL. If that wasn’t enough, he adds variable-height sectors to bring this much closer to a true DOOM clone. By the end, there’s a full game. It’s all up on GitHub on an Apache license.
While this video is not the most gentle introduction to COBOL, it does show you can hack the business-specific language to do whatever you’d like.
Although cryocoolers are capable of pretty impressive cooling, for many of them the underlying working principle is simple enough that you do not need any special skills or a big budget to make your own version. Take the Gifford-McMahon cryocooler for example, which works using nothing more than some kind of coolant gas and a piston in a cylinder that you can even 3D print, as demonstrated by [Hyperspace Pirate] in a recent video.
The lowest temperature reached across the two prototypes was only -84°C, but this was mostly due to some sub-optimal design choices, such as the use of regular air and a clear acrylic tube to get a good glimpse at the inner workings. The trickiest part of this type of cryocooler is probably that you need to move the piston containing the regenerator between both ends of the cylinder to get a cool and a hot side.
That particular problem was solved by using magnets to move the piston externally, which worked beautifully until the problem of using regular compressed air from the shop compressor caused massive ice formation that jammed up the piston. Obviously this was not an unexpected issue, and for the next step the coolant gas will be replaced by helium, as making that gas freeze up requires quite a bit more effort.
A good demo, like [Linus Akesson]’s Sum Ergo Demonstrato, looks like magic to the average hacker. To normies who don’t know the limitations of the RP2350, they don’t see the big deal. To anyone who has spent any time with the chip, though, it’s a series of tricks you cannot help but be amazed by. Fortuanately for us, [Linus] isn’t actually a magician, because while a magician never reveals his tricks, [Linus] has an hour-long video explaining exactly how his demo was accomplished. We’ve embedded both the demo and the explanation below.
Even if you aren’t into YouTube, you should check out the demo video, and again– remember this is all on a Pi Pico with only the extra passives required for video-out. Then you can watch [Linus] explain how he did it, which is really best heard in his own words. There are a couple of bleeding-edge tricks on the RISC V core and peripherals that we would hate to misrepresent– especially the clever hack with the interpolator that he uses for 3D acceleration.
If this sounds a bit familiar, it’s because we were equally impressed by his Kaleidoscopico demo last year. From demos like this to 3D engines on the ESP32, its amazing what you can do on modern micros if you’re willing to hit the limits of the hardware.
If you’ve ever worked with I2C, you know its one of those things that makes working with modern microcontrollers such a pleasure. With a few wires and not many more lines of code, you can communicate with all sorts of hardware such as sensors, displays, and input devices. There are even I2C keyboards out there, although they tend to be a bit pokey — and not in the good way as it pertains to keyboards.
But the bt2i2c project from [Roberto Alsina] promises to improve things. With his firmware flashed to a Pi Pico W, you can establish a connection with any standard Bluetooth keyboard and have the keystrokes sent over the wire via I2C. As far as your project is concerned, the input will appear to be coming from a BlackBerry BBQ20/BBQ10 keyboard using the address 0x1F, which means that there’s already plenty of code out there to work with. While [Roberto] explains its not strictly necessary, connecting a ST7789 display to the Pi Pico over SPI will give you some visual feedback on connection status.
As microcontrollers become increasingly powerful and capable of the sort of thing we would once have done on a “real” computer, a project like this has some fascinating potential. We’ve seen a number of “writerdeck” projects running on chips like the ESP32, and it’s not hard to see the appeal of being able to easily pair your favorite Bluetooth keyboard up to one of them.
The test parts being printed on the Stratasys Fortus 450mc. (Credit: My Tech Fun, YouTube)
Professional Stratasys FDM printers demand a pretty hefty price premium over your typical hobbyist-level machine, with the gold-plating continuing even with the special filament cartridges that you buy for some of their printers.
This raises the question of in how far this eye-watering price tag is justified, and how much is just you paying for support and the brand name. After acquiring a spool of Stratasys ABS filament via a US viewer, [Dr. Igor Gaspar] set to work to try and answer this question.
The viewer had already liberated the spool of ABS+ P430 filament from its cartridge, making it easy to use that directly with the Bambu Lab FDM printer.
To make it a fair comparison, [Igor] also needed to have a sample printed on a real Stratasys printer, for which he used a local company’s services. An interesting sidenote here is that the US viewer’s company moved away from Stratasys to Bambu Lab printers.
[Igor] was able to see his test parts being printed on the Stratasys printer, as said company is in the same city. This showed him that it took 14 hours to print the parts versus 3.5 hours on the Bambu Lab printer, suggesting that his worries about the right printing parameters for the Stratasys filament were warranted. Sussing those out was thus paramount for a fair comparison and warranted some test prints.
From a sheer aesthetic point of view the Stratasys-printed parts looked much cleaner, and their dimensional accuracy was also significantly better due to the slicer adjusting for this. Between the used Stratasys M30 and Bambu Lab ABS filaments there’s no clear winner, with both trading blows. Amusingly enough, the older Stratasys ABS type in the form of the ABS+ P430 filament performed the best of all when printed on the Bambu Lab printer at its preferred temperature setting.
Moral of the story is thus that – unless you really want to pay for that service contract – to loot old Stratasys ABS spool cartridges and use them in your hobbyist FDM printer. As [Igor] says in the conclusion, the nicer looks is probably due to them printing very thin layers, much finer than the 0.2 mm layers he used. This would also match the much longer print time and is thus something we can replicate on any FDM printer with a temperature-controlled printing environment.
