Analog Optical Computer For Inference And Combinatorial Optimization

Although computers are overwhelmingly digital today, there’s a good point to be made that analog computers are the more efficient approach for specific applications. The authors behind a recent paper in Nature are arguing that inference – essential for LLMs – can be done significantly more efficiently using an analog optical computer (AOC).

As the authors describe it, the function of this AOC is to perform a fixed-point search using only optical and analog electronic components. The optics handle the matrix-vector multiplications, while the analog components handle the non-linear operations, subtractions and annealing. This is performed in 20 ns cycles until noise has been reduced to an acceptable level, considering the analog nature of the computer. A big advantage here is that no analog-digital conversions are required as with other (digital) hybrid systems.

So far a small-scale AOC has been constructed for tasks like image classification and non-linear regression tasks, with the authors claiming the AOC being over a hundred times more efficient than current GPU-derived vector processors.

6502 Puts On An SDR Hat

The legendary 6502 microprocessor recently turned 50 years old, and to celebrate this venerable chip which brought affordable computing and video gaming to the masses [AndersBNielsen] decided to put one to work doing something well outside its comfort zone. Called the PhaseLoom, this project uses a few other components to bring the world of software-defined radio (SDR) to this antique platform.

The PhaseLoom is built around an Si5351 clock generator chip, which is configurable over I2C. This chip is what creates the phase-locked loop (PLL) for the radio. The rest of the components, including antenna connectors and various filters, are in an Arduino-compatible form factor that let it work as a shield or hat for the 65uino platform, an Arduino-form-factor 6502 board. The current version [Anders] has been working on is dialed in to the 40-meter ham band, with some buttons on the PCB that allow the user to tune around within that band. He reports that it’s a little bit rough around the edges and somewhat noisy, but the fact that the 6502 is working as an SDR at all is impressive on its own.

For those looking to build their own, all of the schematics and code are available on the project’s GitHub page. [Anders] has some future improvements in the pipe for this project as well, noting that with slightly better filters and improved software even more SDR goodness can be squeezed out of this microprocessor. If you’re looking to experiment with SDR using something a little bit more modern, though, this 10-band multi-mode SDR based on the Teensy microcontroller gets a lot done without breaking the bank.

 

 

Multi-Use Roof Eliminates Roof

One of the biggest downsides of installing solar panels on a rooftop is that maintenance of the actual roof structure becomes much more difficult with solar panels in the way. But for many people who don’t have huge tracts of land, a roof is wasted space where something useful could otherwise go. [Mihai] had the idea of simply eliminating traditional roofing materials altogether and made half of this roof out of solar panels directly, with the other half being put to use as a garden.

Normally solar panels are installed on top of a roof, whether it’s metal or asphalt shingles or some other material, allowing the roof to perform its normal job of keeping weather out of the house while the solar panels can focus on energy generation. In this roof [Mihai] skips this step, having the solar panels pull double duty as roof material and energy generation. In a way this simplifies things; there’s less to maintain and presumably any problems with the roof can be solved by swapping out panels. But we would also presume that waterproofing it might be marginally more difficult.

On the antisolar side of the roof, however, [Mihai] foregoes the solar panels in favor of a system that can hold soil for small garden plants. Putting solar panels on this side of the roof wouldn’t generate as much energy but the area can still be useful as a garden. Of course we’d advise caution when working on a garden at height, but at least for the solar panels you can save some trips up a ladder for maintenance by using something like this robotic solar panel scrubber.

Continue reading “Multi-Use Roof Eliminates Roof”

Round And Round With A Tape Delay Synth

Over the years we’ve been entertained by an array of musical projects from [Look Mum No Computer], and his latest is no exception. It’s a tape delay, loop generator, and synth all in one. Confused? That’s what you get if you position a load of tape heads around a rotating disk with magnetic tape on its perimeter.

Taking a circular piece of inch-thick Perspex, he wraps a length of one inch tape round its perimeter. This is placed as though it were a turntable on a stepper motor with variable speed, and the tape heads are positioned around its edge. Each read head feeds its own preamp which in turn drives a mixer array, and there’s also a record head and an erase head. If you’ve ever played with tape loops you’ll immediately understand the potential for feedback and sequence generation to make interesting sounds. There’s a lot of nuance to the build, in designing the mount for the motor to stop the enclosure flexing, in using a gearbox for increased torque, and in balancing the disk.

The result is as much an effect as it is an instrument in its own right, particularly in its prototype phase when the read head was movable. We’re treated to a demo/performance, and we look forward to perhaps seeing this in person at some point. There’s a future video promised in which a fix should come for a click caused by the erase circuitry, and he’ll make a more compact enclosure for it. Continue reading “Round And Round With A Tape Delay Synth”

Dragon Is The Latest, And Final, Craft To Reboost ISS

The International Space Station has been in orbit around the Earth, at least in some form, since November of 1998 — but not without help. In the vacuum of space, an object in orbit can generally be counted on to remain zipping around more or less forever, but the Station is low enough to experience a bit of atmospheric drag. It isn’t much, but it saps enough velocity from the Station that without regular “reboosts” to speed it back up , the orbiting complex would eventually come crashing down.

Naturally, the United States and Russia were aware of this when they set out to assemble the Station. That’s why early core modules such as Zarya and Zvezda came equipped with thrusters that could be used to not only rotate the complex about all axes, but accelerate it to counteract the impact of drag. Eventually the thrusters on Zarya were disabled, and its propellant tanks were plumbed into Zvezda’s fuel system to provide additional capacity.

An early image of ISS, Zarya module in center and Zvezda at far right.

Visiting spacecraft attached to the Russian side of the ISS can transfer propellant into these combined tanks, and they’ve been topped off regularly over the years. In fact, the NASA paper A Review of In-Space Propellant Transfer Capabilities and Challenges for Missions Involving Propellant Resupply, notes this as one of the most significant examples of practical propellant transfer between orbital vehicles, with more than 40,000 kgs of propellants pumped into the ISS as of 2019.

But while the thrusters on Zvezda are still available for use, it turns out there’s an easier way to accelerate the Station; visiting spacecraft can literally push the orbital complex with their own maneuvering thrusters. Of course this is somewhat easier said than done, and not all vehicles have been able to accomplish the feat, but over the decades several craft have taken on the burden of lifting the ISS into a higher orbit.

Earlier this month, a specially modified SpaceX Cargo Dragon became the newest addition to the list of spacecraft that can perform a reboost. The craft will boost the Station several times over the rest of the year, which will provide valuable data for when it comes time to reverse the process and de-orbit the ISS in the future.

Continue reading “Dragon Is The Latest, And Final, Craft To Reboost ISS”

Scott holding the demo board which has a 7-segment display and keyboard attached

4-bit Single Board Computer Based On The Intel 4004 Microprocessor

[Scott Baker] is at it again and this time he has built a 4-bit single board computer based on the Intel 4004 microprocessor.

In the board design [Scott] covers the CPU (both the Intel 4004 and 4040 are supported), and its support chips: the 4201A clock-generator, its crystal, and the 4289 Standard Memory Interface. The 4289 irons out the 4-bit interface for use with 8-bit ROMs. Included is a ATF22V10 PLD for miscellaneous logic, a 74HCT138 for chip-select, and a bunch of inverters for TTL compatibility (the 4004 itself uses 15 V logic with +5 V Vss and -10 V Vdd).

[Scott] goes on to discuss the power supply, ROM and page mapper, the serial interface, the RC2014 bus interface, RAM, and the multimodule interface. Then comes the implementation, a very tidy custom PCB populated with a bunch of integrated circuits, some passive components, a handful of LEDs, and a few I/O ports. [Scott] credits Jim Loo’s Intel 4004 SBC project as the genesis of his own build.

If you’re interested in seeing this board put to work check out the video embedded below. If you’d like to know more about the 4004 be sure to check out Supersize Your Intel 4004 By Over 10 Times, The 4004 Upgrade You’ve Been Waiting For, and Calculating Pi On The 4004 CPU, Intel’s First Microprocessor.

Continue reading “4-bit Single Board Computer Based On The Intel 4004 Microprocessor”

FreeCAD Foray: Good Practices

Last time, we built a case for a PCB that handles 100 W of USB-C power, an old project that I’ve long been aiming to revive. It went well, and I’d like to believe you that the article will give you a much-needed easy-to-grasp FreeCAD introduction, Matrix knowledge upload style, having you designing stuff in no time.

Apart from my firm belief in the power of open-source software, I also do believe in social responsibilities, and I think I have a responsibility to teach you some decent FreeCAD design practices I’ve learned along the way. Some of them are going to protect your behind from mistakes, and some of them will do that while also making your project way easier to work with, for you and others.

You might not think the last part about “others” matters, but for a start, it matters in the ideal world that we’re collectively striving towards, and also, let’s be real, things like documentation are half intended for external contributors, half for you a year later. So, here’s the first FreeCAD tip that will unquestionably protect you while helping whoever else might work with the model later.

Okay, we’re all hackers, so I’ll start with zero-th FreeCAD tip – press Ctrl+S often. That’ll help a ton. Thankfully, FreeCAD’s autorecovery system has made big leaps, and it’s pretty great in case FreeCAD does crash, but the less you have to recover, the better. Now, onto the first tip.

Continue reading “FreeCAD Foray: Good Practices”