WiFi Router Does Software-Defined Radio

October 12, 2026 by No comments

Perhaps the most famous of software-defined radios is the RTL-SDR, which is very cheap because it’s a digital TV receiver that’s been pressed into service delivering raw I and Q data streams. But the RTL-SDR is certainly not the only chipset that can do this, and today we have a new one.

The MediaTek MT7996 radio chip in the Gemtek W1700K WiFi 7 router can serve as a very competent SDR receiver, and here’s [h0m3us3r] with the code to make it stream over Ethernet to a Linux machine with a web server, through the device’s SoC.

The radio works on the 2.4, 5, and 6 GHz bands, depending upon which version of the device — and thus which radio chip — you have. These may not be entirely useful to many who use lower bands, and given that a similar hack has been found for ESP32 radios on 2.4 GHz it’s likely a demand will arise for a receive up-converter. Until then, you can look at WiFi packets in a waterfall view to your heart’s content.

Since more and more devices now incorporate a radio, we hope that more such undocumented raw streaming modes will be found, and that we’ll see some other new SDRs. Something to look forward to.

The Prusa Core One+ INDX, A Troubled Multi-Extruder Upgrade?

October 12, 2026 by 4 Comments

The Prusa Core One+ INDX is an interesting upgrade to the base FDM printer by Prusa, featuring eight independent extruders – hence the INDX – that in theory should allow for multi-material and multi-color printing with very little purge waste. As [Thomas Sanladerer] found out during his experiences with this printer, there are a lot of issues with this machine, including the ability – or lack thereof – to print with non-Prusa filament.

Although the premise is solid, with a claimed 0.013 grams of purge waste per color or material change along with pretty nice dual-gear Bondtech extruders, the first major hurdle that [Thomas] encountered was that the mandatory calibration sequence on starting a print is designed to only work with Prusa’s own filament and its specific material properties. This resulted in an inability to print with Voxel PLA until [Thomas] removed this sequence manually from the g-code.

This is when [Thomas] found out that the entire print sequence uses the material presets for Prusa’s filament, resulting in a failed print. The results from the manual calibration are also not saved despite the display claiming it does, making this calibration on starting a print absolutely mandatory.

Other issues found are the useless position of the camera as it’s now blocked-off by the INDX hardware, with even manual peeking into the print chamber being heavily obscured by the contraption that holds the extruders. This makes even checking that filament has loaded correctly is borderline impossible, never mind ogling the printing process.

Although there’s a lot to like about this INDX upgrade to the Core One+, the above and other issues discussed in the video make it look somewhat shaky next to competitors like the Bambu Lab H2C and Snapmaker U1, as well as Prusa’s own XL toolchanger printers. During [Thomas]’s print test comparison between the XL and the Core One+ INDX there was no contest that the XL simply works, whereas the INDX even struggles with some Prusa filaments.

Continue reading “The Prusa Core One+ INDX, A Troubled Multi-Extruder Upgrade?” →

The Hexa-Bouncer May Be The Most Overengineered Thing You See This Week

October 12, 2026 by 2 Comments

Sometimes, you want to build a simple project– the minimum viable product that solves your problem and nothing else. Other times you might be feeling a little extra, and want to use oodles of machined aluminum, six steppers, two cameras, a Teensy and a PC with a GPU just to bounce a ball. That’s the mood behind [T-Kuhn]’s hexa-bouncer, a Stewart Platform on steroids.  You can read all about it on his blog or witness it in action in the video below.  Trust us, really have to see it in action.

Continue reading “The Hexa-Bouncer May Be The Most Overengineered Thing You See This Week” →

AVIDdrone's winning entry: a helicopter.

DARPA Lift Challenge Won By Helicopter

October 12, 2026 by 18 Comments

America’s mad scientists all have one dream: DARPA money. DARPA, the Defense Advanced Research Projects Agency, regularly throws round sweet, sweet tax dollars to push the envelope on any number of things, leading to such spinoffs as the very network carrying these words to you. Not every DARPA project ends in revolution though– like the latest DARPA Lift Challenge, whose goal was to see the creation of a vehicle that could perform Vertical Take Off And Landing with a payload 4x its own weight. We’ve embedded the ‘look back’ video below.

We hoped for something revolutionary, something bizzare, something really, really cool– and there were absolutely entries on the field that met all of those superlatives. Unfortunately, sometimes conventional designs are the convention for a reason , and the overall winner by AVIDrone, Inc. was just a very well made electric helicopter. One big lift rotor, one little tail rotor– aside from clever details in the solid-state batteries and carbon-fiber construction, it is a totally normal aircraft… albeit one that could lift 3.84x its dry weight, or 112 lb of payload. Interestingly a battery-powered hexrotor drone took home second place with a 3.64:1 lift ratio. Perhaps the big shock was not seeing combustion engines until third place: another helicopter, this time with a turbine engine. So much for energy density!

Some of the other more bizzare– if less performant–entries did get prizes to further their development, so we’ll likely see them back for the 2028 version of the challenge. Some of the concepts visible were covered here on Hackaday, like turning the drone body into a giant lift rotor , so perhaps one of you will have the bright idea that brings home a million in the next round of the contest.

Continue reading “DARPA Lift Challenge Won By Helicopter” →

Your Computer Memory Is A Lie

October 12, 2026 by 34 Comments

When you first learn to program, memory seems pretty straightforward. You have a bunch of bytes, each with an address. If you want to read or write something, you specify the address, and the computer goes to that place and does your bidding. Sure, some addresses might be reserved for hardware or the operating system, but the general idea is simple: memory is a big array of bytes.

Just like we pretend that a wire is a perfect conductor, the big array of memory model used to be a slight twisting of the truth. On a modern computer, however, it is almost entirely a lie. The address you use probably isn’t a physical address. The data you read might not come from RAM at all. Even if it does, it might be sitting in one of several caches. And if you follow the rabbit hole far enough, you discover that your SSD is arguably just another level in the memory hierarchy.

The Need For Speed

The fundamental problem is that processors have gotten much faster than memory. A modern processor can execute several instructions per clock cycle, but accessing ordinary DRAM can take hundreds of cycles. If the processor had to wait for RAM every time it needed data, much of that impressive computing power would go to waste.

The solution is caching. Keep copies of frequently used data in smaller, faster memories closer to the processor. Most modern desktop processors have at least three levels of cache. The L1 cache is tiny but extremely fast, and often split between instructions and data. L2 is larger and somewhat slower, while L3 is even larger and usually slower as well.

Continue reading “Your Computer Memory Is A Lie” →

Backup Tapes Come Back To The Homelab

October 12, 2026 by 26 Comments

It’s one of the great mysteries of computing, how somehow in the Year Of OLlama 2026, tape remains the god of cheap storage. A while back [Ben Pettis] recounted his experience setting up an LTO-5 tape drive in his homelab, an experience which only becomes more relevant as hard drive prices stay high.

LTO-5 is already well out of date in the professional world, after all– a new server install would be debating the merits of LTO-9 versus LTO-10, but for the homelab it’s a matter of what’s affordable on the used market with decently-sized cartridges. If you can live with 1.5 TB of uncompressed data on a roughly palm-sized Linear Tape, Open cartridge then LTO-5 could be good enough. Yes, the tapes are advertized as 3TB, but that’s assuming a 2:1 compression ratio on the data, which you’re not likely to get on the photos and video that make up much of a data horder’s storage. That’s one thing [Ben] was quick to clear up for us. He also detailed his adventures with the cheaper-to-buy fiber channel model of drive.

The label would have you believe this box stores 30 TB.

Tape drives usually come with an SAS or fiber channel interface– fiber channel being, you guessed it, an optical solution unlike the SCSI-like SAS connection you may be familiar with for block storage devices. With a PCIe adapter it’s not hard to set up, provided you get the cables the right way around! That’s another detail we’re glad [Ben] shared so we can find it again if we have the same problem: with no switch in the circuit, you need a crossover cable to bring TX to RX. Which is obvious in retrospect, but obscure to those of us just learning about the standard.

The advantage, as [Ben] makes clear and everyone has always known is simply price. As of this writing, diskprices.com has 16$/TB as the lowest cost, while [Ben] was able to pick up lightly-used tapes on eBay for only 2$/TB. If you have a lot of data, those savings amortize the drive costs relatively quickly, since a used drive like [Ben]’s can be had for a few hundred bucks.

Is it the ultimate solution for our data storage woes? That depends on your needs, but perhaps not considering we get as many tips on storing data on reel-to-reel as we do on LTO. We even saw a tape-loading robot repurposed to serve bar one time.

A Free Cloud AI

October 12, 2026 by 11 Comments

If you would like an AI “personal agent” but don’t want your data on some opaque cloud computer, there’s a new option that appears to be somewhat nicer: [rociiu] has talorys, an AI agent that runs in your Cloudflare account. True, it is still on a cloud computer, but it is your account on a cloud computer. It is designed so you can use it on Cloudflare’s free account with reasonable limits, and the code running is all on GitHub, so you can — if you are so inclined — look to see what it is doing. The data, too, is all in your cloud instance.

This seems like a reasonable compromise. Cloudflare provides the hosting, and the software in your virtual computer does all the work. At least, that’s how it is promoted. Honestly, we haven’t combed over the code, but the assertion is “Free-tier friendly… no telemetry.”

We are fairly sure they are using Cloudflare’s AI engines, too, so it would be interesting to see how much of your data they could, in theory, harvest in transit. Of course, your data is still in the cloud, so if you are really worried about privacy, this might not be for you. But since everything is in your account and open, you can at least have more control over what’s there. For example, you can easily download or delete your data. You can even see exactly what prompts it uses.

Continue reading “A Free Cloud AI” →