Testing Cheap DC Breakers And How To Not Start Fires

October 28, 2025 by Maya Posch 11 Comments

One characteristic of adding PV solar to homes is a massive increase in high-voltage and high-current DC installations. With this comes a need for suitable breakers, but without the requisite knowledge it can be easy to set up a fire hazard. There is also the issue of online shopping platforms making it easy to get fuses and breakers that may not be quite as capable as they claim, never mind being rated for DC use.

Recently [Will Prowse] had a poke at a range of common purportedly DC-rated breakers from everyone’s favorite US-based seller of tat, to see whether they should be bought or avoided at all cost. Perhaps unsurprisingly the cheap breakers are about as dodgy as you’d imagine. With a hundred plus amps flowing through them they get surprisingly crispy, even if they generally did their job. Minus the few that arrived in a broken condition, of course.

Ultimately [Will] found that the molded case circuit break (MCCB) by one ‘DIHOOL’ performed the best. Compared to the competition, it is much larger and has sizeable terminals that avoid the quaint heat-soaking issues seen with the cheap-and-cheerful rest. At a mere $34 for the 125A-rated version, it’s still a fraction of the cost of a comparable Eaton MCCB, but should upset your insurance company significantly less than the alternatives.

Don’t forget to add in fuses, with [Will] testing a range of cheapo 12V DC fuses, to see which one will prevent fires, and which one cause them. Unsurprisingly, some of them like the Bojack-branded ones ran very hot, making them more of a liability than an asset.

As for what makes DC breakers so different from AC one is that the extinguishing point of a DC arc is much steeper, which means that an AC breaker is likely to fail to extinguish the arc when used for DC applications. This is why a properly rated and ideally certified breaker is essential, and also not really the point where you want to start saving money.

Continue reading “Testing Cheap DC Breakers And How To Not Start Fires” →

Making A Virtual Machine Look Like Real Hardware To Malware

October 27, 2025 by Maya Posch 35 Comments

Running suspicious software in a virtual machine seems like a basic precaution to figure out whether said software contains naughty code. Unfortunately it’s generally rather easy to detect whether or not one’s software runs inside a VM, with [bRootForce] going through a list of ways that a VirtualBox VM can be detected from inside the guest OS. While there are a range of obvious naming issues, such as the occurrence of the word ‘VirtualBox’ everywhere, there many more subtle ways too.

Demonstrated is the PoC ‘malware’ application called Al-Khaser, which can be used to verify one’s anti-malware systems, such as when trying to unleash a debugger on a piece of malware, run it inside a VM, along with many more uses. Among its anti-virtualization features are specific registry key names and values, file system artefacts, directory names, MAC addresses, virtual devices, etc.

In order to squeeze by those checks, [bRootForce] created the vbox_stealth shell script for Bash-blessed systems in order to use the VirtualBox Manager for the renaming of hardware identifier, along with the VBoxCloak project’s PowerShell script that’s used inside a Windows VirtualBox guest instance to rename registry keys, kill VirtualBox-specific processes, and delete VirtualBox-specific files.

Theoretically this should make it much harder for any malware to detect that it’s not running inside Windows on real hardware, but as always there are more subtle ways that are even harder to disguise.

Continue reading “Making A Virtual Machine Look Like Real Hardware To Malware” →

Building A Hydraulic Gear Pump Isn’t So Easy

October 27, 2025 by Maya Posch 8 Comments

The gear pump prototype in action. (Credit: Artisan Makes, YouTube)

Hydraulic gear pumps are deceptively simple: just two gears rotating together, forcing the hydraulic oil from one side to the other where the teeth don’t meet, and thus providing the ability to pressurize said oil to make hydraulic cylinders, final drives, etc. do their thing. As with most machining projects like this, the devil is absolutely in the details, particularly in the tolerances. This is the crash course that the [Artisan Makes] channel on YouTube is currently going through.

In this part one of a series on a DIY gear pump, scrap aluminium is used for the housing, along with 1045 medium carbon steel for the gears and W1A high carbon steel for bearings and other wear surfaces. Since at least one of the gears needs to be driven, a lip seal rated for 10 bar is used to provide a path for the shaft. As noted in the video, this is supposed to be a learning experience, ergo it’s a simplified design that merely targets being functional as a gear pump.

With the basic design figured out, the parts were created on the lathe and mill, followed by assembly. Most of the controversy is about the tolerances within the housing, as any leakage will reduce the efficiency. This means the spacing between the gears and housing, space between the gears and bearings, as well as that provided by the gasket that seals the housing base and top. This is where the comment section somewhat explodes with criticism and advice.

As can be seen in the demonstration with a better gasket, there is absolutely flow when driven at 1200 RPM, but also clearly severe leakage as evidenced by said flow not moving quite as fast as it should. We’re looking forward to the next part, in which addressing these tolerances is tackled, with hopefully a much more performant gear pump resulting.

Continue reading “Building A Hydraulic Gear Pump Isn’t So Easy” →

Satellite Snooping Reveals Sensitive Unencrypted Data

October 27, 2025 by Maya Posch 25 Comments

In an era where running a website without HTTPS is shunned, and everyone wants you to encrypt your DNS queries, you’d expect that the telecommunications back-ends are secured tightly as well. Especially the wireless bits between terra firma and geosynchronous communication satellites.

But as recently discovered by US researchers, the opposite is actually true. The paper by [Wenyi Morty Zhang] et al. (PDF) goes into great detail on how they discovered these unencrypted IP traffic flows and what they found in these captures.

With an off-the-shelf consumer satellite dish mounted to the roof of a university building in San Diego, they performed a scan of IP traffic on 39 geosynchronous satellites. To their surprise, they found unencrypted data that belonged to companies like T-Mobile for their cellular backhaul, Internet traffic targeting airliners, and VoIP communication — all in the clear.

Even more worrying was what looked like military traffic and corporate VPN data containing unencrypted login details, corporate emails and much more. While T-Mobile immediately enabled encryption after this discovery, it remains to be seen whether anyone else will. It’s probably best to assume that any communication can be intercepted and to use e.g. PGP-encrypted emails for anything sensitive.

The researchers have made the IP encapsulation parser (in Python) for DVB-S2(X) captures available for anyone who wants to give this experiment a whirl themselves.

Identifying Fake Small-Signal Transistors

October 25, 2025 by Maya Posch 46 Comments

It’s rather amazing how many electronic components you can buy right now are not quite the genuine parts that they are sold as. Outside of dedicated platforms like Mouser, Digikey and LCSC you pretty much enter a Wild West of unverifiable claims and questionable authenticity. When it comes to sites like eBay and AliExpress, [hjf] would go so far as to state that any of the power transistors available for sale on these sites are 100% fake. But even small-signal transistors are subject to fakes, as proven in a comparison.

Found within the comparison are a Mouser-sourced BC546C, as well as a BC547C, SN3904 and PN2222A. These latter three all sourced from ‘auction sites’. As a base level test all transistors are put in a generic component tester, which identifies all of them correctly as NPN transistors, but the ‘BC547C’ and ‘PN2222A’ fail the test for having a much too low hFE. According to the generic tester at least, but it’s one red flag, along with the pin-out for the ‘BC547C’ showing up as being inverted from the genuine part.

Next is a pass through the HP4145B curve tracer, which confirms the fake BC547C findings, including the abysmal hFE. For the PN2222A the hFE is within spec according to the curve tracer, defying the component tester’s failing grade.

What these results make clear is that these cheap component testers are not a realistic ‘fake’ tester. It also shows that some of the fake transistors you find on $auction_site are clearly fake, while others are much harder to pin down. The PN2222A and 2N3904 used here almost pass the sniff test, but have that distinct off-genuine feeling, while the fake BC547C didn’t even bother to get its pinout right.

As always, caveat emptor. These cheapo transistors can be a nice source for some tinkering, just be aware of possibly wasting hours debugging an issue caused by an off-nominal parameter in a fake part.

The Great ADS1115 Pricing And Sourcing Mystery

October 24, 2025 by Maya Posch 6 Comments

The AdaFruit ADS1115 board hooked up for testing. (Credit: James Bowman)

Following up on the recent test of a set of purported ADS1115 ADCs sourced from Amazon [James Bowman] didn’t just test a genuine Ti part, but also dug into some of the questions that came up after the first article. As expected, the AdaFruit board featuring a presumed genuine Ti ADS1115 part performed very well, even performing significantly better on the tested parameters than the datasheet guarantees.

Thus we can confirm that when you get the genuine Ti part, you can expect very good and reliable performance for your ADC purposes. Which leaves the unaddressed questions about what these cheapo Amazon-sourced ADS1115 ICs are, and how it can be that LCSC has what should be the same parts for so much cheaper than US distributors?

As far as LCSC pricing is concerned, these are likely to be genuine parts, but also the subject of what is known as price discrimination. This involves pricing the same product differently depending on the targeted market segment, with e.g. Digikey customers assumed to be okay with paying more to get the brand name assurance and other assumed perks. Continue reading “The Great ADS1115 Pricing And Sourcing Mystery” →

Making The Smallest And Dumbest LLM With Extreme Quantization

October 23, 2025 by Maya Posch 15 Comments

Turns out that training on Twitch quotes doesn't make an LLM a math genius. (Credit: Codeically, YouTube) — Turns out that training on Twitch quotes doesn’t make an LLM a math genius. (Credit: Codeically, YouTube)

The reason why large language models are called ‘large’ is not because of how smart they are, but as a factor of their sheer size in bytes. At billions of parameters at four bytes each, they pose a serious challenge when it comes to not just their size on disk, but also in RAM, specifically the RAM of your videocard (VRAM). Reducing this immense size, as is done routinely for the smaller pretrained models which one can download for local use, involves quantization. This process is explained and demonstrated by [Codeically], who takes it to its logical extreme: reducing what could be a GB-sized model down to a mere 63 MB by reducing the bits per parameter.

While you can offload a model, i.e. keep only part of it in VRAM and the rest in system RAM, this massively impacts performance. An alternative is to use fewer bits per weight in the model, called ‘compression’, which typically involves reducing 16-bit floating point to 8-bit, reducing memory usage by about 75%. Going lower than this is generally deemed unadvisable.

Using GPT-2 as the base, it was trained with a pile of internet quotes, creating parameters with a very anemic 4-bit integer size. After initially manually zeroing the weights made the output too garbled, the second attempt without the zeroing did somewhat produce usable output before flying off the rails. Yet it did this with a 63 MB model at 78 tokens a second on just the CPU, demonstrating that you can create a pocket-sized chatbot to spout nonsense even without splurging on expensive hardware.

Continue reading “Making The Smallest And Dumbest LLM With Extreme Quantization” →