Most of us using desktop computers, and plenty of us on laptops, have some sort of fan or pump installed in our computer to remove heat and keep our machines running at the most optimum temperature. That’s generally a good thing for performance, but comes with a noise pollution cost. It’s possible to build fanless computers, though, which are passively cooled by using larger heat sinks with greater thermal mass, or by building more efficient computers, or both. But sometimes even fanless designs can benefit from some forced air, so [Sasa] built this system for cooling fanless systems with fans.

The main advantage of a system like this is that the fans on an otherwise fanless system remain off when not absolutely necessary, keeping ambient noise levels to a minimum. [Sasa] does have a few computers with fans, and this system helps there as well. Each fan module is WiFi-enabled, allowing for control of each fan on the system to be set up and controlled from a web page. It also can control 5V and 12V fans automatically with no user input, and can run from any USB power source, so it’s not necessary to find a USB-PD-compatible source just to run a small fan.

Like his previous project, this version is built to easily integrate with scripting and other third-party software, making it fairly straightforward to configure in a home automation setup or with any other system that is monitoring a temperature. It doesn’t have to be limited to a computer, either; [Sasa] runs one inside a server cabinet that monitors the ambient temperature in the cabinet, but it could be put to use anywhere else a fan is needed. Perhaps even a hydroponic setup.

The 555 timer is one of the most versatile integrated circuits available. It can generate PWM signals, tones, and single-shot pulses. You can even put one in a bi-stable mode similar to a flip flop. All of these modes are available by only changing a few components outside of the IC itself. It’s also dirt cheap, so it finds its way into all kinds of applications its original inventors never imagined. There’s a bit of a trope around here as well that you ought not to use a microcontroller when one of these will do, and while it’s a bit of a played-out comment, it’s often more true than it seems. This video shows a few uncommon ways of using these circuits instead of putting a microcontroller to work.

After a brief overview of the internals of the hallowed 555, [Doctor Volt] walks us through some of its uses, starting with applications for digital inputs, including a debounce circuit and a toggle switch. From there, he moves on to demonstrating a circuit that can protect batteries from deep discharge, and a small change to that circuit can turn the 555 into a resetting fuse that can protect against short circuit events. Finally, the PWM capabilities of this small integrated circuit are put to work as an audio amplifier, although perhaps not one that would pass muster for the most devout audiophiles among us.

Even though it’s possible to offload a lot of the capabilities of a 555 onto a microcontroller, there’s certainly an opportunity to offload some things to the 555, even if your project still needs a microcontroller. However, offloading tasks like debounce or input latching to hardware rather than spending microcontroller cycles or pins can make a project more robust, both from reliability and software points of view. For some other useful circuits, some of which have been forgotten in the modern microcontroller age, it’s worth taking a look at some of these antique circuit books as well. While we are sure the 555 designers hoped it would be a big hit, no one imagined this giant one.

For those who have not dealt with the automotive side of electronics before, it comes as somewhat of a shock when you find out just how much extra you have to think about and how tough the testing and acceptance standards are. One particular test requirement is known as the “load dump” test. [Tim Williams] needed to build a device (first article of three) to apply such test conditions and wanted to do it as an exercise using scrap and spares. Following is a proper demonstration of follow-through from an analytical look at the testing specs to some interesting hand construction.

The load dump test simulates the effect of a spinning automotive alternator in a sudden no-load scenario, such as a loose battery terminal. The sudden reduction in load (since the battery no longer takes charging current) coupled with the inductance of the alternator windings causes a sudden huge voltage spike. The automotive standard ISO 7637-2:2011 dictates how this pulse should be designed and what load the testing device must drive.

The first article covers the required pulse shape and two possible driving techniques. It then dives deep into a case study of the Linear Tech DC1950A load dump tester, which is a tricky circuit to understand, so [Tim] breaks it down into a spice model based around a virtual transistor driving an RC network to emulate the pulse shape and power characteristics and help pin down the specs of the parts needed. The second article deals with analysing and designing a hysteric controller based around a simple current regulator, which controls the current through a power inductor. Roughly speaking, this circuit operates a bit like a buck converter with a catch diode circulating current in a tank LC circuit. A sense resistor in the output path is used to feedback a voltage, which is then used to control the driving pulses to the power MOSFET stage. [Tim] does a good job modeling and explaining some of the details that need to be considered with such a circuit.

The third and final article turns what’s been learnt so far into a practical design that can be built, with many extra parts added and explained to make this work in reality. It was nice to see ICs being mostly rejected in favour of a discrete design using transistors and other parts at hand—you can see the individual circuit elements if you know what you’re looking for. That said, the venerable 555 timer is in there, doing one of the things it does best: being a trigger timer.  The physical construction is done Manhattan-style on a couple of board layers, with some hilariously outsized parts bolted on just because. There’s much to learn from this project, although it will be a tough read for any newcomer to electronics.

While we’re considering building our own instruments, here’s an active load build. EMC testing is one of those areas that can really cause problems. Here’s our guide. We don’t see enough discrete components used in projects these days. Here’s a discrete transistor CPU to admire.

If you’re a fan of vintage electronics and DIY tinkering, you’ll find this teardown by [Thomas Scherrer] fascinating. In a recent video, he delves into a rare piece of equipment: the Data Lab Transient Recorder DL 901. This device looks like a classic one-channel oscilloscope, complete with all the knobs and settings you’d expect.

The DL 901, made by Data Laboratories Ltd., is a mystery even to [Thomas], who couldn’t find any documentation online. From the DC offset and trigger settings to the sweep time controls, the DL 901 is equipped to handle slow, high-resolution analog-to-digital conversion. The circuitry includes TTL chips and a PMI DAAC 100, a 10-bit digital-to-analog converter. [Thomas] speculates it uses a successive approximation technique for analog-to-digital conversion—a perfect blend of analog finesse and digital processing for its time.

Despite its intriguing features, the DL 901 suffers from a non-responsive analog input system, limiting the teardown to a partial exploration. For those who enjoyed past Hackaday articles on oscilloscope teardowns and analog tech, this one is a treat. Watch the video to see more details and the full process of uncovering this vintage device’s secrets.

In this episode of [Adrian’s Digital Basement], we dive into the world of retro computing with a focus on diagnosing and repairing an old full-height 5.25-inch floppy drive from an IBM 5150 system. Although mechanically sound, the drive had trouble reading disks, and Adrian quickly set out to fix the issue. Using a Greaseweazle—a versatile open-source tool for floppy disk diagnostics—he tests the drive’s components and explores whether the fault lies with the read/write head or electronic systems.

The repair process provides fascinating insights into the Tandon TM100-1 floppy drive, a key player in vintage computing. Adrian explains how the drive was designed as a single-sided unit, yet hints at potential double-sided capability due to its circuit board, raising possibilities for future tweaks. Throughout the video, Adrian shares handy tips on ensuring proper mechanical maintenance, such as keeping lubrication in check and ensuring correct spring tension. His attention to detail, especially on termination resistors, provided vital knowledge for anyone looking to understand or restore these old drives.

For fans of retro tech, this episode is a must-watch! Adrian makes complex repairs accessible, sharing both technical know-how and nostalgic appreciation. For those interested in similar hacks, past projects like the Greaseweazle tool itself or other Amiga system repairs are worth exploring. To see Adrian in action and catch all the repair details, check out the full video.

There are a lot of fantastic things about Hackaday Supercon, but for me personally, the highlight is always seeing the dizzying array of electronic bits and bobs that folks bring with them. If you’ve never had the chance to join us in Pasadena, it’s a bit like a hardware show-and-tell, where half the people you meet are eager to pull some homemade gadget out of their bag for an impromptu demonstration. But what’s really cool is that they’ve often made enough of said device that they can hand them out to anyone who’s interested. Put simply, it’s very easy to leave Supercon with a whole lot more stuff than when you came in with.

Most people would look at this as a benefit of attending, which of course it is. But in a way, the experience bummed me out for the first couple of years. Sure, I got to take home a literal sack of incredible hardware created by members of our community, and I’ve cherished each piece. But I never had anything to give them in return, and that didn’t quite sit right with me.

So last year I decided to be a bit more proactive and make my own Simple Add-On (SAO) in time for Supercon 2023. With a stack of these in my bag, I’d have a personalized piece of hardware to hand out that attendees could plug right into their badge and enjoy. From previous years I also knew there was something of an underground SAO market at Supercon, and that I’d find plenty of people who would be happy to swap one for their own add-ons for mine.

To say that designing, building, and distributing my first SAO was a rewarding experience would be something of an understatement. It made such an impression on me that it ended up helping to guide our brainstorming sessions for what would become the 2024 Supercon badge and the ongoing SAO Contest. Put simply, making an SAO and swapping it with other attendees adds an exciting new element to a hacker con, and you should absolutely do it.

So while you’ve still got time to get PCBs ordered, let’s take a look at some of the unique aspects of creating your own Simple Add-On.

Low Barrier to Entry

To start with, let’s cover what’s probably the biggest benefit of making an SAO versus pretty much any other kind of electronic device: essentially all the hard work has been done for you, so you’re free to explore and get creative.

Consider the SAO standard, such as it is. You know there’s going to be 3.3 volts, you know physically how your device will interface with the host badge, and should you decide to utilize it, there’s an incredibly common and well-supported protocol (I2C) in place for communication with other devices.

There’s even a pair of GPIO pins thrown in for good measure, which more nuanced versions of the SAO spec explain can be used as the clock and data pins for addressable LEDs. In either event, they provide an even easier way to get your SAO talking to whatever it’s plugged into than I2C if that’s what you’re after.

Not having to worry about power is a huge weight off your shoulders. Voltage regulation — whether it’s boosting the output from a battery, or knocking down a higher voltage to something that won’t fry your components — can be tricky, and has been known to trip up even experienced hardware hackers. There’s admittedly some ambiguity about how much current an SAO can draw, but unless you’re looking to push the envelope, it’s unlikely anything that fits in such a small footprint could pull enough juice to actually become a problem.

Minimal Investment

Another thing to consider is the cost. While getting PCBs made today is cheaper than ever, the cost still goes up with surface area. Especially for new players, the cost of ordering larger boards can trigger some anxiety. Luckily, the traditional SAO is so small that having 20, 30, or even 50 of them made won’t hit you too hard in the wallet. Just as an example, having 30 copies of the PCB for my first SAO fabricated overseas cost me around $12 (shipping is the expensive part).

In fact, an SAO is usually small enough that a quick-turn prototype run with one of the domestic board houses might be within your budget. I’ve been playing around with a new SAO design, and both DigiKey and OSH Park quoted me around $40 to have a handful of boards produced and at my doorstep within 5 to 7 days.

Now assembly of your SAOs, should you outsource that, can still be expensive. Even though they’re small, it’s all going to come down to what kind of parts you’re using in the design. I was recently talking to Al Williams around the Hackaday Virtual Water Cooler, and he mentioned the cost to have just a handful of his SAO made was in the three figures. Then you look at the parts he used in the design, and it was clear this was never going to be a cheap build.

But even if you’ve got deep enough pockets to pay for it, I’d personally recommend against professional assembly in most cases. Which leads nicely into my next point…

A Taste of Mass Production

Being hobbyists, the reality is that most of us never get the opportunity to build more than a few copies of the same thing. For a personal project, there’s rarely the need to build more than one — and even if you count the early prototypes or failed attempts, it’s unlikely you’d hit the double digits.

But for an SAO, the more the merrier. If you’re planning on swapping with others or giving them away, you’ll obviously want quite a few of them. There’s no “right” number here, but for an event the size of Supercon, having 50 copies of your SAO on-hand would be reasonable. As mentioned earlier, I went with 30 (in part due to the per-unit cost) and in the end felt I should have bumped it up a bit more.

But even at 30, it was far and away the largest run of any single thing I’d ever done. After assembling the third or fourth one, I started to pick up on tricks that would speed up the subsequent builds. Where applicable, hand-soldering quickly gave way to reflowing. After some initial struggling, I realized taking the time to make a jig to hold the more fiddly bits would end up saving me time in the long run. Once ten or so were in various states of completion, it became clear I needed some way to safely hold them while in production, so I ended up cutting a couple board holders out of wood on the laser cutter.

Looking back, this part of the process was perhaps what I enjoyed the most. As you might expect, I’ve been involved with  badge production at significant scales in the past. If you have a Supercon badge from the last several years, there’s an excellent chance I personally handled it in some way before you received it. But this was an opportunity to do everything myself, to solve problems and learn some valuable lessons.

Finding a New Community

Finally, the most unique part of making your own SAO is that it’s a ticket to a whole new subculture of hardware hacking.

There are some incredibly talented people making badges and add-ons for the various hacker cons throughout the year, and there’s nothing they like better than swapping their wares and comparing notes. These folks are often pushing the very limits on what the individual hacker and maker is capable of, and can be a wealth of valuable information on every aspect of custom hardware design and production.

When you put your creation up on the SAO Wall at Supercon, or exchange SAOs with somebody, you’re officially part of the club, and entitled to all the honors and benefits occurring thereto. Don’t be surprised if you soon find yourself on a private channel in an invite-only chat server, pitching ideas for what your next project might be.

With a little over a month to go before the 2024 Hackaday Supercon kicks off in Pasadena, and a couple weeks before the deadline on submissions for the Supercon Add-On Contest, there’s still time to throw your six-pin hat into the ring. We can’t wait to see what you come up with.

A few years back NVIDIA created a dedicated cryptocurrency mining GPU, the CMP 170HX. This was a heavily restricted version of its flagship A100 datacenter accelerator, using the same GA100 chip. It was intended for accelerating Ethash, the Etherium proof-of-work algorithm, and nothing else. [niconiconi] bought one to use for accelerating PCB electromagnetic simulations and put a lot of effort into repairing the card, converting it to water-cooling, and figuring out how best to use this nobbled GPU.

Typically, the GA100 silicon sits in the center of the mighty A100 GPU card and would be found in a server rack, cooled by forced air. This was not an option at home, so an off-the-shelf water-cooling block was wedged in. During this process, [niconconi] found that the board wouldn’t power on, so they went on a deep dive into the power supply tree with the help of a leaked A100 schematic. The repair and modifications can be found in the appendix, right down to the end of the article. It is a long read to get there.

NVIDIA has a history of deliberately restricting silicon in consumers’ hands to justify the hefty price tags of its offerings to big businesses, and this board is no different. The plan was to restrict the peak performance of the board to only applications with the same compute requirements as Ethash, specifically memory-intensive algorithms. The FP64 performance was severely limited, but instructions were not removed. This meant the code would run really badly, considering what the GPU is capable of.

The memory was limited to 8 GB, despite some A100 cards hosting a whopping 80 GB. The strategy was to use fuses to limit the crucial instructions, particularly the FP32 FMA and MAD instructions, which are used for multiply-add operations and are crucial for general computing applications. Finally, the PCIe bus was nobbled to run only as a Gen 1 interface with a single lane. They reduced the lane count by removing the coupling capacitors on the PCB, which meant they could just be added later, but it’s still only a slow interface.

[niconconi] went into great detail benchmarking the instruction types, keeping their EM simulation application in mind. After a few tweaks to make it work, they determined it was a good purchase. This article is worth reading for all those hardcore GPU nerds!

If you need a primer on GPU mining, we’ve got you covered. Once you’ve understood proof-of-work crypto, perhaps take a look at Chia?

Thanks to [gnif] for the tip!