Creating Your Alarm On The Fly

November 5, 2022 by Matthew Carlson 4 Comments

We suspect that most of us who use an alarm clock have our particular sound memorized. Common choices are annoying beeping, energetic marimbas, or what used to be your favorite song (which you have now come to despise). [Adam Kumpf] wanted a more pleasant alarm clock and came up with WakeSlow, an alarm clock audio stream, which is a spiritual successor to an earlier project he did called Warmly.

Some might say, “an audio stream? You could create an acceptable alarm tone generator with a 555 and a 2N2222”. The idea behind WakeSlow is to use your existing internet-connected alarm clock that can play an audio stream. You generate a URL using WakeSlow, and it plays the alarm. A custom URL is helpful since it incorporates weather data, letting you know if it’s going to rain, blowing wind, or be sunny that day. It mixes CC0 audio to form the stream, and includes a 5-minute fade to wake you up gradually. After five minutes, it’s jazz time, and it plays a sample of some CC0 jazz.

The code is super simple, and he makes it available on his website under a public domain/CC0 license. The simplicity offers something powerful, making it exactly how you like it. You could incorporate holiday information, a text-to-speech news announcer reading the news of what’s on your calendar that day, or anything you can dream of.

Hackers are generally particular about clocks, and alarm clocks fall under the same umbrella. WakeSlow allows you to skip the hardware part of making your customized alarm, but if you prefer to have the whole thing be custom, we have a few suggestions for alarms to look at.

Compensating For Your TVs Backlight

October 10, 2022 by Matthew Carlson 8 Comments

[Pekka Väänänen] has a Panasonic TV with a broken backlight that creates an uneven pink/green color. While it isn’t a huge deal for most films, black-and-white films tend to show the most effect. So, by modeling the distortion as a function, [Pekka] set out to find an inverse function that corrects the distortion before it gets to the TV.

However, the backlight doesn’t emit enough light for some colors, which means the blue and green channels need to be dimmed. As mentioned earlier, the distortion isn’t even, so the distortion needs to be captured and then calculated.

He took a few pictures with his phone, corrected the perspective, and applied a blur. The camera also has some distortion but works as a first approximation, but that isn’t something he covered here. Next, he set up a webcam and pointed it at the TV, trying to find good gain and offset values with a bit of Python.

Now it just becomes a problem of minimizing the per-pixel difference. Ultimately he just went for a random approach rather than an annealing or hill-climbing approach. Now that he had a function to apply, it was just a matter of adding a custom shader to his video player, which includes a live shader editor. He had to hack in support for an external texture, but he is kind enough to include the shader code and the patch in the article.

The result is excellent, and it’s a great use for an old TV. But perhaps, in some cases, it might be worth replacing the backlight entirely.

Fixing A 30-year Old Roland Bug

October 5, 2022 by Matthew Carlson 12 Comments

The Roland CM-500 is a digital synthesizer sound module released in 1991 that combines two incredibly powerful engines into one unit. However, in 2005 enthusiasts of the Roland MT-25 (one of the engines that went into the CM-500) noticed a difference between the vibrato rate on the MT-25 and the CM-500, rendering it less useful as now midi files would need to be adjusted before they sounded correct. Now thirty-something years later, there is a fix through the efforts of [Sergey Mikayev] and a fantastic writeup by [Cloudschatze].

They reached out to Roland Japan, who decided that since the device’s lifecycle had ended, no investigation was warranted. That led the community to start comparing the differences between the two systems. One noticeable difference was the change from an Intel 8098 to an 80C198. In theory, the latter is a superset of the former, but there are a few differences. First, the crystal frequency is divided by three rather than two, which means the period of the LFO would change even if the crystal stayed the same. Changing the 12 MHz crystal out for 8 MHz gave the LFO the correct period, but it broke the timings on the MIDI connection. However, this is just setting the serial baud rate divisor, which requires changing a few bytes.

Replace the ROM chip with a socket so you can slot your newly flashed PDIP-28 64kx8 ROM into a quick desoldering. Then swap the crystal, and you’ll have a machine that matches the MT-25 perfectly. The forum post has comparison audio files for your enjoyment. Finally, if you’re curious about other fixes requiring an inspiring amount of effort and dedication, here’s a game installer that was brought back from the dead by a determined hacker.

AI Dreaming Of Time Travel

September 28, 2022 by Matthew Carlson 20 Comments

We love the intersection between art and technology, and a video made by an AI (Stable Diffusion) imagining a journey through time (Nitter) is a lovely example. The project is relatively straightforward, but as with most art projects, there were endless hours of [Xander Steenbrugge] tweaking and playing with different parts of the process until it was just how he liked it. He mentions trying thousands of different prompts and seeds — an example of one of the prompts is “a small tribal village with huts.” In the video, each prompt got 72 frames, slowly increasing in strength and then decreasing as the following prompt came along.

There are other AI videos on YouTube, often putting the lyrics of a song into AI-generated form. But if you’ve worked with AI systems, you’ll notice that the background stays remarkably stable in [Xander]’s video as it goes through dozens of feedback loops. This is difficult to do as you want to change the image’s content without changing the look. So he had to write a decent amount of code to try and maintain visual temporal cohesion over time. Hopefully, we’ll see an open-source version of some of his improvements, as he mentioned on Twitter.

In the meantime, we get to sit back and enjoy something beautiful. If you still aren’t convinced that Stable Diffusion isn’t a big deal, perhaps we can do a little more to persuade your viewpoint.

Continue reading “AI Dreaming Of Time Travel” →

Fork And Run: The Definitive Guide To Getting Started With Multiprocessing

September 15, 2022 by Matthew Carlson 23 Comments

Since the early 2000s, the CPU industry has shifted from raw clock speed to core counts. Pat Gelsinger famously took the stage in 2002 and gave the talk the industry needed, stating processors needed specialty silicon or multiple cores to reduce power requirements and spread heat. A few years later, the Core series was introduced with two or four-core configurations to compete with the AMD Athlon 64 x2.

Nowadays, we’re seeing heterogeneous chip designs with big and little cores, chiplets, and other crazy fabrication techniques that are fundamentally the same concept: spread the thermal load across multiple pieces of silicon. This writer is willing to put good money into betting that you’ll see consumer desktop machines with 32 physical cores in less than five years. It might be hard to believe, but a 2013 Intel Haswell i7 came with just four cores compared to the twenty you’ll get in an i7 today. Even an ESP32 has two cores with support in FreeRTOS for pinning tasks to different cores. With so many cores, how to even write software for that? What’s the difference between processes and threads? How does this all work in straight vanilla C98?

Continue reading “Fork And Run: The Definitive Guide To Getting Started With Multiprocessing” →

If The Blade Sees Its Shadow, It’s Another 64th Of Accuracy

September 11, 2022 by Matthew Carlson 43 Comments

If you’ve bought a miter saw in the past few years, you might have noticed the LED “laser” that came with it. The goal was to show where on the piece the saw was going to cut. But over a year or two, you might have found the laser to have drifted or skewed into a crooked line. [Fisher] decided that his after-market laser wasn’t entirely accurate enough and added a shadow line instead. (Video, embedded below the break.)

The blade has a thickness (known as kerf), and with a laser to one side, you can only accurately cut on one side of the line. A shadow line works differently. By shining a line at the top of the blade, you get a mark where the blade will cut precisely. You can also see your marks as the laser doesn’t shine over them. Previously, [Fisher] had tried to use LED strips, but after a comment suggested it, he found a sewing light on a gooseneck. It worked great as a small compact light fitting the blade housing. After some quick modifications, hot glue, and duct tape, the light was installed, and the wires were routed while still allowing the saw its full range of motion.

The result is impressive, with a clear shadow on even darker hardwoods. Just the few test cuts he made seemed entirely accurate. Of course, you can always go deeper down the hole of accuracy and measurement. But overall, [Fisher] has a great little mod that speeds up his workflow more accurately. Continue reading “If The Blade Sees Its Shadow, It’s Another 64th Of Accuracy” →

Tesla’s Dojo Is An Interesting CPU Design

September 6, 2022 by Matthew Carlson 29 Comments

What do you get when you cross a modern super-scalar out-of-order CPU core with more traditional microcontroller aspects such as no virtual memory, no memory cache, and no DDR or PCIe controllers? You get the Tesla Dojo, which Chips and Cheese recently did a deep dive on.

It starts with a comparison to the IBM Cell processors. The Cell of the mid-2000s featured something called the SPE (Synergistic Processing Elements). They were smaller cores focused on vector processing or other specialized types of workloads. They didn’t access the main memory and had to be given tasks by the fully featured CPU. Dojo has 1.25MB of SRAM that it can use as working memory with five ports, but it has no cache or virtual memory. It uses DMA to get the information it needs via a mesh system. The front end pulls RISC-V-like (heavily MIPS-inspired) instructions into a small instruction cache and decodes eight instructions per cycle. Continue reading “Tesla’s Dojo Is An Interesting CPU Design” →