An LLM For The Raspberry Pi

May 10, 2025 by Al Williams 32 Comments

Microsoft’s latest Phi4 LLM has 14 billion parameters that require about 11 GB of storage. Can you run it on a Raspberry Pi? Get serious. However, the Phi4-mini-reasoning model is a cut-down version with “only” 3.8 billion parameters that requires 3.2 GB. That’s more realistic and, in a recent video, [Gary Explains] tells you how to add this LLM to your Raspberry Pi arsenal.

The version [Gary] uses has four-bit quantization and, as you might expect, the performance isn’t going to be stellar. If you are versed in all the LLM lingo, the quantization is the way weights are stored, and, in general, the more parameters a model uses, the more things it can figure out.

Continue reading “An LLM For The Raspberry Pi” →

Best Practices For FDM Printing

May 10, 2025 by Al Williams 38 Comments

If you’ve been designing parts for 3D printing, you probably have some tricks and standards for your designs. [Rahix] decided to write out a well-thought-out set of design rules for FDM prints, and we can all benefit.

One of the things we liked about the list is that it’s written in a way that explains everything. Every so often, there’s a box with a summarized rule for that topic. At the end, there’s a list of all the rules. The rules are also in categories, including part strength, tolerance, optimization, integration, machine elements, appearance, and vase mode.

For example, section two deals with tolerance and finish. So, rule R2.8 says, “Do not use circular holes for interference fits. Use hexagon or square holes instead.”

We also appreciate that [Rahix] touched on some of the counter-intuitive aspects of designing for FDM printing. For example, you might think adding voids in your part will reduce the filament and time required to print it, but in many cases it can have the opposite effect.

Some of these — maybe even most of these — won’t surprise you, but you still might take away a tidbit or two. But having it all down in a checklist and then the ability to scroll up and find the rationale for the rule is great.

Do you have any rules you’d add? Or change? Let us know. Meanwhile, we were eyeing our favorites about adding machine elements to prints.

The Nuclear War You Didn’t Notice

May 9, 2025 by Al Williams 7 Comments

We always enjoy [The History Guy], and we wish he’d do more history of science and technology. But when he does, he always delivers! His latest video, which you can see below, focuses on the Cold War pursuit of creating transfermium elements. That is, the discovery of elements that appear above fermium using advanced techniques like cyclotrons.

There was a brief history of scientists producing unnatural elements. The two leaders in this work were a Soviet lab, the Joint Institute of Nuclear Research, and a US lab at Berkeley.

Continue reading “The Nuclear War You Didn’t Notice” →

Inside A Selective Voltmeter

May 9, 2025 by Al Williams No comments

[Martin Lorton] has a vintage Harmon 4200B selective voltmeter that needed repair. He picked it up on eBay, and he knew it wasn’t working, but it was in good condition, especially for the price. He’s posted four videos about what’s inside and how he’s fixing it. You can see the first installment below.

The 4200B is an RMS voltmeter and is selective because it has a tuned circuit to adjust to a particular frequency. The unit uses discrete components and has an analog meter along with an LCD counter.

Continue reading “Inside A Selective Voltmeter” →

Oscilloscope Digital Storage, 1990s Style

May 9, 2025 by Al Williams 13 Comments

You’re designing an oscilloscope with modest storage — only 15,000 samples per channel. However, the sample rate is at 5 Gs/s, and you have to store all four channels at that speed and depth. While there is a bit of a challenge implied, this is quite doable using today’s technology. But what about in the 1990s when the Tektronix TDS 684B appeared on the market? [Tom Verbure] wondered how it was able to do such a thing. He found out, and since he wrote it up, now you can find out, too.

Inside the scope, there are two PCBs. There’s a CPU board, of course. But there’s not enough memory there to account for the scope’s capability. That much high-speed memory would have been tough in those days, anyway. The memory is actually on the analog board along with the inputs and digitizers. That should be a clue.

Continue reading “Oscilloscope Digital Storage, 1990s Style” →

Scan Your Caliper For Physical Part Copies

May 9, 2025 by Al Williams 60 Comments

We’ve certainly seen people take a photo of a part, bring it into CAD, and then scale it until some dimension on the screen is the same as a known dimension of the part. We like what [Scale Addition] shows in the video below. In addition to a picture of the part, he also takes a picture of a vernier caliper gripping the part. Now your scale is built into the picture, and you can edit out the caliper later.

He uses SketchUp, but this would work on any software that can import an image. Given the image with the correct scale, it is usually trivial to sketch over the image or even use an automatic tracing function. You still need some measurements, of course. The part in question has a vertical portion that doesn’t show up in a flat photograph. We’ve had good luck using a flatbed scanner before, and there’s no reason you couldn’t scan a part with a caliper for scale.

This is one case where a digital caliper probably isn’t as handy as an old-school one. But it would be possible to do the same trick with any measurement device. You could even take your picture on a grid of known dimensions. This would also allow you to check that the distances at the top and bottom are the same as the distances on the right and left.

Of course, you can get 3D scanners, but they have their own challenges.

Continue reading “Scan Your Caliper For Physical Part Copies” →

Understanding Linear Regression

May 8, 2025 by Al Williams 7 Comments

Although [Vitor Fróis] is explaining linear regression because it relates to machine learning, the post and, indeed, the topic have wide applications in many things that we do with electronics and computers. It is one way to use independent variables to predict dependent variables, and, in its simplest form, it is based on nothing more than a straight line.

You might remember from school that a straight line can be described by: y=mx+b. Here, m is the slope of the line and b is the y-intercept. Another way to think about it is that m is how fast the line goes up (or down, if m is negative), and b is where the line “starts” at x=0.

[Vitor] starts out with a great example: home prices (the dependent variable) and area (the independent variable). As you would guess, bigger houses tend to sell for more than smaller houses. But it isn’t an exact formula, because there are a lot of reasons a house might sell for more or less. If you plot it, you don’t get a nice line; you get a cloud of points that sort of group around some imaginary line.

Continue reading “Understanding Linear Regression” →