Bringing XPPen Tablets to FreeBSD: Reverse Engineering a USB Protocol
9 mins
When a graphics tablet works flawlessly on one system but appears completely lifeless on another, the problem is rarely the hardware itself. This article explores how an XPPen tablet - perfectly functional and excellent hardware, yet unusable on FreeBSD - was brought to life through a pragmatic reverse engineering approach. By capturing USB traffic, reconstructing the devices initialization sequence and translating vendor-specific data into standard input events, a fully working userspace solution emerged. Rather than relying on proprietary drivers or kernel modifications, the implementation demonstrates how clean architecture and a structured workflow can bridge compatibility gaps. Along the way, it highlights not only the mechanics of USB protocol analysis and input subsystem integration, but also a broader question: why so many capable devices remain artificially limited by a lack of documentation - when making them work can, in some cases, be surprisingly straightforward.
Programmatic 3D Model Generation with the Tripo3D API
22 mins
Modern generative systems are beginning to reshape how 3D assets are created, lowering the barrier between idea and implementation. While most platforms focus on interactive web interfaces, their real potential emerges when treated as programmable components in an automated pipeline. By combining text or image based generation with structured processing, it becomes possible to create automated, scalable workflows that produce not just individual models, but entire libraries of assets. This article explores how to build such a pipeline using the Tripo3D API, focusing on task-based execution, metadata tracking, deterministic file organization, and flexible export strategies. Rather than replacing traditional CAD or artistic workflows, this approach complements them, bridging the gap between generative models and engineering processes, and turning 3D asset creation into a fully automatable system.
A Better Approach on Billing for Electricity
12 mins
Rethinking electricity billing by aligning costs with load variability rather than just total consumption would allow a better distribution of caused costs and provide an incentive to reduce the strain on the network by reducing load variations.
Using Codex with Hardware In The Loop for Microcontrollers
11 mins
What happens when an AI does not just suggest code, but actually compiles it, flashes it onto a microcontroller, and tests it against real hardware? In this article we explore a workflow where Codex is wired directly into a ModBus RS485 test bench, turning firmware development into a continuous loop of design, implementation, and validation on real silicon. Instead of the usual edit–build–flash cycle, the system behaves more like a self-driven engineer: asking architectural questions, updating documentation, running tests, and iterating until the behavior matches the specification. Using a concrete ATmega2560 + MAX485 setup, we walk through how design documents, TODO tracking, and an explicit agent contract enable this workflow—and what it feels like to debug firmware when the assistant can probe the system itself. From catching subtle protocol violations to restructuring configuration handling in real time, the result is a development process that blends structured engineering with rapid experimentation, all with hardware permanently in the loop.
LLMs Do Not Remember Facts, They Encode Patterns
18 mins
Large language models are often described as systems that store knowledge, but this picture is misleading. In reality, modern AI models do not function like databases filled with facts. Instead they learn complex patterns that describe how ideas, explanations, and symbols tend to relate to each other. When an LLM answers a question, it is not retrieving a stored entry, it is generating the most plausible continuation of a pattern learned from enormous amounts of text. This article explains how those patterns emerge inside neural networks, why LLMs sometimes produce convincing but incorrect answers, and why systems such as RAG and knowledge graphs are essential for reliable AI applications. By understanding how these models actually work, we can stop treating them like encyclopedias and start using them as what they really are: powerful reasoning engines operating on external knowledge systems.
Using MySQL as a Tool for n8n Agents - Flexible Queries without SQL Injection
8 mins
Using large language models together with automation platforms like n8n often requires giving the agent controlled access to structured data. Simply allowing an LLM to generate arbitrary SQL queries is dangerous, but overly restrictive configurations quickly make database tools useless. This article presents a practical pattern that allows flexible database queries while still enforcing strict safety boundaries. The approach combines n8ns MySQL tool node, fromAI parameters and a whitelist based SQL builder to prevent SQL injection while still allowing agents to explore a measurement database intelligently. With a few simple rules - least-privilege database users, identifier whitelists and mandatory query limits - one can safely expose structured data to an AI agent without risking ones database.
The Many Faces of Coherence in Physics (and Beyond)
22 mins
Coherence is one of those words that quietly carries very different meanings depending on context - logical consistency in philosophy, phase stability in optics, superposition in quantum mechanics, and fragile order in the face of decoherence. This article traces the concept across its historical roots and modern usage, starting from classical wave interference and moving through lasers, statistical optics, quantum superposition, coherent states, and decoherence theory. Along the way, it clarifies what physicists actually mean when they speak of temporal coherence, spatial coherence, and quantum coherence as a physical resource. Rather than collapsing these meanings into a single definition, the article shows how they are related by analogy rather than identity. In physics, coherence marks the presence of stable correlations that enable interference and control; in quantum systems, it is the delicate ingredient that makes superposition, entanglement, and quantum computation possible before the environment washes it away. By briefly contrasting this with philosophical notions of coherence as internal consistency, the article highlights why the term is so powerful - and why its meaning may differ when it is used across disciplines.
Schmitt Trigger - OpAmp based switches exhibiting hysteresis
6 mins
A Schmitt trigger is a comparator with positive feedback, yielding a well-defined hysteresis window. This makes it an essential building block whenever slow, noisy, or ambiguous signals must be converted into clean digital transitions. In this article, both the inverting and non-inverting Schmitt trigger configurations are derived step by step using an ideal operational amplifier model, making the underlying mechanism and assumptions transparent. Starting from first principles, the switching thresholds and hysteresis widths are calculated explicitly for both circuit topologies. The analysis shows how the feedback resistor ratios and supply voltage determine the threshold spacing and how the bias voltage sets the center of the hysteresis. The article concludes with a comparison of both configurations, highlighting their different input impedances and back-action characteristics, and places op-amp-based Schmitt triggers in context with dedicated comparator solutions used in practical designs.
Summation Amplifiers - Adding Signals with Operational Amplifiers
7 mins
Summation amplifiers are a fundamental building block in analog electronics, allowing multiple voltage signals to be combined into a single output using operational amplifiers and a small number of resistors. They are widely used in applications such as audio mixing, sensor fusion, and analog computation, where linear superposition of signals is required. By choosing appropriate resistor values, summation amplifiers can implement both weighted and unweighted sums and averages with well defined gain. This article explains the two main summation amplifier configurations - inverting and non-inverting - and derives their behavior step by step using Kirchhoffs laws. It highlights the practical differences between both approaches, discusses common pitfalls such as loading effects and bias currents, and provides guidance on when each topology is appropriate in real-world circuit design.
Growing Out of the 90s: Why Dynamic Web Applications Need to Be Rethought
11 mins
In the early days of the web dynamic websites promised millisecond-accurate freshness and the thrill of a page that seemed alive. But the reality has been decades of wasted resources, fragile scalability, and broad attack surfaces. This article argues that the future lies in static-first design: serving prebuilt content, embracing caching, and letting updates propagate on their own timescales rather than rebuilding pages on every request. Static does not mean outdated though. Modern stacks - from Jamstack to serverless and edge functions - combine static delivery with just enough dynamism where it truly matters: forms, search, personalization, or realtime updates via pub/sub. The result is faster, safer, more resilient websites that scale naturally, without carrying the baggage of 1990s design mistakes.
Measurements from small DIY NMR spectrometer
17 mins
To illustrate a previous theory blog article, this article walks through real nuclear magnetic resonance measurements performed on an extremely simple, home-built spectrometer based on a permanent magnet and minimal RF electronics. Starting from free induction decay and frequency tuning via beatings, it shows how pulsed NMR experiments actually look in practice, how pulse lengths are calibrated using Rabi oscillations, and why raw decay constants extracted from single transients are often misleading. The focus is not on idealized theory or high precision measurements, but on what one observes when working with a small, manually tuned setup and limited automation - and it should illustrate in an educational way how signals and signal shapes look like. Building on this, the article explores echo-based techniques such as the Hahn echo and Carr–Purcell pulse sequences to reveal the difference between reversible dephasing and irreversible loss of coherence. It shows how true transverse relaxation times (T2) can be extracted, why multi-pulse sequences yield shorter effective decay times, and how pulse imperfections shape experimental outcomes. The result is a practical, intuition-driven view of spin coherence that connects textbook concepts directly to measured signals and highlights both the power and the limits of simple NMR experiments.
Playing with ChatGPT RemoteMCP without OAuth
11 mins
ChatGPTs support for remote MCP servers finally makes it possible to connect custom tools and private services directly to the web interface. However, while the MCP protocol itself supports simple authentication methods, the ChatGPT web interface currently requires OAuth for non-public connectors - creating a significant barrier for experimentation, prototypes, and single-user setups. This short tutorial shows a pragmatic workaround: running a remote MCP server with a static shared secret passed via the URL, validated server-side using FastAPI middleware. While explicitly not suitable for production, this approach is ideal for learning MCP mechanics, testing tool design, and exploring integrations without standing up a full OAuth infrastructure.
