Pawan’s Reviews > Practical Electronics for Inventors, Fourth Edition > Status Update

Small parameter based learning, statistics algorithms can be trained on personal machine, and then ported to embedded targets in ONNX or C using emlearn and TFlite. LLMs (AI) come to rescue for generating other C programs (such as spiders or finding correct plaintext for a hashcode from a large list), which are noticeably quicker in low clock rate microcontrollers (compared to C++). VHDL is more familiar (to me).

Although I have tried KiCAD for gerber, LibrePCB, which I have been using for hobby projects, is not mentioned. I have used free version of Quartus Prime for Max10.

There may be more ways to achieve the same task, for instance, I prefer reed relays over BJTs to compartmentalise actuator circuit from controller circuit, especially for one-time events such as unconventional shutdown, ballast blow, spoilers.

Friendly. References to QM, Modern and Thermal Physics makes it more so. Polynomials (quadratic) and complex numbers become important. It is possible to use ICs with OpAmps instead of a signal generator for input. On large circuits, I prioritize checking the outputs (using DMM on breadboard) only around those components that are scarce. I used to write my own programs but the circuit simulators mentioned are good.