It may be “FM” for the less technical. This does not necessarily mean “frequency modulated”. The US government does not allow you to use the first word, while magic is allowed.
Analog circuit design is rather quirky. You’re dealing nonlinear or very severe components. This makes it difficult to accurately model and account for the behavior without simulating.
You won’t need a lot of experience to get unexpected results in complicated analog circuits.
Circuit instability was something I personally experienced while in school. An amplifier was designed to light an LED if my hand passes over a pyroelectric detector. I was too naive to put in three voltage gain stages. We were still using only BJTs, resistors. Bad idea. This is what happens:
They will teach you about the Common emitter
However, they fail to mention that the collector must be reverse-biased to ensure it works properly. If you are using a resistor between ground & emitter to increase gain stability and input impedance in a configuration such as this, it is possible to forget that the base to emitter can also be a diode and could end up forward-biased.
If you set your input voltage too high, in the even split configuration as above, it’s Vdd/2+ Vforward), your negative feedback will change into positive feedback.
My output signal didn’t match the input signal from my pyroelectric when I had three of them lined up in a row. It looked more like this:
It ran at about 1 MHz, with the frequency switching a little bit when I put my hand over the pyroelectric. WTF?
The lab’s graduate student couldn’t understand why it was doing this. That was what I discovered later.
They can teach you about instability, but they cannot remove the black magic.
Plus, you get unintentional capacitance and inductance freaking everywhere in your circuits, and knowing how and where it’s going to come into play and affect stability, signal integrity and other performance aspects requires a lot of experience.
Image source: EDN
There’s also the stuff they sometimes don’t bother teaching you. Let’s suppose you want to put a capacitor in an integrated circuit. There are many reasons why you might want to do that. It is easiest to do this using silicon and a trace over it. It’s simple enough until you examine more of its characteristics.
chart from encyclopedia article – Citizendium
uh huh. Your capacitor will have a who-knows what capacitance.
It hasn’t been dark magic. You learn the most by experiencing things, not by reading.
You can get really into black magic when you actually make devices. The design depends on your expectations of electron and hole mobilities, as well as variability in threshold voltages. These are things you can control, but they are often not as well as you would like.
Source: The Threat of Semiconductor Variability
IEEE Spectrum
Chipmakers noticed a problem about 10 years ago: Even the most advanced manufacturing processes couldn’t produce chips with consistent properties. Two transistors made from the same silicon piece will have different electrical properties.
There’s a black magic to chip designers that they don’t understand.
Engineers often refer to analog circuit design as black magic. This is because there are many things that can happen, mostly because you have simplified your model and overlook a lot of things. The majority of the time.
The same was true about RF engineering. There are many stray fields effects that can occur, which we used to have problems with when we didn’t have the right models.