Years ago, when I first started puttering around with BJTs (we just called them transistors back when) after years messin’ around with vacuum tube circuits, I recall having a hard time understanding how they worked. I knew the base layer must be pretty thin, the emitter heavily doped to provide charge carriers (electrons for NPN transistors, holes for PNP); the base-collector junction needed to be reverse biased, and the base-emitter junction needed to be forward biased. I didn’t much understand anything else since I had neither training nor experience yet.
So I dreamed up an analogy. The base was like a venetian blind holding back the majority carriers in the base-collector junction. You opened the blinds by forward-biasing the base-emitter junction. The majority carriers in the emitter kinda drifted toward the partly opened blinds and then, once on the other side of the base-collector junction, were accelerated in great numbers toward the collector by the base-collector voltage. In my analogy, some carriers from the emitter were intercepted by the partially opened venetian blind slats and became the base current. I had no explanation for how the venetian blinds were “opened” by the forward bias voltage across the base-emitter junction.
For many years I assumed the base current had something to do with this, that is, the base current was the cause rather than the effect. It’s a good thing integrated circuit op-amps came along a few years later, sparing me (at least temporarily) the heavy lifting math and quantum-mechanical theory that better explained what was really going on.
I still puttered around from time-to-time with discrete transistors, using them mostly as switches since the digital bug had bitten me by then, but with op-amps I could go back to regular analog circuit design using passive components. Well, except for a short digression into using transistor-pairs as temperature-compensated logarithmic circuit elements. Back in the day, you could purchase a matched transistor-pair as an integrated circuit, manufactured on a heated, temperature-controlled substrate. This made them ideal for building log amps as well as front-ends for low-drift differential input instrumentation amplifiers.
There were companies, like Philbrick-Nexus and Burr-Brown, that sold hybrid analog function blocks, a mixture of discrete transistor chips hand-soldered to integrated circuits and discrete components to achieve an analog circuit function, all carefully hidden away inside a block of black epoxy resin with gold-plated leads sticking out the bottom. Very expensive, but they featured guaranteed performance specs and saved a lot of time compared to trying to build it yourself.
Alas (or maybe jump for joy), the analog IC manufacturers soon learned how to integrate everything on a single chip, sometimes laser-trimming components during manufacture, and selling a complete solution in a single DIP package. Analog design was forever changed.
Still, if someone is just starting out learning electronics, it is good to know that transistors exist for something else besides no-moving-parts switches. I still think the best way to learn about them is through bench prototyping and carefully maintained notebooks documenting your work. You can read all the theory you think you need, and execute innumerable SPICE models on your PC, but there is just no substitute for learning with real parts driven by real voltages and generating real currents. Well, real except when you get involved with reactances… but that’s another story.