Touching Transistors

After a longish break, thanks in part to my stuff being packed away in boxes I was too lazy to unpack because we’re theoretically house-hunting (that’s going slow), I’m finally back tinkering in my lab. A couple of days ago, I fulfilled a teenage dream from the 1980s: Getting an LED to flash using a 555 timer chip.

This is a weirdly anachronistic circuit to build in this brave new age of Arduinos, RPis, and what I’ve come to think of as microcontroller supremacism. But there’s something very fun about doing something with primitive, simple parts and no code (though wiring up a logic circuit is a kind of coding). Making an LED blink without an Arduino is the engineering equivalent of touching grass. Call it touching transistors.

As my younger and more knowledgeable friends tell me, doing electronics this way isn’t a particularly useful skill in today’s technological environment. It’s like using hand tools for wood-working. Borderline quixotic. The “right” way to make an LED blink in 2023 is to write a “blink” program for a microcontroller. Software ate this older style of electronics sometime in the mid 2000s. “Blink” on an Arduino is now the “Hello world” of electronics (I got past that milestone in my learning curve a couple of years ago). Apparently only a few experimental musicians making weird music synth gadgets do things in this 1980s way anymore.

Still, I was unreasonably pleased with myself at making a 555-blinker, and checking off a 35-year-old to-do item. The experience really took me back, and got me thinking about how electronics has evolved since the 80s.

***

When I was a teenager in small-town India in the 80s, I wanted to get into electronics but it was ended up being too hard. I also wanted to get into model airplane building, amateur astronomy, and DOS-PC-era programming, and was much more successful in those endeavors.

Back then, to get into electronics, you not only had to convince your parents to spend the money, you then had to find the rare shops that stocked a tolerable selection of parts for hobbyists (which for me meant waiting for the rare trip to Calcutta, the nearest big city). Scavenging parts from random broken or old gadgets wasn’t really an option because there weren’t many of those around (the few electronic gadgets people had were simple, rarely broke down, and were repaired and put back into service when they did).

Even the richer kids with a lot of support didn’t get too far. If you had a breadboard, a bit of perf board, a soldering iron, a multimeter, and a modest assortment of resistors, capacitors, and LEDs, you were the absolute envy of the other kids. I knew of only 1 kid who had that kind of comprehensive setup. The peak of achievement I ever heard of was, you guessed it, using a 555 timer chip to build a blinking-LED circuit. This era was so primitive, we didn’t even have blue LEDs (those were invented in 1989 and didn’t become common till later, and this is the reason you don’t see blue blinking lights in Star Trek: TNG). This non-blue era of electronics was like the sepia-tinted era in film photography.

If you did manage to get your hands on a few parts, you then had to figure out where to get the information needed to do anything, and this was usually a bigger, often insuperable obstacle.

There was no easy way to look up the part details, pin-out diagrams, and specs. Identifying random scavenged parts with cryptic part numbers on them was basically impossible. Local store owners could do simple repairs for devices like portable radios (which were called “transistors” back then in India, as distinct from the older tube radios most people still had till the early 80s) or tape recorders, but otherwise knew very little. Our main sources of information were: the few books in the school library (which only covered basic theory), the monthly electronics column in the one science magazine in India, and perhaps most importantly, the few older kids and adults who both knew more than you (usually not by much) and were willing to teach you. Catalogs were scarce: You could browse old copies at stores sometimes.

Not surprisingly, I learned very little of practical utility, and accomplished even less. I had a soldering iron which I think I used once, and only half-finished the only project I actually tried to do: Building a variable voltage power supply (a “battery eliminator”). The kind with a simple transformer, full-wave rectifier, capacitor, and a knob to select a voltage across a voltage divider.

What finally killed my interest was reading one of the monthly columns which featured a circuit using a “Zener diode” and “Darlington pairs.” I couldn’t find the terms in the one beginner book I owned, and nobody around could tell me what they were. At that point I gave up. Electronics is an information intensive field, closer to stamp-collecting than physics. You can’t just figure stuff out intuitively the way you often can with mechanical gadgets. You need specs and datasheets and a bit of a physics explanation for every single step. (If you’re curious, a Zener diode is one that can conduct some current in reverse under a sufficiently high reverse bias, and a Darlington pair is just a pair of regular transistors chained to boost a signal more than a single transistor can).

It wasn’t till I got to engineering college that I got to do a few basic things in a lab course (control a stepper motor, write a bit of machine code for an 8085, that sort of thing). That was an advanced, but narrow experience. You followed instructions from the lab manual, filed your lab report, and you were done. You picked up just enough to do the assignment. You didn’t gain anything like literacy in the parts you touched. I don’t feel like I even picked up much of a vocabulary, let alone literacy, or god forbid, actual fluency.

When I landed in the US in 1997 with little to no lab-grade electronics skills, I was suddenly dumped into a research-university-grade hardware lab featuring an embarrassment of riches which which I almost entirely lacked the literacy to use. There were shelves of parts and manuals, oscilloscopes and data acquisition rigs, a fancy soldering station, heat-shrink tubing (which I didn’t know was a thing). Rigs featuring advanced motors with optical encoders. Digikey and McMaster-Carr paper catalogs with parts you could just order on the phone on the lab budget!

I spent a few months in that lab helping out on a couple of projects and learning some basic skills, but then decided experimental work was too hard, and my illiteracy too deep, and quit. I got dumped off the deep end anyway though. I got assigned to TA a basic general lab course for 3 semesters. Now I’m glad that happened. I learned the basics of using oscilloscopes, data acquisition computers, wind tunnels, and an assortment of other random lab skills. I didn’t learn much electronics, since it was an aerospace engineering lab, but enough that I was no longer embarrassingly illiterate.

So I can’t say I ever got truly literate in electronics skills. I had better theoretical knowledge than typical mech/aero engineers (having self-studied basic texts on circuit analysis and digital logic, which are usually only covered in rudimentary ways in non-EECS engineering curricula) but couldn’t really read or speak the language of circuits, schematics, parts, and pin-outs. Around 2020, when I started getting into Maker stuff seriously, I’d say my electronics literacy was at about the same level as my Spanish. A few words here and there.

***

Now, in 2023, I’m into my fourth run at learning electronics, and this time, I have an insane amount of support and resources compared to my first run in the 1980s. And unsurprisingly, I’m finding much more success. I’m getting past theory to the beginnings of a rudimentary practical literacy. My literacy has jumped from the level of my Spanish to the level of my Tamil (in which I can understand about 30-50% of stuff I hear, and manage some broken pidgin sentences, enough to get around). I’m not going to be composing electronics poetry anytime soon, but I’m not totally lost.

Not only do I have a well-equipped personal lab, I have knowledgeable online peers I hang with, all the pecs and data sheets I could ever need at my fingertips online, and perhaps most importantly, ChatGPT. I’ve mentioned in passing in my last couple of newsletters how I’m using ChatGPT as a basic Jarvis-style tutor and lab-assistant. It truly is a game-changer. I’m considering building myself a proper gptJarvis using GPT. Compared to 2020, when I worked through a bunch of Arduino-centric experiments using an Elegoo kit that came with a manual of experiments, I feel I’m learning much more with my current self-improvised approach, mostly using random parts I’ve bought off Amazon or Adafruit. It’s much harder to work free form than with a curated kit, but whatever you learn, you learn in a deeper way. And having an AI assistant helps a lot.

Right now, I’ve set aside my more ambitious experiments with rovers for a while, to do a bit of touching transistors. Besides my 555-timer blinker, I’ve been building a variety of simple logic circuits (I actually started with analog circuits, but those are harder and require doing some math, so I switched tracks to digital; I’ll get back to analog in a bit).

You’ve probably heard something along the lines of “all computers can be reduced to NAND gates.” Well, here’s what the second simplest electronic NAND gate looks like. There are billions of these things in a modern processor, but this is a very simple chip with just 4 NAND gates, of which I’m using one. It’s the second simplest because I’m using an IC. You can do this with individual transistors too. It would just be more messy wiring.

And here’s a 3-8 demux (it takes an octal number from 000 to 111 and makes one of 8 normally high output pins low; in this picture the one attached to the second LED). This is a slightly less anachronistic thing to be playing with since I see contemporary schematics using such parts in concert with microcontroller-centric designs.

I have a list of other simple circuits I want to mess around with at this level of abstraction: flip-flops, latches, shift registers and so on. Even though digital electronics is a field where the map is very close to the territory (you don’t learn a whole lot more building a simple digital logic circuit on a breadboard than just reading about how it works in a book), there are still enough subtleties to make it interesting. The “grammar” of the physical parts has just enough atoms-related messiness that it makes the literacy more robust. You also learn some explicit things that are generally elided in theoretical treatments, like the need for pull-up/pull-down resistors, leakage currents, ground loops, capacitive coupling between adjacent pins of a chip, and so on.

I’m playing with individual transistors a bit, but playing with logic gates is somewhat more manageable on a breadboard (the complexity of breadboarding rises quadratically with the number of connections you can make, and make wrong). If touching transistors is like touching grass, touching logic gates is like touching asphalt in a parking lot.

I do intend to eventually tackle the two mythical beasts that derailed me as a teenager: Zener diodes and Darlington pairs. After that, I plan to do a little tour of duty through op-amp land, and cross “use an inductor” off my electronics bingo card. Inductors are an oddly specialized bit of vocabulary in basic electronics that don’t seem to enter the picture until you get into RF circuits and signal processing. You deal a bit with unwanted induction effects from parts like motors and solenoids, mostly by throwing in a diode somewhere to stop them, but they’re like a mysterious part of speech off to one side.

And speaking of signal processing, one of the big surprises for me in rigging out my lab was discovering that you can now buy cheap pocket oscilloscopes. This one was around $100 a couple of years ago. It’s a software-eaten oscilloscope and not quite as good as the big bench models I used decades ago (which can cost hundreds to thousands of dollars), but it’s more than enough for the primitive messing around I’m doing.

After that, my main serious learning objective is power electronics, since my next rover-related project is to power it with a solar panel. Power electronics is an interesting mix of mechanical and electrical engineering, and requires some rudimentary fluency in both analog and digital circuitry. Somewhere along the way, I hope to make a PCB of my own, probably using this fancy AI-powered CAD tool called flux.ai that I’ve started playing around with. Once I get there, I’ll have caught up to 2020s tech.

This chapter in my lab experiments is part 80s nostalgia for simpler times before Arduinos, and part remedial education. I’m also waiting on my more knowledgeable friends from my rover group to figure out what the hell is wrong with the Beaglebone Blue single-board computer I’m using to run my rover, so while they do that, I’m indulging a primitivist instinct here.

But there’s also I think a more rational thing going on here. In technology, ontogeny recapitulates phylogeny, and I’m finding that my literacy around contemporary technology is improving thanks to this dabbling in nearly obsolete technologies. For example, I learned why making an LED blink with an Arduino is generally a better strategy than using a 555 chip along with a pair of resistors and a capacitor. Controlling timing with a microcontroller is trivial, and the logic is similar whether you’re doing milliseconds or hours. With hybrid analog and digital hardware though, doing very long-range timing involves all sorts of interesting failures. So microcontrollers are probably the right level of abstraction to work with timing problems. Lower level abstractions leak too much.

This sort of thing is very useful to know, and helps develop technical taste. When I do return to the 2020s after my time-travel vacation to the 1980s, I have a feeling I’ll approach my design and experimentation projects in a more literate and tasteful way.

Maybe I’ll even get to enough fluency in a year or two to design some sort of Evil Product to sell.

Series Navigation

<< The Ribbonfarm LabCan Robots Whittle? >>