Thread by @TubeTimeUS

this afternoon i built a really clever radio transmitter using a circuit i found in a book. it's really quite ingenious, so let's dig into it a little bit...

first, the book. it is "Communications Projects" in the Engineer's Mini-Notebook series by @fmims! i bought this maybe 25 years ago from Radio Shack.

here are the plans. i'll dig into the details of the circuit in a bit, but let's go through the construction a bit.

i couldn't find a straw, so i just wrapped tape around this brush handle and wound the coil on that. 30 turns, then the center tap, then 15 turns *in the same direction* but wrapped back over the existing 30 turns. this is important!

here's the coil. isn't it cute?

and here is the transmitter. i built it a couple of different ways using different transistors and capacitors. in this pic the battery is hooked up backwards. oops.

behind the static you can hear it working: the periodic popping sound.

by changing out the capacitor with a smaller value you can increase the frequency to get a tone.

on the oscilloscope you can see it's not really a sine wave, it's just these little impulses that occur at that particular 400Hz rate.

zooming in on an individual impulse, this is what we see. it looks like a high frequency burst.

you can't really see it too well, but the burst is actually a bunch of really tiny pulses at 6.3MHz. it's got a lot of harmonic content in it, so this transmitter has a lot of spurs. not a good user of the radio spectrum.

ok so how does it work? i've redrawn the circuit to make it a little easier to understand. there are some parasitic components that i've added, and i'll explain...

Cpara is the parasitic capacitance between the windings. remember how the 15 turns are wound over on top of the 30 turns? this increases the parasitic capacitance.

so let's do a quick analysis of the circuit. here's the AC small signal model. i've shorted out the capacitor C1 and replaced the battery with a wire (they're a short at high frequency). the NPN transistor becomes a resistor and a voltage dependent current source.

there are two main current loops in the circuit. 1 is through L1 and driven by the transistor. 2 is driven by L2 and controls the base of the transistor, which controls loop 1...

it's sort of an infinite loop. the transistor inverts the phase by 180 degrees. L1/L2/Cpara invert the phase at their resonant frequency, so we get an oscillator! at any other frequency, the L1/L2/Cpara phase isn't 180 degrees, so it won't oscillate.

i discuss oscillation a bit in this other thread. turns out, to have an oscillator, you need a negative resistance! this was discovered by Hertha Ayrton, the first female electrical engineer.

https://twitter.com/TubeTimeUS/status/1076215701935800320

so where is the negative resistance in this circuit? it is the transistor itself! since Ic=gm*vbe, r=1/gm. and because of the phase inversion, r=-1/gm. there's the negative resistance!
(somewhat simplified)

(have you read Ms Ayrton's paper yet? you should, it's fascinating! here's the link again. ieeexplore.ieee.org/document/53091…)

to recap so far, we have an LC resonant tank (L1/L2/Cpara) and a transistor making a 180 degree phase shift, aka a negative resistor. this topology is known as the Hartley oscillator.

(the Hartley oscillator has what's known as a "dual" which is a mirror-like version of the circuit, with two capacitors and a single inductor instead of two inductors and one capacitor. this "alternate reality" version is known as the Colpitts oscillator.)

but we're not done. this circuit generates *impulses* at that RF frequency. how is that done? you won't believe this, but...

this circuit is actually TWO circuits superimposed on top of each other!!! 🤯

to figure out this mysterious "hidden" circuit, i'm going to look at the DC (low frequency, in this case) equivalent circuit model.

Cpara looks like an open circuit at low frequency, so that's gone. L1 and L2 look like short circuits* so those are gone. moving things around a bit, we get this.

*i'm making a tiny shortcut which i'll discuss later

this circuit operates in two stages. stage 1: capacitor charges up through R1.

stage 2: the transistor turns on HARD, discharges the capacitor and grounds out the whole circuit. *everything* shuts down. then, when there's nothing left to keep the transistor back on, it begins all over again.

what we have here is a relaxation oscillator! instead of a tuned LC circuit and negative resistances, we have a slowly charging capacitor and a comparator. when the voltage gets high enough, you discharge the capacitor and start over.

you don't even need a transistor to make a relaxation oscillator. you can do it with a comparator chip (if you want to get fancy) or even just a humble neon lamp, like the circuit below.

i've always been impressed by clever circuits with components that do double duty. another example is the classic reflex receiver circuit, where a single triode amplifies the radio signal as well as the demodulated audio! this circuit was invented in 1914, so it's been around!

oh yes the tiny shortcut (advanced mode!) i'm pretty sure L1/L2 are used in the relaxation oscillator. when Q1 turns on, dumping current from the battery, L1 resists instantaneous changes in current and a voltage appears across it. this limits the current.

hmm, thinking about it more, L1/L2 are so small they're not going to do a whole lot to the relaxation oscillator. the pulse width is determined more by C1, R1, and the base current in Q1. i'm sure some of you experts will weigh in 😂

but there's one more thing i forgot to discuss: i remember my grandfather showing me the circuit and mentioning that it was originally used for a circuit *inside* a pill that you could swallow in order to monitor your gut!

so i went and looked it up. the paper is from 1962: "Telemetering from within the body using a
pressure-sensitive radio pill"

the circuit is very similar, but it has some important differences. it's not a relaxation oscillator/RF oscillator combo. it's just a simple Colpitts oscillator with two coupled inductors.

(correction: it is a Clapp oscillator. it looks a lot like Colpitts which is why i got confused. same idea though)

the pill monitors pressure changes in the gut, and it records that using a diaphragm that pushes and pulls a tiny piece of ferrite in the middle of the inductor, changing the oscillator frequency.

the researchers could then pick up the signal and record the pressure changes on a chart recorder.

both designs are beautiful examples of minimalist electronic poetry, something you don't see too often nowadays with our fancy chips full of billions of transistors.

so @iMikla pointed out a Scientific American article from 1968 with a *very* similar circuit. it's just missing the resistor.
sites.apam.columbia.edu/courses/apph49…

the article has a few other variations of the circuit.

thinking about it, i realized my explanation of the relaxation oscillator isn't complete: i forgot to discuss hysteresis.

hysteresis means there are two voltage thresholds: high, which turns on the RF oscillator, and low, which turns it off.

to *start* oscillating, the base of the transistor needs a minimum voltage (H on the graph). once it's running, it slowly drains the capacitor until there's not enough voltage to sustain the oscillation (L). then the resistor charges the capacitor back up again and it repeats.

if the circuit had no RF oscillator, the transistor (which has no intrinsic hysteresis) just gets to an equilibrium and the relaxation oscillator doesn't work.

you can make a relaxation oscillator with one transistor, but you have to use avalanche breakdown mode. this exceeds the maximum rating of the transistor! when VCE goes above the avalanche breakdown voltage, the transistor conducts until it discharges C1.