What could be simplest practically buildable oscillator circuit?
The meaning of "practically buildable" here is that it can be built without handpicked parts. For example, a reverse avalanche oscillator is very simple, but behaves differently, or doesn't work at all even with transistors from same batch.
Maybe an unijunction transistor (2N2646) oscillator is one?
Neon bulb, resistor, capacitor, voltage supply high enough to light bulb. The capacitor charges through the resistor, when the voltage across it reaches the neon bulb's breakdown voltage the bulb conducts briefly and partially discharges the capacitor, the bulb goes below it's breakdown voltage and stops conducting and the capacitor begins charging again, repeating the cycle.
This is the simplest I know. Frequency is independent of devices (CMOS with 1/3 2/3 level trigger points) and supply voltage.
simulate this circuit – Schematic created using CircuitLab
There's always the relay oscilator. If you want to make it have a specific frequency you can add resistors and capacitors, but it will oscillate just fine like this, and make quite a racket.
Simplest is a CMOS ring oscillator. You only need a CMOS IC with three (or five, or any odd number of) inverters connected in a ring.
Of course, power supply and bypass caps should be applied, and it is recommended to tie unused inputs to GND or VCC.
Note that the inverter IC used should be unbuffered, like 4069 or 74HCU04, because buffered inverters like the 74HC04 already have three CMOS inverters per single inverter gate, so in that case a buffered inverter gate could simply be connected by itself as a 3 inverter oscillator.
Another suggestion is a relay with normally closed contacts. Apply power to coil through the normally closed contacts, so the moment the contacts open, the coil loses power and contacts close again to power the coil. Buzzing will be heard. Touching of the circuit is not recommended, due to repeated high voltage back-EMF voltage spikes.
For simplicity, to make a sine wave output, you can stick a magnet to a tuning fork and hold a coil of wire next to it after striking. It'll oscillate a while; sometimes just a few seconds is all you need.
This circuit has only two parts plus a delay line which could be a piece of coax or a pattern on a PCB- it depends on an obsolete part (a tunnel diode), but it's not impractical to build as an exercise:
As of this writing, the parts are still available as NOS (not cheap, however).
More practical items are oscillator modules which require nothing other than a power supply, and are available in a wide range of frequencies and multiple technologies (quartz crystal, MEMS, perhaps others). Some can be programmed by the distributor to a specific frequency of your choice. Not only are they cheap, they are widely available and can be had in extreme low power, high accuracy/stability etc. Some also make a great helium detector.
If you want something in the MHz range, a ring oscillator is an option. The circuit below uses just one 74LS04 chip. It consists of 3 inverters connected in a ring. (There are three more inverters that are unused in this circuit, but could be used for some other purposes. The output frequency depends on the delays within the chip. However, the addition of a crystal and two small capacitors can make this a very stable crystal controlled oscillator.
simulate this circuit – Schematic created using CircuitLab
You could use a PUT or DIAC, the modern rough equivalents to the UJT and neon bulb.
PUTs themselves are not particularly modern anymore; they're only available from legacy manufacturers, by the looks of it (e.g. a 2N6028 by Central Semiconductor for $2.50 in singles). They can be synthesized using a complementary pair of BJTs, but admittedly, that may lack the elegance of "simplest". (PUTs are indeed single monolithic devices; they're multilayer semiconductors like BJTs are, but, well, one layer extra, no circuit integration or anything like that.)
That leaves DIACs, basically the same thing but having only two terminals, which switch on suddenly above a trigger voltage, then remain conducting until current falls below the holding threshold. An oscillator uses a large-value pull-up resistor (such that triggering current is crossed, but below holding current), and a parallel capacitor. The advantage over neon bulbs is better availability (neons are hardly manufactured anymore(?)) and lower operating voltage; a downside is no flash of light for visual interest (but, a series LED can be added to make up for it).
The BJT avalanche oscillator is probably the lowest-voltage two-terminal relaxation oscillator, aside from those using a transformer.
If radio frequency is allowed, then it's hard to get simpler than an RF transistor strung up by loose wires and biased -- most likely, a random arrangement of wires will give enough reactance and coupling to make it oscillate, as is detectable with an RF probe. A battery, transistor, several wires, and a resistor, will suffice. A typical (DC wired) circuit might be, collector and base wired together (with sufficient lead length), emitter to resistor to battery to collector. The resistor sets emitter current, while the B-C circuit oscillates as a Hartley or Colpitts style oscillator, more or less. Probably position the resistor by the battery, to give more lead length around C and E.
If low frequencies are preferred, and a transformer is acceptable, then the blocking oscillator is hard to beat:
Source: my site, https://www.seventransistorlabs.com/tmoranwms/Circuits_2010/Blocking_Oscillator.png
Rb is usually not needed, and Rbias and Cbb can be omitted in a pinch (though not both of Rb and Rbias!), though their presence allows the relaxation-oscillator-ness to be exaggerated. I have an example of this built as a relatively high-power "Joule Thief",
Source: my site, https://www.seventransistorlabs.com/Images/JouleThief3.jpg
which I can indeed cause to blink at some Hz by simply touching my fingers across the switch terminals (an equivalent Rbias of some megaohms). A transistor of enough gain, and the transformer's characteristics, are such that the oscillator kicks over from such minuscule bias current; the resulting pulse is only a few microseconds long, but the LED is driven at a couple amperes, enough to light quite visibly.
I nominate the https://en.wikipedia.org/wiki/Gunn_diode
You attach some metal plates (a resonator or antenna), apply power from a battery, and it's oscillating at microwave frequencies.
Since there are no constraints on frequency stability and variability, I'd like to propose a few other alternatives:
A microcontroller with an onboard oscillator, programmed to output a frequency of your choice. One component (blocking caps optional). Frequency should be relatively stable (a few percent) and widely variable. Downside is that it needs tools to setup, including a programmer.
Any high-speed high-gain amplifier will oscillate wildly if untamed. Since taming a high-speed amplifier is its own challenge, I'd say it is easier to build an oscillator than to not build one.
Most easily available components: I suggest the Relaxation oscillator with a Comparator/Opamp. Use any old amplifying element and a flexible selection of R and C to get a clean, buffered oscillation signal.
Hard to beat a neon relaxation oscillator. Bulb, resistor, capacitor, and battery.
Up to you what you think about this but the Soviet amateur circuits frequently depicted an 8 MHz oscillator working on the part of K155LA3 chip (analogue of 7400PC) without any additional elements. It may not work with newer analogs.
While a single element does not generate oscillations, three elements in a ring do.
It is not the most "clean" engineering solution but there was never written that these chips must be hand-picked for it to work.
simulate this circuit – Schematic created using CircuitLab
I would start with the simple neon relaxation oscillator described above and remove all of the unnecessary components.
The resistor and neon tube are removed, and the power supply is replaced with a constant current source. When the capacitor reaches a sufficiently high voltage, a big spark will occur somewhere in the circuit, discharging the capacity, and starting a new cycle. The wiring could be arranged to provide a spark gap in a safe place.
