Before transistors, engineers built oscillators, timers, and even early computer memory using a component that's essentially a glass bulb with two electrodes and a bit of neon gas. The NE-2 neon lamp is a negative-resistance device: below its firing voltage (~65-90V), it's an insulator drawing nanoamps. Once it fires, the gas ionizes and voltage collapses to a maintaining voltage (~55-65V) while conducting milliamps. Below the maintaining voltage, it extinguishes. This hysteresis is the entire oscillator.

The circuit: connect a resistor R from a DC supply (typically 90-150V) to one leg of a neon lamp. Connect a capacitor C from that same node to ground. The other lamp leg goes to ground. That's it — three parts.

How it oscillates:

• C charges through R toward the supply voltage with time constant τ = RC
• When V_C hits the firing voltage V_f, the lamp ionizes and dumps C through the (now low-impedance) lamp
• When V_C drops to the maintaining voltage V_m, the lamp extinguishes
• Cycle repeats

Design rule of thumb: the period is approximately

T ≈ RC · ln[(V_supply − V_m) / (V_supply − V_f)]

Example: V_supply = 120V, V_f = 80V, V_m = 60V, R = 1MΩ, C = 1µF. Then T ≈ 10⁶ · 10^-6 · ln(60/40) = 1 · 0.405 = 0.405 seconds, giving about 2.5 Hz — a slow blink perfect for a "power on" indicator or vintage-style flasher.

Critical constraint: R must be large enough that the supply cannot sustain the maintaining current on its own. If R is too small, the lamp fires once and stays lit as a DC glow discharge. Rule of thumb: pick R so that V_supply/R is well below the lamp's maintaining current (~0.5mA for NE-2). At 120V with 1MΩ, that's 120µA — safely below sustain, so it always extinguishes.

Real-world example: the Atari Punk Console and countless mid-century "electric eye" novelty circuits used exactly this topology. More seriously, neon relaxation oscillators drove early stroboscopes and were the timebase for the first oscilloscopes in the 1930s. They also serve as impromptu high-voltage indicators — put one across a bleeder resistor on a tube amp's B+ rail and you get a free "capacitor still charged, don't touch" warning that consumes no power once the amp is off.

Modern relevance: the physics reappears in UJT (unijunction transistor) oscillators, DIAC-triggered TRIAC dimmers, and any breakdown-based negative-resistance device — the topology is identical, only the trigger element changes.