A different kind of oscillator
The last three guides built oscillators that worship a single frequency: the Wien and phase-shift sine makers, the LC tanks, the crystal. Each one carefully arranges the Barkhausen loop so that exactly one tone has a loop gain of one and the right phase, and everything else dies. The 555 timer throws that elegance away on purpose. It does not make a sine; it makes a square — a brisk on, then off, then on — and it does not care about phase purity at all.
It belongs to a family called the relaxation oscillator. The idea is wonderfully physical: charge a capacitor toward a ceiling, and the instant it reaches that ceiling, snap a switch and let it dump back down to a floor; the instant it hits the floor, snap again and start charging. The capacitor's voltage saws up and down between two levels, and that endless filling-and-emptying is the oscillation. There is no resonance, no tuned tank — just an RC time constant and a circuit with a deadband that remembers which way it is going. If that last phrase reminds you of a Schmitt trigger, hold that thought: the 555 is essentially a Schmitt-trigger oscillator hiding inside a tidy package.
Inside the eight-pin box
Crack the lid and the magic dissolves into four plain parts you already understand. First, three equal resistors (5 k each, which is where the "555" name comes from) stacked across the supply as a voltage divider. They tap off two reference voltages: 2/3 of Vcc and 1/3 of Vcc — the ceiling and the floor. Second, two comparators: an upper one watching whether the capacitor has reached 2/3 Vcc, and a lower one watching whether it has fallen to 1/3 Vcc.
Third, an SR flip-flop — a one-bit memory. The upper comparator's "we hit the ceiling" resets it; the lower comparator's "we hit the floor" sets it. This memory is exactly the hysteresis of a Schmitt trigger made explicit: the two comparators with their two different thresholds give the deadband, and the flip-flop remembers which side you are on so the circuit does not chatter. Fourth, a discharge transistor (an open switch to ground on pin 7) and an output stage (pin 3) that both follow the flip-flop's state.
Astable: the chip that won't sit still
Wire it as an astable and the 555 free-runs forever, with no input needed. You add two external resistors, Ra and Rb, and one capacitor C. The capacitor charges up through both Ra and Rb (the slow filling), and discharges down through only Rb into the discharge pin (the dumping). Tie the threshold and trigger pins together to the top of C so the chip watches its own capacitor. Now the cap saws between 1/3 and 2/3 Vcc, the output (pin 3) flips high while charging and low while discharging, and you have a square-wave generator built from two resistors and one cap.
555 ASTABLE (free-running square wave)
+Vcc
|
[Ra]
|
+-------------------(7) DISCHARGE
|
[Rb]
|
+---(6) THRESHOLD --+-- (2) TRIGGER (pins 6 & 2 tied)
|
=== C
|
GND (3) OUT --> _|''|__|''|__|''|_
charge : Vcc -> Ra -> Rb -> C => OUT high, t_high
discharge: C -> Rb -> pin7 -> GND => OUT low, t_low
C ramps forever between 1/3 Vcc and 2/3 Vcc
t_high = 0.693 x (Ra + Rb) x C
t_low = 0.693 x Rb x C
f = 1.44 / ((Ra + 2 x Rb) x C)
duty = (Ra + Rb) / (Ra + 2 x Rb) (always > 50%)Where do those numbers come from? The 0.693 is just ln(2): charging from 1/3 toward Vcc and stopping at 2/3 covers exactly one factor-of-two slice of the gap, and an RC ramp crosses that in 0.693 time constants. Try real values: Ra = 1 k, Rb = 10 k, C = 100 nF. Then f = 1.44 / ((1000 + 20000) times 100 times 10^-9) = 1.44 / (21000 times 10^-7) = 1.44 / 2.1 times 10^-3, about 686 Hz. The duty cycle is (1 + 10) / (1 + 20) = 11/21, roughly 52 percent — slightly lopsided toward on, and you cannot fix that just by picking nicer resistors.
Monostable: one clean pulse on demand
Rewire the same chip slightly and you get a monostable, also called a one-shot. It has one stable resting state — output low — and it just sits there. Pull the trigger pin briefly below 1/3 Vcc and the lower comparator sets the flip-flop: the output snaps high, the discharge switch opens, and the external capacitor (now charged by a single resistor R from Vcc) begins to ramp up. The instant the cap reaches 2/3 Vcc, the upper comparator resets the flip-flop, the output drops back low, and the cap is dumped to ground, ready for the next trigger. One push in, one fixed-width pulse out.
The pulse width is t = 1.1 times R times C. The 1.1 is ln(3): the cap starts from zero and must climb to 2/3 Vcc, and an RC ramp reaches two-thirds of the way to its target in ln(3) ≈ 1.0986 time constants. Pick R = 100 k and C = 10 uF and you get t = 1.1 times 100000 times 10 times 10^-6 = 1.1 seconds — a tidy, repeatable delay from one resistor and one cap. Want a 1 ms pulse instead? Keep R = 100 k and drop C to 10 nF.
Duty cycle, PWM, and a thousand jobs
The duty cycle is just the fraction of each period the output spends high: duty = t_high / (t_high + t_low). Because the basic astable charges through Ra+Rb but discharges through Rb alone, t_high is always longer than t_low, so the duty is always above 50 percent. To break past that, builders add a diode across Rb: now the charge current detours through the diode and Ra only, while discharge still goes through Rb, so the two halves become independent and you can dial in 50 percent or even less. It is a small, honest hack around a real limitation of the topology.
Steer the duty cycle with a control voltage instead of a fixed resistor and you have pulse-width modulation — the workhorse trick (covered in depth in guide 5) for dimming an LED, controlling a motor, or making a cheap class-D amplifier. The 555's control pin (pin 5) reaches right into the divider and moves the 2/3 Vcc threshold up or down, which slides the timing — a back door that turns the timer into a crude voltage-controlled oscillator. The 555 will not win prizes for PWM precision, but for a hobby motor speed or a flashing light it is perfect.
- Pick the mode: free-running square wave (astable) or one pulse per trigger (monostable).
- Choose C first — a value you have in the drawer, ideally a stable film or ceramic, not a sloppy electrolytic if you can help it.
- Solve for the resistor(s): astable f = 1.44 / ((Ra + 2 x Rb) x C); monostable t = 1.1 x R x C. Keep R above about 1 k so the discharge transistor is not overstressed, and below a few megohm so leakage does not dominate.
- Add the decoupling cap and a 10 nF on pin 5, then verify on a scope — the real frequency will drift a few percent from the math, and that is expected.
Where the 555 bites: honest limits
The classic 555 has a vicious habit at every switching edge. Its output stage briefly turns both transistors on at once, drawing a current spike of tens to hundreds of milliamps straight from the supply for a few hundred nanoseconds. That spike sags the supply rail and can re-trigger the chip's own comparators — the infamous "my 555 resets randomly" bug. The fix is mandatory, not optional: a decoupling capacitor of about 0.1 uF right across the chip's power pins, plus a 10 nF from the control pin to ground to steady the reference.
Be honest about precision, too. The 555's accuracy is set almost entirely by your external R and C, not by the chip — and a cheap electrolytic capacitor can be 20 percent off its marked value and drift with temperature. So timing tolerances of 5 to 10 percent are normal, and the edges carry jitter because the comparators are not infinitely fast or noise-free. If you need a stopwatch-grade interval or a stable clock for digital logic, the 555 is the wrong tool; reach instead for a crystal oscillator, whose quartz tolerance is parts per million. The 555 is convenient, not precise — and knowing that difference is the whole point.
Two more practical notes. There is a CMOS cousin, the 7555, that drives the supply far more gently (almost no shoot-through spike), runs down to 2 V, and draws microamps instead of milliamps — but it sources and sinks less current at the output, so the bipolar original still wins when you need to drive a relay or a small speaker directly. And remember the 555 is fundamentally an audio-and-slower part: above roughly a few hundred kilohertz the internal delays smear the edges and the math falls apart. For anything fast, clean, or truly stable, this is the wrong end of the toolbox — but for blinkers, beepers, delays, and PWM down where life is slow and forgiving, almost nothing beats it.