The Zener diode

Every diode has a reverse voltage that destroys it. A Zener has one that is printed on the label, held to a few per cent, and survivable — so you can run the part backwards on purpose and keep the voltage it sits at.

Two pages back, the reverse rating was the number you stayed under. Push a 1N4007 past 1000 V backwards and the junction breaks down, the current goes from microamps to everything at once, and the part is gone.

A Zener is the same junction with the doping turned up, so that breakdown happens at 5.1 V instead of 1000 — and in a package specified to survive it. Run it backwards past its number and it conducts. Keep pushing, and it goes on sitting at that number while it does.

Here is one with a resistor in front of it and nothing else connected.

Raise the input and watch the output stop

0.00 V in · 0.00 V out

0.00 V

Out

0.00 V

Through the Zener

—

Nothing is happening yet. The Zener is fitted the wrong way round on purpose — stripe to the positive side — so as far as the circuit is concerned it is a blocked diode, and a blocked diode passes nothing.

Breakdown is only fatal when nothing limits the current

What kills an ordinary diode in reverse is not the breakdown. It is the power. The current runs away, the voltage across the part is large, and the heat lands in whichever microscopic spot broke down first — which is far too small to get rid of it.

A Zener is built so that neither of those is true. The breakdown happens evenly across the whole junction rather than at one flaw in it, and the package carries a wattage the way a resistor does: half a watt for a BZX55, one watt for the 1N4728A family, five watts for a 1N5338B in its fat DO-201 body. Stay under the wattage and the part sits in breakdown for years.

Staying under it is your job, not the diode's, and it comes down to one rule:

A Zener never goes straight across a supply. In breakdown it behaves like a very stiff voltage source, so it will pass whatever current the rest of the circuit allows — which, wired directly across 12 V, is everything the supply can give. There is always something in series. Usually a resistor, and the next page is about choosing it.

Two mechanisms, one name

Below about 5 V, the field inside a thin junction gets strong enough to tear electrons straight out of their bonds. That is quantum tunnelling, it is what Clarence Zener described in 1934, and it is the only thing in this chapter actually named after him. Parts like this get lower as they warm up — around −2 mV per °C.

Above about 6 V the junction is wider and the mechanism is avalanche instead: one carrier gets enough energy to knock another loose, and that one knocks two more. Avalanche parts get higher as they warm up.

Between them, near 5.6 V, the two effects roughly cancel and the voltage barely moves with temperature at all. That is not a coincidence in the parts bin — it is why precision reference diodes cluster around 5.6 and 6.2 V.

Everything in both camps is sold as a Zener, on a datasheet that will not tell you which you have bought.

The cliff you are allowed to stand on

1N4733A · 5.1 V · 1 W

Voltage across it−5.23 V

Current

70.8 mA

Burning

0.37 W

Rated for

1 W · 196.1 mA

70.8 mA at 5.23 V, and this is the part doing its job. Take it from 1.00 mA to 196.1 mA — 196 times the current — and the voltage across it moves about 1.0 V in total. That near-vertical line is the whole product.

What the datasheet actually promises

Six numbers, and a beginner needs four of them. These are the 1N4733A, the 5.1 V part in the 1 W family:

On the datasheet	1N4733A	What it means
V_Z at I_ZT	5.1 V at 49 mA	The voltage, and the current it was measured at
Tolerance	±5 %	Anywhere from 4.85 to 5.36 V, part to part
Z_ZT	7 Ω	Near the test current, 1 mA more moves it about 7 mV
I_ZK	1 mA	Below this it is out of breakdown and holding nothing
P_D	1 W	With the derating below
I_R	10 µA at 1 V	Leakage, well before breakdown

The first two lines are the ones that catch people. "5.1 V" is a voltage at one current with a ±5 % band around it: a real 1N4733A passing 20 mA might sit at 4.9 V, and the one next to it in the strip at 5.3 V. That is fine for a bias point and useless as a reference, and no amount of careful design around it changes the number the factory shipped.

Which way round it goes

Backwards, compared with every other diode in this chapter.

The stripe — the cathode, the bar in the symbol — goes to the positive side. Current in normal operation flows into the stripe and out of the other end, which is exactly the direction the last page said current never goes.

The symbol says so if you know to look: a Zener is drawn with the ends of the cathode bar bent, into a shape like a flattened Z. That bent bar means "this part is meant to be run in breakdown". Fit it the way you would fit a rectifier and nothing dramatic happens — it becomes an ordinary diode with a 0.7 V drop, your 5 V rail reads 4.3 V, and the fault takes an afternoon to find.

So what is it for

On its own it is a part that holds a voltage as long as you keep a few milliamps flowing through it. That gets used four ways:

A crude regulator. A resistor and a Zener make a 5 V rail out of anything from 8 to 12 V. It is the next page, and it is the wrong answer more often than people think.
A clamp. Across something that must never see more than a set voltage — a MOSFET gate, an input pin — a Zener does nothing at all until the voltage goes too high, and then it conducts. That is the one job where "it usually passes nothing" is the feature.
A fixed offset. Put one in a signal path and everything downstream sits V_Z lower, at any input voltage.
A reference. A few milliamps through a 5.6 V part, somewhere warm, is a voltage that barely moves. Every bench supply built before cheap references worked this way.

The surge protector on your USB port is the same idea, built heavier and much faster: a TVS diode is a Zener with a large junction, designed to swallow a kilowatt for a microsecond rather than a watt for a year.