3.1 Diodes

A red LED, a 12 V pack and, on the datasheet, "2 V, 20 mA". The reflex is to give it its 2 V. We never do, and the why goes through a lead crystal and the point of a wire.

In 1874, Ferdinand Braun sets a wire on a galena crystal and sees current pass one way and hardly at all the other. Nobody knows why. Jagadish Chandra Bose files a patent for a crystal detector in 1901, Greenleaf Pickard another, in silicon, in 1906, and the crystal set of the 1920s lives on that ignorance: you walk the "cat’s whisker" over the stone until you find the spot that rectifies. It will take Schottky and Mott, in 1938, to explain that metal-on-crystal contact: sixty years of rectifiers without a theory of rectification.

The vacuum tube takes the name (Eccles coins "diode" for it in 1919) and the market. The crystal comes back through Bell Labs. In 1939, Russell Ohl is studying silicon rods for radio; one of them, cracked, separates two zones of different impurities, and in 1940, when light falls on it, a voltage appears across its ends. The P-N junction is found before it is looked for. Shockley writes its theory in 1949, holes on one side, electrons on the other, a zone empty of carriers between the two, and a current that climbs as an exponential of the applied voltage.

At the bench, the exponential says that the voltage moves little and the current a lot: ten times the current for sixty millivolts more in Shockley’s ideal diode, twice that on a real diode at the bench. Hence the numbers we keep. From 0.6 to 0.7 V for silicon; from 0.2 to 0.4 V for a Schottky, a metal-on-silicon contact; from 1.8 to 3.3 V for an LED depending on its colour, because the step the electron drops in the junction comes out as light, red with Holonyak in 1962, blue with Akasaki, Amano and Nakamura, Nobel 2014, and white is a blue under a phosphor.

Back to the LED. At 2.0 V the one on the datasheet draws 20 mA, but the datasheet also says "2.6 V max": its neighbour from the bag draws only 2 at that voltage, and the first draws 40 at 2.2 V, then more still as it warms, because its threshold drops. So we never set a diode’s voltage, we set its current, with a resistor that swallows the rest. On 12 V, a 2 V red LED and 20 mA wanted, (12 − 2) ÷ 0.020 = 500 Ω, 470 Ω in the E12 series, 21 mA, and the resistor dissipates 10 V × 21 mA = 0.21 W, so a half-watt. A pack’s charge indicator is exactly that, and the resistor is the part that gets warm.

The house’s second diode rectifies nothing; it waits. A coil, relay or motor, does not want its current to stop. When the transistor opens it, the coil produces whatever voltage it takes to keep it going, hundreds of volts, and the transistor is what takes them. The flyback diode, in parallel with the coil and blocked as long as the relay is powered, gives the current a path, and it dies out in a few milliseconds instead of blowing something. On the classroom robot’s relay, a 1N4007 is enough. In a motor controller it is a power Schottky, and its 0.5 V at 30 A still make 15 W; the heat sink is there for it.

The classic mistake: hook the "2 V" LED to a 3 V coin cell, see it light, and conclude that it runs on voltage. It lit because the coin cell could not give more. The pack will.