LEDs, photodiodes and special-purpose diodes

A pn junction can do more than rectify. When electrons and holes recombine across the band gap, some materials give off the energy as light, and when light is absorbed near a junction, it creates electron–hole pairs that become a current. These two effects, running in opposite directions, give us light-emitting diodes, laser diodes, photodiodes and solar cells. Varying the junction's construction gives still other devices: Schottky diodes for fast, low-loss rectification and varactors whose capacitance is tuned by voltage. This lesson surveys the special-purpose diodes an electrical engineer meets most often and shows how to design with them.

Light is central to the pedestrian controller. The walk and don't-walk symbols in the signal heads are arrays of LEDs, the cabinet's push-button inputs are isolated through optocouplers, and an ambient-light sensor lets the controller dim the signals at night, as many agencies require to reduce glare. Amara is responsible for the LED drive and the optical inputs.

How an LED makes light

In silicon, recombination releases energy mostly as heat, because silicon has an indirect band gap: an electron at the bottom of the conduction band cannot drop straight into a hole at the top of the valence band without also giving momentum to the crystal. Compound semiconductors such as gallium arsenide, gallium phosphide and gallium nitride have direct band gaps, where that transition happens readily and releases a photon. The photon's energy is close to the band gap, which sets its wavelength:

A band gap of 1.9 eV gives red light near 650 nm; about 2.1 eV gives amber and yellow; about 2.6 to 2.8 eV, from indium gallium nitride, gives blue. White LEDs are usually blue LEDs coated with a phosphor that converts part of the blue into yellow, which the eye sees as white. Because the photon energy is about the band gap energy, the forward voltage is too: red LEDs drop roughly 1.8 to 2.2 V, amber about 2.0 to 2.4 V, and blue, green and white about 2.7 to 3.4 V at their rated current.

Pedestrian signals in North America use a lunar white or white walking person and a Portland orange hand, colours set out in the traffic signal standards adopted by each jurisdiction. The orange hand uses amber-orange AlGaInP LEDs, and the walking person uses white InGaN LEDs, so the two symbols need different drive voltages.

Driving an LED

An LED is a diode, so its current rises exponentially with voltage. Applying a fixed voltage directly is a recipe for burnt parts: a small change in temperature or a part-to-part spread in V_F would cause a large change in current. LEDs must be driven by a current, set either by a series resistor from a voltage supply or by an active current source. The light output is roughly proportional to current over the useful range, so the current is what the designer chooses.

Here N LEDs are in series and I_F is the desired current. Putting several LEDs in series uses the supply voltage efficiently, because less is wasted in the resistor, but leaves less headroom for supply variation. A good rule is to keep at least 15 to 25 percent of the supply across the resistor, so the current stays reasonably steady as the supply and V_F vary. Strings are paralleled to make larger arrays, each with its own resistor; parallel LEDs sharing one resistor will not share current evenly, because their forward voltages differ.

Heat is the main enemy of LEDs. Light output falls and V_F drops as the junction warms, and life shortens sharply above the rated junction temperature. LED datasheets give derating curves for current versus ambient temperature, and signal heads in direct summer sun can get very hot. Traffic signal modules are therefore usually designed with constant-current drivers, temperature compensation and careful thermal paths.

Photodiodes

When a photon with energy above the band gap is absorbed in or near the depletion region of a junction, it creates an electron–hole pair, and the junction's field sweeps the electron to the n side and the hole to the p side. The result is a photocurrent I_ph flowing in the reverse direction, proportional to the light power. The ratio is the responsivity, R_λ, in amperes per watt; for silicon photodiodes it peaks at about 0.5 to 0.6 A/W near 850 to 950 nm. Silicon responds from about 400 to 1,100 nm, beyond which photons have too little energy to cross the 1.12 eV gap.

A photodiode can be used in two modes. In photoconductive mode it is reverse biased, so the depletion region is wide and its capacitance small, which makes it fast; the current is I_ph plus a small dark current. In photovoltaic mode it has no bias, which eliminates dark current and gives the lowest noise, but it is slower. Either way, the photocurrent is usually tiny, nanoamps to microamps, and is converted to a voltage by a transimpedance amplifier: an op amp from ELE-202 with the photodiode at its inverting input and a feedback resistor R_f, giving v_O = I_ph R_f.

Solar cells

A solar cell is a large-area photodiode used in photovoltaic mode to deliver power. With no load, light raises its voltage to the open-circuit voltage V_OC, about 0.6 to 0.7 V for silicon; short-circuited, it delivers the short-circuit current I_SC, proportional to sunlight. Power is the product of current and voltage, which is zero at both ends of the curve and has a maximum in between, the maximum power point. Practical panels connect dozens of cells in series, and solar chargers track the maximum power point electronically. Some remote pedestrian signals and school-zone flashers run on solar panels and batteries for this reason.

import math
VT, n = 0.02585, 1.3
Isc, Is = 3.0, 1.0e-8        # one cell in full sun: 3 A short-circuit current
best = (0, 0, 0)
for k in range(0, 701):
    v = k / 1000
    i = Isc - Is * (math.exp(v / (n * VT)) - 1)
    if i < 0: break
    p = v * i
    if p > best[0]: best = (p, v, i)
Voc = n * VT * math.log(Isc / Is + 1)
p, vm, im = best
print(f"Voc = {Voc:.3f} V, Isc = {Isc:.1f} A")
print(f"maximum power {p:.2f} W at {vm:.3f} V and {im:.2f} A")
print(f"fill factor = {p / (Voc * Isc):.2f}")
print(f"36 cells in series: about {36 * vm:.1f} V at the maximum power point")

Voc = 0.656 V, Isc = 3.0 A
maximum power 1.58 W at 0.559 V and 2.83 A
fill factor = 0.80
36 cells in series: about 20.1 V at the maximum power point

Optocouplers

An optocoupler, or opto-isolator, puts an infrared LED and a phototransistor or photodiode in one package, separated by a transparent insulating barrier. Current in the LED produces light that turns on the output, but there is no electrical connection, so the input and output can be at very different voltages, typically isolated to several kilovolts. The key figure is the current transfer ratio, CTR = I_C / I_F, which for a phototransistor coupler ranges widely, perhaps 50 to 600 percent between parts, and falls as the LED ages. Designs must work at the lowest CTR the part will ever have.

At Cedarline, the push buttons at the street corner connect to the cabinet over long wires that can pick up surges and ground differences. Each button input drives an optocoupler LED through a resistor from the cabinet's 24 V field supply, and the phototransistor pulls a logic input low. If a surge reaches the button wiring, the isolation barrier keeps it out of the logic.

Other special-purpose diodes

Schottky diodes, formed by a metal on lightly doped n-type silicon, conduct by majority carriers, so they have forward voltages of about 0.2 to 0.5 V and almost no reverse recovery. They are the standard choice in low-voltage power supplies and fast clamps, at the cost of higher leakage that rises steeply with temperature. Silicon carbide Schottky diodes extend the idea to hundreds of volts in high-efficiency converters. Varactor diodes are reverse-biased junctions designed for a large change of capacitance with voltage, used to tune oscillators and filters electronically. Laser diodes add an optical cavity to an LED so that light builds up by stimulated emission into a narrow, coherent beam, used in fibre-optic links, barcode scanners and distance sensors. Each of these is still a pn or metal–semiconductor junction, understood with the same physics as lesson 1.