AMAZON multi-meters discounts AMAZON oscilloscope discounts
Introduction
This section will examine the operation of the optical sources and their associated detectors used with fiber optic systems. It will also cover the practical aspects required to turn these devices into fiber optic transmitters and receivers.
Performance issues for these devices will be investigated. Future developments such as the use of optical amplifier systems will also be discussed.
1. Optical sources
Effective optical sources for fiber optic transmission need to have several important properties:
• To be able to effectively couple the small fiber core, as small as 8.5 micrometers for singlemode fibers
• Easily modulated by electrical signals to convey data, with good linearity to prevent harmonics and intermodulation distortion
• Provide high optical output power
• Have high reliability
• Small size and weight
• Low cost
Light emitting junction diodes (LED) and laser diodes (LD) fulfill many of these requirements and we will now examine their properties in detail.
2. Light emitting diodes (LED)
2.1 Construction
A simple light emitting diode is made as a junction of two semiconductor regions, each of which is doped with impurities to give appropriate electrical characteristics. The 'p' type region contains impurities that have fewer electrons than atoms in the crystal lattice and so create atoms with a net positive charge. These are known as 'holes' as they create room for electrons to move in the crystal lattice. Similarly, the 'n' type regions use impurities, which provide more electrons than atoms and effectively donate electrons so that electrons are left floating in the crystal lattice. The most important optical semiconductors are made of elements from groups IIIa and Va of the periodic table as listed in Table 1.
Table 1 Optical semiconductor materials
2.2 Basic LED operating principles
When a positive voltage is applied to the p-region and a negative voltage applied to the n region, electrons and holes flow towards the junction of the two regions where they combine. When an electron combines with a hole, the atom returns to its neutral state and energy is released, having been converted into optical energy in the form of photons. In its simplest form, the radiated energy from the LED is caused by the recombination of the electrons and holes, which are injected into the junction by the forward bias voltage.
FIG. 1 illustrates this process.
FIG. 1 Basic LED operation [ Bias Voltage n-region p-region Light Emission
Recombination Junction]
Band theory provides a simple explanation of the semiconductor emissions. Two allowed bands of energies, separated by a forbidden region called the bandgap, exist, as shown in FIG. 2. At the upper level in the n-region, known as the conduction band, the unbound electrons are free to move, while at the lower level in the p-region, known as the valence band, the unbound holes are free to move.
FIG. 2 Bandgap determines energy emission.
The size of the bandgap determines the energy of the emitted photon. Different semiconductor materials have different bandgap energies and the gap energy (W) in electron volts (eV) can be related to the wavelength (?) by the equation:
The usual LEDs applied in fiber optic systems use gallium aluminum arsenide (GaAlAs) for 800 to 900 nm wavelengths and gallium arsenide (GaAs) for 930 nm. LEDs for use with plastic fibers need to operate at about 660 nm and are produced with gallium arsenide phosphide (GaAsP) compounds. Various indium gallium arsenide phosphide (InGaAsP) compounds are used for longer wavelengths of 1300 and 1550 nm. The semiconductor types, their respective bandgap energies and wavelengths are shown in Table 2.
? = 1240 / W nanometers
Table 2 Semiconductor bandgap energies and wavelengths
2.3 LED geometry
An essential ingredient of optical sources is the ability to couple the light into the small fiber core. Basic LEDs as outlined above emit light in all directions. This makes it difficult to couple the light into the fiber. Various internal structures can be used to concentrate the light into a narrow beam. Two commonly used structures are the Burrus diode and the edge emitting diodes. The Burrus or etched well diode structure uses a hole etched in the substrate coupled with internal structures to confine the light emissions. A fiber can be directly inserted into the hole in the top of the device to collect the light output. Edge emitting diodes (ELED) generate the light in a thin, narrow active junction layer as shown in FIG. 3. The emitting zone can be as small as a few micrometers in thickness and ten micrometers in width. The structure incorporates features designed to confine the light output and guide it to one end of the device. These devices generally produce lower light output than surface LEDs because of the smaller cross-section of the active emitting area, but the beam is more efficiently coupled to the fiber. The brighter and more tightly collimated beams require a more complex construction, associated with higher cost structures and have greater heat generation.
FIG. 3 Edge emitting LED
2.4 Operating characteristics
The output power is generally not greater than 1 mW but can be as low as several microwatts. The output power of an LED is linearly related to the forward driving current. Fiber optic LEDs typically operate at currents of 20 to 100 mA and require forward bias voltages of 1.2 to 1.8 volts. With power dissipation of up to 180 mW from the small device, appropriate heat-sink design is needed to prevent excessive temperature rises and consequently reduced device reliability and lifetime. As LEDs age, their output power decreases.
Spectral widths
The total power emitted by the transmitter is distributed over a range of wavelengths spread about the center wavelength. This is quantified as the spectral width, s?, which is the 3 dB optical power width, usually measured in nm. A typical LED operating at 850 nm will have a spectral width of approximately 40 nm and an LED operating at 1300 nm, a width of approximately 80 nm. Wide spectral widths cause increased chromatic dispersion of the light pulses, as they propagate through the fiber.
Operating Lifetimes
The lifetime of an LED is the time taken for the light output to reduce to half its initial value (i.e. drop by 3 dB at its central operating wavelength). Good LEDs should have a lifetime of around 105 hours (11 years).
Modulation
Digital modulation of the LED output is simply achieved by use of a current source turning the LED on or off. Analog modulation requires LED to have a DC bias applied to ensure the LED is forward biased at all times.
Temperature effects
Operating temperature ranges of -65° to 125°C are possible. Output power decreases as the junction temperature rises, typically at the rate of 0.012 dB/°C.
2.5 Practical LED devices
The LED chips need to be mounted on appropriate packages to dissipate the heat effectively and enable the fiber to be coupled to the light source. Many different approaches are used by manufacturers including transparent windows or lenses in metal caps, holes in packages for the insertion and gluing of fibers, attachment of pig-tails direct to the chip or provision of micro lenses on the chips to collimate the beam. Two of these are illustrated in FIG. 4. The use of a large lens as the device cover is shown in FIG. 4(a). This produces a large beam because of the distance separating the lens and LED and as such, is only suitable for large diameter fibers. FIG. 4(b) shows the use of a micro lens fitted directly to the LED. The beam in this case does not enlarge before being collimated by the lens and so can be effectively coupled to 50 µm fiber core diameters. The losses involved in coupling fibers and sources are discussed in Section 5.
FIG. 4 Practical LED packages
3. Laser diodes
3.1 Basic principles of laser operation
LASER stands for light amplification by the stimulated emission of radiation. LEDs and lasers use very similar principles of operation. In section 2.2, we saw that light is emitted from a LED when an electron drops from a high energy level to a lower one.
When this occurs without outside influence, it is known as spontaneous emission. This occurs in some radioactive material. With the LED discussed in the previous section, a forward bias voltage was used to stimulate the emission. An electron sitting at the upper energy level can also be stimulated to drop to the lower level by a photon with the right amount of energy. In this way, the external photon can stimulate the emission of a second photon at the same wavelength.
Laser action takes place through optical resonance. The laser structure is very similar to an edge LED, having a thin, narrow active region with the addition of reflective end facets and reflective sides as shown in FIG. 5. In this resonator, the light is confined and reflected backward and forward through the excited medium. The laser is biased to begin the emission of photons. The photons reflect backward and forward and stimulate further emission of photons from electrons waiting to recombine. The light traveling back and forth along the axis of the resonator continues this action and builds up in strength until it is strong enough to break through the reflective end and thus, a laser beam is formed.
FIG. 5 Basic laser diode operation
3.2 Operating characteristics
Lasers need to operate at higher drive currents than LEDs so as to generate a sufficiently high density of electrons ready to recombine at the high energy level.
Output power and power consumption
The optical output power of a laser diode shows two distinct slopes as shown in FIG. 6. Below the threshold current, the device operates as an LED with low or no output.
Laser action only occurs above the threshold. The threshold currents are usually in the range of 30 to 250 mA, with forward voltages of 1.2 to 2 volts. Practical devices are usually operated at 20 to 40 mA above the threshold current and can generate optical outputs of 1 to 10 mW continuously, even more if pulsed with low duty cycles. Some lasers operate up to several hundred milliwatts optical output.
FIG. 6 Laser diode power-current-temperature curves
Spectral widths
Laser diodes have typical line widths of 1 nm at 850 nm and 3 nm at 1300 nm and 1550 nm, which is considerably less than those of LEDs. Therefore, they suffer a lot less from chromatic dispersion problems.
Operating lifetimes
Continuously operating laser diodes can have typical lifetimes of 105 hours at room temperature, but they degrade faster at higher temperatures. Commercial laser diodes typically exceed 104 hours at 70°C.
Modulation
Digital modulation of lasers makes use of the threshold current. The laser is biased to just below the threshold current to switch the beam off for logic '0' and it is rapidly switched on by increasing the current above the threshold for the logic '1'. Analog modulation uses a bias current above the threshold so that the operation remains in the linear region of the power-current (P-I) curve, shown in FIG. 6.
Temperature effects Laser diodes are much more temperature sensitive than LEDs, as illustrated in FIG. 6.
The threshold current increases at about 1.5% per °C showing that more current is needed to start laser action at higher temperatures. At constant current, the output power will drop, as the temperature rises. In addition, as the threshold current changes with temperature, this affects the required bias voltage for modulation. Temperature stabilization is therefore needed for practical transmitters as discussed in section 4.
3.3 Practical laser devices
Similar packaging requirements apply, as were discussed for LEDs and these include hermetically sealing all leads, enabling precise alignment of the fiber and laser chip, providing suitable heat-sinking or mounting the chip on a thermoelectric cooler. For active feedback control of the bias current, a photodiode (optical detector) can be mounted inside the package to monitor the power emitted from the rear facet of the laser.
Fiber pig-tails can be fitted by the manufacturer to maximize coupling efficiency. One package configuration is illustrated in FIG. 7. Here a fiber pig-tail is fitted in a grooved block in precise alignment with the laser diode. The user can fit a connector to the pig-tail or splice it direct to the incoming fiber.
FIG. 7 Laser diode package with integral pigtail.
3.4 Advances in laser technology
The more common, 'older' type lasers are called edge emitting diodes, and emit coherent infrared light parallel to the boundaries between the semiconductor layers. This is shown in section 3.1 above.
The latest technology is vertical cavity surface emitting diodes (VCSEL; pronounced 'vixel'). This is a specialized laser diode that promises to revolutionize fiber optic communications by improving efficiency and increasing data speed. Research on VCSELs began in 1993, but the first commercial devices were not released until 2000. The first devices operated in the 850 nm range. In 2002, the first 10 Gbps VSCELs were released, which operated in the 850 nm and 1300 nm range.
The VCSEL emits its coherent light perpendicular to the boundaries between the layers of semiconductors. Until very recently, the VCSELs had only operated in the 850 nm range but recent releases have operated in the 1300 nm range.
The VCSEL has several advantages over edge emitting diodes. The VCSEL is cheaper to manufacture in quantity, easier to test, more efficient and it requires less electrical current to produce a given coherent energy output. The VCSEL emits a narrow, more circular beam than edge emitting diodes. This makes it easier to couple to an optical fiber.
The next challenge is to produce a cost effective and reliable VCSEL that operates in the 1550 nm range where there is the lowest fiber attenuation.
4. Optical transmitter modules
Practical transmitter modules can incorporate a Peltier effect device that is a type of thermoelectric cooler. The laser chip is mounted on the cooler with a temperature monitoring thermistor to control the cooling process. The other approach to stabilization uses a photodiode to monitor the output power radiated at the rear facet of the laser. This is used to control the DC bias current and stabilize the output.
The laser diode in FIG. 8 incorporates a photodiode power monitor measuring the power emitted from the rear facet of the device. Such devices can be mounted in multiple pin industry standard packaging such as the dual inline package (DIP). Individual pins can then be provided for the laser, photodetector, thermoelectric cooler, and temperature sensor.
FIG. 8 Laser diode transmitter module
5. Laser safety considerations
The brightness of some laser beams and their high degree of collimation make them a potential hazard to the human eye, and suitable safety precautions must be taken when working with them. Systems operating at 1300 and 1550 nm use high power lasers for long distance communications. These wavelengths are invisible to the human eye but can cause permanent damage to the retina of the eye. Strict safety codes of practice have been developed governing the use of all laser devices. No optical source or illuminated fiber should ever be viewed by a microscope or by the naked eye.
6. Optical detectors
The function of the optical detector is to efficiently convert the small amount of light energy received from the fiber, as photons, into electrical signals. The detector needs to be a low inherent noise device, incorporating appropriate amplification to generate useful output signals from low level inputs. Two main types of devices are used for practical detectors; PIN diodes and avalanche photodiodes.
7. Pin photodiodes
7.1 Operating principles
Photodiodes convert the optical signals directly into electrical signals, using the reverse of the physical process in the LED. The PIN photodiode has a wide intrinsic semiconductor layer separating the p- and n-regions, as shown in FIG. 9. The diode is reverse biased (5-20 volts) and this helps draw the current carriers away from the intrinsic region.
FIG. 9 PIN photodiode
The width of the intrinsic layer ensures that there is a high probability of incoming photons being absorbed in it rather than in the p- or n-regions. The intrinsic layer has a high resistance because it has no free charges. This results in most of the diode voltage appearing across it, and the resultant electrical field raises the response speed and reduces noise. When light of suitable energy strikes the intrinsic layer, it creates electron-hole pairs by raising an electron from the valence band to the conduction band and leaving a hole behind in the process. The bias voltage causes these current carriers (electrons in the conduction band) to quickly drift away from the junction region, producing a current proportional to the incident light, as shown in FIG. 9.
7.2 Operating characteristics
Cutoff wavelength
The incoming photon must have enough energy to raise an electron over the bandgap to create an electron-hole pair. Different semiconductor materials have different bandgap energies and the gap energy (W) in electron volts (eV) can be related to the wavelength (?) by exactly the same equation as for LEDs:
? = 1240 / W nanometers
For a particular detector, the bandgap W is fixed so the above equation gives the longest wavelength that can be detected, i.e. the cutoff wavelength.
Responsivity
The responsivity ? is the ratio of the output current (i) of the detector to its optic input power (P).
? = i / P amperes per watt
At 800 nm silicon has a responsivity of about 0.5 A/W and InGaAs has peak responsivity of about 1.1 A/W at 1700 nm, reducing to 0.77 A/W at 1300 nm.
Spectral response
The spectral response is shown as the variation of responsivity with wavelength. Typical spectral response curves for silicon and InGaAs PIN diodes are shown in FIG. 10.
FIG. 10 PIN diode spectral responses
Quantum efficiency
The quantum efficiency h of the emitter is defined as the ratio of the number of emitted electrons to the number of incident photons. Silicon and InGaAs have peak quantum efficiencies of about 80%.
Response speed
The speed of response of the detector is limited by the transit time, which is the time it takes for free charges to cross the width of the intrinsic layer. This is a function of the reverse voltage and the physical width. For fast PIN diodes, this ranges from 0.5 to 10 ns.
Capacitance also affects the device response, with the junction capacitance formed by the insulating intrinsic layer between the electrodes formed by the p- and n-regions. High- speed photodiodes can have responses as fast as 10 picoseconds, requiring capacitances of a few picofarads, with very small surface areas.
Current-voltage charac.
The typical current-voltage (I-V) curves for a silicon PIN photodiode are shown in FIG. 11. It can be seen that even when there is no optic power, a small reverse current flows called the dark current. This is caused by the thermal generation of free charge carriers, typically doubling with each subsequent 10°C increase in temperature after 25°C.
FIG. 11 Silicon PIN photodiode I-V characteristics
Dynamic Range
The linear relationship between voltage and optic power shown in FIG. 11 is maintained typically over six decades giving a dynamic range of some 50 dB.
7.3 PIN photodiode packaging
PIN photodiode packaging is similar to those used for LEDs and lasers, but the optical requirements are less critical. The detectors' active area is usually much larger than the core of the fiber, so lateral misalignment is less of a problem.
8. Avalanche photodiodes
8.1 Operating principles
Avalanche photodiodes use semiconductor junction detectors with internal gain through avalanche current multiplication. A very high reverse bias voltage (50-300 volts) is applied to a p-n junction. A photon is absorbed in the depletion region, creating a free electron and a free hole. These charges accelerate in the strong electric field. When they collide with neutral atoms in the crystal lattice, their kinetic energy is sufficient to raise electrons across the bandgap and create additional electron-hole pairs. These secondary charges also accelerate creating more electron-hole pairs. In this way, the current produced by one photon is multiplied.
8.2 Avalanche photodiode structure
One form of avalanche diode is the reach-through diode, as illustrated in FIG. 12. The p+ and n+ layers are highly doped regions with very small voltage drops. The depletion region is lightly doped, almost intrinsic. Most of the photons are absorbed in this area, forming electron-hole pairs. The electrons move to the p-region that has been depleted of free charge by the large reverse voltage. The depletion region at the p-n+ junction effectively reaches right through the p layer. The strong electric fields across the p-layer cause avalanche multiplication of the electrons. The holes produced drift across the p layer to the p+ electrode but do not cause further multiplication. Because this structure limits the charge carrier multiplication to electrons only, it has better noise performance.
FIG. 12 Avalanche photodiode
8.3 Operating characteristics
Cutoff wavelength
Avalanche photodiodes are variations of PIN diodes so the materials, spectral ranges and cutoff wavelengths are the same as for PIN diodes.
Current-voltage characteristics
The typical current-voltage (I-V) characteristic curves for an avalanche photodiode are shown in FIG. 13.
FIG. 13 Avalanche photodiode characteristic curves
Response speed
The response speed is limited by the charge carrier transit time and the RC time constant as for PIN diodes. Transit-time-limited avalanche diodes are produced with rise times of the order of a few tenths of a nanosecond. Rise times less than 100 picosecond are achievable.
Dynamic Range
Avalanche diodes are similar to PIN diodes in having excellent linearity, over a wide dynamic power range, typically from fractions of a nanowatt to several microwatts.
Gain
The gain of an avalanche photodiode is temperature dependent, generally decreasing as temperature rises. Temperature stabilization may be required for devices operating over extended temperature ranges. Gains from 20 to 150 are typical.
8.4 Avalanche photodiode applications
The avalanche photodiode requires a stable high voltage power supply and more complex bias circuitry. This increases costs and decreases the reliability. The avalanche diodes are generally less reliable than standard PIN diodes. It follows that PIN diodes are generally the preferred devices for normal applications. Avalanche diodes provide increased sensitivities of 5 to 10 dB and half the rise time of standard PIN diodes. Such avalanche diodes are required when the system has high losses and needs to work at low signal to noise ratios, such as on long distance communications links. On such systems, the savings in providing increased repeater spacing outweigh the disadvantages.
9. Optical receiver modules
9.1 Basic elements of a ractical receiver
The basic elements of a ractical receiver are: (a) a detector to convert the received optical signal to an electrical signal (b) an amplifier to increase the electrical signal to a level where it can be processed (c) a demodulator or decision circuits to recover the original data from the input signal.
We have seen that avalanche photodiodes incorporate internal amplification, so these devices combine the functions (a) and (b).
9.2 Amplifiers
The typical optical signals reaching a fiber optic receiver can be as low as one microwatt. Using a PIN diode with a 0.6 A/W responsivity to detect such a signal would produce an output of around 600 nanoamps. This output needs to be amplified and converted to a voltage for subsequent processing. Amplifiers can be divided into four types depending upon the type of input transistor (FET or bipolar) and the amplifier configuration, irrespective of high-impedance or transimpedance. The high impedance amplifier gives high gain, but it can saturate for a relatively low input signal. The transimpedance amplifier uses negative feedback to reduce the gain as shown in FIG. 14. This gives the amplifier an increased dynamic range, wider bandwidth, but less sensitivity.
FIG. 14 Amplifier configurations
The pre-amplifier characteristics at different data rates with PIN and avalanche photodetectors are shown in FIG. 15. It can be seen that the avalanche diode has greater sensitivity at all data rates. A FET amplifier provides greater sensitivity at lower data rates, while the bipolar amplifiers have superior performance at higher data rates.
The quantum limit shown is the minimum detectable power caused by the statistical nature of the photon detection and carrier pair generation mechanisms.
FIG. 15 Pre-amplifier characteristics.
Some amplifier configurations make use of an amplifier mounted on the same chip as the photodetector. This increases the signal level without lowering the ratio of signal power to noise power. External amplifiers by contrast always add noise and reduce the signal to noise ratio.
9.3 Receiver packaging
Receiver packaging needs to address mechanical, electronic, and optical issues. The main mechanical concern is how to mount the fibers with the receiver module itself, following typical industry standard packaging such as dual inline packages (DIP). The electronic interfacing needs to address the bias voltages required by the photodetector, amplifier and the detector circuitry, as well as the gain controls and output requirements. The optical interfacing is simpler than transmitters because the detector areas are larger than the core of most fibers and butting the fiber against the detector collects sufficient light. Some modules use fiber connectors on the case to link direct to the detector, or use a large multimode fiber to collect all the light from the small core of a singlemode fiber.
10. Optical amplifiers
Optical amplifiers increase the signal strength without converting the signal into electrical form. They work on the stimulated emission principle as specialized forms of lasers, which amplify the light in a single pass through the amplifier. There are currently two basic types of optical amplifiers in use; doped fiber amplifiers and semiconductor laser amplifiers.
10.1 Doped fibers
Doped fiber amplifiers make use of fibers whose cores are doped with elements, which can amplify light at particular wavelengths. Fibers doped with the rare earth erbium are effective at wavelengths between 1520 and 1560 nm and praseodymium doped fibers operate at the 1300 nm wavelengths. The dopant atoms in the fiber are excited by a pump laser, operating at a shorter wavelength. Light at the signal wavelength can stimulate these energetic atoms to emit their excess energy, as light at the signal wavelength, in phase with the signal pulses. The operating principle of these devices is illustrated in FIG. 16.
FIG. 16 Doped fiber amplifiers
10.2 Semiconductor laser amplifiers
Semiconductor laser amplifiers operate like ordinary lasers without the mirrors. Light from the external source makes one pass through the active area and stimulates further photon emissions. One major difficulty with these devices involves the coupling of light into the laser, as shown in FIG. 17. The semiconductor laser has an active area several microns wide and less than a micron thick. The light exiting a singlemode fiber will produce a beam of at least 9 microns diameter. Clearly, most of this light from the fiber will miss the active input layer of the laser and be lost. This high transfer loss from fibers offsets most of the gain produced by the laser amplifier. However, these devices can be usefully integrated onto chips with other semiconductor optical devices, such as laser sources, to avoid such transfer losses.
FIG. 17 Semiconductor laser amplifiers