Fiber Optic Essentials

76 Fiber Optic Essentials

The shot noise in an APD is that of a PIN diode multiplied by M times

an excess noise factor, denoted by the square root of F, where

FM = M + 1 −2 −

1

M

 (3.22)

In this case,  is the ratio of the ionization coefficient of the holes divided

by the ionization coefficient of the electrons. In III-V semiconductors

F = M. It is also important to remember that the dark current of APDs

is also multiplied, according to the same equations as the shot noise.

Because the accelerating forces must be strong enough to impart ener-

gies to the carriers, high bias voltages (several hundred volts in many

cases) are required to create the high-field region. At lower voltages,

the APD operates like a PIN diode and exhibits no internal gain. The

avalanche breakdown voltage of an APD is the voltage at which colli-

sion ionization begins. An APD biased above the breakdown point will

produce current in the absence of optical power. The voltage itself is

sufficient to create carriers and cause collision ionization. The APD is

often biased just below the breakdown point, so any optical power will

create a fast response and strong output. The tradeoffs are that dark cur-

rent (the current resulting from generation of electron-hole pairs even in

the absence of absorbed photons) increases with bias voltage, and a high-

voltage power supply is needed. Additionally, as one might expect, the

avalanche breakdown process is temperature sensitive, and most APDs

will require temperature compensation in datacom applications. For these

reasons, APDs are not as commonly used in datacom applications as PIN

diodes, despite the potentially greater sensitivity of the APD. However,

they are currently being used in applications where sensitivity is very

important, such as WDM.

3.4.2 PHOTODIODE ARRAY

Photodiode arrays are beginning to find applications as detectors in par-

allel optics for supercomputers and enhanced backplane or clustering

interconnects for telecommunication products. They are also used for

spectrometers (along with CCD arrays), which makes them useful to

test fiber optic systems. Photodiode arrays are also being considered for

WDM applications.

In the previous sections (Sections 3.2 and 3.3), we have discussed the

physics guiding photodiode operation. In a photodiode array, the individ-

ual diode elements respond to incident flux by producing photocurrents,

3. Detectors and Receivers 77

which charge individual storage capacitors. InGaAs photodiode arrays

are the materials used in spectroscopic applications. Cross-talk, or sig-

nal leakage between neighboring pixels, is normally a concern in array

systems. In InGaAs photodiode arrays it is limited to nearest-neighbor

interactions. However, the erbium doped fiber amplifiers (EDFA) do

exhibit cross-talk, so there is some discussion of using different ampli-

fiers, such as trans-impedance amplifiers (TIA arrays). It is expected that

this type of detector will be able to operate at speeds up to 10 Gbit/s

(Figure 3.10) [13].

While some WDMs filter out the wavelength dependent signal, another

common way to separate the signal involves using a grating to place

different wavelengths at different physical positions, as is done in spec-

trometers. In ultra-dense WDM, as many as 100 optical channels can

be used. Photodiode arrays are an obvious detector choice for these

applications. In addition to detecting the signal, they offer performance

feedback to the tunable lasers. At this time, photodiode arrays have not

been commercially used in WDM, but their use would be similar to that

in spectrometer design.

Figure 3.10 An optical demultiplexing receiver array, where data channels are focused

onto an array of high-speed InGaAs photodiodes. The signals are then amplified using a

trans-impedance amplifier (TIA) array.