Question 5 of 6: Receiver Front-End Choice and the Erbium-Doped Fiber Amplifier
Nivaar worked solution (AI-drafted; not reviewed by a licensed engineer)
Notes on this paper
Paper format. 98-Phys-B2 Electro-Optical Engineering, National
Examination May 2015 — a three-hour closed-book examination (one 8.5×11 inch
double-sided handwritten note sheet permitted). The cover page states any five of
the six questions constitute a complete paper and only the first five as they appear
in the answer book are marked; every question is nonetheless answered in full below so the
paper remains a complete study resource. Figure-based Question 4 is solved against the actual
photodiode responsivity curve printed on the exam, not an
assumed shape.
Reference texts. G. Keiser, Optical Fiber Communications, 4th ed.
(fiber modes and dispersion, link power and risetime budgets, LED/laser and photodiode
characteristics, EDFA); B. E. A. Saleh and M. C. Teich, Fundamentals of Photonics,
2nd ed. (LED spectral width, laser diode rate equations, photodiode noise); E. Hecht,
Optics, 5th ed. (waveguiding and dispersion background).
Question 5: Receiver Front-End Choice and the Erbium-Doped Fiber Amplifier (equal value)
Find. (a) The unequalized 3 dB bandwidth of each front-end option,
which to choose, and what an equalizer buys option B. (b) EDFA operating principle,
characteristics, and its advantages/disadvantages in a WDM system.
Approach. Each front end is, without equalization, a single-pole RC
low-pass whose resistance is the total resistance the detector's current source sees:
$R_b\parallel R_{in}$ for the plain high-impedance amplifier, and the much smaller
feedback-reduced input resistance $R_f/(1+A_{ol})$ for the transimpedance amplifier.
Part (a)
Option B — high-input-impedance front end. The detector's bias
resistor and the amplifier's input resistance appear in parallel to the signal current:
$$R_{eq,B}=\frac{R_bR_{in,B}}{R_b+R_{in,B}}=\frac{(680\times10^3)(1\times10^6)}{680\times10^3+1\times10^6}=405\ \text{k}\Omega.$$
$$f_{3dB,B}=\frac{1}{2\pi R_{eq,B}C_T}=\frac{1}{2\pi(4.05\times10^5)(5\times10^{-12})}=\boxed{78.6\ \text{kHz}}.$$
Option A — transimpedance amplifier. Shunt-shunt feedback around the
open-loop gain $A_{ol}$ reduces the resistance the detector sees to approximately
$$R_{eq,A}=\frac{R_f}{1+A_{ol}}=\frac{220\times10^3}{1+500}=439\ \Omega,$$
$$f_{3dB,A}=\frac{1}{2\pi R_{eq,A}C_T}=\frac{1}{2\pi(439)(5\times10^{-12})}=\boxed{72.5\ \text{MHz}}.$$
Choice. Without any equalization, the TIA (option A) delivers roughly
900× more bandwidth (72.5 MHz vs. 78.6 kHz) because feedback divides the effective
input resistance by $(1+A_{ol})$. On noise, feedback lowers the input impedance without
adding the noise of a physical 439 Ω resistor. The TIA's thermal-noise current is set
by the feedback resistor in parallel with the bias resistor, $R_f\parallel R_b=166$ kΩ.
Option B's is set by $R_b\parallel R_{in}=405$ kΩ. The TIA's mean-square noise current
$4kTB/R$ is therefore about $405/166=2.4$ times larger (about 3.9 dB), a modest penalty that
is far outweighed by not needing post-detection equalization at all — option A is the correct
choice for any link needing more than a few hundred kHz of bandwidth. Option B's huge
$R_b\parallel R_{in}$ gives excellent noise performance (small $4kT/R_{eq}$) but only inside
its very narrow unequalized bandwidth.
Benefit of adding an equalizer after amplifier B. An equalizer with a
rising ($+20$ dB/decade) response beyond $f_{3dB,B}$ can flatten the receiver's overall
frequency response well past 78.6 kHz, recovering usable bandwidth from the high-$R$, low-noise
front end without giving up its excellent low-frequency SNR. The cost is that the equalizer
boosts noise at high frequencies right along with the signal, so the net SNR improvement is
smaller than the raw bandwidth gain suggests — a classic high-impedance-front-end/
integrate-and-differentiate trade-off, whereas the TIA achieves its bandwidth without paying
this noise-boost penalty at all.
Final results — part (a)
Quantity
Value
Option B unequalized bandwidth
78.6 kHz
Option A (TIA) unequalized bandwidth
72.5 MHz
Recommended design
Option A (transimpedance amplifier)
Part (b) — erbium-doped fiber amplifier (EDFA)
An EDFA is a length of silica fiber whose core is doped with trivalent erbium ions
(Er$^{3+}$), optically pumped (typically at 980 nm or 1480 nm, via a wavelength-division
coupler) into an excited metastable state. Signal photons near 1550 nm — matching the
Er$^{3+}$ transition from the metastable $^4I_{13/2}$ level down to the ground $^4I_{15/2}$
level — stimulate emission from the population-inverted ions as they pass through,
providing gain directly in the optical domain with no optical-electrical-optical conversion.
Key characteristics: gain typically 20–30 dB over a usable band of roughly 30–40 nm
(conventional C-band, extendable with L-band doping profiles), noise figure typically 4–6
dB (set by amplified spontaneous emission, ASE), and gain that is largely polarization- and
bit-rate-independent (a slow, population-based gain medium, unlike a semiconductor optical
amplifier). In a WDM system, the chief advantages are that a single EDFA
amplifies all wavelength channels simultaneously (no per-channel electronics), is transparent
to modulation format and bit rate, and introduces comparatively low noise and no electronic
bandwidth bottleneck. The main disadvantages are gain non-uniformity across
the amplification band (requiring gain-flattening filters in long cascades), gain saturation
and cross-gain effects when channels are added/dropped (transient power excursions on
surviving channels), accumulated ASE noise limiting cascadability, and the fact that EDFAs
only operate in the 1530–1610 nm window (no equivalent standard rare-earth amplifier for
the 850 nm or 1310 nm bands).
EDFA schematic: erbium-doped fiber pumped at 980/1480 nm via a WDM
coupler provides population inversion, amplifying the co-propagating 1550 nm signal band
directly in the optical domain.