Jet Propulsion Laboratory

California Institute of Technology

Flicker noise of
high-speed p-i-n photodiodes
#% @% % %
E. Rubiola , E. Salik , N. Yu , L. Maleki
# FEMTO-ST Institute, Besançon, France
% JPL/CALTECH, Pasadena, CA, USA
@ Dept. of Physics, California State Polytechnic University, Pomona, CA, USA

Outline
• introduction
• method
• background noise
• results
Work carried out at the JPL/CALTECH
1 under NASA contract, with support from ARL and AOSP/DARPA
 p-i-n InGaAs photodiode
II − forbidden iF iF I − forward−bias
region region (not used)
light
vF

p layer vF
bias −Vb P=0

i depletion region dark

photovoltaic
cell P=−vF iF
n substrate light P=P max
photoconduction

vF =0
bright (virtual gnd)
light loa
dl
ine
short current
III − photoconductive Isc = Vb /R IV − photovoltaic
region region
! "
vF
iF = Is exp − 1 − iP kT /q ! 25.6 mV at 300 K
kT /q P
iP = ηΦ = η photocurrent
hν
= ρP ρ = responsivity

2 photoconductive region => lowest C => high speed

 Signal and noise
microwave-modulated IR Pλ (t) = P λ [1 + m cos 2πν0 t]
microwave photocurrent iac (t) = ρP λ m [1 + α(t)] cos [ω0 t + φ(t)]
with AM and PM noise

white noise Si = 2qı

Virtually no information on AM/PM flicker is available

Motivations
• frequency distribution systems
deep space network, VLBI, inter-lab link
• laser metrology
• photonic oscillators (Leeson effect)
(E. Rubiola, The Leeson effect, arXiv:physics/0502143)
3
 Experimental method (1)
• the photodiode output is insufficient to saturate a mixer
• a preliminary survey suggests that the photodiode phase
flickering is lower than that of a microwave amplifier
2
(typical amplifier flicker -105 dBrad /Hz at 1 Hz)
• we choose some photodiodes similar to one another, with a max
speed of 12-15 GHz
(Discovery Semiconductors, Fermionics, Lasertron)
• a single-photodiode interferometric (bridge) scheme can’t work
because the equilibrium condition is difficult
hybrid
−90° Δ
laser EOM
RF
0° 0°
IF
Σ LO
−90°
synthes. carrier suppr. adj.

(detection of α or ϕ)

4
 Experimental method (2) 2

infrared
1.32 µ m
iso P! Pµ r(t) hybrid %
YAG (13dBm) EOM !90° =6dB
laser (!3dBm) (!26dBm) RF v(t)

0° 0°
photodiodes g=37dB FFT
50% coupler under test IF analyz.
iso s(t) & LO
22dBm g’=52dB
monitor !90°
output power phase & aten. phase $
meter
(carrier suppression) (detection of " or #)

100 power
MHz 9.9GHz ampli
PLL
synth. microwave near!dc

Fig. 1. Scheme of the measurement system.

• bridge (interferometric) scheme

analyzer measures the output spectrum, S (f ) or S (f ). The where q is the electron charge, % is the detector responsivity,
# low phase noise, limited ϕ by
α the noise figure of the ∆ amplifier
gain, defined as kd = v/α or kd = v/ϕ, is m the index of intensity modulation, and P λ the average
# carrier ! rejection " in ∆# => the ∆ amplifier does
optical power. This not flicker
is proved by dividing the spectrum density
# rejection gP R
µ of dissipative
−the source noise Si = 2qı = 2q%P λ of the the output current i by the average
0
kd = loss , (3) 2
# square microwave current i2ac = %2 P λ 12 m2 . The background
Rev. Sci. Instr. 73 6 p. 2445 (2002), and arXiv:physics/0503015
where g is the amplifier gain, P the microwave power, R = amplitude and phase white noise take the same value because
µ 0
50 Ω the characteristic resistance, and # the mixer ssb loss. they result from additive random processes, and because the
Under thethe
• noise
conditions of of
our the
setup ∑
(seeamplifier is isnot
below) the gain 43 detected
instrument gain kd is the same. The residual flicker noise is
dBV[/rad], including the dc preamplifier. The notation [/rad] to be determined experimentally.
5 Electron. Lett. 39 19 p. 1389 (2003) The differential delay of the two branches of the bridge is
means that /rad appears when appropriate.
 Background noise (1)
• well understood:
! " # g power gain
gPµ R0 dissip.
• phase-to-voltage kd =
!
− loss
Pµ
(∆ ampli)
microw. pow.
gain [V/rad] R0 charact. resist.
! " (50 Ω)
2F kT0 dissip. ! ssb mixer loss
Sφ t = + loss F noise figure
Pµ
(∆ ampli)
• thermal noise
=
2F kT0
+
!
dissip.
"
kT0 thermal energy
2 2 2
R0 ρ P λ m loss (4×10−21 J)
q electron charge
(1.6×10−19 C)
4q ρ responsivity [A/W]
Sφ s =
ρm2 P λ m modulation index
• shot noise Pλ optical power

• experimentally determined or up-bounded:

• contamination from AM noise (RIN)

6
 Background noise (2)
low optical power => thermal noise >> shot noise
1. replace the detectors with microwave signals
hybrid
−90° Δ
synthes. RF

0° 0°
IF
Σ LO
−90°
carrier suppr. adj. (detection of α or ϕ)

2. terminate the input of the delta amplifier

hybrid
iso −90° Δ
laser EOM
RF

synthes.
50% coupler
iso 0° 0° IF
Σ LO
−90°
carrier suppr. adj. (detection of α or ϕ)

7 ... and take the worst case

 Technical difficulties (1): crosstalk
• high EOM driving power
(22 dBm)

• low photodiode output

power (-26 dBm)

• finite isolation
(100-120 dB?)

• even small fluctuations of

the environment induce
noise as a consequence of
the fluctuating crosstalk
W:
waving a hand
0.2 m/s,


3 m far from the system
• work nighttime, when B:
background noise
nobody is around P:
photodiode noise
8
 Technical difficulties (2): reflections

• back reflection causes the

spectrum to be polluted

• flares appear at random in

some spectra, as shown

• unexplained physical
mechanism

S:
example of single spectrum, with

optical connectors and no isolators
B:
background noise
P:
photodiode noise
9
 Technical difficulties (3): reflections

• back reflections causes

spectra to be polluted at
random

• the average spectrum is

smooth

• wrong slope

• it is difficult to identify
and to discard polluted
spectra A:
average spectrum, with optical

connectors and no isolators
Figure 3:B:
background
Examples noise
P:
photodiode noise
of environment
eff
10 the corner. All the plots show the inst
 Technical difficulties (4): fibers

• the path of the optical

fibers affects the internal
stresses, and in turn the
reflections

• unpredictable effect on
noise, which is not the
photodiode noise

• trimming the system

takes patience F:
after bending a fiber, 1/f noise can

increase unpredictably
amples of environment effects and B:
background
experimental noise
P:
photodiode noise
mistakes around
ll 11the plots show the instrument Background noise (spectrum B)
 Example of photodiode noise

Figure 2: Example of measured spectra Sα (f ) and Sϕ (f ).

... after patient adjustement

modulator (EOM) is rejected. The amplitude noise of the source is rejected
12 to the same degree of the carrier attenuation in ∆, as results from the general
 Some results
all the pairTable
of two different photodiodes are
1: Flicker noise of the photodiodes.
compared

photodiode Sα (1 Hz) Sϕ (1 Hz)

estimate uncertainty estimate uncertainty
−7.1 −8.6
HSD30 −122.7 +3.4 −127.6 +3.6
−3.1 −1.8
DSC30-1K −119.8 +2.4 −120.8 +1.7
−1.5 −1.7
QDMH3 −114.3 +1.4
−120.2 +1.6
unit dB/Hz dB dBrad2 /Hz dB

estimated uncertainty
measured 0.5
in adB
second test, by
random, restoring
affects the the photodetectors and breaking the
differences
path from
the


hybrid junctionby

(amplified to the
the three-corner
∆ amplifier, and terminating the two
method)
free ends. The worst case is used as the background noise. The background
1 dB

systematic, affects all values in the same way
thereby obtained places an upper bound for the 1/f noise, yet hides the shot
13 This
is
correct
noise.

(non amplified
because by the
the shot noisethree-corner
arises in the method)
photodiodes, not in
 Conclusions

• the photodetectors we measured are similar in

AM and PM 1/f noise

• 2
the 1/f noise is about -120 dB[rad ]/Hz
• other effects are easily mistaken for the
photodetector 1/f noise
• environment and packaging deserve attention in
order to take the full benefit from the low noise
of the junction

www.arxiv.org, read the document arXiv:physics/0503022v1

14