IEEE PHOTONICS TECHNOLOGY LETTERS, VOL. 4, NO.

8, AUGUST 1992 847

Output Power Characteristics of

Erbium-Doped Fiber Ring Lasers
Th. Pfeiffer, H. Schmuck, and H. Biilow

Abstract-The output power of an erbium-doped fiber ring Wavelength

laser has been measured as a function of pump power, emission Pump Laser Diode Selective
1 = 980 nm
wavelength, doped-fiber length, and reflectivity of the output
coupler. A theoretical model for the calculation of the output
. Coupler
(980/1550 nm)
Er3+ Doped Fibre

power is introduced that is based on a previously derived gain

equation for erbium-doped fiber amplifiers. The calculated val-
ues are in good agreement with the experimental data, confirm-
ing the applicability of the theory to a reliable modeling of the
( Polarization
Controler Variable Flbre ODtlCal Filter
output power of erbium fiber ring lasers. 00

INTRODUCTION Isolator Laser Output

E RBIUM-DOPED fiber lasers are attracting an in-

creasing interest for application in telecommunica-
tion systems. They are tunable over a large wavelength
Fig. 1. Experimental setup of the erbium-doped fiber ring laser. Labels
1-3 indicate reference points used in the theoretical analysis.

range around 1550 nm and their emission spectrum is I/ and VI are the transmission factors describing the
perfectly matched to the gain spectrum of erbium-doped roundtrip losses in the passive fiber cavity (from 1 to 2)
fiber amplifiers [l], [2]. The mode of operation can be and the losses between the end of the active fiber and the
chosen to be single line [ll, [21, pulsed [31, [41 or multiline output connector (from 1 to 31, respectively. They include
[ 5 ] , [6]. The design of the cavity is of great importance for all losses of passive components like filter, isolator, cou-
the output power and the tuning range in all these cases. plers, and splices. The reflectivity of the output coupler is
Some of the parameters influencing the performance are R. The intracavity power P F ( G ) at wavelength A, can be
the output coupling ratio, the intracavity losses, and the analytically expressed by the gain G and pump power Pp
doped fiber length. We have investigated experimentally at wavelength A, [7]:
and theoretically the output power of a 980 nm pumped
erbium-doped fiber ring laser as a function of these pa-
rameters. For the theoretical calculation of the output
power we used an analytical gain equation that was previ-
ously derived for erbium-doped fiber amplifiers [71.
THEORETICAL MODEL
The setup of the ring laser is depicted in Fig. 1. Due to
the optical isolator the laser operates in a traveling wave
Here G,,, = exp ( ( a p / 6 - a L ) .L ) describes the maxi-
mode. The optical filter selects one oscillation wavelength
mum available gain at AI* that is only reached for arbitrar-
A,. Above threshold the laser output power P F t is then
ily high pump powers. L is the active fiber length and
given by
6 = ( A L . P,""'>/(A, * P F t ) denotes the ratio of saturation
Pyt = P F ( G ) . G * V , . (1 - R ) powers at A, and A,. The saturation powers P,: and the
(1)
small signal absorption coefficients ( Y ~ ,can ~ either be
where P p is the intracavity power at A, at the entrance measured [8] or calculated [7]. Combining (1)-(3) leads to
of the amplifier fiber (reference point 2 in Fig. 1). The the single-line output power of the erbium-doped fiber
gain G between point 2 and 1 is determined by the ring laser above threshold
oscillation condition

with pump threshold

Manuscript received April 24, 1992. This work was supported in part
by the European RACE program.
The authors are with Alcatel SEL Research Centre, D-7000 Stuttgart
40, Germany.
IEEE Log Number 9202018.

1041-1135/92$03.00 0 1992 IEEE

848 IEEE PHOTONICS TECHNOLOGY LETTERS, VOL. 4, NO. 8, AUGUST 1992

and slope efficiency

h 1-R R=10 %
7)=Y.Jl,. * ( 1- (V.R*G,,,,,)-'). (6) A R=27 %
A, 1- V.R

Both pump power and threshold power are to be under-

stood as given for the doped fiber (point 1 in Fig. 1).
EXPERIMENTAL
In the experiments we used an erbium-doped fd/Si/Ge
fiber with a core diameter of 2.3 p m and a numerical 0 1 ' , c
0 10 20 30 40 50 80
aperture of 0.30. The erbium concentration was estimated
PUMP P O W E R , mW
to be about 50 ppm. The fiber was optically pumped by a Fig. 2. Ring laser output power at A, = 1533 nrn versus pump power
980 nm laser diode launching 52 mW power into the for R = 10, 27, 57%. Active fiber length L = 20 m. Straight lines are
doped fiber. Traveling wave operation of the ring laser calculated using (4)-(6).
was accomplished by a polarisation independent optical
isolator (isolation > 60 dB). The lasing wavelength was 12- ' " " ' " '
selected by means of an angle-tuned optical etalon with a E
m L=10 m
bandwidth of 2.6 nm. The output coupler was a variable u 10-

ratio coupler made of standard single-mode fibers. With

this device, the reflectivity could be continuously varied
between R = 0.001 and R = 1.0 with an insertion loss of
0.5 dB. The excess loss at the output arm (point 3 in Fig. t-
1) varied as V,,, = 0.75 + 0.15 R. At R = 1.0 the total 3
a
ring loss was 6.81 dB, the loss VI between the exit of the t-
3
active fiber and the output connector was 2.86 dB without 0
Kxc. 0 10 20 30 40 50 80 70 80 90 100

The output power of the ring laser with active fiber REFLECTIVITY, 70
length L = 20 m at wavelength A, = 1533 nm is shown in (a)
Fig. 2 as function of pump power for reflectivities R = 10, 12

27, and 57%. The straight lines represent the result of a E

m
theoretical analysis using (4)-(6) with a,, = 0.42/m, q;at U 10

= 0.42 mW, a, = 0.611/m, PTt = 0.103 mW. These val-

ues have all been measured separately as described in [8].
The agreement between theoretical prediction and experi-
mental results is good, except for a slight deviation in the
t - 4
R = 57% curve. However, the calibration error of the 2
a
variable ratio coupler is up to *8% due to a limited I - 2
3
reproducibility of the screw setting. So a larger R value
might have been used in the calculation, resulting in a O O
0 10 20 30 40 50 80 70 80 90 100
better agreement. REFLECTIVITY, %
The variation of the output power with reflectivity R is (b)
shown in Fig. 3 for different fiber lengths ( L = 10-30 m)
12
at A, = 1533 nm. At a fiber length of 10 m the laser does c
L
I
not oscillate for a reflectivity below 2%. In this case the m
U
fiber is not long enough to compensate for the high ring U
losses. For L L 20 m the laser oscillates at all values of R
with a maximum output power around R = 5-10%. The $
agreement between experimental results and theoretical f
calculation, indicated by the solid lines, is good. There I-
3
are, however, some experimental uncertainties that are a
!-
responsible for the slight deviations between the calcu- 5
lated curves and the experimental points. Besides the 0
0 10 20 30 40 50 60 80 90 100
above mentioned uncertainty of the reflectivity R , the 70

REFLECTIVITY, Yo
second source of error is the loss of the fusion splice at
(C)
point 2 in Fig. 1. It was set constant in the calculation,
Fig. 3. Output power as function of reflectivity R for A, = 1533 nm.
although a slight variation is to be expected during the Pump power 52 mW, active fiber length L = 10-30 m. Solid lines are
cutback of the active fiber from L = 30 to 10 m. For long calculated using (4)-(6).
PFEIFFER et al.: POWER CHARACTERISTICS OF ERBIUM-DOPED LASERS 849

fibers ( L = 30 m) the theoretical model tends to overesti- 12-

mate the output power at high reflectivities. The reason E
m
for this is presumably the linear fiber background loss, u 10-

like Rayleigh scattering, that was not included in the w

U /\*-*W,
theoretical analysis above. The attenuation at 1314 nm \A
-,.-
3 ..--.-.-A--.-.\
due to Rayleigh scattering was measured with an OTDR
f ’I -\
setup and was extrapolated to be 12.5 dB/km at 1533 nm,
using a dependence on wavelength. Nevertheless, the I3
1
Ai - t -
1.
- 4- L=20rn
a
agreement between measured and calculated laser output I- 2- -A- L = 1 0 rn
3
power is good, especially if one takes into account that
O O
none of the many parameters used in the calculation was I

fitted. Instead all parameters have been determined in

separate measurements.
At low reflectivity ( R < 30%) the stability of the output
power is poor. So higher reflections are preferable, both
for good stability and for a large wavelength tuning range.
This is shown in Fig. 4 where the output power of the ring
laser for R = 42% and pump power of 52 mW is depicted
as a function of the emission wavelength. The tuning
range is determined by different factors. For L = 10 m REFERENCES
the largest available gain G,,, at the edges of the tuning
J. L. Zyskind, J. W. Sulhoff, J. Stone, D. J. Digiovanni, L. W. Stulz,
range is small compared to the respective values in the H. M. Presby, A. Piccirilli, and P. E. Pramayon, “Electrically
range 1530-1560 nm. Therefore the pump threshold is tunable, diode-pumped erbium-doped fiber ring laser with fiber
increased and the slope efficiency is decreased according Fabry-Perot etalon,” Electron. Lett., vol. 27, pp. 1950-1951, 1991.
H. Schmuck, Th. Pfeiffer, and G. Veith, “Widely tunable narrow
to ( 5 ) and (6), resulting in a decreased output power. The linewidth erbium doped fiber ring laser,” Electron. Lett., vol. 27, pp.
characteristic becomes flat for longer fibers ( L = 20 m) 2117-2119, 1991.
and is limited by the tuning range of the filter (1520-1570 Th. Pfeiffer and H. Schmuck, “Widely tunable actively mode-
locked erbium fiber ring laser,” in Proc. 2nd Top. Meet. Opt.
nm). For pump powers well above threshold the output Ampl@ers Appl., Snowmass Village, CO, 1991, pp. 116-119.
power is mainly determined by the slope efficiency which R. P. Davey, K. Smith, and A. McGuire, “High-speed, mode-locked,
is almost independent of the laser wavelength for long tunable, integrated, erbium fiber laser,” Electron. Lett., vol. 28, pp.
482-484, 1992.
fibers [see (611. At fiber lengths of L > 20 m, a multiline H. Schmuck and Th. Pfeiffer, “Fiber-pigtailed FabIy-Perot filter
output around 1530 nm was observed for filter settings used as tuning element and for comb generation in an erbium
below 1528 nm. We suppose that this was due to Rayleigh doped fiber ring laser,” in Proc. 17th ECOC, Paris, 1991, pp.
145- 148.
scattering inside the erbium-doped fiber which in conjunc- H. Okamura and K. Iwatsuki, “Simultaneous oscillation of wave-
tion with the large available gain around 1530 nm leads to length tunable, singlemode lasers using an Er-doped fiber ampli-
lasing inside the active fiber. This could not be controlled fier,” Electron. Lett., vol. 28, pp. 461-463, 1992.
Th. Pfeiffer and H. Bulow, “Analytical gain equation for erbium
by the filter. doped fiber amplifiers including modefield profiles and dopant
distribution,” IEEE Photon. Technol. Lett., vol. 4, pp. 449-451, May
CONCLUSION 1992.
A. A. M. Saleh, R. M. Jopson, J. D. Evankov, and H. Aspell,
We have measured and theoretically modeled the out- “Modeling of gain in erbium doped fiber amplifiers,” IEEE Pho-
put power of an erbium fiber ring laser under various ton. Technol. Lett., vol. 2, pp. 714-717, 1990.