Not Applicable.
The present application relates to optical fiber depolarizers and depolarized optical sources incorporating the same.
In many optical systems, such as pumps in optical communication systems, fiber gyros, control channels in optical amplifiers, sensors, and inteferometers, it is desirable to have a depolarized source to reduce polarization effects. In particular, Raman amplifiers are becoming increasingly used for distributed or remote amplification. These amplifiers are useful because they may be implemented in the normal fiber that carries the optical communications signal, and as a result, systems may be designed where amplification occurs in the transmission link itself, allowing greater transmission distances between amplifiers, higher receiver sensitivities, and lower transmission powers. Lower transmission powers have the added advantage of reducing nonlinear effects in the fiber. Furthermore, the wavelength at which the fiber Raman amplifier operates is determined by the wavelength of the pump light, unlike a rare earth-doped fiber amplifier whose gain bandwidth is limited by the rare earth species doped in the fiber. Raman is also a very low noise process, limited primarily at low powers by pump noise and quantum noise effects.
One particular concern with Raman amplification is the polarization dependence of the amplification process. The Raman gain coefficient when the pump light is polarized parallel to the signal polarization is about an order of magnitude greater than when the pump polarization is orthogonal to the signal polarization.
A long-haul single mode optical fiber is usually non-polarization preserving due its lower cost and superior performance compared to polarization maintaining fiber. Therefore, the polarizations of the pump and signal to change relative to each other as they propagate along the length of the fiber amplifier. In practice, Raman amplifiers are long, up to several kilometers in length, which permits sufficient variation in the relative polarizations of the signal and the pump that the polarization-dependent gain (PDG) effect is averaged out.
However, there remains a possibility that the pump and signal maintain the same relative polarizations for a significant length of the fiber amplifier. In such situations, the amplifier gain may be strongly dependent on the relative polarizations of the signal and the pump. This leads to uncertainty in amplifier performance, which results in increased errors in signal detection or increased system margin requirements.
One approach to reducing the effect of PDG is to use pump source light that is “unpolarized,” or that has a low degree of polarization. Unpolarized light is light that does not have a dominant polarization at an instantaneous point in time, or for which the dominant polarization changes more quickly than an amplifier response time.
Unpolarized light is further characterized as light which, when passed through a linear or circular polarization analyzer and incident on an optical power meter, shows no appreciable variation in transmitted power with analyzer azimuthal direction, and has no phase relationship between the two orthogonal polarization states.
As understood in this application, light is considered depolarized if the intensity of the light output from the depolarizer is substantially equally divided between orthogonal polarization states with no phase correlation between these two components. The use of unpolarized light to pump an optical gain medium alleviates the PDG effects, making it desirable for many optical pumping applications.
In the past, different techniques have been used to depolarize the light from a polarized light source. One conventional method for converting polarized light to depolarized light is to launch the polarized beam into a single piece of polarization maintaining (PM) optical fiber so that the launch angle is substantially at an angle of 45° relative to the birefringence axes of the PM fiber. Such a device is called a single-stage fiber Lyot depolarizer. Certain variations on this method are shown in U.S. Pat. No. 5,692,082 by N. Fukushima of Fujitsu Limited. The length of the PM fiber, as taught by Fukushima, is set so that an optical path length difference for the two polarization modes propagating through the PM fiber is greater than the coherence length of the incident light. As such, the two polarization modes are phase decorrelated, and the polarization state of the light output from the fiber is effectively randomized. However, the depolarizer in accordance with the teaching of Fukushima is costly because it requires a very long piece of PM fiber (e.g., ˜100 meters for a longitudinal mode linewidth of the order of 10 GHz). In addition it has proved highly temperature sensitive. Indeed it was observed that the degree of polarization (DoP) fluctuates rapidly and with a large amplitude (˜15-20%) over a temperature range as small as a few degrees. Thus this approach is only suitable for pump applications that can tolerate a DoP of approximately 20%.
In an article entitled “Degree of Polarization in the Lyot Depolarizer”, published in the Journal of Lightwave Technology, Vol. LT-1, No. 3, September 1983, William K. Burn teaches the calculation of optimum birefringent fiber lengths for a two-stage Lyot depolarizer for depolarizing the output of a laser source having a single longitudinal mode. In the case of a single longitudinal mode, the coherence function of the spectrum decreases monotonically with increased fiber length. Thus, as long as a minimum depolarizing fiber length is provided, a precise length is not necessary.
More recently, Hiroyuki Koshi of The Furukawa Electric Co. Ltd discloses in U.S. published application Ser. No. 0,025,111 A1 published Feb. 28, 2002, the use of a fiber depolarizers with Fabry-Perot lasers in a multiple pump architecture. Each depolarizer is a two-stage fiber Lyot depolarizer the lengths L1:L2 having a ratio of 1:2. The lengths of the PM fiber are adjusted in accordance with the coherence length of the incident light. Koshi states that “the effect of canceling the degree of polarization of the pumping source does not change even if the lengths of the polarization-maintaining optical fibers are set longer than the optimal lengths.” The Fabry-Perot lasers used in the Koshi device produce a multiple longitudinal mode spectral output, not a single longitudinal mode as assumed by Burn. As shown by the present invention, the DoP can actually be made worse by extending the length of the PM fibers beyond the optimum length. Koshi did not realize this. Accordingly, the temperature instability exhibited in the prior art is not adequately corrected in the Koshi disclosure.
A low cost, low loss temperature insensitive depolarizer is highly desired for Raman pumping architectures based on high power laser diodes, particularly a depolarizer capable of providing a DoP of less than 10%.
The present invention has found that a Lyot depolarizer for depolarizing a laser with multiple longitudinal modes comprises two segments of PM fiber optimized for length as a function of the longitudinal mode spacing of the laser source, and not as taught in the prior art as a function of the laser coherence length. Furthermore, the present invention has found secondary coherence peaks in the coherence function of a multiple longitudinal mode source, which can significantly worsen the DoP if the PM fiber lengths are significantly extended beyond optimum, contrary to the teaching in the prior art.
The present invention makes use of the periodic structure of the source spectrum, assuming substantially equal power in alternate longitudinal modes, by converting half the longitudinal modes to an orthogonal polarization state. As a result a significantly shorter fiber depolarizer is needed than for a single longitudinal mode source. And furthermore, greater temperature stability is achieved by properly adjusting the lengths of both segments of PM fiber.
Thus an aspect of the present invention provides a depolarizer for converting a linearly polarized laser output having multiple longitudinal modes into light in which the intensity is substantially equally divided between orthogonal polarization states comprising:
A further aspect of the invention provides a depolarized optical source comprising:
In a further preferred embodiment a depolarizer for depolarizing a laser source having multiple longitudinal modes comprises:
Further features and advantages of the present invention will become apparent from the following detailed description, taken in combination with the appended drawings, in which:
It will be noted that throughout the appended drawings, like features are identified by like reference numerals.
There is no consensus in the prior art regarding the optimization of optical fiber Lyot depolarizers. Recently at OFC '02, S. Matsushita et al. from Fitel published a paper entitled “Design of Temperature Insensitive Depolarizer for Raman Pump Laser Diode”, paper WB3, OFC '02, Mar. 17-22, 2002. This is a mostly empirical paper which does not provide a methodology nor any formula for an optimum depolarizer length. For reducing temperature sensitivity, however, Matsushita suggests a pigtail length that is ⅓ of the depolarizer length. By contrast, J. S. Wang et al. teach that the longer section, i.e. the depolarizer length, should be about ⅔ the length calculated from the laser mode spacing and the fiber birefringence, that is L2˜2λ2/(3.Δλ.Δn2), and the shorter section, i.e. the pump pigtail, should be between ½ and ⅓ the longer section, that is λ2/(3.Δλ.Δn1)<L1<4λ2/(9.Δn1), in an article entitled “Reduction of the Degree of Polarization of a Laser Diode with a Fiber Lyot Depolarizer” published in IEEE Photonices Technology Letters, Vol. 11, No. 11, November 1999, p. 1449-51. The authors achieved a DoP of approximately 20%.
In order to be practical, especially for a co-propagating Raman pump application, it is desired to achieve a DoP of less than 10%.
With reference to an article by Kiyofumi Mochizuki, entitled “Degree of polarization in jointed fibers: the Lyot depolarizer,” published in Applied Optics, Vol. 23, No. 19, 1 Oct. 1984, p. 3284-3288, the DoP at the output of L2 can be calculated as follows:
Γ(τ) is the coherence function of the light source, that is, the Fourier transform of its frequency spectrum, where τ1 and τ2 represent the polarization mode dispersion in fibers L1 and L2 respectively.
It is assumed that α is nominally zero degrees (or equivalently 90 degrees), that is, the slow axis X1 (or equivalently the fast axis Y1) of the laser pigtail L1 is aligned with the linear polarized beam exiting the laser diode. In practice there could be an offset as high as, for example, 5 degrees due to alignment error with the laser output. As shown further below this misalignment contributes significantly to the DoP peak. It should be minimized as much as possible.
α=εp<<1 Equation 2
It is also assumed that β is approximately 45 degrees, that is, the birefringent axes X2 and Y2 of the depolarizer fiber L2 are oriented at approximately 45 degrees relative to the axes X1, Y1 of the pigtail fiber L1.
β=π/4+εd, with εd<<1 Equation 3
Then a good approximation for the DoP is:
using the following notations:
Γ1≡Γ(τ1); Γ2≡Γ(τ2); Γ2+1≡Γ(τ1+τ2) and Γ2−1≡Γ(τ2−τ1)
In practice, the length L1 of the first birefringent fiber corresponds to many beat lengths, L1>>(λ/Δn1). Hence both cos(L1Δβ1) and cos(2L1Δβ1) vary rapidly with temperature due to minute changes in the mean wavelength λ, or the filter length L1 or the birefringence Δn1. Therefore, an optimum design of the depolarizer should aim at minimizing the peak DoP value which is (from equation 4).
DoPpeak≅[4εd2+Γ22+2εp·Γ2·|(Γ2−1−Γ2+1)+εp2·{2Γ12+2|Γ12−Γ2+1·Γ2−1|+Γ2+12+Γ2−12}]1/2 Equation 5
Expressions εd and εp from the angular orientation of the PM fibers should be minimized as much as possible. In practice εd can be kept below 0.02 (<1 degree), with a good PM fusion splicer. But εp may be as high as 0.09 (5 degrees) depending on the laser packaging technology. Hence the first design objective is to minimize Γ2.
Γ2≅0, for L2=λr2/(2Δλ·Δn2) Equation 8
where Δλ is the laser longitudinal mode spacing and Δn2 is the birefringence of the fiber used in L2.
Once equation 8 is satisfied, then the maximum DoP is limited to:
DoPpeak(Γ2≅0)≅[4εd2+εp2·{2Γ12+2|Γ12−Γ2+1·Γ2−1|+Γ2+12+Γ2−12}]1/2 Equation 9
Then the optimum length for the first birefringent fiber pigtail L1 is such that:
{2Γ12+2|Γ12−Γ2+1·Γ2−1|+Γ2+12+Γ2−12} is minimum Equation 10
As seen in
The fine curve is used for optimizing the length L1 of the first PM-fiber section by minimizing the quantity {2Γ12+2|Γ12−Γ2+1|·Γ2−+Γ2+12+Γ2−12}. Per the invention, this is achieved in the range, 2λ2/(9.Δλ.Δn1)<L1<λ2/(3.Δλ.Δn1), and in general the optimum corresponds to: (0.4 L2<L1<0.6L2), as indicated in Equation 12.
It is found that condition 10 is generally satisfied when:
Γ12−Γ2+1·Γ2−1≅0 Equation 11
This latter condition also provides minimum DoP fluctuations once condition 8 is also satisfied (Γ2≅0) see Equation 4.
It is also found that condition 10 is generally satisfied for
0.4 L2<L1<0.6 L2 Equation 12
Using the method outlined above, a DoP better than 5% over a 0-70 degree C. temperature range was demonstrated, as shown in FIG. 6.
The present invention has demonstrated that temperature sensitivity can be reduced to maintain less than 5% DoP in a PM optical fiber based Lyot depolarizer for use with 14xx nm pump laser diodes. This is quite important for Raman amplifier architectures since the depolarizer of the present invention can reduce insertion loss over previously used depolarizers, and reduce overall costs of an amplifier, by utilizing lower cost components.
Most Raman amplification modules contain several pump lasers to provide multiple pumping wavelengths, typically between 2 and 5. It is preferable to combine all of the pump signals before transmission through a single depolarizer in order to reduce the component cost.
The amplifier architecture illustrated in
In order to effectively depolarize the combined pump lasers, two conditions must be met. The polarization of the pump lasers must be maintained prior to launch into the depolarizer. And the depolarizer must be insensitive to the pump wavelengths. The first condition is met by utilizing all polarization maintaining components before the depolarizer.
It can be shown in the following analysis that the depolarizer is not very sensitive to the pumping wavelengths. It has been shown that the optimal fiber lengths for the depolarizer are as follows:
L1˜λ2/(4.Δλ.Δn1)
L2˜λ2/(2.Δλ.Δn2)
The mode spacing in the laser diode is given by:
Δλ=λ2/2ng1
where ng is the group index in the laser diode, and 1 is the diode cavity length. Using this expression, the optimal fiber lengths are then both proportional to ng/Δn. The primary wavelength dependence cancels out. Over the wavelength range (˜100 nm maximum) of a multi-pump Raman amplifier, the wavelength dependence of ng and Δn. are expected to be very small, on the order of a few percent.
The length of the first depolarizing fiber segment L1 shown in
The embodiment(s) of the invention described above is(are) intended to be exemplary only. The scope of the invention is therefore intended to be limited solely by the scope of the appended claims.
This application claims priority from provisional application No. 60/393,728 filed Jul. 8, 2002, for the present invention.
Number | Name | Date | Kind |
---|---|---|---|
5488683 | Michal et al. | Jan 1996 | A |
5692082 | Fukushima | Nov 1997 | A |
6522796 | Ziari et al. | Feb 2003 | B1 |
6603593 | Fidric et al. | Aug 2003 | B2 |
6618521 | Koshi | Sep 2003 | B2 |
20020025111 | Koshi | Feb 2002 | A1 |
Number | Date | Country | |
---|---|---|---|
20040005106 A1 | Jan 2004 | US |
Number | Date | Country | |
---|---|---|---|
60393728 | Jul 2002 | US |