Optical: systems and elements – Optical amplifier – Optical fiber
Reexamination Certificate
1999-05-10
2001-04-10
Hellner, Mark (Department: 3662)
Optical: systems and elements
Optical amplifier
Optical fiber
C359S199200
Reexamination Certificate
active
06215584
ABSTRACT:
FIELD OF THE INVENTION
This invention relates to gain flattened broadband optical amplifiers and more particularly to fiber amplifiers used for multi-channel Dense Wavelength Division Multiplexing (DWDM) based light wave communications systems and networks.
BACKGROUND OF THE INVENTION
Optical amplifiers, in particular Erbium doped fiber based amplifiers (EDFAs) are now the most commonly used optical amplification devices to amplify weakened optical signals. These are used by network elements such as Add/Drop multiplexers, optical cross-connects, switches, and routers. Largely used for both short and long haul optical communication networks, they can also be used for CATV broadcasting. The newest wavelength division multiplexing (WDM) systems, with all optical channels carried simultaneously by the same fiber as different wavelengths of light, require even more uses of EDFAs. This is primarily because the presence of multiple wavelengths, with each wavelength as a separate channel, offers great data carrying capacity. Not only that, but the use of multiple wavelengths on a single fiber also offers a flexibility for network planning and data/voice traffic routing.
The conventional EDFA bandwidth has been extended recently from the standard 1525-1565 nm area to the new 1565-1605 nm band (see U.S. Pat. No. 6,104,527 issued Aug. 15, 2000 to Dan Yang). More bandwidth is therefore now available, giving rise to more channels. This allows optical networks to transport information at rates of up to a terabit per second.
The physics of Erbium doped fiber is such that signal amplification or gain varies, depending on signal wavelength. Because of this, feeding several wavelengths of light into an amplifier, with each wavelength representing a separate channel, can produce problems. Since each channel is transmitted as a separate wavelength, and since gain is dependent on wavelength, different channels receive different amplifications. As consequence, the output of optical amplifiers used in multi-channel transmission systems must always be gain flattened. Gain flattening is accomplished by using gain flattening filters such as fiber gratings or by choosing more naturally flat Erbium fiber. Regardless of the gain flattening method used, the configuration of the passively gain flattening device of optical amplifiers is well defined for a given total input signal power and does not change once the amplifier package is closed. A passively gain flattened amplifier is therefore always designed for a specific gain and input power. For each specific gain and input power, the amplifier gain profile is constant. It must be noted that during amplifier operation the amplifier gain is locked to a constant value for most of WDM operation.
Given the above, another problem arises when the total amplifier input level changes. This condition can occur when the channel load varies due to events such as the dropping of channels at a node or the adding of channels at a junction. When this condition occurs, the pump power of the amplifier has to be adjusted in order to keep the gain constant. This is done to keep the amplifier gain profile unchanged. This method is suitable for applications such as changing the number of channels with each channel having a constant power level. By adjusting the pump power to keep the amplifier gain constant, each channel will be amplified evenly no matter how many channels are loaded. An apparatus for accomplishing this is disclosed in U.S. patent application Ser. No. 09/282206 filed Mar. 31, 1999.
Unfortunately, the invention contained in the above noted application has a shortcoming that renders it unsuitable for some applications. Specifically, it cannot maintain the amplifier gain profile when channel load remains the same but channel power varies. This condition normally occurs in instances such as a wave division multiplexed (WDM)ring configuration where all or some of the channels may be replaced at a node.
For applications such as the above, locking the amplifier gain to a constant is no longer practical. Locking the gain to a specific value produces a small output signal when the input signal is small. This is due to gain being the ratio between output and input. If this ratio is fixed, then having a small input produces a small output. Similarly, a large input will produce a large output for a fixed gain. This relation causes difficulty in optical system design as transmission fiber loss, or the link loss, is always constant. Thus, if a channel output is low, it may not be strong enough to reach the receiver.
To possibly solve this, it is possible to not lock the gain to a specific value. In this case, if gain is not locked, pump power is not adjusted depending on input. Amplifier output may remain about the same especially if the amplifier is working in deep saturation regime. However, because input level has changed, gain (or atomic level population inversion) becomes different from the designed (and optimized) one. The gain profile which was originally corrected by an embedded filter or some other method may be destroyed. This can cause a large gain variation across the wavelengths, with some wavelengths being amplified more than others and with some wavelengths having more amplified power than others. As a result, some channels reach the receiver while others may become too weak to be detected. This phenomenon is known as amplifier gain tilt.
FIG. 1
shows this occurring. As can be seen from the plot, the gain is and the gain flatness is dependent on the input signal level.
To solve the amplifier gain tilt problem, there are known amplifiers that have a specific characteristic. These amplifiers have a small input dynamic range within which the gain profile can remain unchanged if pump power is not adjusted. However, this small input dynamic range limits the use of such amplifiers. What is needed is an optical amplifier that not only has a large input dynamic range but also minimizes the gain tilt effect. Such an amplifier must respond evenly to different input levels without changing the output gain response profile. This amplifier must also allow the user to control the output signal power level without worrying about gain tilt.
To accomplish all of the above, the amplifier must be actively gain controlled. For active gain control, the gain flattening device must be adjusted for each new input/output power condition.
Pursuant to the above need, two active gain flatness control methods have been suggested. These use either actively controlled acousto-optic filters or a combination of optical switches and gain flattening filters, with each switch/filter combination corresponding to a different required gain. Both methods use different gain flattening filter profiles for each new signal and/or amplified spontaneous emission (ASE) condition. Given that the Erbium fiber gain spectrum has a complex shape, the acousto-optic method requires using several acousto-optic filters. Such a method is clearly expensive. Furthermore, each acousto-optic filter requires complex RF electronics and specialized programs for control, as well as a complex feedback mechanism to adjust and optimize the combination. Also, the large insertion loss of acousto-optic devices prevent them from being used outside of research labs.
The second method, using active gain flatness control, consists of using two optical switches, one at the input and one at the output, in conjunction with a number of gain flattening filters in parallel. Each time the control circuit detects an input level, it compares the input level with a preprogrammed table and switches to the corresponding gain flattening filter. This method, unfortunately, has some shortcomings. Specifically, it lacks flexibility in that for N filters have to be put in place. If N is a large number, this can be very costly.
Accordingly, what is required is an optical amplifier with a large dynamic input that provides a user adjustable power output while providing a constant gain profile, thereby preserving the gain dependence on signal wavelen
Dignam Marc M.
Yang Dan Dan
Hellner Mark
JDS Uniphase Inc.
Teitelbaum Neil
LandOfFree
Input independent tilt free actively gain flattened... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Input independent tilt free actively gain flattened..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Input independent tilt free actively gain flattened... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2478990