US 20020005981 A1 Abstract Compound gain flattening filters of optical amplification include a series of filter components with spectral responses that are combined to approach a target loss spectrum. Small variations in the central wavelengths of the component responses can produce large errors in the combined response of the compound gain flattening filter. However, the components can be divided into sub-components with balanced central wavelength deviations to achieve the desired response goals despite exhibiting otherwise detrimental wavelength deviations.
Claims(61) 1. A compound gain flattening filter comprising:
at least first and second sets of filter sub-components together approximating a single set of filter components defined by a best fit algorithm that determines central wavelengths of response for the filter components to approach a desired spectral filter response; at least one sub-component of said first set exhibiting a central wavelength of response shifted in a positive direction from the central wavelength of a corresponding component determined by the best fit algorithm; at least one sub-component of said second set exhibiting a central wavelength of response shifted in a negative direction from the central wavelength of the same corresponding component determined by the best fit algorithm; and said at least first and second sets being combined for producing an actual spectral filter response that is closer to the desired filter response than a filter response of the single set of filter components in which the corresponding filter component exhibits an actual central wavelength similarly shifted in either direction from the central wavelength determined by the best fit algorithm. 2. The filter of 3. The filter of 4. The filter of 5. The filter of 6. The filter of 7. A compound gain flattening filter of an optical amplification system comprising:
a series of filter components that produce a combined filter response approximating a desired spectral filter response; each of said filter components exhibiting a substantially different filter component response than other of said components of the series; said one component being divided into at least two sub-components having similar response profiles that are summed for producing one of the different filter component responses; and said response profiles of the sub-components being relatively shifted in wavelength in opposite directions to better approximate the desired spectral filter response than if the filter component response of the one component were similarly shifted in a single direction. 8. The filter of 9. The filter of 10. The filter of 11. The filter of 12. The filter of 13. The filter of 14. The filter of 15. An optical amplifying system comprising:
an optical amplifier exhibiting a variation in gain over a domain of wavelengths; a pair of first and second filters exhibiting substantially similar response profiles that contribute to reducing the variation in gain over the domain of wavelengths; each of said first and second filters exhibiting at least one peak wavelength of loss that differs from a common peak wavelength of loss that minimizes the variation in gain with a single filter; and said peak wavelengths of the first and second filters being offset in opposite directions from the common peak wavelength for reducing the variation in gain with respect to a similar offset of the peak wavelengths in the same direction from the common peak wavelength. 16. The system of 17. The system of 18. The system of 19. The system of 20. The system of 21. The system of 22. The system of 23. An optical amplifying system comprising:
an optical amplifier exhibiting a variation in gain over a domain of wavelengths; a plurality of pairs of long period gratings; each pair including two long period gratings that exhibit spectral responses with central wavelengths matched to a desired central wavelength of the pair for reducing gain variations of the optical amplifier; one long period grating member of each pair belonging to one of two similar sets of long period gratings and the other long period grating member of each pair belonging to the other of the two similar sets of long period gratings; and said central wavelengths of the two long period gratings of a first of said pairs being offset in opposite directions with respect to the desired central wavelength for reducing the variation in gain with respect to similar offsets of said central wavelengths in the same direction with respect to the desired central wavelength. 24. The system of 25. The system of 26. The system of clam 23 in which the long period grating members of each set are offset in the same direction. 27. The system of 28. The system of 29. The system of 30. A method of reducing gain variation in an optical amplification system comprising the steps of:
determining a desired filter loss spectrum; fitting the desired filter loss spectrum with a series of filter components having individual spectral responses centered at different specified wavelengths; fashioning at least a first of the filter components from at least two filter sub-components with similar spectral responses; and arranging the filter sub-components so that the central wavelengths of the two filter sub-components are offset in opposite directions with respect to the specified wavelength of the first filter component for better matching the desired loss spectrum despite variations in the central wavelengths of the filter sub-components. 31. The method of 32. The method of 33. The method of 34. The method of 35. The method of 36. The method of 37. The method of 38. The method of 39. The method of 40. The method of 41. A method of making a compound gain flattening filter comprising the steps of:
approximating a predetermined filter loss spectrum with a plurality of spectral response profiles having prescribed central wavelengths; fabricating a plurality of filter components having similar response profiles and central wavelengths that at least approximately correspond to the prescribed central wavelengths; and forming at least a first of the filter components with at least two filter sub-components having similar response profiles and central wavelengths that are offset in opposite directions from the prescribed central wavelength of the first filter component. 42. The method of 43. The method of 44. The method of 45. The method of 46. The method of 47. The method of 48. The method of 49. A compound optical filter comprising:
at least first and second sets of filter sub-components together approximating a single set of filter components defined by a best fit algorithm that determines central wavelengths of response for the filter components to approach a desired spectral filter response; at least one sub-component of said first set exhibiting a central wavelength of response shifted in a positive direction from the central wavelength of a corresponding component determined by the best fit algorithm; at least one sub-component of said second set exhibiting a central wavelength of response shifted in a negative direction from the central wavelength of the same corresponding component determined by the best fit algorithm; and said at least first and second sets being combined for producing an actual spectral filter response that is closer to the desired filter response than a filter response of the single set of filter components in which the corresponding filter component exhibits an actual central wavelength similarly shifted in either direction from the central wavelength determined by the best fit algorithm. 50. The filter of 51. The filter of 52. The filter of 53. The filter of 54. The filter of 55. A compound optical filter comprising:
a series of filter components that produce a combined filter response approximating a desired spectral filter response; each of said filter components exhibiting a substantially different filter component response than other of said components of the series; said one component being divided into at least two sub-components having similar response profiles that are summed for producing one of the different filter component responses; and said response profiles of the sub-components being relatively shifted in wavelength in opposite directions to better approximate the desired spectral filter response than if the filter component response of the one component were similarly shifted in a single direction. 56. The filter of 20 components each contribute approximately one-half of a total attenuation exhibited by the filter component response of the one component. 57. The filter of 58. The filter of 25 components include respective peak losses that are shifted in wavelength by less than one nanometer. 59. The filter of 60. The filter of 61. The filter of Description [0001] Optical amplifiers produce spectrally related variations in gain that can be reduced by gain flattening filters having compensating spectral responses. The gain flattening filters, which are sometimes required to exhibit complex response profiles, can be assembled from filter components that exhibit simpler response profiles. [0002] The complex spectral responses that are sometimes required of gain flattening filters can be approximated by concatenating filter components with simpler filter responses. Such compound gain flattening filters can be formed by concatenating conventional filters with spectral response profiles distinguished by their peak losses and their central wavelengths. [0003] Best fit algorithms solve for the peak loss and central wavelength characteristics of the filter components to approximate a target loss spectrum. However, manufacturing even simple filter components with Gaussian response profiles can be difficult to achieve to desired accuracy. The resulting filter errors (i.e., insertion loss error function), which are largely due to unintended shifts in the central wavelengths of the filter components, can leave excessive gain ripple representing uneven amplification of the desired transmission spectrum. [0004] Long period gratings with Gaussian response profiles can be used as filter components of compound gain flattening filters. Manufacturing these filter components to central wavelength accuracies less than one-half nanometer is difficult, and errors of only one Angstrom can produce significantly deteriorated results. Discarding gratings outside of tolerance and trial and error tuning are two approaches to meeting desired tolerances. Both are time consuming and expensive——the first due mainly to wasted efforts and the second due mainly to repeated efforts. [0005] We have discovered that compound gain flattening filters can be assembled from a series of simpler filter components having wider tolerances than those normally prescribed but without the expected errors. Residual gain ripple from new amplifier and filter combinations can be minimized by approaching a best fit loss spectrum for the filters. Higher manufacturing yields are possible by utilizing filter components that would normally be considered outside acceptable tolerance. [0006] Our new compound gain flattening filters can be manufactured using certain conventional steps including determining a filter desired loss spectrum and using a conventional algorithm to fit the desired loss spectrum with a series of filter components having individual spectral responses with peak losses specified at different wavelengths. However, instead of manufacturing a single set of filter components matching these specifications, at least a first of the filter components is fashioned from a pair of filter sub-components with similar spectral responses. The peak losses of the filter sub-components sum to the intended peak loss of the first filter component, and the central wavelengths of the filter sub-components also average to the specified central wavelength of the first filter component. However, the central wavelengths of the filter sub-components are offset in opposite directions with respect to the specified central wavelength of the first filter component for utilizing filter sub-components with wider range tolerances. The resulting response of the compound gain flattening filter is closer to the desired loss spectrum than if the central wavelength of the first filter were similarly offset in either direction. [0007] Compound gain flattening filters can be made in accordance with our invention by dividing an entire series of filter components into two sets of filter sub-components. The central wavelengths of all of the filter sub-components of the first set are shifted in a positive direction and the central wavelengths of the filter sub-components of the second set are shifted in a negative direction, both with respect to the prescribed central wavelengths of the filter component series. Together, the two sets of filter sub-components produce an actual spectral filter response that is closer to the desired loss spectrum than if the central wavelengths of the series of filter components were similarly shifted in either direction. [0008] The two sets of filter sub-components are particularly useful for reducing gain ripple of paired amplifiers having similar gain outputs, especially amplifiers separated by a fiber span. One set is positioned within one of the amplifiers, and the other set is positioned within the other amplifier. The gain ripples of the two amplifiers are separately reduced, avoiding amplifier-to-amplifier transmissions that are not appropriately corrected. [0009] The filter sub-components of our invention are preferably made according to usual manufacturing practices, but are sorted following manufacture according to their shift from prescribed central wavelengths. For example, those shifted in a positive direction can be paired with those shifted by a similar amount in a negative direction so that the average central wavelength of the pair approaches the prescribed central wavelength. A similar balance can be achieved by larger combinations of sub-components having central wavelengths distributed about a prescribed mean. [0010] Instead of or in addition to sorting by pairs or other combinations, tuning can be used to intentionally offset the central wavelength of one member of a combination to balance an opposite direction of offset exhibited by one or more other members of the combination. Tuning can also be practiced upon the one member to balance the filter central wavelengths in a static or dynamic fashion to accommodate changing conditions of use. [0011] The sensitivity of the actual filter response to central wavelength errors varies among the series filter components. Central wavelength errors of filter components with particularly sharp profiles or large losses can contribute to insertion loss errors more than other filter components with similar wavelength errors but with flatter profiles. Thus, wider tolerances of central wavelength can be accommodated for filter components with flatter profiles. In the practice of our invention, just the filter components having the sharper profiles can be subdivided into balanced wavelength sub-components for reducing the number of required filter components and sub-components without requiring stricter tolerances for the sharper profile filter components. [0012] The desired loss spectrum for the compound gain flattening filters represents the best fit of a combination of theoretical filter components varying in peak loss and central wavelength. However, the best fit of the series of filter components is still an approximation of a target loss spectrum required to entirely eliminate gain ripple. [0013]FIG. 1 is a graph of individual spectral responses of individual filter components intended to provide an approximate fit to a target loss spectrum. [0014]FIG. 2 is a graph of an insertion loss error function representing a difference between the target loss spectrum and the sum of individual spectral responses of the filter components. [0015]FIG. 3 is a graph of two different insertion loss error functions contrasting the effects of shifting central wavelengths of the component responses in opposite directions by approximately one Angstrom. [0016]FIG. 4 is a diagram comparing a conventional compound gain flattening filter to a new compound gain flattening filter with components subdivided into pairs. [0017]FIG. 5 is a graph that compares the insertion loss error functions resulting from various central wavelength shifts of pairs of filter sub-components replacing the filter components. [0018]FIG. 6 is a diagram of an optical amplification system showing two sets of filter sub-components located between amplifier coils. [0019]FIG. 7 is a diagram of an optical amplification system showing sets of filter components located within amplifier stages separated by a fiber span. [0020]FIG. 8 is a diagram similar to FIG. 6 in which one of the filter sets is tunable. [0021]FIG. 9 is a diagram of an optical amplification system in which one of several filter components is divided into a pair of sub-components between amplifier coils. [0022] A typical compound gain flattening filter includes a series of filter components that exhibit individual spectral responses such as the spectral responses [0023] Also shown in the graph of FIG. 1 is a target loss spectrum [0024] The insertion loss error function [0025] Designing and manufacturing even simple Gaussian response filters, such as long period gratings, to central wavelength accuracies less than one Angstrom is presently difficult, requiring, for example, large numbers of filters to be discarded or extensive trial and error efforts to achieve the required accuracy. However, we have found that central wavelength errors of one Angstrom or more can be accommodated by dividing the filter components into pairs of sub-components that are sorted or otherwise arranged to balance the central wavelength errors. [0026]FIG. 4 depicts an example of a conventional gain flattening filter [0027]FIG. 5 compares an insertion loss error function [0028] Thus, pairing sub-components (e.g., [0029] The remaining drawing figures depict optical amplification systems incorporating compound gain flattening filters with wavelength balanced filters. For example, an amplification system [0030]FIG. 7 depicts an optical amplifying system [0031]FIG. 8 depicts an optical amplifying system [0032]FIG. 9 depicts an optical amplifying system [0033] The balancing of any one or more pairs of sub-components can be also taken with regard to variations in the peak losses of the sub-components. For example, a sub-component that exhibits a peak loss of [0034] Both the components and the sub-components proposed for use in this invention are preferably long period gratings having Gaussian response profiles. Generally, a good fit to the target loss spectrum can be achieved with a combination of four independent Gaussian response profiles. These, of course, can be subdivided in accordance with our invention to loosen tolerances. Best fit algorithms are known to those of skill in the art for fitting Gaussian profiles to curves with more complex shapes. [0035] Our new compound gain flattening filters can also be constructed with other filter types and response profiles including fiber Bragg gratings and thin film gratings. The amplifiers of choice presently are erbium amplifiers, but other amplifiers working in the same or different spectrums could also be used. Classifications
Legal Events
Rotate |