WO2007134638A1

WO2007134638A1 - Optical band splitter/combiner and apparatus comprising the same

Info

Publication number: WO2007134638A1
Application number: PCT/EP2006/062500
Authority: WO
Inventors: Marco Piazza; Marco Zanatta
Original assignee: Pirelli & C. S.P.A.
Priority date: 2006-05-22
Filing date: 2006-05-22
Publication date: 2007-11-29

Abstract

An optical device (300) adapted to split/combine a first and a second wavelength bands, comprising a first optical band splitter (100) having and two output ports (102,103), five optical couplers (106-109) optically coupled in cascade between the input port and the output ports, and intercalated by four optical differential delay devices (111-114). The device further comprises a second optical band splitter (310;310a;310-l b;310-l c;310-2d) comprising an input port (335, 335a-c; 1910) coupled to one output port of the first optical band splitter, two output ports (345, 345a; 1505; 1705, 345d) and three optical couplers (506-508) optically coupled in cascade between the input port and the output ports of the second optical band splitter, and intercalated by two optical differential delay devices (511, 512), wherein an isolation, at one output port of the second optical band splitter, of the second wavelength band with respect to the first wavelength band is higher than an isolation at the other output port of the second optical band splitter the first wavelength band with respect to the second wavelength band.

Description

OPTICAL BAND SPLITTER/COMBINER AND APPARATUS COMPRISING THE SAME

§ § § § § DESCRIPTION

Background of the invention Field of the invention

The present invention generally relates to the field of optical telecommunications, and to devices employed in optical telecommunications networks. More particularly, the invention concerns optical wavelength band splitters/combiners, and it specifically relates to Fiber-To-The- Premises (FTTP) network equipments comprising optical wavelength band splitters/combiners. Description of related art

Currently deployed FTTP networks include Gigabit Passive Optical Networks (GPONs) for broadband delivery of voice, video and high-speed data directly to the home or broader community through optical fibers. Converged voice, video and data services networks are also known as "triple play networks". These networks support two signals in downstream direction (from a central station to the user) and one signal in upstream direction. A first downstream signal delivers analog television and a second downstream signal delivers digital voice and data services, such as for example telephone and/or Internet connection. The upstream signal is typically a digital signal delivering voice and data from the user to the central station or the service provider. Typically, FTTP systems deliver voice, video and data over a PON using the ITU-T GPON standard. The system supports Radio Frequency (RF) analog video delivery using a 1555 nm wavelength overlay. High-quality video on a PON is achieved with a high-power 1555 nm signal, and the power requirement at 1555 nm is greatly demanding. The second downstream signal (for digital voice and data) uses a 1490 nm wavelength. The upstream digital signal is typically allocated at 1310 nm wavelength. In FTTP networks, as well as in many other applications, a key aspect is signal splitting and/or combining. Signal multi/demultiplexing has to fulfill very demanding requirements such as, among the others, wide bandwidths and small cross-talk over a wide temperature range (typically from -4O⁰C to +85⁰C). Many applications require an almost rectangular wavelength response in order to maintain a low-loss and wavelength-independent transmission in a pass-band and a high- level rejection to all wavelengths in a stop-band. Low cost components, such as light emitters, are used over a wide temperature range, possibly generating a large wavelength drift. For example, anticipated telecommunications applications seek a 1.3/1.55 μm Wavelength Division Multiplexing (WDM) filter having a flat and low-loss pass-band at 1.280-1.335 μm and a -50 dB stop-band at 1.525-1.575 μm.

Various, more or less satisfactory devices have been proposed to comply with these demanding requirements. A critical review of these devices can be found in the published International application No. WO 2005/124412, in the name of the present Applicant, that is incorporated herein by reference.

In that document, it is pointed out that guided-wave type optical devices (e.g., splitters/combiners), also referred to as Planar Lightwave Circuit (PLC) devices or Integrated Optical Circuit (IOC) devices, have the advantage, over devices in different technologies, that they can be constructed on flat substrates in large quantities through processes like photolithography. Hence, they attract attention as a promising type of splitting/combining components which can be reproduced and integrated as compact parts.

In the cited International application, it is disclosed that the Applicant found that an optical band splitter/combiner comprising five cascaded splitting devices and four differential phase delay devices alternated with the splitting devices, as depicted in Figure 1, provides a high on-chip- density PLC device, particularly suitable for FTTP networks, having wide flattened continuous bands and strong tolerance to fabrication errors. In particular, the Applicant found that a splitter/combiner for a first and a second continuous optical bands wider than 10 nm may be obtained by designing the five splitting devices and the four differential delay devices so as to send the power input at an operating wavelength within a first optical band substantially in one of the two output ports, and the power input at an operating wavelength within a second optical band substantially in the other of the two output ports. A preferred choice of the splitting coefficients of the five splitting devices may result in a splitter/combiner of a first and a second continuous optical bands wider than 10 nm having particularly wide flattened bands and very strong tolerance to fabrication errors.

For a thorough discussion of the splitter/combiner disclosed in WO 2005/124412, reference is made to the cited document in its entirety.

Figure 1 shows a symbolic diagram of the optical device disclosed in WO 2005/124412. The device, globally denoted as 100, comprises a first input port 101, and a first and a second output ports 102, 103 optically coupled to the first input port 101. Conventionally, in an optical device made up of two waveguides, the output port that is on the same waveguide as the input port is denoted "bar" output port, or shortly "bar" port, whereas the output port that is on the other waveguide is denoted "cross" output port, or, shortly, "cross" port. This, is the device 100 is made up of two waveguides, the output port 102 is the bar output port, and the output port 103 is the cross output port. The first input port 101 is adapted to receive an input optical radiation Pj_n. The device 100 may optionally comprise a second input port 104. With respect to the second input port 104, bar and cross designations are reversed. The optical device 100 is adapted to split the input optical radiation Pj_n received at the input port 101 into two output optical radiations Pbar and Pcross, outputting respectively from the bar port 102 and the cross port 103 and having each a respective optical power spectrum. A first, a second, a third, a fourth and a fifth optical splitting devices, represented respectively by blocks 106, 107, 108, 109 and 110, are optically coupled in cascade, and a first, a second, a third and a fourth optical differential delay devices, represented respectively by blocks 111, 112, 113 and 114, are optically coupled and intercalated to the optical splitting devices 106, 107, 108, 109 and 110. Summary of the invention

Figure 2 is a diagram showing a simulated transmittance spectral response of an optical splitter/combiner designed according to the teachings of WO 2005/124412, in an attempt to meet the ITU-T specification in terms of ripple and isolation in the two prescribed wavelength bands. Curve 205 represents the response at the cross output port 103, whereas curve 210 is the response at the bar output port 102. From Figure 2 it can be appreciated that the single device so designed is not suitable to be used in GPONs, because it would be not compliant with the requirements in terms of cross-talk of the ITU-T prescriptions, which specify, for the level of crosstalk of the digital signal, comprised within the wavelength band 1490 nm ± 10 nm, on the analog signal, comprised within the wavelength band 1555 nm ± 5 nm, a value of at least -28 dB, and a value of at least -32 dB for the level of cross-talk of the analog signal on the digital signal.

From Figure 2 it can also be appreciated that the optical response of the device so designed is substantially symmetric: considering two equally wide optical bands respectively centered at the wavelength of symmetry of the response curve of the respective pass-band output port, the optical characteristics are substantially equal for the two bands. For example, considering two bands respectively centered at approximately 1490 nm and 1558 nm, and 10 nm wide (depicted as shaded areas in Figure 2) the optical splitter/combiner flattens the two pass-bands substantially equally, being the in-band ripples less than 0.1 dB, and isolates the two bands substantially equally, being the isolation value approximately -22 dB for the first band, centered at 1490 nm, and approximately -19 dB for the second band, centered at approximately 1558 nm (the isolation values are denoted IM and I1-2 in the drawing).

Considering a first and a second wavelength bands which are continuous, have equal widths ranging from approximately 5 nm to approximately 40 nm, and with respective first and second central wavelengths in a range from approximately 600 nm to approximately 2000 nm, separated so that the two bands do not overlap, for the purpose of the present description, an optical band splitter device is considered to have a symmetric response when the difference between the isolation of the first band compared to the second band, and of the second band compared to the first band does not exceed approximately 5 db, otherwise the optical band splitter is considered to have an asymmetric response.

In WO 2005/12441 it was proposed that, in order to comply with the cross-talk requirements of the ITU-T specifications, two or more devices of the type depicted in Figure 1 can be cascaded. In particular, a device was proposed consisting of three optical splitters/combiners of the type depicted in Figure 1, in a tree-cascade configuration, featuring a simulated response complying with the above ITUT specifications.

Starting from the consideration that a single optical splitter/combiner as disclosed in WO 2005/12441 cannot, perse, satisfy the ITU-T requirements, and that a cascade of at least two such devices is needed to that end, the Applicant has observed that the resulting device is affected by some drawbacks.

Firstly, the resulting device dimensions are relatively large, due to the necessity of providing five plus five directional couplers, intercalated by four plus four differential delay devices for each of the two optical paths. This contrasts with the trend in reducing the devices' size, so as to achieve higher integration levels. Furthermore, the relatively high number of directional couplers to be provided introduces non-negligible insertion losses, and renders the final device more sensitive to manufacturing errors. The production yield is thus negatively affected.

The Applicant has tackled the problem of providing an optical device adapted to be used as a splitter/combiner device in PONs, particularly in GPONs complying with the ITU-T prescriptions, and not, or at least less affected by the drawbacks of the known devices.

The Applicant has found that the asymmetry in the ITU-T requirements in terms of cross- talk between the analog and digital signals may be exploited to design an optical device adapted to be used as a splitter/combiner device in an FTTP, providing better performances than the known devices.

According to an aspect of the present invention, an optical device as set forth in appended claim 1 is provided.

The optical device is adapted to split/combine a first and a second wavelength bands, the first wavelength band having a first central wavelength, and the second wavelength band having a second central wavelength, wherein:

- the first and second central wavelengths are in a range of wavelengths from approximately 600 nm to approximately 2000 nm;

- the first and second wavelength bands are continuous and have equal widths ranging from approximately 5 nm to approximately 40 nm, and

- a separation between the first and the second central wavelengths is such that the first and second wavelength bands do not overlap. The optical device comprises a first optical band splitter comprising an input port, a first output port and a second output port, and five optical couplers optically coupled in cascade between the input port and the first and second output ports, and intercalated by four optical differential delay devices, wherein the first optical band splitter is adapted to split the first and the second wavelength bands, when input at the input port, and to make them available respectively at the first and second output ports, the first optical band splitter having a respective power response at the first and second output ports wherein the central wavelength of each wavelength band corresponds to the maximum of the power response at the respective output port and an isolation, at the first output port, of the second wavelength band with respect to the first wavelength band that is essentially equal to an isolation, at the second output port, of the first wavelength band with respect to the second wavelength band.

A second optical band splitter is provided, comprising an input port coupled to the first output port of the first optical band splitter, a first output port and a second output port, and three optical couplers optically coupled in cascade between the input port and the first and second output ports, and intercalated by two optical differential delay devices, wherein the second optical band splitter is adapted to split the first and the second wavelength bands, when input at the input port of the second optical band splitter, and to make them available respectively at the first and second output ports thereof, the second optical band splitter having a respective power response at the first and second output ports thereof, wherein the central wavelength of each wavelength band corresponds to the maximum of the amplitude response at the respective output port of the second optical band splitter, and wherein an isolation, at the first output port of the second optical band splitter, of the second wavelength band with respect to the first wavelength band is higher than an isolation at the second output port of the second optical band splitter of the first wavelength band with respect to the second wavelength band.

Brief description of the drawings

The features and advantages of the present invention will be made apparent by the following detailed description of some embodiments thereof, provided merely by way of example, description that will be conducted making reference to the attached drawings, wherein:

Figure 1 shows a schematic diagram of a known optical device, disclosed in WO 2005/124412;

Figure 2 shows a simulated spectral response of an optical device according to Figure 1; Figure 3 shows, in terms of functional blocks, an optical band splitter/combiner according to an embodiment of the present invention;

Figure 4 shows a symbolic diagram of an optical coupler used in the construction of the optical band splitter/combiner of Figure 3;

Figure 5 shows a general scheme of an optical device adapted to be used in the optical band splitter combiner of Figure 3, according to an embodiment of the present invention;

Figure 6 schematically shows a possible layout arrangement of an optical device according to the general scheme of Figure 5;

Figure 7 is a schematic cross-sectional view of the device of Figure 6, taken along the line VII-VII of Figure 6; Figure 8 schematically shows a layout arrangement of an optical coupler used in the device of Figure 6;

Figure 9 schematically shows a layout arrangement of a differential optical delay device of the device of Figure 6;

Figures 1OA and 1OB show, respectively, a spectral response, and the isolation between the two channels of a first example of optical device according the general scheme of Figure 5; Figure 11 schematically shows a layout arrangement of the optical band splitter/combiner of Figure 3, in an embodiment of the present invention;

Figure 12 is a diagram showing an experimental spectral response of the optical band splitter/combiner of Figure 11 ; Figure 13 is a diagram showing the isolation between the two channels of a second example of optical device according to the general scheme of Figure 5;

Figure 14 shows, in terms of functional blocks, a second embodiment of a device according to Figure 3;

Figure 15 is a diagram showing the spectral response of the device of Figure 14; Figure 16 shows, in terms of functional blocks, a third embodiment of a device according to Figure 3;

Figure 17 is a diagram showing the spectral response of the device of Figure 16;

Figure 18 shows, in terms of functional blocks, a fourth embodiment of a device according to Figure 3; Figure 19 is a diagram showing the spectral response of the device of Figure 18;

Figure 20 shows, in terms of functional blocks, an optical device according to an embodiment of the present invention, including a device according to Figure 3; and

Figure 21 is a diagram showing a spectral response of a 1300 nm/1500 nm symmetric optical filter included in the device of Figure 20, according to an embodiment of the present invention.

Detailed description of the preferred embodiments

Figure 3 shows a schematic diagram of an exemplary optical device 300 adapted to be used as an optical band splitter/combiner, particularly, albeit not limitatively, in a GPON FTTP. The device 300 comprises a cascade of a first and a second optical devices. The first optical device is the device disclosed in the above cited International application

WO 2005/124412, and shown with the reference numeral 100 in Figure 1, i.e. a filter having a symmetric response in respect of the equally-wide bands centered around two prescribed central wavelengths, characterized by a substantially equally flat response at both the bar and the cross output ports 102 and 103, and isolation values of one band with respect to the other at the two ports differing of less than approximately 5 dB. This kind of optical device will be hereinafter also shortly referred to as a "symmetric filter".

The second optical device, denoted 310 in the drawing, is a device according to the general scheme of Figure 5, and has an input port 335 optically coupled to the bar output port 102 of the first optical device 100, and an output port 345.

In use, an input optical radiation or signal Pi_n inputted to the device 300 may be an optical signal comprising a signal Pin,i at a first wavelength λi and a signal Pin,2 at a second wavelength λ2 comprised within a first and a second optical bands, respectively.

For the purpose of the present description, the first and the second wavelength bands are assumed to be continuous and to have equal widths, ranging from approximately 5 nm to approximately 40 nm. The first and the second wavelength bands have respective first and second central wavelengths, both the first and the second central wavelengths being in a range of wavelengths from approximately 600 nm to approximately 2000 nm, and being separated from each other of an amount such that the first and second wavelength bands do not overlap. Each of the first wavelength λi and the second wavelength λ2 making up the input optical signal Pi_n may be close to the respective central wavelength. The first and the second signal wavelengths λi, λ2 are typically widely spaced, for example the spacing between the first and the second wavelengths may be greater than about 10 nm, or even greater than about 30 nm.

The optical device 300 is adapted to separate the input optical signal Pi_n into two components P_out,i and P_out,2, corresponding to the two signals Pin,i and Pin,2, outputted at the output port 345 of the second device 310 and at the cross port 103 of the first device 100.

In the following, the two devices 100 and 310 are described in detail.

With reference to Figure 1, the optical device 100 can be characterized, in terms of its optical behavior, as being adapted to receive, at its input port 101, a relatively broad-band input optical signal Pi_n, and being adapted to split a first and a second optical wavelength bands comprised within the input optical signal, making them available respectively at the bar and cross output ports 102 and 103. The expression "split a first and a second optical wavelength bands" means that, comparing the two output power spectra Pbar and P_cross normalized at their respective maximum intensity, the optical power in the cross output radiation Pcross is, at each wavelength of one of the two optical bands, greater than the optical power at the same wavelength in the power spectrum of the bar output radiation Pbar. Similarly, the optical power in the bar output radiation Pbar is, at each wavelength of the other of the two optical bands, greater than the optical power at the same wavelength in the power spectrum of the cross output radiation P_cross. In other words, at any wavelength of, e.g., the first optical band, the optical power outputting from one of the output ports 102, 103 of the device 100 is more than half of the total output power outputting from the output ports of the splitter/combiner at the given wavelength.

It is observed that, dually, the device 100 may be used as a wavelength bands combiner, instead of a splitter.

It is convenient to specify a predetermined respective level of cross-talk for each of the first and second optical bands. The term 'cross-talk' refers to a relative-power level Xi and X2 (in dB) set for the first and the second optical bands, respectively. The output power spectrum outputting from that of the two output ports 102 and 103 at which the considered band is suppressed exhibits, at each wavelength within said band, a power level below the predetermined level of cross-talk. Exemplarily, at the bar output port 102, the power at any wavelength within one of the two bands (called "stop-band") should be Xi dB below the power of any wavelength within the other of the two bands (called "pass-band"). On the other hand, at the cross output port 103, the power at any wavelength within the other of the two bands should be X2 dB below the power of any wavelength within the one of the two bands. It is observed that, in a similar way, cross-talk values can be defined for each of the devices making up the optical device 300, such as the device 310, as well as for the device 300 considered in its entirety. In the first optical device 100 of the band splitter 300, each of the optical splitting devices

106, 107, 108, 109 or 110 may be any kind of device adapted to split an input optical radiation into at least two optical radiations outputting from at least two separate output positions of the device. The optical splitting devices 106, 107, 108, 109 or 110 may be for example N x W-port devices, wherein M is at least equal to 2 for all devices and N is at least equal to 2 for devices 107, 108, 109 and 110. For example, each of the optical splitting devices 106, 107, 108, 109 or 110 may be a multi-layer beam splitter or a Fabry-Perot cavity or any mirror (e.g. a Brewster angle window). The generic optical splitting device may be an MZI (Mach-Zehnder Interferometer) splitter/combiner or an optical coupler, such as for example an MMI coupler or a PLC directional coupler, particularly a single 2-port PLC directional coupler. The optical splitting device 106 may also be a Y-branch coupler, with one input and two outputs.

Referring to Figure 4, which shows a symbolic diagram of an optical coupler adapted to be used as an optical splitting device, independently from the structure or the number of ports, each of the optical splitting devices 106, 107, 108, 109 and 110 may be viewed as an optical device comprising a respective first input port 407, a first (bar) output port 408 and a second (cross) output port 406; the optical splitting devices 107, 108, 109 and 110 also have a second input port 405.

At a given wavelength, the power coupling ratio C of a generic optical coupler is defined according to the following equation (1):

si P cross

P cross + P bar

Generally, the power coupling ratio C of an optical coupler may be expressed by the following equation (2):

C=sin2 Q(K), where Q(K) is the coupling angle of the power coupling ratio, and typically depends on wavelength.

Referring back to Figure 1, the first and fifth optical splitting devices 106 and 110 have substantially the same associated coupling angle, equal to a first coupling angle value QA(K). The second and fourth optical splitting devices 107 and 109 have substantially the same associated coupling angle, equal to a second coupling angle value θe(λ). The third optical splitting device 108 has an associated coupling angle equal to a third coupling angle value Qc(K), and may comprise two distinct optical splitting devices having substantially the same associated coupling angle, equal to about QA(K), and optically connected to each other. In this case, the overall coupling angle value Qc(K) is substantially equal to about twice QA(K).

In order to take into account the unavoidable deviations due to fabrication errors, as well as those due to the measurement errors, a generic optical splitting device will be regarded, for the purpose of the present invention, as having an associated generic coupling angle of value θ(λ) if its actual coupling angle does not depart from the exact coupling angle value θ(λ) by more than

1% of the respective value θ(λ). As an example, the first and fifth optical splitting devices 106, 110 will be regarded as having the same associated coupling angle value θ_A(λ) even when their actual coupling angles deviate from each other by at most 2%. In this case, the associated coupling angle value θ_A(λ) is the average of the two values.

The first, the second, the third and the fourth optical differential delay devices 111, 112, 113 and 114 have a respective associated differential delay, φx(λ), x=A, B, C, D. Each of the first, second, third and fourth optical differential delay devices 111, 112, 113 and 114 comprises at least a first and a second optical branches arranged in parallel configuration and having different optical- path lengths, in order to introduce the differential delay φ_x(λ), x=A, B, C, D, between the optical radiations propagating through the two optical branches. Generally, the differential delay changes when changing the wavelength of the optical radiation propagating therethrough. For a generic differential optical path delay, typically the longer the wavelength, the smaller the corresponding absolute value of the differential delay, as expressed by the following equation (3):

.

A generic optical differential delay device 111, 112, 113 or 114 may be viewed as an optical device comprising a first input port and a first output port optically connected through the first optical branch, and having a second input port and a second output port optically connected through the second optical branch.

In one embodiment of the present invention, the optical differential delay devices 111, 112, 113 or 114 may comprise a pair of planar waveguides having different optical paths. The differential delay can be obtained, for example, by adiabatically changing the waveguide shape (e.g. enlarging or narrowing its width) on one arm or by providing an extra length on one arm or by thermally heating one arm with respect to the other. By placing the physical modification of the waveguide on the opposite arm, the actual differential delay changes the sign.

The differential delay φ may be given by the relation φ= β₂ L₂- βi L₁, wherein L_x, x=1,2, is the physical length of the optical path, β_x, x=1,2, is the propagation constant of the optical radiation and the suffixes 1 and 2 refers conventionally to the first branch and the second branch, respectively. As a result, a "positive" differential delay will correspond to an optical path of the second branch longer than that of the first branch, while a "negative" differential delay means the opposite.

Once the first and the second branches of the first optical differential delay device 111 are arbitrarily selected, so as to determine the sign of the respective differential delay φ_A(λ .thfc first and the second branch of the remaining optical differential delay devices 112, 113 and 114 are consequently determined by the optical connection between said differential delay devices 112, 113 and 114 and the first optical differential delay device 111. More in detail, the optical splitting device

107 has a first input port optically coupled to the first output port of the first differential delay device 111 and a second input port optically coupled to the second output port of the first differential delay device 111. The subsequently cascaded second optical differential delay device 112 has its first input port optically coupled to the first (bar) output port of the preceding splitting device 107 and its second input port optically coupled to the second (cross) output port of the preceding splitting device 107. Hence the first and the second optical branches of the second optical differential delay device 112 are determined. The subsequently cascaded third optical splitting device 108 has its first input port optically coupled to the first output port of the preceding differential delay device 112 and its second input port optically coupled to the second output port of the preceding differential delay device 112. The subsequently cascaded third optical differential delay device 113 has its first input port optically coupled to the first (bar) output port of the preceding splitting device 108 and its second input port optically coupled to the second (cross) output port of the preceding splitting device 108. The subsequently cascaded fourth optical splitting device 109 has its first input port optically coupled to the first output port of the preceding differential delay device 113 and its second input port optically coupled to the second output port of the preceding differential delay device 113. The subsequently cascaded fourth optical differential delay device 114 has its first input port optically coupled to the first (bar) output port of the preceding splitting device 109 and its second input port optically coupled to the second (cross) output port of the preceding splitting device 109. The subsequently cascaded fifth optical splitting device 110 has its first input port optically coupled to the first output port of the preceding differential delay device 114 and its second input port optically coupled to the second output port of the preceding differential delay device 114.

Each of the differential delay devices 111, 112, 113 and 114 is adapted to introduce a differential delay which is substantially an even multiple of π at least at a first wavelength λi_op within the first optical band (φ(λi_op)= 2nπ, n integer) and substantially an odd multiple of π at least at a second wavelength λ2_0p within the second optical band (φ(λ2_0p)= (2n'+1)π, n' integer).

In case a differential delay device comprises two optical paths which differ only in the respective lengths, in order to ensure that the differential delays at the two respective wavelengths λi_op and λ2_0p differ by an odd multiple of π, the length difference ΔL= |l_2 - Li| should be according to the following expression (4):

AL = (2m + l)π / | β (λ_2op ) - β (\_op ) | , _m mteger;

with m=0, ΔL =π / | β (λ_2op)- β (λ_lop) | .

The absolute values of the differential delays of the differential delay devices 111, 112, 113 and 114 may be substantially the same (|φA(λ)Hφ_B(λ)Hφc(λ)Hφ_D(λ)|). The third differential delay device 113 may have an associated differential delay φc(λ) which is opposite in sign with respect to the differential delay φβ(λ) of the second differential delay device 112. The fourth differential delay device 114 may have an associated differential delay φo(λ) which is opposite in sign with respect to the differential delay φA(λ) of the first differential delay device 111. In particular, the first and the third differential delay devices 111 and 113 may have substantially the same differential delay, and the second and the fourth differential delay devices 112 and 114 may have substantially the same respective differential delay, opposite in sign to the differential delay of the first and the third differential delay devices 111 and 113. Alternatively, the first and the second differential delay devices 111 and 112 may have substantially the same differential delay, and the third and the fourth differential delay devices 113 and 114 may have substantially the same respective differential delay, opposite in sign to the differential delay of the first and the second differential delay devices.

In general, a difference between two optical devices due to the unavoidable fabrication errors does not depart from the optical devices being substantially identical. Accordingly, a deviation of the actual value of a generic differential delay from a respective nominal value φ by at most 1% does not depart, for the purpose of the present invention, the actual differential delay device from having an associated differential delay equal to about the nominal value φ.

The optical splitting devices 106, 107, 108, 109 and 110 have respective associated coupling angles 0A(λ), θβ(λ) and θc(λ) adapted to direct, in combination with above described differential delay devices 111, 112, 113 and 114, more than half of the total power of an optical radiation (like the signal Pi_n,i in Figure 3) inputting at the first input port 101 and having any wavelength within the first optical band to one of the first and second output ports 102, 103 (e.g., to the bar port 102), and more than half of the total power of an optical radiation (like the signal Pin,2 in Figure 3) inputting at the first input port 101 and having any wavelength within the second optical band to the other of the first and second output ports 102, 103 (e.g., to the cross port 103).

The coefficients 0A(λ), θβ(λ) and θc(λ) comply with the following relations at least at a third wavelength λ3_0p within the first optical band and at least at a fourth wavelength λ4_0p within the second optical band (relations (5)): 2θ,(λ_4oJ+θ_c(λ_4oJ-2θ_β(λ_4oJ« ^ + Λπ

2θ,(λ_3OJ+θ_c(λ_3OJ+2θ_β(λ_3oJ« (l-0| + mπ ^'

where te {0,1} , k is an integer and m is a non-negative integer; /( and m may be selected in order to have 0A(λ), Θβ(λ) and θc(λ) positive.

Nominally, the first wavelength λi_op coincides with the third wavelength λ3_0p and the second wavelength λ2_0p coincides with the fourth wavelength λ4_0p, but wavelength deviations may occur because of the fine tuning of the overall structure. In fact, when designing an optical device in accordance with the present invention, the designer needs to make the pass-band and the stop- band of each output port of the device to fit with the desired optical bands. In addition, the designer needs to take into account the specific desired level of cross-talk of each stop-band, which in general may vary with the first and the second optical band. As a consequence, a fine-tuning process is usually required to achieve the desired specifications and it is generally driven by the optical band having the most stringent specification for the cross-talk. Exemplarily, one may start from an ideal condition wherein λi_op is equal to λ3_0p and is at the center of the desired optical band, and λ2_0p is equal to λ4_0p and is at the center of the respective desired optical band. After calculating, e.g. by computer simulation, the corresponding spectral response, one may seek an optimal configuration by stepwise changing, for example, λi_op and λ3_0p independently from each other. Typically, λ2_0p and λ4_0p move accordingly. The process ends when an optimal solution is found.

It is noted here that, strictly speaking, the above combination of the optical splitting devices and differential delay devices of Figure 1 does not direct an optical radiation inputting at the input port 101 and having a specific wavelength within the first optical band totally at one of the first and second output port 102, 103 and an optical radiation inputting at the input port 101 and having a specific wavelength within the second optical band totally in the other of the first and second output ports 102, 103. This is because of the deviation of λi_op from λ3_0p and λ2_0p from λ4_0p at the end of the fine tuning process described above. Nevertheless, the optical device 100 is adapted to direct an optical radiation inputting at the input port 101 and having a wavelength within the first optical band (like the signal Pin,i) substantially at one (e.g., the bar port 102) of the first and second output ports 102, 103, and an optical radiation inputting at the first input port 101 and having a wavelength within the second optical band (like the signal Pin,i) substantially in the other (e.g., the cross port 103) of the first and second output ports 102, 103, wherein "substantially" means at least the 85% of the total power outputting from the device 100 at the respective wavelength. For similar reasons, the deviation of λi_op from λ3_0p and λ2_0p from λ4_0p at the end of the fine-tuning process causes the response of the first optical device 100 not to be exactly symmetric in the first and second bands, but only approximately symmetric, as shown in Figure 2. For the purpose of the present invention, the deviation from exact symmetry in the response of the optical device 100 due to the deviation of λi_op from λ3_0p and λ2_0p from λ4_0p are nonetheless neglected, and the response is considered symmetric even in presence of such deviations.

It is appreciated that in equations (5) above the wavelength dependence of the optical splitting devices in the optical bands of interest is taken into account.

Aiming to optimize the optical device 100 with respect to tolerance to fabrication errors, it is advantageous to choose the indexes t, m and k in order to keep the coupling angle values θ_A(λ), θ_B(λ) and θc(λ) as low as possible. This in turn means that, according to relation (2), also the lengths of the splitting devices 106, 107, 108, 109 and 110 are as low as possible, thus reducing the wavelength dependence of the coupling angle itself. In addition, in view of a high density optical circuitry, it is advantageous to minimize the physical length of the constituent parts of the optical device. The reduced wavelength dependence of the splitting devices is particularly critical in high index contrast technology.

In particular, t, m and /(should advantageously satisfy the following selection rules (6):

m ≥ I k I for t = 0 m ≥ k + l foτ t = l ANO k ≥ O . m ≥ -k for t = 1 AND k < 0

Table 1 below shows some of the possible configurations corresponding to the lower-order choices of indexes t, m and k in accordance to the rules (6):

The last column of Table 1 shows the respective sum of all the coupling angles of the band splitter/combiner 100 at the third wavelength λ3_0p within the first optical band, according to the second relation of equations (6). The sum is a good indication of the total length of the optical splitters 106, 107, 108, 109 and 110 in the particular case of PLC optical splitters. The last column is advantageously sorted in ascending order. The last column is particularly useful when optimizing the splitter/combiner 100, in that it shows how a choice of t, m and k corresponding to a given value of 2θ_A+θc+2θ_B is preferred over a choice of t, m and k corresponding to a higher value of 2θ_A+θc +2Θ_B . In fact, a PLC splitter/combiner comprising longer optical couplers will occupy more space on the substrate of fabrication, thus reducing the yield of the manufacturing process. Moreover, longer optical couplers will generally lead to narrower rejection bands and to higher sensitivity of the overall spectral response of the respective splitter/combiner with respect to small fluctuations of the structural parameters arising from the manufacturing process. This is especially true in the context of high index contrast PLC splitters/combiners, wherein the tolerance of the optical devices to manufacturing errors worsen with the increasing of the refractive index contrast.

It is preferable to choose the parameters t, m, k according to the first ten rows of Table 1 , which corresponds to a choice, in addition to rules (6), of m smaller than or equal to INT[(3+t)/2], wherein INT is the integer part of the argument. In other words, the coupling angles ΘA, ΘB and θc satisfy the relation 2Θ ^ (A₃₀^ )+ θ _c (A₃₀^ ) +2Θ_B (A₃₀^ )< 2π at least at a third wavelength λ3_0p within the first optical band. More preferably, in addition to rules (6), m should be advantageously chosen smaller than or equal to INT[(2+t)/2], corresponding to the first six rows of Table 1, even more preferably m should be smaller than or equal to INT[(1+t)/2], corresponding to the first three rows of Table 1. In other words, the coupling angles ΘA, ΘB and θc more preferably satisfy the relation 2 'θ_A(λ_3op )+θ_c(λ_3op )+ 2'θ_B(λ_3op)≤ 3/2π , even more preferably they satisfy the relation 2B_A (λ_3op )+θ_c (λ_3op )+ 2Θ_B (λ_3op ) ≤ π , at least at a third wavelength λ3_0p within the first optical band. The preferred lowest-order choice corresponds to the first raw of Table 1, wherein the choice (f,m,/()=(0,0,0) represents the smallest overall sum of the coupling angles 2θ_A+θc+2θ_B =π/2, at least at a wavelength within the first optical band, five time smaller than the sum of the coupling angles of, for example, the case (f,m,/()=(0,2,-1).

Referring back to Figure 3, the first optical device 100 may be a filter adapted to perform a first, relatively coarse splitting of the input optical signal Pin,i into two components at the first wavelength λi and at the second wavelength λ2, corresponding to the signals Pin,i and Pin,2. By "relatively coarse splitting" there is meant that the separation of the first and second optical bands, operated by the first optical device 100 is typically not perfect, and a certain crosstalk between the two bands exists at both of the two output ports 102 and 103 of the first device 100.

In order to reduce the cross-talk between the two bands at the output port 102 of the first optical device 100, the signal outputted therefrom is fed to the second optical device 310, which further filters out the undesired signal component, thereby providing, at the output port 345 thereof, a signal P_out,i, corresponding to the input signal Pin,i in the first optical band, which is less affected by cross-talk with the signal Pin,2 in the second optical band.

The second device 310 in Figure 3 is, as mentioned in the foregoing, an optical device according to the general scheme of Figure 5.

The optical device according to the general scheme of Figure 5, globally denoted 500, comprises at least a first input port 501 and at least a first and a second output ports 502 and 503 optically coupled to the first input port 501. In case the device 500 is made up of two optical waveguides, the output port 501 is the bar output port, whereas the output port 503 is the cross output port. The optical device 500 may optionally comprise a second input port 504. With respect to the second input port 504, bar and cross designations are reversed. Similarly to the device 100, the first input port 501 is adapted to receive an input optical radiation, denoted P'in in the drawing, including two optical bands, and the optical device 500 is adapted to split the input optical radiation P'in into two output optical radiations PW (bar output radiation) and P'cross (cross output radiation), outputting respectively from the bar port 502 and the cross port 503, and having each a respective optical power spectrum. The optical device 500 is in particular adapted to split a first and a second optical wavelength bands as defined in the foregoing, i.e. two continuous wavelength bands having equal widths, ranging from approximately 5 nm to approximately 40 nm, having respective first and second central wavelengths in a range of wavelengths from approximately 600 nm to approximately 2000 nm and being separated of an amount such that the first and second wavelength bands do not overlap.

It is pointed out that it is not essential that the first and second bands that are split by the first optical device 100 be identical to the first and second bands that are split by the second optical device 310; deviations (e.g. of about ±5 nm) in the respective central wavelengths, separation therebetween and width may be envisaged, for example in order to optimize the design of the two devices.

The optical device 500 comprises a first, a second and a third optical splitting devices, represented respectively by blocks 506, 507 and 508, optically coupled in cascade, and a first and a second optical differential delay devices, represented respectively by blocks 511 and 512, optically coupled, and intercalated to the optical splitting devices 506, 507 and 508. The optical splitting devices 506, 507 and 508 may be similar in structure to the optical splitting devices 106, 107, 108, 109 and 110 of the device depicted in Figure 1, and the same considerations made above hold; likewise, the differential delay devices 511 and 512 may be similar in structure to the differential delay devices 111, 112, 113 and 114 of the device of Figure 1, and the same considerations made above hold.

The optical splitting device 506 has an input port optically connected to the first input port 501 of the device 500. The optical splitting device 508 has a first output port optically connected to the first output port 502 of device 500 and a second output port optically connected to the second output port 503 of device 500. As indicated in Figure 5, the first and third optical splitting devices 506 and 508 have substantially the same associated coupling angle equal to about 0A'(λ). The second optical splitting device 507 has an associated coupling angle equal to about θB'(λ), as shown in Figure 5.

The first and the second optical differential delay devices 511 and 512 have a respective associated differential delay, φx'(λ), x=A, B as indicated in Figure 5. Each of the first and second optical differential delay devices 511 and 512 is as described in the foregoing in connection with the device if Figure 1. The same considerations given above for the sign of φ_x(λ), x=A, B, C, D hold for φ_x'(λ), x=A, B. As a consequence, once the first and the second branches of the first optical differential delay device 511 are arbitrarily selected, so as to determine the sign of the respective differential delay φA'(λ), the first and the second branches of the other optical differential delay device 512 are consequently determined by the optical connection between the second differential delay device 512 and the first optical differential delay device 511. More in detail, the optical splitting device 507 has a first input port optically coupled to the first output port of the first differential delay device 511 and a second input port optically coupled to the second output port of the first differential delay device 511. The second optical differential delay device 512 has its first input port optically coupled to the first (bar) output port of the preceding optical splitting device 507 and its second input port optically coupled to the second (cross) output port of the preceding optical splitting device 507. Hence the first and the second optical branches of the second optical differential delay device 512 are determined. The subsequently cascaded third optical splitting device 508 has its first input port optically coupled to the first output port of the preceding differential delay device 512 and its second input port optically coupled to the second output port of the preceding differential delay device 512.

Advantageously, each of the differential delay devices 511 and 512 is adapted to introduce a differential delay which is substantially an even multiple of π at least at a first wavelength λ'i_op within the first optical band (φ(λ'i_Op)= 2nπ, n integer) and substantially an odd multiple of π at least at a second wavelength λ'2_0p within the second optical band (φ(λ'2op)= (2n'+1)π, n' integer).

Exemplarily, in case a differential delay device comprises two optical paths which differ only in the respective lengths, in order to ensure that the differential delays at the two respective wavelengths λ'i_Op and λ'2_0p differ by an odd multiple of π, the length difference ΔL= |L2 - L11 should be according to the following expression:

AL = (2m + I)K I | β (λ\_op ) - β (λ\_op ) | ,_m integer.

Preferably, m=0 and ΔL = π / | β (V_2op ) - β (λ\_op ) | .

In a preferred configuration, the absolute values of the differential delays of the differential delay devices 511 and 512 are substantially the same (|φΑ(λ)| ≡ |φ'B(λ)|), and they may have same or opposite sign. Advantageously, the optical splitting devices 506, 507 and 508 have respective associated coupling angles 0A'(λ) and θβ'(λ) adapted to direct, in combination with the above described differential delay devices 511 and 512, more than half of the total output power of an optical radiation inputting at the first input port 501 and having any wavelength within the first optical band to one of the first and second output ports 502, 503, and more than half of the total output power of an optical radiation inputting at the first input port 501 and having any wavelength within the second optical band to the other of the first and second output ports 502, 503.

Advantageously, the coefficients 0A'(λ) and θβ'(λ) should substantially comply with the following relations, at least at a third wavelength λ'3_0p within the first optical band and at least at a fourth wavelength λ'4_0p within the second optical band (relations (7)):

where t'e {0,1} , k' is an integer and m' is a non-negative integer. Advantageously, k' and m' are selected in order to have 0A'(λ) and θβ'(λ) positive. Nominally, the first wavelength λ'i_Op coincides with the third wavelength λ'3_0p and the second wavelength λ'2_0p coincides with the fourth wavelength λ'4_0p, but wavelength deviations may occur because of the fine tuning of the overall structure, as discussed above.

In fact, when designing an optical wavelength band splitter/combiner in accordance with the present invention, including, as shown in Figure 3, a device of the type shown in Figure 1, i.e. a symmetric filter, coupled to a device according to the general scheme of Figure 5, one needs to make the pass-band and the stop-band of each output port of the device to fit with the desired optical bands. In addition, the designer needs to take into account the specific target level of crosstalk of each stop-band, which in general may be different for the first and the second optical bands. As a consequence, a fine tuning process is required to achieve the desired specifications, and it is driven by the optical band having the most stringent specification for the cross-talk. Exemplarily, one may start from an ideal condition wherein λ'i_op is equal to λ'3_0p and is at the center of the desired optical band, and λ'2_0p is equal to λ'4_0p and is at the center of the respective desired optical band. After calculating, e.g. by computer simulation, the corresponding spectral response, one may seek an optimal configuration by stepwise changing, for example, λ'i_op and λ'3_0p independently from each other; λ'2_0p and λ'4_0p move accordingly. The process ends when an optimal solution is found.

It is underlined that the combination of the optical splitting devices 506, 507 and 508 and of differential delay devices 511 and 512 to form the device of Figure 5 does not direct an optical radiation inputting at the first input port 501 and having a specific wavelength within the first optical band totally at one of the first and second output ports 502, 503 and an optical radiation inputting at the first input port 501 and having a specific wavelength within the second optical band totally in the other of the first and second output port 502, 503. This is because of the deviation of λ'i_op from λ'3o_P and λ'2o_P from λ'4_0p at the end of the fine tuning process described above. Nevertheless, the optical device 500 is adapted to direct an optical radiation inputting at the first input port 501 and having a wavelength within the first optical band substantially at one of the first and second output ports 502, 503, and an optical radiation inputting at the first input port 501 and having a wavelength within the second optical band substantially in the other of the first and second output ports 502, 503, wherein "substantially" means at least the 85% of the total power outputting from the device 400 at the respective wavelength. It can be appreciated that in relations (7) account is taken of the wavelength dependence of the splitting devices in the optical bands of interest.

The Applicant has found that with the optical device depicted in Figure 5, by suitable design it is possible to obtain an asymmetric filter, flattening the response at either one or the other of the two output ports 502, 503. In particular, in a design aiming to optimize the optical device 500 with respect to tolerance to fabrication errors, the Applicant has found that it is advantageous to choose the above discussed indexes P, m' and k' in order to keep the coupling angle values 0A'(λ) and Θβ'(λ) as low as possible. This in turn means that, according to the relation expressing the coupling angle of an optical coupler like the one shown in Figure 4 as a function of the length Lc of the coupler straight coupling region:

θ (λ) = κ (λ) [^ +δ Z(λ)] ,

wherein κ(λ) is the wavelength-dependent coupling per unit length in the straight part of the coupler and δl_(λ) is an equivalent effective interaction length accounting for the wavelength- dependent coupling contribution of the coupler input and output curved sections, also the lengths of the splitting devices 506, 507 and 508 are as low as possible; this allows reducing the wavelength dependence of the coupling angle itself. In addition, in view of a high density optical circuitry, it is advantageous to minimize the physical length of the constituent parts of the optical device. The reduced wavelength dependence of the splitting devices is particularly critical in high index contrast technology.

In particular, the Applicant has found that the parameters t', m' and k' should advantageously satisfy the following selection rules:

[W > I it¹1 for f= 0

Jw'≥ JtM-I for t= \ AND Jt¹ > 0

L'≥ -it¹ for f= \ AND Jt¹ < 0

In particular, the Applicant has found that the set of values for the three parameters t\ m' and k' producing the shortest configuration for the optical device 500 is f = m' = k' = 0.

As in the case of the device 100 of Figure 1, reducing the lengths of the optical splitters 506, 507 and 508 is particularly advantageous in PLC technology; in fact, a PLC splitter/combiner comprising longer optical couplers occupies more substrate space, thus reducing the yield of the manufacturing process. Moreover, longer optical couplers generally lead to narrower rejection bands and to higher sensitivity of the overall spectral response of the respective splitter/combiner with respect to small fluctuations of the structural parameters arising from the manufacturing process. This is especially true in the context of high index contrast PLC splitters/combiners, wherein the tolerance of the optical devices to manufacturing errors worsen with the increasing of the refractive index contrast.

The Applicant has found that by choosing t'= m' = k' = 0, the flattening of the response of the device 500 at the bar port 502 can be achieved by causing the two differential delay devices 511 and 512 introduce a phase shift of the same sign, whereas the flattening of the response at the cross port 503 can be achieved by making the two differential delay devices 511 and 512 introduce shifts having opposite sign.

In operation, an input optical signal P'm including signals at a first and second wavelengths λi and λi respectively within the first and the second optical bands is fed to the input port 501 of the optical device 500. The optical device 500 splits the optical radiation into two optical radiations PW and P'cross outputting respectively from the first output port 502 and the second output port 503. More particularly, at the bar port 502 the output signal PW comprises most of the optical power at one of the first and second wavelengths λi and λi, whereas at the cross port 503 the output signal P'cross comprises most of the optical power at the other of the first and second wavelengths λi and λi. In case a predetermined level of cross-talk is set, the output signal at the respective output port comprises an optical power at the considered wavelength below the specified cross-talk level.

As mentioned in the foregoing, the first device 100 and the second device 310 of Figure 3 can be realized in the form of PLC optical devices. A possible layout of the device of Figure 1 realized as a PLC device is disclosed in the cited International application, and is also depicted in Figure 11.

One or both the optical devices 100 and 500 may advantageously be formed on a (preferably common) substrate (520 in Figure 5), such as for example a silicon or oxide substrate. In a more preferred configuration, the optical devices 100 and 500 are PLC optical devices, more preferably PLC optical devices each comprising a pair of optical waveguides, even more preferably high density PLC optical devices. Advantageously, the core-to-cladding refractive index contrast of the waveguides is greater than about 1%, preferably equal to or greater than about 2%, preferably equal to approximately 2.5 % (at 1550 nm). Advantageously, the refractive index contrast of the waveguides is lower than or equal to about 4.5%, preferably lower than or equal to about 3%. The above-mentioned waveguide core-to-cladding index contrast may be obtained with a convenient doping of boron and phosphorus. The Applicant has found that these values of index contrast are a good trade-off between the degree of on-wafer integration (which increases with the increase in the refractive index) and the tolerance to manufacturing errors (which increases with the decrease in the refractive index), and are in particular an optimal choice in order to guarantee bending radii of the order of 1.5 mm with acceptable bending loss. An advantage of this solution may be the possibility to achieve smaller devices, i.e. a higher density on a wafer. On the other hand, high wafer density does not necessarily mean higher yields, because smaller features and higher index contrast, in general, worsen both fabrication and coupling tolerances. Figure 6 is a schematic top-plan view of an exemplary PLC optical device 600 in accordance with the general scheme of Figure 5 described in the foregoing, in an embodiment of the present invention.

The optical device 600 comprises two substantially identical planar optical waveguides

621 and 622 which are advantageously put in close proximity to each other at three different locations, in order to form three optical couplers 606, 607 and 608, corresponding to the optical splitting devices 506, 507 and 508 of Figure 5. Two optical differential delay devices 611 and 612, corresponding to the devices 511 and 512 of Figure 5, are intercalated to the couplers 606, 607 and 608. The two optical-path lengths of each of the optical waveguides 621 and 622 in a region between two successive couplers 606, 607 and 608 are made different from each other in order to form the optical differential delay devices 611 and 612. In particular, the first optical differential delay device 611 comprises a first and a second (arbitrarily selected) optical arms 611' and 611", formed on the waveguides 621 and 622, respectively. The second optical arm 611" is, in the shown example, longer than the first optical arm 611' and hence, according to the convention introduced above, the associated differential delay ψΑ is positive in sign.

The sign of the differential delays associated with the remaining optical differential delay device 612 is determined by its connections, as described in greater detail above. Specifically, the optical arm of the optical differential delay device 612 laying on the first optical waveguide 621 is regarded as the first optical arm, and the optical arm of the optical differential delay device 612 laying on the second optical waveguide 622 is regarded as the second optical arm. For example, the lower arm 612" of the second optical differential delay device 612 is regarded as the second arm, being optically connected to the second output port of the optical splitter 607, which in turn is the cross output port of the optical splitter 607 connected to the second arm 611" of the first differential delay device 611. In the exemplary embodiment of Figure 6, the first and the second optical differential delay devices 611 and 612 have respective associated differential delays φΑ and Φ'B which are equal and opposite in sign, the first optical differential delay device 611 having a positive differential delay φΑ, whereas the second optical differential delay device 612 has a negative differential delay -φΑ. Thus, as mentioned in the foregoing, the device of Figure 6 is adapted to flatten the response at the cross port 602.

The optical waveguides 621 and 622 may be realized on a flat substrate 630. The overall length Lt of device 600 may be for example equal to about 3800 μm, and the overall width Wt may be for example equal to about 2000 μm. One end of the optical waveguide 621 forms a first input port 601 , adapted to receive the input optical signal P'in, and the other end of the optical waveguide 621 forms a bar output port 602 with respect to the input port 601, adapted to emit the bar optical signal PW. One end of the optical waveguide 622 forms a cross-output port 603 adapted to emit the cross optical signal P'cross; the other end of the optical waveguide 622 forms an optional, second input port 604. With respect to this second input port, bar and cross designations are reversed.

By way of example, the two waveguides 621 and 622 may be buried, ridge or rib waveguides on a substrate material, or they may be photonic crystal waveguides on a substrate material. Advantageously, the core-to-cladding structure of the two waveguides 621 and 622 may be made of a combination of materials such as SiU2, Ge:Siθ2, BPSG, BPTEOS, GeBSG, SiON, SΪ3N4, Si, SiGe, Al_xGai_-xAs, ln_xGai_-xAsP, Cd_xZni_-xTe, GaN or the like, or polymeric materials such as polyimides, acrylates, polycarbonates, silicones, benzocyclobutene (BCB), epoxy resins or the like. In particular, the optical device 600 may be fabricated using silica on silicon technology, with a waveguide core layer made of germanium-doped silica, formed for example by means of PECVD (Plasma Enhanced Chemical Vapor Deposition) process; the lower cladding may consist of silica, obtained by means of superficial thermal oxidation of a silicon substrate. The upper cladding layer may be a BPTEOS glass (boron and phosphorus doped silica from liquid precursors), deposited chemically at air pressure condition through an APCVD (Atmospheric Pressure CVD) process.

Figure 7 is a schematic cross-sectional view of an exemplary configuration of the waveguide 621 taken along the line VII-VII in Figure 6, wherein the same reference numerals are used where appropriate. The planar optical waveguide 621 is for example an optical waveguide buried into a silica (Siθ2) layer 740 on the silicon substrate 630, such as for example a silicon wafer having a thickness T of 600 μm (Figure 7 may not be to scale).

The waveguide 621 has preferably a square core having W x W cross-sectional side of, for example, 2.6 μm x 2.6 μm, in order to satisfy the requirement of monomodality in the lower end of the low-wavelength band (e.g. at 1490 nm-10 nm=1480 nm) for the chosen index contrast. Quotes A and B in Figure 7 are for example 5 μm and 10 μm, respectively. Figure 8 shows a schematic (not to scale) diagram of an exemplary optical directional coupler 700. In the case of the directional coupler of the type shown in Figure 8, θfλj also depends on the length Lx of the straight coupling region, on the wall-to-wall distance S2 between the waveguides at the coupling region over the length Lx, on the shape, width and depth of the waveguides, on the refractive index difference between waveguide core and cladding, on the geometry of the input and output curves, etc.

The optical splitting devices 506, 507 and 508 of Figure 5 are preferably based on the same design scheme as the optical coupler 800 of Figure 8, provided however that a proper length L_x is chosen for the respective straight coupling region length.

Similar considerations apply to the device 100 of Figure 1, and, in particular, the optical splitting devices 106, 107, 108, 109 and 110 may be based on the design scheme of Figures 7 and 8. In a possible embodiment of the optical device 500, the first and the third optical couplers 606 and 608 forming the first and the third optical splitting devices 506 and 508 are advantageously substantially identical and have advantageously the same coupling lengths LA. The second optical coupler 607, forming the second optical splitting device 507, has a coupling length LB. The straight coupling lengths L_x, x=A, B, the wall-to-wall waveguide separations S1 and

S2, the radius R and the angle α of the input and output curves of the optical couplers are selected in order to direct, in combination with the optical differential delay devices 511 and 512, more than half of the output power of a first optical radiation inputting at the first input port 501 and having any wavelength within the first optical band to either one of the first and second output ports 502, 503, and more than half of the output power of a second optical radiation inputting at the first input port 501 and having any wavelength within the second optical band to the other of said first and second output ports 502, 503.

In the exemplary case of optical bands respectively centered at 1560 nm and 1490 nm and approximately 10 nm wide, refractive index difference equal to about 2.5% and size WxW of about 2.6x2.6 μm, a possible set of values for the above parameters is the following: S1 = 29.5 μm,

S2 = 2.6μm, R = 2000μm, α = 4.7° ; the coupling lengths L_x, x=A, B, depend on the desired splitting ratio for each coupler.

Figure 9 shows a schematic (not to scale) diagram of an exemplary design of the generic optical differential delay device 611 or 612, particularly the optical differential delay device 612. Both of the optical differential delay devices 611 and 612 are preferably based on the same design scheme of Figure 9, provided however that the respective longer optical-path lays on the proper optical waveguide 621 or 622, in order to achieve the proper sign of the respective differential delay.

The optical differential delay device 612 of Figure 9 is obtained by shaping the two optical waveguides 621 and 622 in the form of two arcs having the same angle β and different radii n and r2 in order to provide the first arm 612' with an extra length ΔL with respect to the second arm

612". The angle β and the radii n and Ti are conveniently selected in order to achieve a differential delay which is substantially an even multiple of π at a first wavelength (λ'i_Op) within the first optical band and substantially an odd multiple of π at a second wavelength (λ'2_0p) within the second optical band. Exemplarily, the angle β and the radii n and Ti (measured with respect to the waveguide axis) are 20.935°, 1700 μm and 1732.1 μm, respectively. Accordingly, the difference ΔL between the lengths of the second and the first arm 612", 612' is set equal to about 11.729 μm in order to achieve a differential delay, in absolute value, of about 22π at a wavelength λ'i_op equal to 1556 nm and of about 23π at a wavelength λ'2_0p equal to 1490nm.

Similar considerations apply to the device of Figure 1, and the differential delay devices thereof may be based on the layout scheme of Figure 9.

For example, let it be assumed that the layout scheme of Figure 6 is used to make an optical device adapted to split two optical bands as defined above, with central wavelengths of approximately 1490 nm and 1560 nm, and featuring a flat response, at the bar output port 603, for a band centered at the 1560 nm wavelength (a device hereinafter referred to as a "1560 nm flat filter"), wherein, for the purposes of the present invention, by "flat" there is meant that the in-band ripple in the 1560 nm band is less than the in-band ripple of the 1490 nm band. The length of the straight coupling region of the first and third optical couplers 606 and 608 is chosen equal to about 88.201 μm, the extra length of the waveguide 622 compared to the waveguide 621 in the differential delay device 611 is chosen equal to approximately 278.131 μm, the extra length of the waveguide 621 compared to the waveguide 622 in the differential delay device 612 is chosen equal to approximately 11.729 μm, and the length of the straight coupling region of the second optical coupler 607 is chosen equal to approximately 278.131 μm.

Figure 10A is a diagram showing the ideal simulated response (transmittance in dB - on the ordinate - vs. wavelength in μm - on the abscissa) of the 1560 nm flat filter discussed above. It can be appreciated that the response at the cross output port 503, represented by the dotted curve R1 (corresponding to the wavelength band centered at 1560 nm) is flattened (the ripple being less than 0.1 dB), whereas the response at the bar output port 502, represented by curve R2 (corresponding to the wavelength band centered on 1490 nm) is scarcely flattened (the ripple is approximately equal to 0.5 dB). Figure 10B is the same diagram as that of Figure 10A with broader ordinate axis. Figure 10B shows the simulated value of isolation between the two bands (centered on 1490 nm and 1560 nm, respectively) at the cross output port 503 (dotted curve R3) and at the bar output port 502 (curve R4). It can be appreciated that the isolation I2-1 between the two channels at the bar port 502 is significantly higher than the isolation I1-2 at the cross port 503, so the difference between the isolation values exceeds 5 dB, and the device exhibits an asymmetric response. In other words, the output port exhibiting the higher in-band ripple, in the considered example the bar port 502, is also the output port at which the higher isolation between the two bands is achieved. In an embodiment of the present invention, the second optical device 310 of Figure 3 consists of a device of the type shown in Figure 5, designed to have a flat response in the band being the stop-band for the output port 102 of the device 100.

Figure 11 shows schematically a possible physical layout of the cascade of the two devices 100 and 500 of Figures 1 and 5, forming the optical band splitter 300 of Figure 3, particularly designed to form a symmetric filter in respect of two bands centered around 1490 nm and 1555 nm (a "symmetric 1490 nm/1555 nm" filter), and of an asymmetric filter designed to split these two bands, flattening (at the output port 502) the band centered on 1555 nm (a "1555 nm flat" filter). The two devices, denoted globally 1105 and 1110, are integrated on a common substrate 1115, and are realized as a PLC optical device, as discussed above. The geometrical dimensions are the following: L1_A = 17.792 mm, L2_A = 86.449 mm, L3A = 2 * L1_A = 35.585 mm; L1_B = 88.201 mm, L2B = 278.131 mm; ΔLA = ΔLB = 11.729 mm. It is pointed out that the extra length values are the same for both the two devices 1105 and 1110, because the same spectral periodicity is required. It can be readily appreciated that the dimensions (Ls, Ws) of the symmetric 1490 nm/1555 nm filter 1105 are significantly greater than those (L1555, W1555) of the 1555 nm flat filter 1110, due to the fact that only three cascaded optical couplers, and two differential delay devices are used in the latter, compared to the five couplers and three differential delay devices used in the former. For example, in a practical implementation of the invention, the length Ls = 5825 mm, the width Ws = 2020 mm, the length L1555 = 3760 mm, the width W1555 = 630 mm. In this example, the resulting space saving compared to the cascade of two symmetric filters of the type shown in Figure 1, as proposed in the cited prior art International application, is relevant, and is equal to approximately 2065 mm on the total length, and approximately 1390 mm on the total width. Also, avoiding two directional couplers means avoiding extra losses due to bends and curvature discontinuities. Figure 12 plots (transmittance in dB - on the ordinate - vs. wavelength in μm - on the abscissa) experimental data related to the measured spectral response of the device of Figure 11, compared to a comparative prior art example based on a cascade of two 1490 nm/1555 nm symmetric filters of the type shown in Figure 1. In Figure 12, the black dots (interpolated by curve R12a) relate to the measured TE (Transverse Electric) field, and the black squares (interpolated by curve R12b) relate to the measured TM (Transverse Magnetic) field in a cascade of two symmetric 1490 nm/1555 nm filters of the type shown in Figure 1, as proposed in the cited International application; the empty dots (interpolated by curve R12c) and the empty squares (interpolated by curve R12d) relate respectively to the measured TE and TM fields in a cascade of one symmetric1490 nm/1555 nm filter and one 1555 nm flat filter according to the general scheme of the present invention shown in Figure 3. Looking at the enlarged detail in Figure 12, it can be appreciated that, despite the higher in-band ripple in the 1490 band of the device 1110, the measured in-band losses are even lower in the case of the cascade of a symmetric 1490 nm/1555 nm filter and a 1555 nm flat filter, except at the band edges: this may be due to the fact that only eight, instead of ten couplers are used overall, which brings about a lower extra loss. Thus, cascading one symmetric 1490 nm/1555 nm filter and one 1555 nm flat filter seems to represent an excellent trade-off between optical performances and device dimensions, because in addition to featuring a lower on-wafer occupation and similar in-band losses compared to the cascade of two symmetrical filters, the isolation value is even higher.

Designing an optical device having the general scheme of Figure 5 and featuring a flat response for a specified wavelength band is a matter of appropriate choice of the device parameters, particularly layout parameters like the bending radii of the couplers, the coupling lengths, the extra lengths for the differential delay devices; for example, like the discussed selection of parameters is adapted to give the 1555 nm flat filter discussed above, a "1490 nm flat" filter can be obtained based on the general scheme of Figure 5, i.e. an optical device adapted to split two optical bands as defined above, with central wavelengths of approximately 1490 nm and 1555 nm, and featuring a flat response at the cross port for the wavelength band around 1490 nm. The diagram of Figure 13 plots (transmittance in dB - on the ordinate - vs. wavelength in μm - on the abscissa) the isolation between the two channels of one such device; by comparison with the diagrams of Figures 1OA and 1OB, it can be appreciated that although the response at the cross port 503 (dotted curve R5) in the 1490 nm band has a lower ripple compared to the response in the same, 1490 nm band exhibited by the 1555 nm flat filter of Figures 10A and 10B (less than 0.1 dB of in-band ripple), the isolation I2-1 of the band of interest with respect to the other band is worse (approximately -11 dB instead of -28 dB).

It is to be underlined that a skilled person could be induced to connect, at the output port 102 of the first device 100 of Figure 3, i.e. at the port that passes one of the first and second optical bands and stops the other band, an asymmetric filter that flattens the said one band; for example, in case the first device 100 is designed to pass, at the port 102, a signal in the band centered on 1490 nm, stopping a signal in the band centered on 1555 nm, the skilled person might considered cascading to the port 102 of the first device 100 a 1490 nm flat filter; this solution might prejudicially be regarded as preferred, because the 1490 nm flat filter flattens (i.e. has lower in-band ripple) the band of interest (in this example, 1490 ±10 nm). Contrary to this, in view of the description above, such a solution reveals to be less performing, because, as mentioned, with the general scheme of Figure 5 it is possible to obtain filters that exhibit a higher isolation at the port affected by a higher ripple. In other words, better performance are achieved putting, in cascade to the symmetric 1490 nm/1555 nm filter, a filter according to the general scheme of Figure 5 and designed to flatten the other band (1555 nm).

Several embodiments are possible falling within the device scheme of Figure 3.

In particular, in Figures 14, 16 and 18 three invention embodiments are depicted wherein, in addition to the first and second devices 100 and 310, a third optical device is provided, denoted 1405, having an input port 1410 coupled to the output port 103 of the first device 100, and an output port 1415. The signal made available at the second output port 103 of the first optical device 100 is fed to the third optical device 1405, adapted to further filter out the undesired component, thereby providing, at an output port 1415 thereof, a signal less affected by cross-talk with the signal in the other band. The device 1405 may either be an asymmetric filter according to the general scheme of Figure 5 and designed to flatten the response to the other band (e.g., a

1490 nm flat filter), or a symmetric one of the type shown in Figure 1 (e.g., a symmetric 1490 nm/1555 nm filter).

The signal outputting from the cross output port 103 of the first device 100 is fed to the input port 1410 of the third device 1405 and is then taken from the output port 1415 thereof (corresponding to either the bar output port 102 of the device of Figure 1, or to the output port 502 of the device of Figure 5).

In greater detail, in Figure 14 an embodiment 300b of the device 300 of Figure 3 is shown, wherein to the first optical device 100 two devices 310-1b and 310-2b are cascaded: the device 310-1 b is an asymmetric filter according to the general scheme of Figure 5 and designed to split the first and the second bands defined in the foregoing, and featuring (at an output port thereof) a flat response in respect of the second band (e.g., a 1555 nm flat filter), whereas the device 310-2b is an asymmetric filter according to the general scheme of Figure 5, designed to split the first and the second bands defined in the foregoing, and featuring (at an output port thereof) a flat response in respect of the first band (e.g., a 1490 nm flat filter). The input signal Pi_n, composed of the signals Pin,i and of the signal Pin,2, is fed to the input port 101 of the first device 100; the signal taken from the bar output port 102 of the first device 100 is fed to an input port 335b of the asymmetric filter 310-1 b (corresponding to the input port 501 of Figure 5), and taken from the bar output port 1420 thereof (corresponding to the output port 502 of Figureδ); the signal is then fed to an input port 1420 of the asymmetric filter 310-2b (corresponding to the input port 501 of Figure 5), and the signal P_out,i is finally taken from the bar output port 345b of the asymmetric filter 310-2b (corresponding to the output port 302 of Figure 5). This allows increasing the optical isolation of the digital channel with respect to the analog channel, and to increase the on-wafer yield.

The curves R7 and R8 plotted in Figure 15 (transmittance in dB - on the ordinate - vs. wavelength in μm - on the abscissa) are the results of experimental measures conducted by the Applicant on a device according to Figure 14, designed to split a first and a second band centered on about 1490 nm and 1555 nm. It is observed that the Insertion Losses (IL) include two contributes of coupling losses between waveguide and single-mode fiber and extra propagation losses due to the longer waveguide layout designed for device testing. The curves on the left side of the diagram of Figure 15 relate to the measured response of the device in respect of a third wavelength band, centered on about 1310 nm, and the isolation I1310-1490 and I1310-1555 with respect to the first and second bands centered on 1490 and 1555 are indicated, as will be explained with reference to Fig.20.

The embodiment of Figure 16 is similar to that of Figure 14, with two devices 310-1c and 310-2c cascaded to the bar output port 102 of the first device 100, but the asymmetric filter 310-2a of Figure 14 has been replaced by an asymmetric filter 310-2c designed to split the first and the second bands defined in the foregoing, and featuring (at an output port thereof) a flat response in respect of the second band [i.e., in the above considered example, two 1555 nm flat filters have been cascaded along the digital signal path); the signal outputting from the bar output port 102 of the first device 100, fed to the input port 335c of the first asymmetric filter 310-1c, outputted from the bar output port 1605 thereof, fed to the input port 1610 of the second asymmetric filter 310-2c, is finally taken from the bar output port 345c of the second asymmetric filter 310-2c.

The curves R9 and R10 plotted in Figure 17 are the results of experimental measures conducted by the Applicant on a device according to Figure 16, designed to split a first and a second bands centered on about 1490 nm and 1555 nm. It can be appreciated that, compared to the embodiment of Figure 14, higher isolation values on the band centered on 1490 nm can be obtained, but the in-band ripple is higher. As for Figure 15, the curves on the left side of the diagram of Figure 17 relate to the measured response of the device in respect of the third wavelength band, centered on about 1310 nm, and the isolation I1310-1490 and I1310-1555 with respect to the first and second bands centered on 1490 nm and 1555 nm are indicated.

In Figure 18 a still further embodiment 30Od of the device 300 of Figure 3 is shown, wherein two devices 310-1 d and 310-2d are cascaded (to the bar output port 102) of the first device 100: the device 310-1 d is a symmetric filter like the first device 100 (e.g., a symmetric 1490 nm/1555 nm filter), whereas the device 310-2d is an asymmetric filter according to the general scheme of Figure 5 designed to split the first and the second bands defined in the foregoing, and featuring (at an output port thereof) a flat response in respect of the second band (i.e., in the above considered example, a 1555 nm flat filter). The signal outputted from the bar output port 102 of the first device 100, is fed to the input port 335d of the symmetric filter 310-1d, and taken from the bar output port 1805 thereof; the signal is then fed to input port 1810 of the asymmetric filter 310-2d, and the signal P_out,2 is finally taken from the bar output port 345d of the asymmetric filter 310-2d.

The curves R11 and R12 plotted in Figure 19 are the results of experimental measures conducted by the Applicant on a device according to Figure 18, designed to split a first and a second bands centered on about 1490 nm and 1555 nm. As for Figures 15 and 17, the curves on the left side of the diagram of Figure 19 relate to the measured response of the device in respect of the third wavelength band, centered on about 1310 nm, and the isolation I1310-1490 and I1310-1555 with respect to the first and second bands centered on 1490 nm and 1555 nm are indicated.

From the discussion of the embodiment depicted in Figure 19, it can be appreciated that the good performance, especially in terms of isolation between the bands, obtained by cascading a symmetric filter and an asymmetric, filter can be achieved even if these two filters are not consecutive one to another, and one or more other optical devices are inserted therebetween.

Also, the order of the devices 310-1 b and 310-2b in Figure 15 and of the devices 310-1 d and 310-2d in Figure 19 may be reversed. It is pointed out that, similarly to the path of the signal in the first band, two or more filters can be cascaded along the path of the signal in the other band for achieving the desired performance, e.g. in terms of isolation. In particular, the device 1405 may be, as already mentioned, either a symmetric filter, or an asymmetric filter flattening the band stopped at the cross port 103 of the first device 100, or an asymmetric filter flattening the band passed at the cross port 103, or a cascade of two or more of these components.

The optical device 300 may be used in a PON, particularly a GPON compliant with the ITU-T specifications, and in this case the first and the second optical wavelength bands contain the wavelengths of 1490 nm and 1555 nm, respectively or in inverse order. The optical device 300 is in this case adapted to receive the input optical signal Pi_n comprising a downstream analog TV and a digital data signal, and made up for example of the two channels at approximately 1490 nm (1490 nm ± 10 nm - the signal denoted P1490 in the drawing) and at approximately 1555 nm (1555 ± 5 nm - the signal denoted P1555 in the drawing), and to split the two channels, making them available at the output ports 330 and 345, respectively.

For example, referring to the diagrams of the measured response of Figures 15, 17 and

19, it can be appreciated that in the devices of Figures 14, 16 and 18 every ITU - T isolation requirement is fulfilled. Also, Polarization Dependent Loss (PDL) requirement is fulfilled (the ITU-T specifications impose PDL < 0.5 dB), being in-band polarization behavior negligible (no differences between the TE and TM polarization modes in terms of maximum in-band losses are observed).

In Figure 20 a schematic diagram of an optical device 2100 is depicted, making use of the optical device 300 according to the present invention. The optical device 2100 includes a first device 2105, with cascaded the device 300 of Figure 3, realized for example according to any one of the embodiments described above.

The device 2100 may be used as an Optical Network Unit (ONU) in a PON, i.e. as a terminal apparatus of an FTTP network, particularly as a triplexer in triple-play networks, to be installed at the customer side. More in detail, the optical device 2100 comprises the optical band splitter/combiner device

300 of Figure 3, for splitting/combining a first and a second optical band, like for example the 1490 ± 10 nm band and the 1555 ± 10 nm.

The optical band splitter/combiner 300 has the input port 101 adapted to receive a signal composed by the P1490 signal and the P1555 signal, to separate the two components and to make each of them available at the output ports 345 and 103 (or 1415), respectively. The optical device 2100 may also comprise a first optical receiver 2110, adapted to receive a first optical wavelength λi within the first optical band (e.g. near 1490 nm), and optically connected, for example by means of a waveguide, to the output port 345 of the optical band splitter/combiner 300, and a second optical receiver 2115, adapted to receive a second optical wavelength λ2 (e.g. near 1555 nm) within the second optical band and optically connected, for example by means of a waveguide, to the output port 103 (or 1415) of the optical band splitter/combiner 300. The optical device 2100 further comprises an additional optical band splitter/combiner 2105, adapted to split/combine a third optical band from/to the first and the second optical band. The device 2105 has an output port 2130 that is connected to the input port 101 of the optical band splitter/combiner 300, for example through an optical waveguide 2120. The optical device 2100 may also comprise an optical transmitter 2125, adapted to emit an optical radiation having a third wavelength λ3 within the third optical band (e.g. near 1310 nm), and optically connected to a port 2135 of the optical band splitter/combiner 2105, for example through an optical waveguide. The optical device 2100 has a port 2140 adapted to be connected to an optical transmission line such as for example a fiber optic transmission line. In an embodiment of the present invention, the optical band splitter/combiner 2100 is realized according to the general scheme of Figure 1, i.e. it is a symmetric filter, and is designed in such a way to be adapted to split/combine optical signals in a wavelengths band around 1300 nm and optical signals in a wavelengths band around 1500 nm. The layout of the optical band splitter/combiner 2100 may be similar to the layout of the symmetric 1490 nm/1555 nm filter of Figure 1, depicted in Figure 11, with the following dimensions: L1 = 31.552 mm, L2 = 110.797 mm, L3 = 2 * L1 = 63.104 mm, ΔL = 3.138 mm. It can be appreciated that the extra length is in this case shorter than the one used in the 1490 nm/1555 nm symmetric filter, because of the higher channel spacing required by a 1300 nm/1500 nm splitter/combiner. The simulated spectral response of this filter is shown in Figure 21: it can be appreciated that, on the bar port 2130 (curve R13), the filter 2105 provides more than 100 nm flat transmittance band around λ = 1310 nm, that is the desired pass band width for the upstream data signal. At the same time, the response at the cross port 2135 (curve R14) exhibits a flat high transmittance over a wide band (of about 80 nm) centered at 1520 nm, that is, the downstream pass-band.

In use, an optical signal P₁₃₁₀ having the third wavelength λ3, emitted by the optical transmitter 2125, is directed into the port 2135 of the device 2105. The signal P₁₃₁₀ is then mostly directed to the optical port 2140 of the device 2105, in order to be fed into the optical transmission line in an upstream direction (signal P_up). An optical signal P_down comprising a signal P₁₄₉₀ at a first wavelength λi within the band 1490 ± 10 nm, and a signal P₁₅₅₅ at a second wavelength λ2 within the band 1555 ± 5 nm, propagating through the optical transmission line in a downstream direction, is fed to the port 2140 of the device 2105. the signal P_down is then mostly directed to the output port 2130 of the device 2105. The optical signal P_down is then fed into the input port 101 of the device 300, which splits the optical signal P_down into the two component optical signals P₁₄₉₀ and P₁₅₅₅, outputting them from the two output ports 345 and 103 (or 1415), respectively. Each of the two optical radiations is subsequently received by the respective optical receiver 2110 and 2115.

The optical devices in accordance with the present invention find particularly useful applications in optical networks adapted to distribute telecommunication services to a plurality of customers. For example, optical networks may be triple play networks or passive optical networks or fiber-to-the-premises networks or, more generally, access networks. The services are delivered using at least a first and a second signal having respectively a first and a second wavelength within respectively a first and a second optical band. Exemplarily, the optical network comprises a central station, a plurality of terminal stations, and a plurality of optical links connecting each terminal station to the central station. Each optical link may comprise cabled fibers and may include passive or active optical devices apt to branch, add, route, amplify, attenuate or switch the optical signals propagating through the link. The terminal station may be of the kind described in Figure 20.

Claims

1. An optical device (300) adapted to split/combine a first and a second wavelength bands, the first wavelength band having a first central wavelength, and the second wavelength band having a second central wavelength, wherein: - the first and second central wavelengths are in a range of wavelengths from approximately 600 nm to approximately 2000 nm;

- a separation between the first and the second central wavelengths is such that the first and second wavelength bands do not overlap, the optical device comprising a first optical band splitter (100) comprising an input port (101), a first output port (102) and a second output port (103), and five optical couplers (106-109) optically coupled in cascade between the input port and the first and second output ports, and intercalated by four optical differential delay devices (111-114), wherein the first optical band splitter is adapted to split the first and the second wavelength bands, when input at the input port, and to make them available respectively at the first and second output ports, the first optical band splitter having a respective power response at the first and second output ports wherein the central wavelength of each wavelength band corresponds to the maximum of the power response at the respective output port and an isolation, at the first output port, of the second wavelength band with respect to the first wavelength band that is essentially equal to an isolation, at the second output port, of the first wavelength band with respect to the second wavelength band, characterized by further comprising: a second optical band splitter (310;310a;310-1b;310-1c;310-2d) comprising an input port 335;335a-c;1910) coupled to the first output port of the first optical band splitter, a first output port (345;345a;1505;1705;345d) and a second output port, and three optical couplers (506-508) optically coupled in cascade between the input port and the first and second output ports of the second optical band splitter, and intercalated by two optical differential delay devices (511,512), wherein the second optical band splitter is adapted to split the first and the second wavelength bands, when input at the input port of the second optical band splitter, and to make them available respectively at the first and second output ports thereof, the second optical band splitter having a respective power response at the first and second output ports thereof, wherein the central wavelength of each wavelength band corresponds to the maximum of the amplitude response at the respective output port of the second optical band splitter, and wherein an isolation, at the first output port of the second optical band splitter, of the second wavelength band with respect to the first wavelength band is higher than an isolation at the second output port of the second optical band splitter of the first wavelength band with respect to the second wavelength band.

2. The optical device of claim 1 , wherein: the first optical band splitter has a difference between the isolation, at the first output port thereof, of the second wavelength band with respect to the first wavelength band and the isolation, at the second output port of the first optical band splitter, of the first wavelength band with respect to the second wavelength band which is less than approximately 5 db.

3. The optical device of claim 1 or 2, wherein: the second optical band splitter has a difference between the isolation, at the first output port thereof, of the second wavelength band with respect to the first wavelength band and the isolation, at the second output port of the second optical band splitter, of the first wavelength band with respect to the second wavelength band which is higher than approximately 5 db.

4. The optical device of claim 1 , 2 or 3, wherein an in-band ripple, at the first output port of the second optical band splitter, is higher than an in-band ripple at the second output port of the second optical band splitter.

5. The optical device of any one of the preceding claims, wherein said five optical couplers optically coupled in cascade of the first optical band splitter comprise a first (106), a second (107), a third (108), a fourth (109) and a fifth (110) optical couplers, the first and the fifth optical couplers having an associated first coupling angle Θ_A, the second and the fourth optical couplers having an associated second coupling angle Θ_B, and the third optical coupler having an associated third coupling angle θ_c, the first, second and third coupling angles satisfying, at least at a wavelength λ_3op within the first optical band, the following relation:

2θAJ₊θ_c(λ₃J₊2θ_β(λ₃J^<2π .

6. The optical device of claim 5 wherein the first, second and third coupling angles satisfy, at least at said wavelength λ3_0p within the first optical band, the following relation:

.

7. The optical device of claim 5 wherein the first, second and third coupling angles satisfy, at least at said wavelength λ3_0p within the first optical band, the following relation:

2θAJ+θ_c(λ₃J+2θ_β(λ₃J<π ■

8. The optical device of claim 5 wherein the first, second and third coupling angles satisfy, at least at said wavelength λ3_0p within the first optical band, the following relation:

2θA₀>θ_c(λ₃J+2θ_s(λ₃J = !.

9. The optical device of any of the preceding claims, wherein said three optical couplers optically coupled in cascade of the second optical band splitter comprise a sixth (506), a seventh (507) and an eighth (508) optical couplers, the sixth and the eighth optical couplers having an associated fourth coupling angle Θ'_A, the seventh optical coupler having an associated fifth coupling angle Θ'_B, the fourth and the fifth coupling angles satisfying, at least at a wavelength λ'_3θp within the first optical band, the following relation:

2θ;(λ^' _3o,)+θ_β(λ^' _3o,) = | .

10. The optical device of any of the preceding claims, wherein said four optical differential delay devices of the first optical band splitter comprise a first (111), a second (112), a third (113) and a fourth (114) differential delay devices optically coupled in cascade, the first and the second optical differential delay devices having differential delays equal in sign, the third and fourth differential delay devices having respective differential delays equal in sign and opposite in sign to the differential delay of the first and second differential delay device.

11. The optical device of claim 10, wherein said two optical differential delay devices of the second optical band splitter comprises a fifth (511) and a sixth (512) differential delay devices optically coupled in cascade, the fifth and sixth optical differential delay devices having differential delays opposite in sign.

12. The optical device of any of the preceding claims, wherein the optical device is a planar optical device.

13. The optical device of claim 12 as depending on claim 9 or 11, wherein the optical device comprises a first pair of optical waveguides forming the first, second, third, fourth and fifth optical couplers, and a second pair of optical waveguides forming the sixth, seventh and eighth optical couplers, the first and second pair of optical waveguides having an optical waveguide in common.

14. The optical device of claim 13 when depending on claim 11, wherein the first pair of optical waveguides forms the first, second, third and fourth optical differential delay devices, and the second pair of optical waveguides forms the fifth and sixth optical differential delay devices.

15. The optical device of claim 13 or 14, wherein the optical waveguides have an index contrast higher than about 1%.

16. The optical device of any of the preceding claims, further comprising: a third optical band splitter (1405) comprising an input port (1410), a first output port (1415) and a second output port, wherein the input port of the third optical band splitter is optically coupled to the second output port of the first optical band splitter.

17. The optical device of any of the preceding claims, further comprising a fourth optical band splitter (310-2b;310-2c) comprising an input port (1520;1710) coupled to the first output port of the second optical band splitter, a first output port (345b;345c) and a second output port.