CA2336787C - Method for polarization mode dispersion compensation, and a polarization mode dispersion compensator - Google Patents

Abstract

The polarisation transformer (TRFI) on a birefractive substrate material (SU B) has a greater number of cells (Pj), each with two mode converter electrodes (Eij, i=1, 2; j=1, 2,... n) which can be controlle d individually or individually in more than two groups in order to compensate for polarisation mode dispersion.

Description

iii H~~t~~505/PCT/DE99/01846 Description Method for polarization mode dispersion compensation, and a polarization mode dispersion compensator The invention relates to a method for polarization mode dispersion compensation as claimed in claim 1, and to a polarization mode dispersion compensator as claimed in claim 5.
Optical waveguide transmis:aion paths which are as long as possible are used in optical transmission technology. The optical waveguide~; are not completely isotropic owing to the way in which they are produced.
As a result of the long 'transmission paths, birefringence of the optical signals which are transmitted results in frequency-dependent polarization transformation - referred to a~; polarization mode dispersion PMD, or polarization dispersion for short.
As a result of the change in polarization and the different frequency-dependent propagation times resulting from this, this leads to transmitted pulses being broadened, which makes it harder to identify them at the receiving end, thus limiting the data rate which can be transmitted.
The polarization mode dispersion is also dependent on the temperature or on mechanical stress.
Adaptive PMD compensators are therefore required, which are inserted in the transmission path. The aim of a PMD
compensator is to make the polarization transmission response of the overall system comprising the transmission path and the compensator approximately (to a first or, if required, even higher order) frequency-independent at at least one optical carrier frequency in the region of the transmission band. Modulated signals can thus be transmitted without distortion.
For wavelength division multiplexing WDM, it is desirable to achieve this frequency independence for the individual transmission bands (transmission wavelengths) at least in each of the individual ~ CA 02336787 2001-O1-08 August 11, 2000 DE 009901846 GR 98 P 8057 Foreign version channels. The requirements for such a transformer/compensator are low insertion loss, compatibility with optical wavegui.des, that is to say low coupling loss and mechanical compatibility, and a polarization behavior which is frequency-dependent and is as variable as possible.
PMD emulators whose adjustment can be varied but which are nevertheless low in price and low in attenuation, and which can simulate in a significant manner the frequency-dependent polarization transmission behavior of optical waveguide distances of lengths of up to several thousand kilometers and in widely differing conditions (for example with temperature fluctuations? are required in order to develop PMD compensators and in order to check the PMD
tolerance of uncompensated transmission systems.
A "Transmission System and Receiver with Polarization Control" is known from Patent Application WO 95/34141, which uses an LiNB03 cz-ystal whose XY plane runs transversely to the propagation direction Z. The propagation speeds in the X and Y' axes are initially constant. However, they become slightly different by application of control voltages, so that the polarization can be varied. However, as a rule, polarization transformers are unsuitable for PMD
compensation since, in this case, it is necessary to correct for major propagation time differences between the individual modes. These may amount to a duration of from approximately 1/3 of a bit up t:o several bits.
Compensators whose transmission characteristics are therefore the inverse of a transmission path are suitable to compensate for PMD. Compensators are known from the literature, which are in the form of retarders/polarization rotators and are arranged between relatively strongly birefringent pieces of optical waveguide. Retarder is a. generic germ for optical transmission elements which transmit two mutually orthogonal eigen modes AMENDED SHEET

' CA 02336787 2001-O1-08 August 11, 2000 DE 009901846 GR 98 P 8057 Foreign version - 2a -attenuation, [lacunas but with phase delays which are in general different.
The strongly birefringeni: optical waveguide sections maintain or ensure two mutually orthogonal main polarizations and are thus polarization maintaining optical waveguides PMF (polarization maintaining fibers). These PMFs are strongly polarization-dispersive, that is to say different polarizations lead to widely differing propagation times. An appropriate example is described in the proceedings of the "Optical Fiber Communication Conference", 1995, OFC'95, the Optical Society of America, as article WQ2 on pages 190 to 192.
AMENDED SHEET

iii ~ 02336787 2001-O1-08 . ~. WO 00,/03505/PCT/DE99/01846 An integrated optical single-sideband modulator and phase shifter is described .in "IEEE Journal of Quantum Electronics", Volume 18, No. 4, April 1982, pages 767 to 771. This device contains, on a lithium-niobate substrate, a ground electrode which is in the form of a comb and is drawn aver the chip, and electrodes which are in the form of combs, lie in a row and whose tines are interleaved with the tines of the ground electrode, and every alternate one of which is respectively connected to a first control voltage or a second control voltage, respectively. In this polarization transformer, the TE-TM mode conversion can be preset only with ~45° linear-polarized eigen modes, or with circular-polarized eigen modes. In this case, the TE-TM phase shift element is cfioverned by the chip length and the chip temperature and. cannot be varied by an electrical voltage. Another disadvantage of this arrangement is that a preset polarization transformation is effective for only one specific optical frequency, that is to say the frequency dependence of polarization transformers cannot be preset freely.
An integrated optical polarization transformer which uses lithium niobate LiNb03 ar lithium tantalate LiTA03 as the substrate is described in "IEEE Journal of Quantum Electronics", Volume 25, No. 8, August 8, 1989, pages 1898 to 1906. This :requires only three different control voltages, one phase-shifter voltage and two mode-converter voltages in order to produce any desired polarization change. The phase-shift voltage produces a phase delay between TE (transverse electrical) and TM (transverse magnetic) waves, which are at the same time the eigen modes, but does not produce any conversion between them. One of the two mode-converter voltages produces TE;-TM mode conversion with linear polarization at ~45° elevation angles as eigen modes, and the other produces TE=TM mode conversion with circular polarization as eigen modes.
However, a predetermined polarization transformation is .v WO 00/03505/PCT/DE99/01846 - 3a -effective for only specific optical frequency. At other optical frequencies, the polarizat_Lon transformation is dependent ' CA 02336787 2001-O1-08 -~~ August 11, 2000 DE 009901846 GR 98 P 8057 Foreign version on the polarization transformation which is set for this specific optical frequency.
A TE-TM transformer with simple electrode shapes is described in the Proceedings of the Fourth European Conference on Integrated Optics ECIO 87, Glasgow, Scotland, Editors Wilkinson and Lamb, pages 115 to 118.
These known arrangements are used for polarization transformation, for example as polarization compensators and receivers. They are not planned for use as PMD compensators.
A method for PMD compensation is likewise described in Electronics Letters, February 17, 1994, Volume 30, No. 4, pages 348 and 349. In this case, a number of sections of polarization-maintaining fibers (PMF) are used as transmission elements and are connected by means of polarization transformers, with a polarization transformer with a downstream polarization-maintaining fiber being used for PMD
compensation. The attenuation which occurs may be very high owing to the spliced joints wh:Lch are required.
The connection of a PMD compensator to an optical receiver and the process of obtaining a control criterion for setting the compensator are also described here. A functionally similar device has also been described on pages 258-259 of the Proceedings of the OEC'94 (Makuhari Exhibition, Japan), Article 14E-12, which achieves time differences between the modes of 28 ps.
In practice, the compensators described in the cited references are limited to very short sections of PMF. In consequence, when optimizing the control criterion, it is possible for a secondary optimum to be arrived at, so that the compensator is not set optimally.
AMENDED SHEET

,.: .s.~....._ ......~.... <'. ~~: ~~.~;~3.-=~ ~'? ~'- ~ 3;.. _. _ ~.,_.aaa.-~---~- ~ ~ i.

WO 00,/03505/PCT/DE99/01846 '. _ p, pMD compensator which comprises only PMFs is described in German patent application file reference 19816178.6. However, this require~~ mechanically moving parts.
The object of the present invention is thus to specify a method for PMD compensation and a P1~
compensator which has numerous setting options and, within the transmission band in use, as far as possible allows desired frequency-dependent polarization ZO transformation for the purpose of specific compensation.
The advantage of the PMD compensator according to the invention is that it can be used universally.
The phenomenon of polarization mode dispersion, which 15 is governed by a number of parameters, can in principle also be compensated for only by suitable adjustment options corresponding to a large number of degrees of freedom. The individual control voltages allow so many frequency-dependent polarization transformations to be 20 set and carried out during operation that any required PMD for even those which ,are higher than first order, can be formed with very high accuracy, both for the purpose of PMD compensation and fc>r PMD emulation.
Further advantages are th.e small physical size 25 which is normal with integrated optical components, and the fact that the functions of polarization transformation and the production of different propagation times for different polarizations can be integrated on one component, a chip, in the PMD
30 compensator according to the invention.
The PMD compensator can likewise be used as a PMD emulator or else as a polarization transformer.
Variants of the PMD compensator allow even more compact methods of construction.

~~?~H:~'~,~:.~F~~-H_~h,~,~:Y'J--~~:~w.v. fs.xmw.mr_~~...._.

'. - 6 -The invention will be explained in more detail with reference to exemplary embodiments. In the figures:
Figure 1 shows the construction o:E a polarization mode dispersion compensator according to the invention, Figure 2 shows a variant with crossing electrical lines, Figure 3 shows a variant with el<~ctric fields running at right angles to the waveguide direction, Figure 4 shows a variant with a folded beam path, Figure 5 shows an adaptive compen;aator, Figure 6 shows an emulator, Figure 7 shows a variant of a polarization mode dispersion compensator according to the invention, with a different crystal cut, Figure 8 shows a further variant without any electrode crossings, Figure 9 shows a variant with a ground electrode without tines, Figure 10 shows a variant with optical amplifiers, Figure 11 shows a detail of this, Figure 12 shows a section view with a waveguide embedded in a substrate, and Figure 13 shows a variant with a t;wo-mode waveguide.
Figure 1 shows a PMD comp.°nsator TRF1 according to the invention. This is in the form of a chip CH1 with a lithium-niobate substrates SUB. Other possible materials are lithium tantalate or similar highly birefringent materials.
' The crystallographic axes Y and Z lie in the plane of the drawing, while the crystallographic axis X
points into the plane of the drawing. A waveguide WG is produced on the surface of the crystal by titanium diffusion. The waveguide WG is a monomode waveguide so that TE (transverse electrical;f and TM (transverse magnetic) waves can propagate with a refractive-index difference of about 0.07.

i i, _.~___ ...
- . CA 02336787 2001-O1-08 ' WO 00,~03505/PCT/DE99/01846 -- - _ Electrically conductive int:erdigital electrodes which are in the form of a comb whose tines (spur lines, fingers) are arranged tran~wersely with respect to the waveguide are vapor-deposited on the surface of the substrate SUB. An electrode M with tines that are likewise arranged transversely with respect to the waveguide, runs in a meandering shape over the entire chip and can be connected to ground (ground electrode).
The other mode converter electrodes Eij (i= 1,2;
l0 j=1,2....n) which are in the form of combs, are electrically isolated from one another. A voltage Uij on one electrode produces an electric field in the waveguide wG, which runs periodically in the propagation direction Y, or in the opposite direction to this. Directly underneath the electrodes, it runs periodically in the direction X, or in the opposite direction to this. The optical wave, or the optical signal OS passes through the chip from the input IN to the output OUT. The period length L between the individual electrode tines is chosen to be approximately equal to a beat wavelength between the TE
and TM waves. A beat wavelength is that length for which the retarder with TE and TMi waves as eigen modes has a phase delay of precisely 360° between these eigen modes. With mixed polarizations, this is therefore repeated periodically at intervals of multiples of a beat wavelength. For an optical wavelength of 1550 nm (nanometers), the beat wavelength in lithium niobate corresponds to approximately 21. ~.m. The tine width (finger width) and the electrode separations are therefore expediently each chosen to be approximately equal to L/4. This results in a uniform structure in which tine widths and intermediate spaces are of equal size. In order to allow TE-T'M conversion with a variable phase to be carried out, additional intervals of alternately about L/4 and 3L/4 are in each case provided after a number of tine periods. This results in phase delays between TE and TM waves of 90° and 270°, respectively - the latter of which cancels out !ii -'r"~".~ CA 02336787 2001-O1-08 WO 00,/03505/PCT/DE99/01846 - 7a -the former once again - thus resulting in TE/TM
conversion with a different phase angle. This will now be explained in more detail: when. there is a pure TE
wave at the input of the PMD compensator, or WO 00,/03505/PCT/DE99/01846 _ g _ of a part of it, this results in a pure TE wave at the output, provided there are no control voltages on the electrodes. For suitably selected actuation voltages, the TE wave on the input side is converted into a pure TM wave. If these actuation voltages are reduced, then this results in a mixture of TE sand TM waves at the output, whose phase difference is freely variable. In particular, actuation voltage levels of about half result in an output polarization state which can be varied between 45° linear, right-hand circular, -45°
linear, left-hand circular, and 45" linear once again.
The ground electrode M in each ease has an overall width of approximately L/2 or L at these points (the ground electrode between the mode electrodes and cells could in principle also be made narrower, or could be replaced by bonding wires).
At least two mode converter electrodes, Elj and E2j (j - 1,2...n), including the' associated ground electrode segment interleaved in electrodes Elj and E2j, may be formed into cells Pj. A cell may likewise -like a mode converter electrode - be formed into a mode converter. If the mode-converter electrodes have positions differing by modulo half the beat wavelength of the substrate material, this is then a mode converter is which not only the intensity and mathematical sign, but also the phase of the mode conversion can be selected. In the <:ase of cells formed with two electrodes, these can be actuated using a pair of control voltages which are dependent on two parameters. The exemplary embodiments are limited to cells of this type. However, in principle, cells of different size with different electrodes and/or different numbers of electrodes, or different arrangements of the mode-converter electrodes, are possible. Each polarization transformer produced in the birefringent substrate material may be a cell.
In the exemplary embodimE~nt, n cells are provided, with values from about 25 to 50 being realistic for the number n. The voltages ' ' CA 02336787 2001-O1-08 w. WO 00/03505/PCT/DE99/01846 _ g _ on the electrodes Elj and E2j of the cells are, however, not in each case identical. or, as described in the Proceedings of the 4th European Conference on Integrated Optics (ECIO '87), Glasgow, UK, May 11-13 1987, pages 115-118, variable sinusoidally as a function of the longitudinal posit=ion coordinate, but are freely variable, individually o:r in groups.
Freely variable in groups .means the following:
for example, electrode voltages of Ulj, U1(j+1), U1 (j+2) , . . . to U1 (j+9) , (j= 1, 11, 21, . . . . ) may be identical, in the same way as the' electrode voltages U2j, U2 (j+1) , U2 (j+2) , . . . to U2 (j+9) , (j= 1, 11, 21,....). Thus, in this example, the type Elj electrodes are supplied with the same voltage by ten adjacent cells in each case, and, in the same way, the type E2j electrodes are each supplied by ten adjacent cells. This can easily be ach_Leved by means of conductive connections between t:he electrodes. The variability of the PMD compensator is limited by the withstand voltage of the spaces between the electrodes.
Ideally, it should be possib:Le to carry out polarization transformations over a:n indefinitely short length, but this would necessitate correspondingly high voltages. Attention must therefore be paid to a high withstand voltage during production. This can be done, for example, by applying insulating layers to the crystal surface.
While, in the prior art, only one polarization transformation, which has two degrees of freedom, can be specified for one optical frequency by two variable mode-converter voltages or mode:-converter voltage parameters, while the polarization transformations for all other optical frequencies result from this, the invention means that, by cascading a number of mode-converter electrodes, cell or cell groups which can be actuated individually or on a group-specific basis with, in general, in each case two variab3e mode-converter voltages, polarization transformations can be predetermined independently of one another within wide .-. WO 00/03505/PCT/DE99/01846 - 9a -limits for a number of optical frequencies. This also applies to the other ' CA 02336787 2001-O1-08 exemplary embodiments of the invention, with in each case three actuation voltages, instead of two, and thus three degrees of freedom being pro~Tided per cell in the example illustrated in Figure 3. In that example, the additional degree of freedom does not, however, relate to the mode conversion but to a differential phase shift. The latter can be used to select the operating wavelength. It also influences the PMD which is produced or compensated for, of course, but the PMD
changes which can be achieved in the vicinity of an optimally set operating wavelength as a result of the differential phase shift can, in general, also be achieved by actuation of the mode-converter electrodes.
One major reason for also using differential phase shifting is thus matching of the PLAID compensator to the desired operating wavelength.
A single mode converter, which is formed in a birefringent substrate material and whose mode-conversion phase can be set by means of two parameters to be in phase or in quadrature, can favorably influence, or even compensate for, PMD only in special cases. It is not suitable for technically worthwhile PMD compensation.
However, according to the invention, at least one further mode converter is added, which has at least one further control voltage, in which case the total of at least three control voltage's used for mode conversion in all the mode converters which are used are governed by at least three parameters. In the simplest case, the parameters are' identical to the control voltages being used. Th~~ sensible minimum number of control voltages is 4 to ~S thus covering, for example, at least two cells each having two mode-converter electrodes. Expedient :refinements operate with at least 10 different control voltages.
If one is prepared to accept. complicated wiring on the chip, for example using electrical lines that cross over, then the variant TRF2 shown in Figure 2 can be produced by means of insulating intermediate III
' CA 02336787 2001-O1-08 ' layers. The tines of the mode electrodes E11 and E12;
E21 and E22, ... to Enl and En2 are in this case each located successively between two tines of the ground electrode M. For the same maximum strength of the longitudinal electric field, this variant can carry out . polarization transformations over a somewhat shorter distance than the PMD compensator shown in Figure 1, and thus offers greater variability for polarization transformation, with the chip CIi2 having the same overall length. The periodicity of the electrode tines is still L. Their width and separations are approximately L/6. There is no need to introduce any greater separations or widths for the ground electrode.
A further exemplary embodiment TRF3 of the transformer is shown in Figure 3. It may also be advantageous to introduce a transverse electrical/transverse-magnetic phase shift, as is known from "IEEE Journal of Quantum Electronics", Volurne 25, No. 8, pages 1898 to 1906, by mean;a of electric fields running at right angles to the waveguide along the crystallographic Z-axis. The elliptical polarization transformation which can be achieved in this way, is characterized by having three degrees of freedom, can produce a control response that can be coped with somewhat more easily for compensation of PMD and, above all and as, already mentioned, allows the PMD
compensator to be matched to the' desired operating wavelength is advantageous, while t:he disadvantage, on the other hand, is the lower electrooptical coefficients, which allow only sm<~ller phase shifts.
The cells PPj (j=1,2,...n) on the chip CH3 likewise contain mode-converter electrodes Ell, E21, ..., which are actuated with voltages U11, U21., ... Phase-shifter electrodes EP1, ..., which are acauated by voltages U3j (j=1, 2...n) are arranged between the mode-converter electrodes. These can be chosen freely individually or individually in groups, that is to say, for example, the voltages U3j, U3 (ji-1) , U3 (j+2) , . . . to U3 (j+9) , - lla -(j= 1, 11, 21, . . . ) of the groups identified by different indices j can be chosen t,o be identical within each group. The phase-shifter sections of the phase-shifter electrodes have different lengths which are each a quarter of a beat wavelength greater or less than an integer multiple of the beat wavelength, that is to say L (N+1/) or L (K-1/) N, K = l, 2, 3, . . . . In this way, as in Figure 1, the electrodes, which are in the form of combs, form two groups E11 to Eln and E21 to E2n, which each produce TE-TM conversion with the same phase. Thus, if necessary, a number of voltages which are applied to a group of electrodess in adjacent cells, that is to say, for example, Ul,l. to U1,10, U1,11 to Ul, 20, . . . . , that is to say Ulj t.o U1 (j+9) , j= 1, 11, 21,....) may once again be identical, in the same way as the voltages U2j to U2 (j+9) , (j== 1, 11, 21, . . . . ) . If one wishes to exploit this <~dvantage, all the electrodes can be actuated individually, as is also possible in Figure 1 or 2. .
In integrated optics, a change between different crystal cuts and materials is normal in all cases. For example, IEEE J. Lightwave Technology, LT-5, No. 9, 1987, pages 1229-1238 desnribes, in Figure 3 there, how an integrated-optical component, in this case a modulator, can be produced equally well with an X-cut and with a Z-cut . In that example, the electrode configuration is changed on the transition from the Z-cut to the X-cut since, furthermore, the same electrooptical coefficient is intended to be used, and the electric field is therefore intended to run along the same crystal axis, in that example the Z-axis.
The exemplary embodiments in Figures l to 3 can also be modified in an analogous manner to that shown in Figure 7. In Figures 1 to 3, mode converters are intended to carry out mode conversion between Z and X-polarization. In lithium niobate, this is done using the electrooptical coefficient r51 :by means of a quasi-static electric field in the X-direction which is periodic at the beat wavelength L -- w CA 02336787 2001-O1-08 - : WO 00/03505/PCT/DE99/01846 of the birefringent waveguide WG. '.thus, while Figures 1 to 3 can be produced, for example, using lithium niobate with an X-cut and a Y-propagation direction, the exemplary embodiment TRF5 in Figure 7 can be implemented, for example, using lithium niobate with a Z-cut and a Y-propagation direction. This crystal-axis orientation is shown in Figure 7. While, in Figures 1 to 3, the periodic electric fields that are at right angles to the plane of the drawing, that is to say running along the X-axis, produce mode conversion between the X-axis and the Z-axis, in Figure 7 these are the periodic electric fields which run transversely with respect to the waveguide WG in the plane of the drawing, that is say once again running along the X-axis. Figure 7 shows a suitable arrangement, as a detail of the substrate SUB. The illustration shows a mode converter PMj (j - 1 ... n) which, for group-specific actuation, is part of a larger mode converter.
Electrodes EMCllj, EMCl2j, EMC2lj, EMC22j (j - 1 ... n) in the form of combs are fitted on each side of the waveguide WG. The electrodes EMCl.lj, EMC2lj, EMCl2j, EMC22j end transversely approximately at the edges of the waveguide WG in order to achieve as large as possible an overlap integral of optical and electric fields. The normal insulating buffer layers, for example of Si02, can be applied between the chip surface and the electrodes to provide the required insulation between crossing electrodes. The mode-converter voltages Vlj or -Vlj, respectively, is applied to the electrode EMCllj, E~MCl2j, on one side.
The voltages V2j and -V2j are applied to the electrodes EMC2lj, EMC22j on the other side. With respect to the electrodes on the first-mentioned side, these are offset through a quarter L/4 of one beat wavelength L
of the birefringent crystal SUB in the propagation direction Y. At intervals other than L/4, Vlj and V2j are not at 90°, but produce mode conversions affset at corresponding different phase angles to one another.
The electrode separations on one side of the waveguide - 13a -and the electrode width are also approximately L/4. The ' CA 02336787 2001-O1-08 two electrodes in the form of connbs on each side are isolated from one another by an insulating intermediate layer at the crossing points, each, have periods of one beat wavelength L, and are offset with respect to one another through half L/2 a beat wavelength L. Mode conversion can be carried out in phase or at 90° by mode-converter voltages Vlj and U'2j and the inverted voltages -Vlj, -V2j that are dependent on them (that is to say voltages which are actually inverted with respect to the voltages Vlj and V2j), thus allowing endless polarization transformation, and endless phase shifting of signals whose mode has been converted.
In another exemplary embodiment TRF7, illustrated in Figure 8; one of the mode-converter electrodes on one side of the wavelength is omitted. On the other side, both electrodes are replaced by a ground electrode EMC. In order to improve the efficiency, the electrode EMC may, but need not, have tabs which overlap the waveguide (Figure 9, TRF7). The arrangement has the advantage that the remaining electrodes EMCllj, EMC2lj, EMC do not cross. This exemplary embodiment of a mode converter PMj (j - 1 ...
n) likewise allows mode conversion in both quadratures, but only owing to the fact that a second mode-converter electrode EMC2lj with a voltagE~ V2j is provided alongside a first mode-converter electrode EMCllj with a voltage Vlj. The two electrodes are applied on a chip SUB, offset with respect to one another through an odd multiple 3L/4 of one quarter L/4 of a beat wavelength L
,30 in the propagation direction Y. high mode-conversion levels, which are freely variable in phase and in quadrature, are achieved by using a number, or a large number, of mode converters PMj (j - 1 ... n) preferably at intervals of an integer (or else half-integer) multiple of the beat wavelength L as the distance between an electrode EMCllj and the corresponding electrode EMC11(j+1) of the next mode converter PM(j+1). This cascading is implemented in an analogous manner to that in Figure 1.

CA 02336787 2001-O1-08 ' A greater bandwidth than in lithium niobate, but with a reduced PMD compensation capability, can be achieved, for example, by using lithium tantalate, which is less birefringent and, accordingly, has a longer beat wavelength L.
An alternative embodiment of mode converters, which is similar to the embodiment: in Figures 7 and 8 but can be used specifically for III/V semiconductors such as GaAs and InP, is described in the proceedings of the European Conference on Ini=egrated Optics ECIO
1987, Glasgow, May 11-13, 1987; ~>ages 115-118 and in the Proceedings of the European Conference on Optical Communications 1990, pages 309-312. Optical amplifiers can also be provided in III/V semiconductors. This has the advantage that the attenuation of the PMD
compensator can be compensated fo:r once again on the chip itself. It could be regarded as a disadvantage that the gain in III/V semiconductors is generally dependent on the polarization. However it is possible, for example by means of differently biased quantum wells, to form amplifiers which amplify TE to a greater extent than TM, or vice versa. Amplification independent of the polarization can thus be achieved by cascading two opposite amplifier:a, whose respective gain can be controlled by a pumping current . Since the attenuation in III/V materials is also more strongly polarization-dependent than, for example, in lithium niobate, that polarization dependence can be corrected by means of opposite polarization. dependence in the optical amplifier or amplifiers which is or are installed.
Polarization-dependent losses PDL also occur on optical transmission paths. These result, for example, from polarization-dependent isolators or couplers. If PDL occurs in the presence of PMD, 'then this results in additional losses during transmission. In one specific examplary embodiment of a polarization mode dispersion compensator TRF8 as shown in Figure 10, optical amplifiers OVj (j - l...n) are thus also provided on the substrate SUB as well as v ~ WO 00,/03505/PCT/DE99/01846 mode converters PHLj (j - l...n). As a detail from Figure 10, Figure 11 shows a spe~~ific mode converter PHLj and an upstream or downstream optical amplifier OVj. The former receives control signals VPHLjp (j -l...n; p=l...f). The index p, with a maximum value f, relates to the number of mode converter control signals which are available in each mode converter PHLj. The latter receive pumping signals OVI?mj (m - 1...2; j -l...n) .
The pumping signal connections OVMlj result in the TE gain being influenced more strongly, and the pumping signal connections OVM2j rE=_sult in the TM gain being influenced more strongly. The signals OVMlj, OVM2j can thus be produced by suitable combinations, for example linear combinations, of signals OVMVj, OVDVj, one of which OVMVj at least approximately influences only the mean gain, and the other, OVDVj, at least approximately influences only the differential TE-TM gain. As shown in Figure 10,, these combinations can be formed in an arithmetic unit RW which is produced, for example, from resistors and operational amplifiers, or is in the form of a microprocessor program. According to the invention, by varying the signals OVDVj in addition to the mode-converter signals VPHLij of the mode converters PHLj, it is possible to compensate not only the PMD, but also for any PDL which is present, at the same time.
In this exemplary embodiment, instead of the reflectors shown in Figure 4 and in Figure 10, bends KR1, KR2 can be provided in the waveguide WG, for beam deflection. As described on pages 654 and 655 of Electronics Letters, April 15, 1999, Volume 35, No. 8, such bends can be provided by means of photonic band-gap materials.
In general, the compensation for or production of PDL together with the compensation for or emulation of PMD is not just limited to sem_Lconductor materials or to materials with optical amplifiers. Even in passive materials such as lithium ni.obate, dichroitic - 16a -elements OVj, such as all polarization-dependent couplers whose coupling _ 1~.
can be varied electrooptically, or polarization-dependent interferometers, can be used such that the difference between the attenuation levels of specific waves which are polarized orthogon:ally with respect to one another, in particular TE and TM, can be adjusted by means of control signals OVDVj. In this case, Figure also shows a block diagram of such exemplary embodiments of the invention.
Figure 4 shows an exemplary embodiment TRF4 of 10 the PMD compensator with a folded beam path in the waveguide WG. The ends of a chip C'H4 are provided with metallic or dielectric mirrors MI. Shortly before arrival at a mirror, the waveguide WG is bent, thus resulting in reflection and feeding into the next section in as ideal a manner as possible. The mirror system can be replaced by couplers in whose center the mirror is located. This reduces t:he requirements for the accuracy of the longitudinal position of the mirrors MI, but results in stronger undesired reflections if not manufactured exa~~tly.
In the region of the input IN and of the output OUT, the waveguide ends are cut off at an angle of, for example, 6° to 8°, thus making it possible to produce a very low-reflection junction between the waveguide WG
and an external quartz-glass waveguide. The angled cut could also be used for the chips in Figures 1 to 3.
Other low-reflection junctions ;such as dielectric dereflection, are also possible. The individual cells P1, P2 , Pj , Pn can be conf figured as shown in Figures 1 to 3.
It should also be statec! that it is also possible to vary the transformation response by varying the temperature of the chip. 7:f this effect is undesirable, the temperature of thE= chip must thus be kept constant, or must be taken into account by inclusion in a control system.

!li ~ 02336787 2001-O1-08 ' ' a August 11, 2000 DE 009901846 GR 98 P 8057 Foreign version _ 18 _ Figure 5 shows the use of an adaptive PMD
compensator TR. The illustration shows a transmission path with an optical transmitter TR, a transmission path provided by an optical waveguide LWL, and an optical receiver RX. The receiver contains a photodiode PD, followed by a decision maker DFF. The transmitted data signal is emitted at the output OD of the decision maker.
The PMD compensator to compensate for the polarization mode dispersion of the optical signal OS
is positioned upstream of the photodiode. Further details, such as amplifiers, have been omitted for reasons of clarity.
The control criterion is obtained, for example, from the baseband signal BB emitted at the output of the photodiode. This is done, for example, via a filter FI1, which is in the form of a ba:ndpass filter with a resonant frequency at about half the symbol rate of the transmitted (binary) data. This is expedient if the maximum PMD values that occur are less than one symbol period of the data signal. Alternatively, the filter may be in the form of a low-pass filter whose cut-off frequency is approximately 1/10 to 1/4 of the symbol rate. This allows even relatively large PMD values to be detected clearly. The filter is followed by a first detector DET1, for example a rectifier. The detector output voltage U1 obtained in this way is supplied to a regulator MP, preferably a microprocessor with an analog/digital and digital/analog converters, and operational amplifiers for actuating the PMD
compensator C.
In order to allow both small and large PMD
values to be detected clearly, further filters and detectors FI2 and DET2 may also be provided.
When using the bandpass filter, the output voltage U1 is at a maximum when they eye opening of the baseband signal BB is a maximum. This corresponds to undistorted AMENDED SHEET

a WO 00/03505/PCT/DE99/01846 optimum. transmission. If the e:ye opening of the baseband signal is a maximum, the detector output voltage U2 is also a maximum when a low-pass filter is used.
The output signals from tile regulator are the voltages Uij (i=1, 2, 3; j - 1, 2, ...n) which the PMD
compensator requires. The regulator MP varies the voltages Uij individually or in <3roups. The voltages which are retained are those whic3z allow the greatest possible detector output voltages to be achieved. This allows the PMD compensator to provide optimum compensation for the polarization mode dispersion on the transmission path.
Figure 6 shows the use of the PMD compensator as an emulator EMU. A microprocessor MP1 receives signals via an adjustment device ER., and converts them into corresponding control voltage's Uij. In a simple version, the control voltages Uij could also be set by means of a potentiometer.
The optical signals fed in by the transmitter TR are emitted in transformed form at the output OE of an amplifier OV. The latter may also be omitted.
Apart from the materials lithium niobate, lithium tantalate and III/V semiconductors which have been mentioned, there are also a 1<~rge number of other possible ways to produce PMD compensators on a chip or substrate SUB. The chip/substrate may be birefringent, so that the waveguide WG is birefr.ingent, although the chip/substrate may also be non-birefringent (isotropic), provided the waveguide WG itself is composed of birefringent material.
For example, the waveguide WG may be composed of a liquid crystal, or a liquid or solid polymer. To produce it, a waveguide channel is formed out in a substrate material. If necessary, initial orientation of these waveguide matesrials is produced by electric fields.
The exemplary embodiment i.n Figure 12 shows a chip/substrate SUB, preferably composed of polymer, with a cover plate CD placed on it., in a cross-section from one end surface. The crystal axes X and Z of the waveguide WG lie in corresponding directions to those in Figures 1 to 3; the latter figures show a plan view of the exemplary embodiment as in Figure 11, looking through the substrate cover CD. I:n other embodiments, for example as shown in Figures 7 to 10, the crystal axes may also lie in different directions. The cover plate CD may also be composed of pc>lymer; however it is also feasible to use materials such as quartz glass (silicon dioxide) and silicon or other materials with low optical attenuation for the substrate and cover plate.
The manufacture of waveguide and V-grooves for optical waveguide coupling by means of hard dies which are pressed into polymers is cited on pages 1396 to 1398 of Electronics Letters, July 9, 1998, Volume 34, No. 14, and in the references mentioned there.
Solid polymers with a good electrooptical coefficient r33 are mentioned in the proceedings of the 24th European Conference of Optical Communications, Madrid, September 20 to 24, 1998, pages 501 to 502.
However, different electrooptica7. coefficients are required for the application required here.
In a further exemplary embodiment, as shown in Figure 11, the waveguide WG consists of a ferroelectric liquid crystal, which is birefringe:nt. The waveguide is incorporated in a cutout on the surface of the polymer substrate. The cutout can be produced by a metal die during manufacture of the substrate, or afterwards.
Electrodes ELij, M are applied to t:he cover plate, for example by vapor-deposition. If a considerable proportion of the optical field is transmitted into the electrodes, as is the case in the drawing, an optically low-loss material such as indium-tin oxide (ITO) must be used for the electrodes. The electrodes also have supply lines, so that an external voltage can be applied to them.
The ferroelectric liquid crystal is oriented such that, when no electric fields are applied between the electrodes, it has a different refractive index for transverse electromagnetic fields in the Z-direction (in the plane of the drawing in th.e right-hand part of the illustration) than in the X-direction (see the left-hand part of the illu.>tration). However, electrostatic field components in the Y-direction change the major axes of the birefringence. The change in the angle of the major axes is referred to as the tilt angle. In the case of Clark-Lagerwall switches, this can be varied to a major extent but not without hysteresis, in the case of deformed-helix ferroelectric liquid crystals (DH-FLC), it can likewise be varied to a major extent, and in the case of electroclinic liquid crystals (electroclinic FLC) it c:an be varied to a lesser extent, but relatively qu~.ckly. Electroclinic and DH-FLC appear to the most suitable candidates for this component.
The electroclinic effect. is described, for example, in Physical Review Letters, Volume 38, 1977, from page 848. The deformed-helix ferroelectric effect is described in Liquid Crystals, Volume 5, 1989, from page 1171. A broad overview of f:erroelectric liquid crystals can be found in "'Ferroelectric Liquid Crystals - Principles, Properties and Applications,", Volume 7, in the series "Ferroelectricity and Related Phenomena", Editor George W. Taylor, Gordon and Breach Science Publishers, ISBN 2-88124-282-0.
The rest of the operation of this exemplary embodiment of the invention has already been described in conjunction with Figures 1 to 3.
Since the difference in the refractive index between the two eigen modes (in this case X-polarized and Z-polarized) may be greater than in lithium WO 00,/03505/PCT/DE99/01846 - 21a -niobate, a greater differential group propagation time is obtained for a given physical length, which is advantageous for PMD cvmpensator. A further advantage is the high a CA 02336787 2001-O1-08 electrooptical coefficient of a F:LC. This allows very small and/or short electrodes ELij to be used, which widens their bandwidth and, for example, can render temperature stabilization superfluous.
Electrodes or electrode parts, in this case including electrode arrangements other than those in Figures 1 to 3, may also be applied to the substrate SUB.
Furthermore, additional electrodes may be applied which simplify initial polarization of the liquid crystal. In order to produce an electric field running in the X-direction, large-area electrodes PE1, PE2 are provided on the outsides of the substrate halves SUB, CD in Figure 11, or t:he component can be placed on one large-area electrode, or between two Large-area electrodes. The bottom of the waveguide can also be occupied by a continuous electrode. In order to produce a field running in the Z-direction, electrodes are provided on the left and right alongside the waveguide on the substrate parts SUB or CD. If they are accommodated on the substrate part SUB, the initial polarization can be produced even without the substrate part CD being fitted, so that the electrodes located there cannot interfere with the Z-profile of the polarization field.
One promising possibility for polarization is to treat or to coat one or more walls of the waveguide chemically, thermally or mechanically. Further design freedom is obtained by the large electrooptical effect in an FLC. The electrode combs can, in some circumstances, in consequence be ?-educed down to one tine each. This widens the optical :bandwidth. A further consequence of the high electrooptical coefficient is that only parts of the waveguide~ are required for polarization transformation, for e:~ample 10 to 200 ~m every 100 to 2000 Vim. In consequence, other materials, for example materials which acre more strongly birefringent or which produce less attenuation, can be used for the waveguide parts located. between - 22a -polarization !II

transformers: Possible materials include solid or liquid polymers, nematic liquid crystals and similar materials.
If polymers or silicon arcs used as substrate, V-grooves may be stamped in or anisotropically etched in in order to connect the invention to optical waveguides composed of quartz glass, which means there is no need for active adjustment, and which thus allow low-cost production. The technique to produce waveguides on silicon or in quartz glass is described, for example, in the Proceedings of the 24th European Conference of Optical Communications, Madrid, September to 24, 1998, pages 319 to 328, and in the references there.
15 In the exemplary embodiments in Figures 1 to 3 the electrooptical coefficient r51 is required, which can vary the polarization by means of a vertical electric field. It would also be possible to use r52, which is used by a longitudinal electric field. If 20 required, the crystal axes can be oriented in order to use other electrooptical coefficients, for example r42 or r41 for a Y-cut and X-propagation direction, r43 or r41 for a Z-cut and X-propagation direction, or r53 or r52 for a Z-cut and a Y-propagation direction.
Until now, birefringent: materials have essentially been discussed for PMD compensation.
Waveguide birefringence is, of course, often achieved simply by substrate birefringence. It is sufficient for the substrate material or the wave=guide itself to be birefringent.
In a generalization of the principle of operation, any arrangement in which mode conversion between two modes that have different propagation speeds may be used for PMD compensation or emulation for PMD compensation. Principal me>des thus occur, in general, instead of principle states of polarization.
These are those modes which have they greatest -20365-4352(S) principal modes (and principle states of polarization) are identical to the eigen modes.
The exemplary embodiment TRF9 in Figure 13 shows a lithium-niobate crystal with a X-cut and a Z-propagation direction. Other crystal cuts or materials may also be used. This has a two-mode waveguide WG. For its part, the waveguide WG contains two waveguides WG1, WG2. These have different widths, so that they have different group propagation times. They run in proton-replaced material, so that only one polarization can propagate in each case in WG1, WG2 and WG thus has only two modes, namely its two principal modes. Mode converters (P1, ...Pj, ...,Pn) are distributed over WG. With this crystal cut, these each have two electrodes Elj, E2j (j - l...n) which run essentially parallel to the waveguides WG1, WG2. In the area of WG1, WG2, these each produce horizontal electric fields, resulting in a differential phase shift between the two waveguide modes. The coupling between the two waveguides is thus varied by varying the control voltage between the electrodes Elj, E2j, so that the waveguide modes can be converted to one another, to a greater or lesser extent. In order to compensate the PMD, polarization beam splatters PBSl, PBS2 are positioned upstream or downstream of the arrangement.
In accordance with one aspect of this invention, there is provided a method for compensation for polarization mode dispersion (PMD), in which an optical signal (OS) to be compensated is supplied to an arrangement having an optical waveguide (WG) with principal modes having two different propagation constants and with a number of electrooptical mode converters (P1, ...Pj, ...,Pn; E11, E21,...) which are used for converting the principal modes of the waveguides to 20365-4352(S) 24a one another, and in which more than two electrooptical mode converters (P1, ...Pj, ...Pn; E11, E21,...) or more than two groups of mode converters are individually actuated by more than two control voltages which are governed by more than two parameters, in such a manner that the polarization mode dispersion of the optical signal (OS) is compensated for.
In accordance with another aspect of this invention, there is provided a polarization mode dispersion (PMD) compensator (TRF1 ... TRF8) having a waveguide (WG) with principal modes which have two different propagation constants, through which waveguide (WG) an optical signal (OS) passes, having a number of electrooptical mode converters (P1, ...Pj, ...,Pn: E11, E21, ...) which are used for converting the principal modes of the waveguide to one another, of which more than two electrooptical mode converters (P1, ...Pj, ...,Pn; E11, E21,...) or more than two groups of mode converters can be actuated in order to compensate for the polarization mode dispersion.

Claims (22)

Claims

1. A method for compensation for polarization mode dispersion (PMD), in which an optical signal (OS) to be compensated is supplied to an arrangement having an optical waveguide (WG) with principal modes having two different propagation constants arid with a number of electrooptical mode converters (P1, ...Pj, ...,Pn; E11, E21,...) which are used for converting the principal modes of the waveguide to one another, and in which more than two electrooptical mode converters (P1, ...Pj, ...,Pn; E11, E21,...) or more than two groups of mode converters are individually actuated by more than two control voltages which are governed by more than two parameters, in such a manner that the polarization mode dispersion of the optical signal (OS) is compensated for.

2. The method as claimed in claim 1, characterized in that the two principal modes are orthogonal main polarizations and the waveguide (WG) is birefringent.

3. The method as claimed in claim 1 or 2, characterized in that, in an arrangement. having additional differential phase shifters (EP1, .., Epn) they are also actuated individually and/or in groups individually in such a manner that the polarization mode dispersion is compensated for and/or the operating wavelength is selected.

4. The method as claimed in one of claims 1 to 3, characterized in that at least one voltage (U1, U2), which is used as a control criterion, is obtained by filters and rectification from the baseband signal (BB) in a receiving device (RX) for compensation.

5. A polarization mode dispersion (PMD) compensator (TRF1 ... TRF8) having a waveguide (WG) with principal modes which have two different propagation constants, through which waveguide (WG) an optical signal (OS) passes, halving a number of electrooptical mode converters (P1, ...Pj, ...,Pn; E11, E21,...) which are used for converting the principal modes of the waveguide to one another, of which more than two electrooptical mode converters (P1, ...Pj, ...,Pn; E11, E21,...) or more than two groups of mode converters can be actuated in order to compensate for the polarization mode dispersion.

6. The PMD compensator (TRF1 ... TRF8) as claimed in claim 5, characterized in that at least one mode converter contains a mode converter electrode (Eij, EMCkij; k=1,2; i=1,2;
j=1,2..n; M, EMC) which is designed like a comb with tines running transversely with respect to the waveguide (WG) .

7. The PMD compensator (TRF1 ... TRF5) as claimed in claim 5 or 6, characterized in that cells (Pj) are in each case formed from at least two mode converter electrodes (Eij; i = 1, 2;
j = 1, 2, . . . , n) .

8. The PMD compensator (TRF1 ,... TRF4) as claimed in one of claims 6 or 7, characterized in that alternating separations are provided between successive mode converter electrodes (E1j and E2j, or E2j and E1 (j+1)).

9. The PMD compensator (TRF1 .... TRF5) as claimed in one of claims 6 to 8, characterized in that the tines of at least two mode converter electrodes (E12, E22, ...Eij; i = 1, 2; j = 1, 2, ..., n) in a cell (Pj) in each case engage in one another.

10. The PMD compensator (TRF1 ... TRF6) as claimed in one of claims 5 to 9, characterized in that a ground electrode (M) which is likewise designed like a comb is provided, in which case it times and the times of the mode converter electrodes (E12, E22, ...Eij; i = 1, 2; j = 1, 2, ..., n) engage in one another.

11. The PMD compensator (TRF2) as claimed in claim 10, characterized in that two tines - one each from one mode converter electrode (Eij and E2j) - are in each case arranged between two times of the ground electrode (M) in a cell (PVj) comprising two mode converter electrodes (Eij, i = 1, 2: j = 1, 2.

12. The PMD compensator (TRF3) as claimed in one of claims 5 to 11, characterized in that broader phase-shifter electrodes (EPj) are arranged between the mode converter electrodes (E1j, E2j) and can be actuated individually or in a number of groups, on a group-specific basis.

13. The PMD compensator (TRF1 ... TRF6) as claimed in claim 12, characterized in that a phase-shifter electrode (EPj) in each case covers an area between the mode converter electrodes of one cell and an area between two cells (P1, P2; ...).

14. The PMD compensator (TRF8) as claimed in one of claims 5 to 13, characterized in that said PMD compensator has at least one optical amplifier (OVj; j=1...n) .

15. The PMD compensator (TRF1 ... TRF6) as claimed in one of claims 5 to 14, characterized in that a dichroitic element (OVj; j=1...n) is provided, which has a controllable difference between the gains or attenuations of two orthogonal polarizations.

16. The PMD compensator (TRF1 ... TRF6) as claimed in claim 15, characterized in that polarization-dependent attenuation or amplification of an optical medium can be compensated for by controlling a difference between the gains or attenuations of two orthogonal polarizations.

17. The PMD compensator (TRF1 ... TRF6) as claimed in one of claims 5 to 16, characterized in that said PMD compensator is in the form of a chip or is produced on a substrate (SUB) composed of lithium niobate, lithium tantalate, III/V semiconductors, silicon, silicon dioxide or polymer.

18. The PMD compensator (TRF1 ... TRF6) as claimed in claim 17, characterized in that the waveguide (WG) contains a polymer or a liquid crystal.

19. The PMD compensator (TRF4) as claimed in one of claims 5 to 18, characterized in that the waveguide (WG) is folded at least once by reflection.

20. The PMD compensator (TRF1 ... TRF6) as claimed in one of claims 5 to 19, characterized in that said PMD compensator is provided in a receiving device (RX), and in that at least one voltage (U1, U2) which is used as a control criterion is obtained by filters and rectification of the baseband signal (BB).

21. The PMD compensator (TRF1 ... TRF6) as claimed in one of claims 5 to 20, characterized in that said PMD compensator is provided as an externally adjustable emulator (EMU).

22. The PMD compensator (TRF1 ... TRF8) as claimed in one of claims 5 to 21, characterized in that the two principal modes are orthogonal main polarizations, and the waveguide (WG) is birefringent.