JP2008035265A - Multi-wavelength carrier-suppressed optical pulse signal generator - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a multi-wavelength carrier-suppressed optical pulse signal generator which is small in device size, reduced in power consumption and can generate a wavelength-multiplexed CS-RZ (Carrier-suppressed-Return to Zero) optical signal. <P>SOLUTION: The multi-wavelength CS optical pulse signal generator includes a reference clock signal generator 102, a two-branch electric distributor 104, an optical pulse strings generation part 134, an optical pulse signal generation part 136, and a multi-wavelength optical multiplexer 122. The reference clock signal generator generates a reference clock signal 103, The two-branch electric distributor takes the reference clock signal as an input and branches it into two to output a first clock signal 105a and a second clock signal 105b. The optical pulse strings generation part generates and outputs CS optical pulse stringths having mutually different oscillation frequencies. The optical pulse signal generation part takes the plurality of CS optical pulse strings as inputs and generates and outputs CS optical pulse signals having mutually different oscillation frequencies. The multi-wavelength optical multiplexer multiplexes and outputs CS optical pulse signals having mutually different CS optical pulse frequencies, which are outputted from the optical pulse signal generation part. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to an apparatus for generating a multi-wavelength carrier-suppressed RZ optical pulse signal by intensity modulation in a carrier-suppressed RZ (Return to Zero) format.

  In optical communication networks, transmission distance and capacity have been increased. Various formats of optical signals used in the optical communication system constituting this optical communication network have been proposed, and some of them have been put into practical use. A typical optical signal format in practical use is an intensity modulation format that represents a binary digital signal by the intensity of light. This intensity modulation format is roughly divided into two types: NRZ (Non Return to Zero) format in which the light intensity is maintained between consecutive “1” signals, and continuous “1” signal. RZ format in which the light intensity once becomes zero.

  An optical signal in the RZ format is generated by optical intensity modulation of individual optical pulses constituting the optical pulse train by an optical intensity modulator with respect to the optical pulse train arranged regularly at regular intervals on the time axis. . Optical intensity modulation of the individual optical pulses constituting the optical pulse train is to selectively block or transmit the optical pulses constituting the optical pulse train. That is, the RZ format optical signal is generated as a binary digital signal by selectively blocking or transmitting the optical pulses constituting the optical pulse train. Therefore, in order to generate an optical signal in the RZ format, an optical pulse train is required in advance, and a light source that generates this optical pulse train is essential.

  As described above, the optical signal in the RZ format is a binary digital signal obtained by optically modulating an optical pulse train that is regularly arranged at regular intervals on the time axis. The expression is used in the following meaning. That is, the expression optical pulse signal is used only when it means an optical pulse train as a binary digital signal obtained by optically modulating optical pulse trains regularly arranged at regular intervals on the time axis. To do. On the other hand, the expression “optical pulse train” is used to indicate the total of optical pulses arranged without being lost at regular intervals on the time axis.

  Since the RZ format is a format in which the light intensity once becomes zero even between consecutive “1” signals, the wavelength band of light as an optical carrier is generally wider than the NRZ format, as described below. Become. Hereinafter, the wavelength band of light as an optical carrier may be referred to as a wavelength spectrum band of an optical pulse signal or an optical pulse train.

  An optical pulse signal in the RZ format always has an optical pulse indicating a bit representing “1” on the time axis, so this optical pulse signal must be configured as a set of optical pulses with a narrow half-value width. become. On the other hand, the optical pulse signal in the NRZ format is configured as a continuous wide optical pulse while “1” continues when bits representing “1” appear continuously. For this reason, the half width of the optical pulse constituting the optical pulse signal in the NRZ format is, on average, wider than the half width of the optical pulse constituting the optical pulse signal in the RZ format.

  Accordingly, the frequency band occupied by the optical pulse signal in the RZ format (hereinafter sometimes referred to as a frequency spectrum band) is wider than the frequency spectrum band occupied by the optical pulse signal in the NRZ format. In the following description, when it is not necessary to distinguish between a spectrum expressed by frequency and a spectrum expressed by wavelength, it may be simply referred to as a spectrum.

  When the spectrum band is widened, the effect of so-called waveform distortion, in which the half-value width of the optical pulse on the time axis is broadened due to the group velocity dispersion of the optical fiber, which is the signal transmission medium, appears prominently. Distance is limited. Second, considering the increase in capacity by wavelength multiplexing, it is necessary to increase the wavelength difference allocated to adjacent channels in order to suppress crosstalk between channels to which adjacent wavelengths are allocated. In any case, an optical pulse signal having a wide spectrum band is not preferable from the viewpoint of efficient use of the frequency band of the optical communication network in which it is used.

  Therefore, a method for narrowing the spectrum band of an RZ format optical pulse signal has been proposed. A typical method among them is to adopt a so-called carrier suppression RZ format in which an optical pulse train having an inverted phase as an optical carrier wave is RZ-formatted between adjacent optical pulses on the time axis (for example, Non-patent document 1). Reversing the phase as an optical carrier wave between adjacent optical pulses on the time axis is synonymous with the fact that the phase difference between adjacent optical pulses is π. Hereinafter, the carrier suppressed RZ format may be referred to as a CS-RZ format (Carrier-suppressed-RZ format).

  Reversing the phase as an optical carrier between adjacent optical pulses on the time axis means that the phase as an optical carrier is not continuous and the phase jumping portion where the phase of the optical carrier changes abruptly by π It means to exist between pulses. Therefore, the effect of interference occurring between adjacent optical pulses is an effect of canceling out the mutual amplitude. On the other hand, when the phase as an optical carrier wave between adjacent optical pulses on the time axis is the same phase, the effect of interference generated between these optical pulses is an effect in which the mutual amplitudes are added.

  For this reason, in the CS-RZ format, even when the duty ratio of the optical pulse signal is increased, waveform distortion due to interference between adjacent optical pulses on the time axis is suppressed as compared with the normal RZ format. Here, the duty ratio of an optical pulse is the half-value width of the optical pulse with respect to the interval between adjacent optical pulses arranged on the time axis (it is a time width per bit and is sometimes called a time slot). The ratio of Therefore, increasing the duty ratio means that the half-value width of the optical pulse becomes wider with respect to the time slot. That is, if the time slot is fixed to widen the half width of the optical pulse, or if the time slot is narrowed by fixing the half width of the optical pulse, the duty ratio becomes high.

  That is, according to the CS-RZ format, the width of the optical pulse constituting the optical pulse signal on the time axis can be made wider than that of the normal RZ format, and as a result, the spectral band of the optical carrier can be reduced. Incidentally, the CS-RZ format can reduce the spectral bandwidth by about 25% compared to the normal RZ format where the phase as an optical carrier between adjacent optical pulses on the time axis is the same phase (see Non-Patent Document 1). Are known.

  As described above, the CS-RZ format is excellent in resistance to waveform distortion due to group velocity dispersion of an optical fiber, and is excellent in frequency utilization efficiency. That is, by adopting an optical pulse signal in the CS-RZ format, it is possible to realize an optical communication system with excellent long-distance transmission characteristics and frequency utilization efficiency.

Conventionally, a Mach-Zehnder interferometer is a typical method for generating a carrier-suppressed optical pulse train (hereinafter sometimes referred to as a “CS optical pulse train”) that is required to generate an optical pulse signal in the CS-RZ format. A method using a type of LiNbO ₃ light intensity modulator is known (for example, see Non-Patent Document 1). Hereinafter, the LiNbO ₃ light intensity modulator may be referred to as an LN light intensity modulator.

This method will be described by taking CS optical pulse train generation with a repetition frequency of 40 GHz as an example. First, CW light generated from a continuous wave (CW) light source is input to an LN light intensity modulator. Then, the DC bias level of the LN light intensity modulator is set to the minimum voltage value of the transmittance, the repetition frequency is 20 GHz, and the voltage difference between the maximum and minimum (peak-to-peak voltage, below) V _pp .) Is driven by an LN light intensity modulator with an electrical modulation signal (often a sine wave) that is twice the half-wave voltage V _π . An optical pulse train is output from the LN optical intensity modulator.

  According to this method, since the characteristic change of the optical pulse is small even if the wavelength of the CW light source is changed, a wavelength variable CS optical pulse train generating light source having a small wavelength dependence can be provided. This method also has an advantage that the repetition frequency can be easily changed.

  Therefore, an example of a DWDM (Dense Wavelength Division Multiplexing) transmission / reception device using a CS optical pulse train generator configured using an LN optical intensity modulator (see Non-Patent Document 2) has been proposed. ing. With reference to FIG. 1, a DWDM transmitter constituting the transmission side of the DWDM transmission / reception apparatus will be described. The DWDM transmitter means a multi-wavelength carrier-suppressed optical pulse signal generation device (hereinafter sometimes abbreviated as “multi-wavelength CS optical pulse signal generation device”).

  FIG. 1 is a schematic block diagram of a conventional multi-wavelength CS optical pulse signal generation device configured using an LN optical intensity modulator. Non-Patent Document 2 discloses an outline of a DWDM transmitter / receiver, but in FIG. 1, it is disclosed in Non-Patent Document 2 so that it can be easily compared with the multi-wavelength CS optical pulse signal generator of the present invention. The part of the DWDM transmitter (multi-wavelength CS optical pulse signal generation device) of the device is extracted and reconfigured by supplementing necessary components, and is shown as a conventional multi-wavelength CS optical pulse signal generation device.

  The conventional multi-wavelength CS optical pulse signal generator shown in FIG. 1 includes a reference clock signal generator 102, a two-branch electrical distributor 104, an optical pulse train generator 130, an optical pulse signal generator 132, and a multi-wavelength optical multiplexer. And a waver 122.

The reference clock signal generator 102 generates a reference clock signal 103 having a frequency of f _CLK that is synchronized with the system clock. The two-branch electrical distributor 104 branches the reference clock signal 103 into a first clock signal 105a and a second clock signal 105b. The first clock signal 105a and the second clock signal 105b are input to the optical pulse train generation unit 130 and the optical pulse signal generation unit 132, respectively.

The optical pulse train generator 130 includes a 1/2 frequency divider 106, a first electrical distributor 108 that is an N branch electrical distributor, N electrical amplifiers 112-1, 112-2, ... and 112-N, Continuous-wave light sources 118-1, 118-2, ... 118-N, N LN light intensity modulators 116-1, 116-2 with wavelengths λ ₁ , λ ₂ , ... and λ _N respectively. , ... It is composed with 116-N. Here, N is an integer of 2 or more, and the same applies to the following description.

  The optical pulse signal generation unit 132 includes a second electric distributor 110, which is an N-branch electric distributor, N electric signal generators 114-1, 114-2, ... 114-N, and N optical intensity modulators. 120-1, 120-2, ... 120-N.

The clock signal 105a is input to the 1/2 frequency divider 106, converted to a clock signal 107 having a frequency of _fCLK / 2, output, and input to the first electric distributor 108. The clock signal 107 input to the first electric distributor 108 is output with N branches.

The clock signals output from the electrical distributor 108 are respectively input to the electrical amplifiers 112-1, 112-2,... 112-N, and the output voltage value is amplified to V _pp = 2V _π by each electrical amplifier and output. Is done.

On the other hand, CW lights having wavelengths λ ₁ , λ ₂ ,... Λ _N are output from the continuous wave light sources 118-1, 118-2,. ₁ wavelength lambda, lambda _2, CW light is ... lambda _N are, LN optical intensity modulator 116-1 and 116-2, respectively, are input to ... 116-N. Further, LN optical intensity modulator 116-1, 116-2, ... 116-N are each of electric amplifiers 112 - 1 and 112 - 2, ... 112-N are output from a frequency f _CLK / 2 The clock signals 113-1, 113-2,... 113-N are respectively supplied.

Here, the reason why the 1/2 frequency divider 106 is necessary will be described. In order to output a CS optical pulse train having a frequency of f _CLK from the LN optical intensity modulator, the frequency is f _CLK / 2, as described with reference to generation of a CS optical pulse train having a repetition frequency of 40 GHz. This is because a modulated electric signal may be applied. That, LN optical intensity modulator 116-1 and 116-2, the respective ... 116-N, the clock signal 113-1 and 113-2 a frequency of f _CLK / 2, by supplying ... 113-N CS optical pulse trains 117-1, 117-2,..., 117-N having a repetition frequency of f _CLK are output.

Here, V _π is a half value of the voltage difference V _pp between the maximum and minimum of the amplitude of the electric signal input to modulate the LN light intensity modulators 116-1, 116-2,... 116-N. means. As a result, LN optical intensity modulator 116-1 and 116-2, ... from 116-N, ₁ wavelength lambda, respectively, lambda _2, in ... lambda _N, CS optical pulse train of repetition frequency f _CLK 117-1, 117-2, ... 117-N is output. .., 117-N are respectively input to light intensity modulators 120-1, 120-2,... 120-N included in the optical pulse signal generation unit 132. As the light intensity modulators 120-1, 120-2,... 120-N, the above-described Mach-Zehnder interferometer type LN light intensity modulator is used.

On the other hand, the second clock signal 105b of the other system output from the two-branch electric distributor 104 is N-branched by the second electric distributor 110, which is an N-branch electric distributor, and the clock signals 111-1, 111- 2, ... 111-N is generated and output. By driving the electric signal generators 114-1, 114-2, ... 114-N in phase synchronization with the respective clock signals output from the second electric distributor 110, the electric signal generators 114-1, 114-N generates and outputs electrical data signals 115-1, 115-2,... 115-N having a bit rate of f _CLK , respectively. The electrical data signals 115-1, 115-2,... 115-N are input to the light intensity modulators 120-1, 120-2,. 120-N outputs CS-RZ optical signals 121-1, 121-2, ... 121-N.

These CS-RZ optical signals 121-1, 121-2,... 121-N are combined by a multi-wavelength optical multiplexer 122 configured using a WDM coupler or the like, so that the number of wavelength multiplexing is finally reduced. A wavelength multiplexed CS-RZ signal 123 of N is obtained.
A. Hirano, Y. Miyamoto, S. Kuwahara, M. Tomizawa, and K. Murata, “A novel Mode-Splitting Detection Scheme in 43-Gb / s CS- and DCS-RZSignal Transmission,” IEEE J. Lightwave Technology. , vol. 20, No. 12, pp.2029-2034, 2002. 40 Gb / s CS-RZ LiNbO3 External Modulator [online] January 2004 Fujitsu Limited Product Catalog] [Search July 13, 2006] Internet <http://telecom.fujitsu.com/en/products/ device / pdf / 40g-csrz.pdf>

  However, the wavelength multiplexing CS-RZ optical signal generator according to the prior art described above has the following problems to be solved.

  The first problem is that the size of the wavelength multiplexing CS-RZ optical signal generator becomes large. This requires an LN optical intensity modulator and a separate continuous light source for the generation of a CS optical pulse train, and the size of the apparatus increases to accommodate these. Furthermore, according to the conventional apparatus, the amount of heat generated from the electric amplifier is not negligible. If the heat generated from the electric amplifier is large, it adversely affects the characteristics of the continuous light source, optical components such as a light intensity modulator, and electronic components such as an electric signal generation circuit. Therefore, a device design that takes into account efficient heat dissipation is required, and as a result, the device size increases.

The second problem is that power consumption increases. Required to generate a CS optical pulse train, V _pp of the modulation voltage of the LN optical intensity modulator is 2V _[pi. The half-wave voltage V _[pi in general LN optical intensity modulator, a 5 V~10 V, therefore, V _pp of the required modulation voltage becomes 10 V~20V. This is a large value of 24 dBm to 30 dBm in terms of electric power when the impedance of the LN light intensity modulator is 50Ω. Incidentally, in the current development status of electronic devices, the saturation output of an electric amplifier is about 30 dBm.

  Accordingly, the number of LN optical intensity modulators that can be driven by one electric amplifier for generating a CS optical pulse train is from 1 to 6 at most. That is, in order to construct a wavelength division multiplexing CS-RZ optical signal generation device having N wavelength multiplexing numbers, N or a number of electrical amplifiers close to it are required. That is, large power consumption is required.

  The problem described above becomes more prominent as the number of multiplexed wavelengths increases. In optical communication systems, there is a strong demand for practical miniaturization, configuration simplification, and low power consumption, and a wavelength-multiplexed CS-RZ optical signal generator that can meet these demands is desired. ing.

  Accordingly, an object of the present invention is to provide a multi-wavelength CS optical pulse signal generation device capable of generating a wavelength-multiplexed CS-RZ optical signal with a small device size and low power consumption.

  In order to achieve the above object, according to the gist of the present invention, a multi-wavelength CS optical pulse signal generation device having the following configuration is provided.

  The basic configuration of the multi-wavelength CS optical pulse signal generation device according to the present invention includes a reference clock signal generator, a two-branch electrical distributor, an optical pulse train generator, an optical pulse signal generator, and a multi-wavelength optical multiplexer. It is a prepared structure. The reference clock signal generator generates a reference clock signal. The two-branch electrical distributor receives a reference clock signal, branches the first clock signal and the second clock signal into two, and outputs them.

  The optical pulse train generation unit generates and outputs CS optical pulse trains having different oscillation frequencies. The CS optical pulse train is generated by a plurality of mode-locked lasers provided in the optical pulse train generator. As the mode-locked laser, it is preferable to use a distributed feedback semiconductor laser described later.

  When each of a plurality of CS optical pulse trains is input, the optical pulse signal generation unit generates and outputs carrier-suppressed optical pulse signals having different oscillation frequencies. The carrier-suppressed optical pulse signal is generated and output by a plurality of optical intensity modulators provided in the optical pulse signal generation unit.

  The multi-wavelength optical multiplexer combines and outputs carrier-suppressed optical pulse signals with different oscillation frequencies output from the optical pulse signal generation unit.

  The above-described multi-wavelength CS optical pulse signal generation device is preferably configured by using a mode-locked semiconductor laser (MLLD) as the above-mentioned mode-locked laser. The oscillation frequencies of the first to Nth N MLLDs included in the optical pulse train generation unit are different from each other.

  Further, the above-described optical pulse train generation unit preferably includes a first electric distributor. A first clock signal is input to the first electrical divider, and N branches (N is an integer equal to or greater than 2) are output to the (1-1) to (1-N) clock signals. .

  In the optical pulse train generation unit, the (1-1) to (1-N) clock signals are input to the first to N-MLLD, respectively. Thus, the first to N-MLLDs operate in mode synchronization, and the first to NCS optical pulse trains are output from the first to N-MLLDs, respectively.

  The optical pulse signal generation unit described above may include a second electrical distributor, first to Nth electrical signal generators, and first to Nth N light intensity modulators. A second clock signal is input to the second electric distributor, and is branched into N (2-1) to (2-N) clock signals for output. The first to Nth electrical signal generators generate and output the first to Nth electrical signals, respectively, when the (2-1) to (2-N) clock signals are input.

  In the optical pulse signal generation unit, the first to N-CS optical pulse trains and the first to Nth electrical signals are input to the first to Nth optical intensity modulators, respectively. In the first to Nth N light intensity modulators, the first to Nth CS pulse trains are optically modulated by the first to Nth electrical signals, respectively. First to Nth carrier suppression optical pulse signals are generated and output from the optical intensity modulator, respectively.

  The first to Nth carrier suppression optical pulse signals are input to the above-described multiwavelength optical multiplexer, and the first to Nth carrier suppression optical pulse signals are combined to generate a multiwavelength carrier suppression optical pulse signal. Generated and output.

Each of the first to N-MLLDs described above is preferably a first distributed feedback semiconductor laser having the following configuration. That is, the N first distributed feedback semiconductor lasers having different Bragg frequencies are represented as i-MLLDs (i is an integer from 1 to N). The i-MLLD includes an optical modulation region having a function of modulating the light intensity, the gain region population inversion is formed, and the phase control region is effective index variable grating Bragg frequency is f _i This is a distributed feedback semiconductor laser comprising a distributed reflector region in which is formed. In the i-MLLD, an optical modulation region, a gain region, a phase adjustment region, and a distributed reflector region are arranged in series and housed in a resonator.

With respect to the light modulation region, the transmittance can be modulated with the frequency f _rep by injecting an alternating current or applying an alternating voltage. Under conditions repetition frequency to output an optical pulse train is f _rep, so that the oscillation longitudinal mode of the i-MLLD becomes _{f i ± q (f rep /} 2) (q is an odd number.) Cavity length Has been adjusted. In addition, by modulating the transmittance of the light modulation region with the frequency f _rep , it is possible to output a light pulse train that is carrier-suppressed and has a repetition frequency of f _rep by mode-locking operation. Here, the longitudinal mode refers to the spectrum of the oscillating light and is sometimes called a resonator mode.

  Further, it is more preferable that each of the first to N-mode-locked semiconductor lasers described above is a second distributed feedback semiconductor laser having the following configuration. The difference between this second distributed feedback semiconductor laser and the above-mentioned first distributed feedback semiconductor laser is that, in addition to the light modulation region, gain region, phase adjustment region, and distributed reflector region, the light absorption coefficient is further increased. The light absorption coefficient adjustment region for adjustment is provided. That is, the value of the light absorption coefficient in the light absorption coefficient adjustment region is set to a sufficient size so that the reflected light reflected from the resonator end face on the light absorption coefficient adjustment region side does not reach the distributed reflector region. Is a feature.

  The light absorption coefficient adjustment region is disposed between the distributed reflector region and one of the resonator end faces adjacent to the distributed reflector region. That is, the light modulation region, the gain region, the phase adjustment region, the distributed reflector region, and the light absorption coefficient adjustment region are arranged in series in this order, and the end surface of the light modulation region and the outside is the first end surface, The end face between the absorption coefficient adjustment region and the outside is the second end face. One resonator end face close to the distributed reflector region means a second end face.

  In the multi-wavelength CS optical pulse signal generation device according to the present invention, the optical pulse train generation unit that generates the CS optical pulse train is generated by a plurality of mode-locked lasers, and the oscillation frequencies of the plurality of mode-locked semiconductor lasers are different Yes. That is, the CS optical pulse train is generated by a mode-locked semiconductor laser instead of being generated by a combination of a CW light source and an LN optical intensity modulator as in the conventional example.

  For this reason, unlike the conventional multi-wavelength CS optical pulse signal generation device that requires a separate continuous light source and an LN optical intensity modulator, the function of generating a CS optical pulse train can be realized by a mode-locked semiconductor laser. Therefore, it is possible to omit the power required in the LN light intensity modulator, and the power consumption can be reduced accordingly. Further, since an incidental element such as an electric amplifier required for driving the LN optical intensity modulator is not required, the size can be reduced accordingly.

  Further, by using a distributed feedback semiconductor laser as a mode-locked semiconductor laser, an overwhelming reduction in size can be realized.

  As described above, according to the multi-wavelength CS optical pulse signal generation device of the present invention, it is possible to generate a wavelength multiplexed CS-RZ optical signal with low power consumption and small device size.

By using the above-described first or second distributed feedback semiconductor laser as the distributed feedback semiconductor laser, the CS optical pulse train having a repetition frequency of f _rep can be output. In other words, N first or second distributed feedback semiconductor lasers each having a diffraction grating having different Bragg frequencies f _i (i is an integer from 1 to N). Is installed in the optical pulse train generation unit, it becomes possible to generate and output CS optical pulse trains having different oscillation frequencies from this optical pulse train generation unit.

The first or second distributed feedback semiconductor laser outputs a CS optical pulse train with a repetition frequency of f _rep by mode-locking operation in the oscillation longitudinal mode given by f _i q (f _rep / 2) To do.

The first distributed feedback semiconductor laser can output an SC optical pulse train having a repetition frequency of f _rep by operating as follows.

By injecting current into the gain region, an inversion distribution, which is a condition for oscillation, is formed, and the first distributed feedback semiconductor laser enters an oscillation state. Next, the resonance condition of the first distributed feedback semiconductor laser is determined by adjusting the effective refractive index of at least one of the effective refractive index of the phase adjustment region and the distributed reflector region, and the oscillation longitudinal mode is f. _The condition necessary for mode synchronous operation of _i ± q (f _rep / 2) is prepared. That is, when the oscillating a first distributed Bragg reflector semiconductor laser in this state, but no correlation between oscillation longitudinal mode can output oscillation light is an oscillation longitudinal mode is _{f i ± q (f rep /} 2) It becomes the state which is.

Next, the first distributed feedback semiconductor laser can be mode-locked by modulating the transmittance of the light modulation region with the frequency f _rep . That is, a certain correlation is introduced between the oscillation longitudinal modes. By oscillating the first distributed feedback semiconductor laser with a constant correlation between the oscillation longitudinal modes, a CS optical pulse train having a repetition frequency of f _rep can be output.

The second distributed feedback semiconductor laser can output an SC optical pulse train having a repetition frequency of f _rep by operating as follows. In the second distributed feedback semiconductor laser, as in the case of the first distributed feedback semiconductor laser, it is repeatedly performed by mode-locking operation in the oscillation longitudinal mode given by f _i ± q (f _rep / 2). A CS optical pulse train having a frequency of f _rep can be generated.

  The difference from the case of the first distributed feedback semiconductor laser is that the second distributed feedback semiconductor laser has a light absorption coefficient adjustment region. The value of the light absorption coefficient in the light absorption coefficient adjustment region can be set to a sufficient size so that the reflected light at the resonator end surface on the light absorption coefficient adjustment region side does not reach the distributed reflector region. When this state is realized, the boundary condition for determining the oscillation longitudinal mode of the second distributed feedback semiconductor laser is determined by the reflection from the first end face and the reflection from the distributed reflector region, and the second condition The end face has no influence on the determination of the oscillation longitudinal mode of the second distributed feedback semiconductor laser.

  That is, the fact that the reflected light does not reach the distributed reflector region is equivalent to the fact that the reflectance of the resonator end face (cleavage surface) on the distributed reflector region side is zero. Therefore, the CS optical pulse train can be reliably generated without being affected by the residual reflection of the resonator end face on the distributed reflector region side. If there is residual reflection at the cavity end face on the distributed reflector region side, it becomes difficult to generate a CS optical pulse train as a regular optical pulse train. Therefore, the second distributed feedback semiconductor laser does not require an advanced manufacturing technique for bringing the reflectance of the resonator end face (cleavage face) to 0 as much as possible.

When the above-described first or second distributed feedback semiconductor laser is used in the multi-wavelength CS optical pulse signal generator of the present invention, the repetition frequency f _rep of the optical pulse train is equal to the frequency f _CLK of the reference clock signal. You only have to set it.

  Hereinafter, embodiments of the present invention will be described with reference to FIGS. Each figure shows one configuration example according to the present invention, and only schematically shows the arrangement relationship of each component to the extent that the present invention can be understood. It is not limited to. In the following description, specific materials and conditions may be used. However, these materials and conditions are only one preferred example, and are not limited to these. In addition, overlapping description of the same components in each drawing including FIG. 1 described above may be omitted.

<Multi-wavelength CS optical pulse signal generator>
With reference to FIG. 2, the configuration and operation of the multi-wavelength CS optical pulse signal generation device of the present invention will be described. FIG. 2 is a schematic block diagram of the multi-wavelength CS optical pulse signal generation device of the present invention.

  The multi-wavelength CS optical pulse signal generator of the present invention includes a reference clock signal generator 102, a two-branch electrical distributor 104, an optical pulse train generator 134, an optical pulse signal generator 136, and a multi-wavelength optical multiplexer 122. And is composed. The reference clock signal generator 102 generates a reference clock signal 103. The two-branch electric distributor 104 receives the reference clock signal 103, branches it into a first clock signal 105a and a second clock signal 105b, and outputs them.

  The reference clock signal generator 102 can be configured using a pulse pattern generator or the like, and if necessary, a basic clock signal in an optical communication system in which the multi-wavelength CS optical pulse signal generation device of the present invention is used. A clock signal synchronized with the signal is generated.

  The optical pulse train generator 134 generates and outputs CS optical pulse trains having different oscillation frequencies. Specifically, the optical pulse train generator 134 includes an electric amplifier 124 that receives and amplifies the first clock signal 105a. The electric amplifier 124 outputs a clock signal 125 obtained by amplifying the first clock signal 105a. The clock signal 125 is input to the first electric distributor 126 and is divided into N branches (N is an integer of 2 or more), and the (1-1) th clock signal (127-1) and the (1-2th) ) The clock signal (127-2),... Is output as the (1-N) th clock signal (127-N). The (1-1) clock signal (127-1), the (1-2) clock signal (127-2),..., The (1-N) clock signal (127-N) are respectively the first This is input as a mode synchronization signal to the MLLD (128-1), the second MLLD (128-2),..., The N-MLLD (128-N).

The oscillation frequency values of the 1st-MLLD (128-1), 2nd-MLLD (128-2), ..., N-MLLD (128-N) are f ₁ , f ₂ , ...・, F _N. In subsequent description, certain oscillation frequency f _1, f _2, · · ·, a f _N, the wavelength lambda ₁ in a vacuum, respectively, lambda _2, · · ·, also be expressed in terms of lambda _N. A relationship of f _i · lambda _i = c the velocity of light in vacuum as c (all integers i from 1 N).

The multi-wavelength CS optical pulse signal generator shown in FIG. 2 is a multi-wavelength CS optical pulse signal generator suitable for use in an N-channel wavelength multiplexing optical communication device. In N-channel wavelength division multiplexed optical communication system, each channel, different wavelengths lambda _1, respectively, lambda _2, · · ·, identified by lambda _N. That is, by using the multi-wavelength CS optical pulse signal generator shown in FIG. 2, for example, the wavelength λ ₁ is assigned to the first channel, the wavelength λ ₂ is assigned to the second channel,... By assigning a wavelength λ _N to a channel, it is possible to construct an N-channel wavelength multiplexing optical communication system.

Here, the meanings of the oscillation frequency values f ₁ , f ₂ ,..., F _N will be described. The oscillation frequency value f _i means that the oscillation longitudinal mode spectrum is given by f _i ± q (f _rep / 2) on the frequency axis centered on the frequency f _i (q is an odd number). Means. After that, within a range that does not cause misunderstanding, the oscillation longitudinal mode spectrum is given by f _i ± q (f _rep / 2) on the frequency axis centered on the frequency f _i , and the oscillation frequency value is simply f _i Suppose there is.

The (1-1) th clock signal (127-1), the (1-2) th clock signal (127-2),..., The (1-N) th clock signal (127-N) is a mode synchronization signal. Are input to the 1st MLLD (128-1), the 2nd MLLD (128-2),..., And the Nth MLLD (128-N), respectively. As a result, each of the first 1-MLLD (128-1), the second 2-MLLD (128-2),..., The N-MLLD (128-N) operates in mode synchronization, and the oscillation frequency is f ₁ , f ₂ ,..., f _N being the first 1-CS optical pulse train (129-1), the second 2-CS optical pulse train (129-2),..., the N-CS optical pulse train (129 -N) is generated and output.

  For example, the process until the first 1-CS optical pulse train (129-1) is generated and output by the CS optical pulse signal generation device of the present invention shown in FIG. 2 is the same as the conventional CS optical pulse signal shown in FIG. Compared with the process until the CS optical pulse train 117-1 (corresponding to the first 1-CS optical pulse train (129-1)) generated by the generator is output from the LN optical intensity modulator 116-1. Then it becomes as follows.

CW light having a wavelength of λ ₁ (f ₁ in terms of frequency) that is the basis of the CS optical pulse train 117-1 is output from the CW light source 118-1. The CW light having the wavelength λ ₁ is input to the LN optical intensity modulator 116-1, and is generated and output as the CS optical pulse train 117-1. That is, in order to generate and output the CS optical pulse train 117-1, the CW light source 118-1 and the LN optical intensity modulator 116-1 are necessary. In contrast, the first 1-CS optical pulse train (129-1) is generated and output only by the first 1-MLLD (128-1). The same applies to the steps until the second-CS optical pulse train (129-2),..., The N-CS optical pulse train (129-N) is generated and output.

  For this reason, in order to generate and output a CS optical pulse train with a conventional CS optical pulse signal generator, drive these including the CW light sources (118-1, 118-2, ..., 118-N) A power supply is required to Also, including LN light intensity modulators (116-1, 116-2,..., 116-N), electric amplifiers 112-1, 112-2,. N is required. The reason why the electric amplifier is required is that the modulation voltage of the LN optical intensity modulator is as high as several tens of volts.

  Therefore, according to the conventional CS optical pulse signal generation device, two components, that is, the CW light source and the LN optical intensity modulator are necessary, and the size becomes large. As N, which is the wavelength multiplicity, increases, the problem of an increase in size becomes more prominent. In addition, the need for an electric amplifier to supply the modulation voltage required for the LN optical intensity modulator also causes an increase in size. Moreover, electric power for driving the electric amplifier is required, power consumption is large, and measures against heat generation from the electric amplifier are also required.

On the other hand, according to the CS optical pulse signal generation device of the present invention, the CW light source and the electric amplifier for the LN optical intensity modulator are not required, and the driving power of the MLLD that generates the CS optical pulse train is the same as that of the LN optical intensity modulator. Small enough compared to driving power. The injection power required for laser oscillation of MLLD is almost equal to the drive power for driving the CW light source. However, compared with the modulation voltage V _pp required for the LN optical intensity modulator being 10 V to 20 V, the modulation voltage required for intensity modulation required for the mode synchronization operation is sufficiently small, such as several volts or less.

Each of the plurality of CS optical pulse trains is input to the optical pulse signal generation unit 136, and carrier-suppressed optical pulse signals having different oscillation frequencies are generated and output. More specifically, the optical pulse signal generation unit 136, _first oscillation frequency is lambda, lambda _2, the 1-CS optical pulse train (129-1) is · · · lambda _N, the 2-CS optical pulse train (129 -2), ..., a N-CS optical pulse train (129-N) of respective, respectively is input, the oscillation frequency is different from lambda ₁ from each other, lambda _2, first a ... lambda _N A -CS optical pulse signal (121-1), a second CS optical pulse signal (121-2), ..., an N-CS optical pulse signal (121-N) are generated and output.

  The first 1-CS optical pulse signal (121-1), the second 2-CS optical pulse signal (121-2), ..., the N-CS optical pulse signal (121-N) are respectively generated by this optical pulse signal. Are generated and output by the first light intensity modulator 120-1, the second light intensity modulator 120-2,..., And the Nth light intensity modulator 120-N included in the unit 136. The multi-wavelength optical multiplexer 122 includes a first 1-CS optical pulse signal (121-1) output from the optical pulse signal generation unit 136, a second 2-CS optical pulse signal (121-2),. -CS optical pulse signal (121-N) is multiplexed and generated and output as wavelength multiplexed CS-RZ signal 123 with N wavelength multiplexing.

  In order to realize the above function, the optical pulse signal generation unit 136 has the same configuration as the optical pulse signal generation unit 132 of the conventional multi-wavelength CS optical pulse signal generation device shown in FIG. That is, the optical pulse signal generation unit 136 includes the second electric distributor 110, which is an N-branch electric distributor, N electric signal generators 114-1, 114-2,... 114-N, N light Intensity modulators 120-1, 120-2,... 120-N are provided. Therefore, since the operation is the same as that of the optical pulse signal generation unit 132 of the conventional multi-wavelength CS optical pulse signal generation device, the description thereof is omitted.

<First distributed feedback semiconductor laser>
With reference to FIG. 3, the configuration of the first distributed feedback semiconductor laser suitable for use in the multi-wavelength CS optical pulse signal generation device of the present invention and the mode-locking operation principle thereof will be described. FIG. 3 is a schematic configuration diagram of the first mode-locked semiconductor laser.

  The first mode-locked semiconductor laser shown in FIG. 3 employs a method for realizing a change in the effective refractive index of the optical waveguide in the phase adjustment region and the distributed reflector region by the plasma effect or the Pockels effect. In the following description, the effective refractive index of the optical waveguide in the phase adjustment region or the distributed reflector region may be simply referred to as the effective refractive index of the phase adjustment region or the distributed reflector region.

The first distributed feedback semiconductor laser is a mode-locked semiconductor laser in which a natural number multiple of the resonator circulation frequency approximates the repetition frequency f _rep of the generated optical pulse train. Approximation here means that the difference between the natural frequency multiple of the cavity frequency of the distributed feedback semiconductor laser and the repetition frequency f _rep of the optical pulse train is necessary for mode-locking the distributed feedback semiconductor laser. This means that the frequency pulling is small enough to occur. When this first distributed feedback semiconductor laser is used in the multi-wavelength CS optical pulse signal generation apparatus of the present invention, the repetition frequency f _rep of the optical pulse train is set equal to the frequency f _CLK of the reference clock signal.

  The first distributed feedback semiconductor laser 100 includes an optical modulation region 10, a gain region 12, a phase adjustment region 14, and a distributed reflector region 16 connected in series in this order. In each region, optical waveguides 10a, 12a, 14a and 16a having a double hetero structure sandwiched between the p-side cladding layer 20 and the n-side cladding layer 22 are formed. In the distributed feedback semiconductor laser, the intensity of oscillation light is modulated (loss modulated) by propagating through the optical waveguide 10a, and is amplified by stimulated emission by propagating through the optical waveguide 12a, and propagates through the optical waveguide 14a. As a result, the phase velocity is changed and Bragg reflected in the optical waveguide 16a. The oscillation light can be modulated by loss modulation as described above, but can also be gain-modulated in the optical waveguide 12a.

  The light modulation region 10, the gain region 12, the phase adjustment region 14, and the distributed reflector region 16 are each configured by sandwiching an optical waveguide between the p-side cladding layer 20 and the n-side cladding layer 22, and the p-side and n-side This means a structure including the electrode. That is, the light modulation region 10 will be described as an example. The light modulation region 10 includes the p-side electrode 24 in the light modulation region, the light modulation region portion of the p-side cladding layer 20, the optical waveguide 10a in the light modulation region, and the n side. It indicates the whole including the cladding layer 22 and the light modulation region portion of the n-side common electrode 32. The same applies to the gain region 12, the phase adjustment region 14, and the distributed reflector region 16.

  A current is injected into the gain region 12 by the constant current source 38 via the p-side electrode 26 and the n-side common electrode 32, thereby forming an inversion distribution necessary for laser oscillation and generating a gain. In the light modulation region 10, light modulation necessary for the expression of mode synchronization is performed. In order to cause light modulation in the light modulation region 10, there are a method of injecting a current into this region and a method of applying a voltage.

In order to implement the method of injecting current, the constant current supplied from the constant current source 34 and the frequency supplied from the AC power source 36 are f _rep between the p-side electrode 24 and the n-side common electrode 32. This can be done by supplying a current obtained by adding the AC current to the coupler 58.

In order to implement a method of applying a voltage as a method of causing light modulation in the light modulation region 10, a constant voltage supplied from a constant voltage source 34 between the p-side electrode 24 and the n-side common electrode 32 is used. And an AC voltage having a frequency f _rep supplied from the AC power supply 36 is added by the coupler 58 and applied.

  The phase adjustment region 14 and the distributed reflector region 16 are made of a material that is transparent to the laser oscillation wavelength. For example, in the case of an InGaAsP semiconductor laser having a laser oscillation wavelength of 1.55 μm, the phase adjustment region 14 and the distributed reflector region 16 are configured by an InGaAsP layer having a band gap wavelength of 1.3 μm.

  Further, the laser oscillation wavelength is not limited to this example, and may be 1.1 μm or 1.48 μm. In that case, the InGaAsP mixed crystal ratio is selected so as to be transparent corresponding to the laser oscillation wavelength, and the phase adjustment region 14 and the distributed reflector region 16 are formed. It is also possible to use a material other than InGaAsP according to the oscillation wavelength of the distributed feedback semiconductor laser 100. According to the oscillation wavelength of the distributed feedback semiconductor laser 100, by selecting a material having a band gap corresponding to the oscillation wavelength, it is possible to configure the distributed feedback semiconductor laser 100 that can obtain the required oscillation wavelength. Is possible.

  The p-side cladding layer 20 near the interface between the optical waveguide 16a and the p-side cladding layer 20 in the distributed reflector region, or the n-side cladding layer near the interface between the optical waveguide 16a and the n-side cladding layer 22 in the distributed reflector region A diffraction grating 18 is formed on either side of 22. FIG. 3 shows an example in which the diffraction grating 18 is formed on the p-side cladding layer 20 side in the vicinity of the interface between the optical waveguide 16a and the p-side cladding layer 20 in the distributed reflector region. Here, the vicinity of the interface means a range where an evanescent field of light guided through the distributed reflector region 16 reaches.

As described above, by modulating the light modulation region 10 and the gain region 12 with current or voltage, the distributed feedback semiconductor laser 100 operates in a mode-locked manner, and generates an optical pulse train with a repetition frequency f _rep . As shown in FIG. 3, the phase adjustment region 14 and the distributed reflector region 16, respectively, via the p-side electrode 28 in the phase adjustment region, the p-side electrode 30 in the distributed reflector region, and the n-side common electrode 32, respectively. Constant current injection or reverse bias voltage can be applied using constant current sources or constant voltage sources 40 and 42. Thus, in the case of constant current injection, the effective refractive index of the phase adjustment region 14 and the distributed reflector region 16 can be adjusted using the plasma effect. In the case of applying a reverse bias voltage, the effective refractive index of the phase adjustment region 14 and the distributed reflector region 16 can be adjusted using the Pockels effect or the like.

Of the oscillation longitudinal modes of the distributed feedback semiconductor laser 100, two longitudinal modes close to the Bragg frequency f ₀ of the distributed reflector region 16 are f ₀ + (f _rep / 2) and f ₀ − (f _rep / 2). That is, it is possible to adjust the effective refractive indexes of the phase adjustment region 14 and the distributed reflector region 16 so that both longitudinal modes have a symmetrical relationship with the Bragg frequency on the frequency axis.

  Since the CS optical pulse train is output from both the resonator end surface 44 on the distributed reflector region 16 side and the resonator end surface 46 on the light modulation region 10 side of the distributed feedback semiconductor laser 100, the CS output from either side The optical pulse train may be put to practical use. Further, it is desirable that the resonator end face 44 on the distributed reflector region 16 side is provided with a non-reflective film.

  In order to explain the operating principle of the first distributed feedback semiconductor laser, the characteristics of the time waveform of the CS optical pulse train will be described first with reference to FIG. FIG. 4 is a diagram illustrating a time waveform of an amplitude envelope of an optical carrier wave forming a CS optical pulse train, where the horizontal axis indicates time and the vertical axis indicates light intensity. However, the vertical axis indicates the light intensity of the first optical pulse, which will be described later, in the positive direction, and the optical intensity of the second optical pulse train whose phase as the optical carrier is shifted by π from the optical pulse constituting the first optical pulse train. Is shown for the negative direction.

  An optical pulse observed as a change in light intensity is expressed as an envelope of an amplitude waveform of an electric field vector of light as an optical carrier wave. Accordingly, in the following description, the time waveform of the light pulse represents an envelope of the amplitude waveform of the electric field vector of light.

  The CS optical pulse train is an optical pulse train in which the phases as optical carrier waves between adjacent optical pulses arranged on the time axis are opposite to each other. That is, the phase difference between the two adjacent optical pulses arranged on the time axis as an optical carrier is equal to π.

If the period T _rep of the time waveform of the CS optical pulse train on the time axis is 1 / f _rep (repetition frequency is f _rep ), the time waveform of this CS optical pulse train has a period of 2T _rep (= 2 / f _rep ) And a second optical pulse train having a period of 2T _rep (= 2 / f _rep ).

Here, as shown in FIG. 4, the first optical pulse train is an optical pulse train having a cycle of 2T _rep (repetition frequency f _rep / 2) and an optical phase as an optical carrier wave between adjacent optical pulses. It is. Also, the second optical pulse train has the same period of 2T _rep , the optical phase as an optical carrier between adjacent optical pulses is in phase, and is shifted from the first optical pulse train by T _rep (= 1 / f _rep ) And the optical phase as the optical carrier wave is inverted with respect to the first optical pulse train (the phase difference is π).

The frequency spectrum of the CS optical pulse train will be described with reference to FIGS. Figure 5 (A) is a diagram showing a frequency spectrum of the first optical pulse train repetition frequency of f _rep / 2, Fig. 5 (B), the repetition frequency is f _rep / 2, the first optical pulse train FIG. 5C shows a frequency spectrum of a second optical pulse train whose phase as a carrier wave is shifted by π with respect to the optical pulse train obtained by combining the first optical pulse train and the second optical pulse train. It is a figure which shows the frequency spectrum. 5A to 5C, the horizontal axis represents the frequency as an optical carrier wave. The vertical axis indicates the magnitude of each amplitude of the frequency spectrum component, the amplitude of the frequency spectrum component of the first optical pulse is in the positive direction, and the frequency spectrum whose phase is shifted by π from this frequency spectrum component The amplitude of the component is shown for the negative direction.

The amplitude waveform of the frequency spectrum of the first optical pulse train, as shown in FIG. 5 (A), when the frequency of the optical carrier to f _0, around the longitudinal modes of the frequency f _0, the longitudinal mode of the amplitude E _m However, on the frequency axis, the waveforms are separated by f _rep / 2 and all exist in the same phase and discretely. Here, m is an integer, and the amplitude E ₀ where m = 0 corresponds to the amplitude component of the frequency as the optical carrier of the first optical pulse train. Here, the longitudinal mode or vertical mode of amplitude E _m of the frequency f _0, the frequency respectively the amplitude component or amplitude of the frequency spectrum which is f ₀ which means the amplitude component of the frequency spectrum is E _m.

The amplitude waveform of the optical spectrum of the second optical pulse train, as shown in FIG. 5 (B), like the spectrum of the first optical pulse train, about a longitudinal mode frequency f _0, the absolute value of the amplitude is E _m The longitudinal mode becomes a waveform in which the frequencies are discretely present with a frequency of f _rep / 2. Here, m is an integer, and the amplitude E ₀ where m = 0 corresponds to the amplitude component of the frequency as the optical carrier wave of the second optical pulse train. However, the second optical pulse train is generated at a time shifted from the first optical pulse train by T _rep (= 1 / f _rep ) (first request), and the phase as the optical carrier is changed to the first optical pulse train. On the other hand, it is reversed (second request). Reversing the phase is synonymous with the phase difference between the two being π.

Because of the first request that the second optical pulse train is generated at a time shifted by T _rep (= 1 / f _rep ) from the first optical pulse train, the phase of each longitudinal mode of the second optical pulse train is an odd m Is shifted by π with respect to the phase of the longitudinal mode of the first optical pulse train. That is, the positive / negative of the amplitude is reversed for the longitudinal mode where m is an odd number. Further, due to the second request that the phase of the second optical pulse train as the optical carrier is inverted with respect to the first optical pulse train, the phase of the entire longitudinal mode of the second optical pulse train is the longitudinal of the first optical pulse train. It is shifted by π with respect to the phase of the mode.

As a result of satisfying the first and second requirements, the amplitude waveform of the optical spectrum of the second optical pulse train is reversed in the magnitude of the amplitude of the longitudinal mode of the first optical pulse train with respect to the spectrum of the first optical pulse train. The resulting waveform. Accordingly, the amplitude waveform of the optical spectrum of the CS optical pulse train, which is a combined waveform of the first optical pulse train and the second optical pulse train, is a waveform as shown in FIG. That is, the even-order longitudinal mode including the frequency component as the optical carrier disappears, and only the odd-order longitudinal mode has a spectrum waveform in which the frequencies exist discretely with a frequency _frep .

  The CS optical pulse train is named by eliminating a specific-order longitudinal mode (here, even-order longitudinal mode) from the frequency components of the optical pulse train as an optical carrier, that is, carrier suppressed. This is because the optical pulse train is generated.

As described above, the time waveform of the CS optical pulse train has a relationship in which the time waveform and the frequency spectrum are given in FIG. 4 and FIG. 5 (C). That is, as shown in FIG. 4, the optical pulses are arranged in a relationship in which the phase difference between them as an optical carrier between adjacent optical pulses is equal to π. Furthermore, longitudinal mode component constituting the frequency spectrum of the CS optical pulse train, as shown in FIG. 5 (C), around the longitudinal modes of the frequency f _0, the even-order longitudinal modes including the frequency components of the optical carrier Only the odd-order longitudinal mode disappears, and the frequencies exist discretely with a frequency of f _rep . That is, the frequency spectrum of the CS optical pulse train is composed of frequency components (longitudinal mode components) given by f ₀ ± q (f _rep / 2) where q is an odd number.

  Next, from the viewpoint of generating a CS optical pulse train having such a time waveform and a frequency spectrum, the relationship between the time waveform and the frequency spectrum waveform of the CS optical pulse described above will be described as follows.

The time waveform e _cs (t) of the CS optical pulse train amplitude is given by the following equation (1).

The first term of equation (1) gives the time waveform of the amplitude of the first optical pulse train, and the second term gives the time waveform of the amplitude of the second optical pulse train. Here, the period of the optical pulse, which is the interval on the time axis between adjacent optical pulses, is represented by T _rep . That, T _rep is in inverse relationship to each other between the repetition frequency f _rep of the optical pulse train, a T _rep = 1 / f _rep.

  Equation (1) can be transformed into the following equation (2).

Here, (1-exp (imπ)) appearing in the equation (2) becomes 0 when m is an even number, and therefore, the term in which the m is an even number in the equation (2) disappears. As a result, the CS optical pulse train E _cs (t) that gives a time waveform of the following amplitude is expressed by the following equation (3).

That is, the even-order longitudinal mode corresponding to the spectrum waveform shown in FIG. 5C disappears, and a time waveform consisting only of the odd-order longitudinal mode is obtained.

  Here, the optical spectrum and time waveform of the CS optical pulse train represented by FIG. 5 (C) and Equation (3) will be further considered.

The above equation (3), by introducing a sequence A _m is defined by the relationship given by the following equation (4), deformed in the form of the following equation (5).

  Here, when conversion is performed to replace (m + 1) with n for the second term of Expression (5), the following Expression (6) is obtained.

  Equation (6) can be interpreted as follows. Hereinafter, a description will be given with reference to FIGS. 6 (A) to (C) and FIGS. 7 (A) to (C).

  FIGS. 6A to 6C are diagrams illustrating frequency spectra of third and fourth optical pulse trains and a combined optical pulse train, which will be described later, FIG. 6A is a frequency spectrum of the third optical pulse train, and FIG. FIG. 4C is a diagram showing the frequency spectrum of a combined optical pulse of a third optical pulse train and a fourth optical pulse train; FIG. 6 (A) to 6 (C), the horizontal axis represents frequency and the vertical axis represents amplitude magnitude on an arbitrary scale.

  FIGS. 7A to 7C are diagrams showing time waveforms of the third and fourth optical pulse trains and the combined optical pulse train, FIG. 7A is a time waveform of the third optical pulse train, and FIG. FIG. 4C is a diagram showing a time waveform of a combined optical pulse of a third optical pulse train and a fourth optical pulse train; FIG. 7A to 7C, the horizontal axis represents time and the vertical axis represents light intensity on an arbitrary scale. However, the vertical axis of FIG.7 (C) is that the light intensity of the third and fourth optical pulse trains to be described later is in the positive direction, and the phases of the optical pulses and the optical carrier waves constituting the third and fourth optical pulse trains are as follows. The light intensity of a light pulse that is shifted by π is shown for the negative direction.

According to Equation (6), the optical spectrum of the CS optical pulse train is given by a combined waveform of the third optical pulse train and the fourth optical pulse train. Here, as shown in FIG. 6 (A), the third optical pulse train has a center frequency of f ₀ + (f _rep / 2) as its optical carrier, and the m-th longitudinal mode starts from this frequency. (M is an integer) has an amplitude of 2A _m and has a spectrum waveform whose frequencies are discretely present on the frequency axis by f _rep apart. In other words, the frequency of the m-th longitudinal mode with an amplitude of 2A _m is f ₀ + q (f _rep / 2) where q is an odd number (where q = 2m + 1). In addition, as shown in FIG. 6 (B), the fourth optical pulse train has a center frequency f _0- (f _rep / 2) as an optical carrier, and the amplitude of the mth longitudinal mode starting from this frequency. there is a 2A _-m, has a spectrum waveform frequency is present discretely separated by f _rep. The frequency of the m-th longitudinal mode with an amplitude of 2A- _m is f ₀ -q (f _rep / 2) where q is an odd number (where q = 2m + 1).

  As given by Equation (6), or as shown in FIGS. 6 (A) and 6 (B), the optical spectrum of the third optical pulse train and the optical spectrum of the fourth optical pulse train have their longitudinal mode amplitudes. Have spectral shapes that are anti-symmetric with respect to frequency. In each of the third and fourth optical pulse trains, the phases of the longitudinal modes specified by the value of m are in phase.

  This means that the phases are aligned between the optical pulses constituting the third optical pulse train, and that the phases are aligned between the optical pulses constituting the fourth optical pulse train. Further, when the third optical pulse train and the fourth optical pulse train are compared, the phases of the components as the respective optical carriers are also in phase with each other. This means that the optical pulses constituting the third optical pulse train and the fourth optical pulse train are generated simultaneously.

  Accordingly, the frequency spectrum of the combined optical pulse train of the third and fourth optical pulse trains is as shown in FIG.

When the contents described above are shown as time waveform relationships, the relationships shown in FIGS. 7A to 7C are obtained. 7A to 7C show time waveforms of the optical pulse train. That is, the optical carrier frequency shown in FIG. 7 (A) is f ₀ + (f _rep / 2), the third optical pulse train that is a normal RZ optical pulse train with a repetition frequency of f _rep , and FIG. 7 (B) f light carrier frequencies shown in ₀ - (f _rep / 2) a and, when the repetition frequency is the fourth and the optical pulse train is a normal RZ optical pulse train is f _rep, occurring at the same timing, a synthesis The time waveform is a CS-RZ optical pulse train shown in FIG.

  In the present invention, the above situation is realized by mode-locking a distributed feedback semiconductor laser having a distributed reflector region. Hereinafter, the relationship between the characteristics of the distributed reflector region, which plays a major role in realizing the above situation, and the longitudinal mode (sometimes referred to as a resonator mode) of the distributed feedback semiconductor laser is shown in FIG. This will be described with reference to FIG.

  FIG. 8 is a diagram for explaining the relationship between the reflectance spectrum of the distributed reflector region (thick solid line), the penetration depth (broken line), and the longitudinal mode (thin solid line), with the horizontal axis indicating the wavelength on an arbitrary scale. It is displayed. In addition, although the vertical axis is omitted, in the vertical axis direction, the reflectance spectrum and the vertical mode indicated by the solid line are indicated by light intensity, and the penetration length indicated by the broken line is indicated by an arbitrary scale. .

Here, an explanation will be given assuming an ideal situation where the end face reflectance at the resonator end face 44 on the distributed reflector region 16 side is zero. If the end face reflectivity at the resonator end face 44 is 0, there is no need to consider the influence of the end face reflection at the resonator end face 44 for the oscillation operation of the distributed feedback semiconductor laser, and the reflection characteristics by the pure diffraction grating The situation will be determined only by. The penetration length means L _eff when the phase Φ of the reflected light from the distributed reflector region 16 is defined by the following equation (7).
Φ = 2n _DBR k ₀ L _eff (7)
Here, n _DBR is the effective refractive index of the distributed reflector region, and k ₀ is the wave number. As shown in FIG. 8, if the end face reflectivity at the resonator end face 44 is 0, the reflectivity and penetration length profile (wavelength dependence) is symmetric with respect to the Bragg wavelength λ _B.

Here, of the longitudinal modes of the distributed feedback semiconductor laser, the two longitudinal modes (values converted to wavelengths) closest to the Bragg wavelength λ _B (converted to frequency f ₀ ) are related to the wavelength axis shown in FIG. Consider a case where the light beam exists at a position symmetrical to the Bragg wavelength λ _B. The frequency interval between the two longitudinal modes matches the resonator circular frequency, and it approximates the repetition frequency (f _rep ) of the generated CS optical pulse train. The approximation here is that the difference between each of the two resonator frequencies in the longitudinal mode and f _rep is such that frequency pull-in occurs when mode-locked operation occurs in the distributed feedback semiconductor laser. It means small.

  In this state, it is assumed that the distributed feedback semiconductor laser oscillates. In this case, in the above two longitudinal modes, the maximum of the reflectance in the distributed reflector region and the minimum of the penetration length are located symmetrically with respect to the Bragg wavelength, and both extreme values approximate the Bragg wavelength. Therefore, the laser oscillation conditions are the same in the two longitudinal modes, and this distributed feedback semiconductor laser generates laser oscillation in the two longitudinal modes. When mode-locking operation is caused in this state, this laser has two-wavelength mode-locking operation having a frequency corresponding to the wavelength of each of the two longitudinal modes.

Furthermore, the frequency profile of the reflectivity and penetration depth in the distributed reflector region shows that the maximum of the reflectivity and the minimum of the penetration length exist at positions symmetrical to each other with respect to the frequency converted from the Bragg wavelength. The profile itself has the same shape. In addition, when mode-locking operation occurs, frequency pull-in occurs between the light of these two wavelengths (light of the frequency corresponding to the above-described two longitudinal modes) via the modulation sideband. As a result, the frequency interval as the optical carrier wave of these two lights is fixed to the modulation frequency, that is, f _rep and also the phase synchronization is realized. Therefore, these two-wavelength mode-locked pulse trains have the same wavelength difference with the repetition frequency (f _rep ) and are phase-synchronized, so that the generated time also matches.

  As described above, the distributed feedback semiconductor laser that has been mode-locked under the condition that the end face reflectivity at the resonator end face 44 shown in FIG. 8 is 0 is shown in FIGS. ) And the CS optical pulse train generation condition described with reference to FIGS. 7A to 7C, the CS optical pulse train is generated.

In order to generate a CS optical pulse train, the two longitudinal modes closest to the Bragg wavelength must be set to be arranged symmetrically with respect to the Bragg wavelength. The longitudinal mode wavelength of the laser is determined by the following equation for the distributed feedback semiconductor laser having the structure shown in FIG.
mλ = 2n _mod L _mod + 2n _gain L _gain + 2n _pc L _pc + 2n _DBR L _eff (8)
Here, n _mod , n _gain , n _pc , and n _DBR are effective refractive indexes of the light modulation region 10, the gain region 12, the phase adjustment region 14, and the distributed reflector region 16, respectively. L _mod , L _gain , and L _pc are the lengths of the light modulation region 10, the gain region 12, and the phase adjustment region 14, respectively. L _eff is the penetration _depth of the distributed reflector region 16.

As shown in Expression (8), it is possible to change the longitudinal mode wavelength by changing at least one of n _pc and n _DBR .

In order to generate a CS optical pulse train from the distributed feedback semiconductor laser having the configuration shown in FIG. 3, the phase adjustment region 14 and the distributed reflector region 16 are respectively connected via the p-side electrodes 28 and 30 and the n-side common electrode 32. Then, a constant current or a reverse bias voltage is applied using a constant current source or a constant voltage source 40 and 42, respectively, and n _pc and n _DBR are changed. That is, the effective refractive index of n _pc and n _DBR is set to two longitudinally close to the Bragg wavelengths of the distributed reflector region 16 by a plasma effect when a constant current is applied and by a Pockels effect when a reverse bias voltage is applied. It is possible to adjust the mode so that it is placed symmetrically to the Bragg frequency. Arranging symmetrically with respect to the Bragg frequency means that one of the frequencies corresponding to the two longitudinal modes is smaller than the Bragg frequency, the other is larger than the Bragg frequency, and the difference between both frequencies is equal to the Bragg frequency.

  By applying a DC voltage as a bias voltage to the optical modulation region 10 and applying the modulation voltage in an overlapping manner, a CS optical pulse train having a variable pulse width can be easily generated. That is, by making the optical modulation region 10 function as a semiconductor electroabsorption modulator (also referred to as an EA modulator), a CS optical pulse train having a variable pulse width can be easily generated. The reason for this will be described with reference to FIGS. 9 (A) to 9 (C).

  9A to 9C are diagrams for explaining the relationship between the characteristics of the EA modulator and the optical gate waveform, and FIG. 9A is a diagram illustrating the transmittance with respect to the voltage of the EA modulator. Fig. 9 (B) is a diagram for explaining the optical gate characteristics when the applied DC voltage is set to the forward bias side and the modulation voltage is superimposed, and Fig. 9 (C) is the reverse bias applied to the applied DC voltage. It is a figure where it uses for the description of the optical gate characteristic when it sets to the side and a modulation voltage is superimposed.

  9 (A) to (C), the horizontal axis indicates the voltage applied to the EA modulator on an arbitrary scale, (+) indicates that it is a forward voltage, and (-) Indicates a reverse voltage. The vertical axis shows the transmittance on an arbitrary scale. In FIGS. 9B and 9C, when the EA modulator is regarded as a gate element, the time waveform of the intensity of the output light output from the EA modulator is shown on the right side as a gate waveform. .

  As shown in FIG. 9 (A), when a reverse bias voltage is applied to the EA modulator, the effect of shifting the band gap wavelength of the semiconductor constituting the EA modulator to the long wavelength side appears, and the absorption increases. The absorption increase is not linear with the applied voltage, but increases more rapidly, so the EA modulator has a sudden change in transmittance with respect to the applied voltage when a reverse bias voltage is applied. It shows the characteristic that On the other hand, when a forward bias voltage is applied to the EA modulator, the band gap wavelength of the semiconductor constituting the EA modulator is set to a wavelength shorter by 50 nm to 100 nm than the operating wavelength, so the change in transmittance is minute. .

  Here, when the DC voltage applied to the EA modulator is set to the forward bias side and the modulation voltage is superimposed, the optical gate characteristics as the gate element of the EA modulator are as shown in FIG. It is possible to obtain an optical gate characteristic with a wide pulse width that has a characteristic that the time during which is turned on (time during which light is transmitted) is longer than the time during which light is turned off (time during which light is blocked). If a distributed feedback semiconductor laser having an EA modulator structure is driven as the light modulation region 10 under this driving condition, a CS optical pulse train having a wide pulse width can be generated. Since the optical gate characteristics change when the DC voltage and the modulation voltage are changed, the pulse width of the generated CS optical pulse train can be made variable.

  On the other hand, when the DC voltage applied to the EA modulator is set to the reverse bias side and the modulation voltage is superimposed, the optical gate characteristics as the gate element of the EA modulator are as shown in FIG. An optical gate characteristic with a narrow pulse width having a characteristic that the OFF time is longer than the ON time can be obtained. Under this driving condition, a CS optical pulse train having a narrow pulse width can be generated by driving a distributed feedback semiconductor laser having an EA modulator structure as the optical modulation region section 10. Further, the pulse width of the generated CS optical pulse train can be varied by changing the DC voltage and the modulation voltage.

  Since the mode-locked semiconductor laser has a characteristic that the mode-locked light pulse circulates in the laser resonator, the gain or absorption modulation effect generated in the light modulation region occurs multiple times for one light pulse. Therefore, even if the gain or absorption modulation is not strong once, an optical pulse grows every time it circulates in the resonator due to the multiple modulation effect described above. In the case of an active mode-locked semiconductor laser, the gain or the intensity of absorption modulation changes according to the applied modulation voltage. This means that in an active mode-locked semiconductor laser, a sufficient gain or absorption modulation effect for generating an optical pulse can be secured even if the modulation voltage is small due to the multiple modulation effect.

  Next, experiments and results for demonstrating that a CS optical pulse train is generated by mode-locking a distributed feedback semiconductor laser will be described. The demonstration experiment was performed in the literature (S. Arahira and Y. Ogawa, “40 GHz actively mode-locked distributed Bragg reflector laser diode module with an impedance-matching circuit for efficient RF signal injection,” Jpn. J. Appl. Phys., Vol. 43, No. 4B, pp. 1960-1964, 2004.) was performed by operating the distributed feedback semiconductor laser in an active mode synchronization. This distributed feedback semiconductor laser employs an EA modulator structure in the light modulation region.

  The distributed feedback semiconductor laser used in the demonstration experiment has a light modulation region length of 165 μm, a gain region length of 610 μm, a phase adjustment region length of 110 μm, and a distributed reflector region length of 50 μm. This is an InP-based multi-electrode semiconductor laser device. The resonator length is 1065 μm, and the resonator circular frequency is about 40 GHz. The gain region has a quantum well structure in which the quantum well layer is a 0.6% compressive strained InGaAsP layer and the barrier layer is a non-strained InGaAsP layer. The composition ratios and thicknesses of the quantum well layer and the barrier layer are set so that the photoluminescence peak wavelength of this multiple quantum well structure is 1562 nm.

A bulk InGaAsP layer in which the composition ratio and thickness of each layer are set so that the photoluminescence peak wavelength is 1465 nm is applied to the waveguide layer in the light modulation region, phase adjustment region, and distributed reflector region. Using. In addition, the resonator end face on the distributed reflector region side was coated with a non-reflective coating with an reflectance of about 2% using an Al ₂ O ₃ thin film. The oscillation threshold for laser oscillation by injecting current into the gain region of the distributed feedback semiconductor laser was about 30 mA, and the slope efficiency, which is the ratio of the output light intensity to the injection current, was about 0.1 W / A. These oscillation threshold values and slope efficiency values are typical values for a semiconductor laser.

  With reference to FIGS. 10A and 10B, an experimental result of CS optical pulse train generation of the first mode-locked semiconductor laser will be described. FIGS. 10A and 10B are diagrams for explaining the experimental results of the CS optical pulse train generation of the first mode-locked semiconductor laser, and FIG. 10A shows the time waveform of the generated CS optical pulse train. FIG. 10 (B) is a diagram showing a spectrum waveform. In FIG. 10 (A), the horizontal axis indicates time for one scale of 10 ps, and the vertical axis indicates light intensity in arbitrary units. In FIG. 10B, the horizontal axis indicates the wavelength in units of nm, and the vertical axis indicates the laser oscillation spectrum intensity in units of dBm.

  10A and 10B show the time waveform of the CS optical pulse train and the optical spectrum waveform of the oscillation light, respectively. The time waveform and the optical spectrum of the oscillation light are observed under the following conditions. That is, a current of 71 mA is injected into the gain region, a current of 3.25 mA is injected into the phase adjustment region, a forward bias voltage of +0.39 V, a repetition frequency of 39.81312 GHz, and a modulation voltage into the optical modulation region A modulation voltage having an intensity of +2.4 dBm was applied.

First, as shown in FIG. 10 (B), the optical spectrum of the oscillation light has a minimum center spectral component (the part corresponding to the Bragg wavelength λ _B of the spectrum curve), and left and right with the Bragg wavelength as the center. It is symmetrical and shows an optical spectrum shape peculiar to the CS optical pulse train. From this, it can be seen that the CS optical pulse train is generated. That is, longitudinal mode spectrum peaks are arranged symmetrically about the Bragg wavelength λ _B.

  From the time waveform of the CS optical pulse train shown in FIG. 10 (A), the pulse width was estimated to be 14.9 ps. In this case, the duty ratio, which is a value obtained by dividing the pulse width by the pulse period, was estimated to be 59.3% (14.9 / 25.1 = 0.593). Despite such a high duty ratio, the light intensity falls to 0 between successive light pulses (light intensity is 0 in the middle of the peak position of adjacent light pulses). This is one of the features of the CS optical pulse train, that is, since the phase is inverted between successive optical pulses, it indicates that the intensity becomes 0 due to interference in the middle of successive optical pulses.

  With reference to FIG. 11, an experimental result of the variable pulse width characteristic when the bias voltage and the modulation voltage intensity applied to the light modulation region are changed will be described. FIG. 11 is a diagram showing the dependence of the optical pulse width of the CS optical pulse train output from the first distributed feedback semiconductor laser on the modulator bias voltage and the modulation voltage. In FIG. 11, the horizontal axis shows the modulator bias voltage scaled in V units, and the vertical axis shows the half-value width of the optical pulse constituting the optical pulse train output from the distributed feedback semiconductor laser, ps (picoseconds). Scaled in units. In FIG. 11, (a) shows the RF (Radio Frequency) signal strength of -1.1 dBm, (b) shows the RF signal strength of +2.4 dBm, (c) shows the RF signal strength of +7.4 dBm, and (d) shows the RF signal strength. The signal intensity is +15.9 dBm, and (e) shows the case where the RF signal intensity is +18.1 dBm.

  In the example shown in FIG. 11, the half width of the optical pulse could be varied in the range from 3.3 ps to 15.9 ps. From the results shown in FIG. 11, a duty ratio of 50% (in this case, a pulse width of 12.55 ps) generally used in an optical communication system using an optical pulse signal in an RZ format generated by modulating a CS optical pulse train is obtained. Assuming that a CS optical pulse train is generated, the RF signal intensity required to modulate the transmittance of the light modulation region is +7.4 dBm or less. As shown in FIGS. 11 (a) to 11 (c), it can be seen that +7.4 dBm or less is sufficient for generating a CS optical pulse train having a pulse width of 12.55 ps.

  This is a value approximately 1/50 to 1/200 compared with the value (24 dBm-30 dBm) in the CS optical pulse train generation method using the LN optical intensity modulator disclosed in Non-Patent Document 1, which is a conventional example. It is. That is, according to the CS optical pulse generation method described here, the CS optical pulse train generation method using the conventional LN optical intensity modulator means low power consumption.

  With reference to FIGS. 12A and 12B, the characteristics of the time waveform and wavelength spectrum waveform of the CS optical pulse train output by the mode-locked operation of the first distributed feedback semiconductor laser will be described. 12 (A) and 12 (B) are diagrams showing a CS optical pulse train having an optical pulse width of 3.3 ps, FIG. 12 (A) shows a time waveform, and FIG. 12 (B) shows a wavelength spectrum. . The horizontal axis of FIG. 12A shows the time scaled in units of ps, and the vertical axis shows the light intensity scaled arbitrarily. In addition, the horizontal axis of FIG. 12B shows the wavelength scaled in nm units, and the vertical axis shows the wavelength spectrum intensity scaled in dBm units.

  As shown in FIG. 12B, the wavelength spectrum has no spectral component at the center (wavelength position indicated by an upward arrow in FIG. 12B) and has a symmetrical shape. That is, the intensity of the wavelength spectrum is minimal at the wavelength position indicated by the upward arrow in FIG. 12B, and the wavelength spectrum is symmetrical with respect to this minimal position. From this, it can be seen that the CS optical pulse train having the time waveform shown in FIG. 12A is generated from the distributed feedback semiconductor laser.

  As described above, a CS optical pulse train having a variable optical pulse width can be generated by using a multi-electrode type distributed feedback semiconductor laser. Further, the modulation voltage required for modulating the transmittance of the light modulation region, which is required for generating the CS optical pulse train, can be lower than that of the conventional method. In other words, by using a multi-electrode type distributed feedback semiconductor laser, the device itself can be made compact, it can be driven with low power consumption, and the width of the optical pulse constituting the CS optical pulse train to be generated is adjusted. Is possible.

<Second distributed feedback semiconductor laser>
With reference to FIG. 13, the configuration of the distributed feedback semiconductor laser 300, which is the second distributed feedback semiconductor laser, and the mode-locked operation principle of the distributed feedback semiconductor laser 300 will be described. FIG. 13 is a schematic enlarged cross-sectional view of the distributed feedback semiconductor laser 300. FIG.

  The difference between the distributed feedback semiconductor laser 300 and the distributed feedback semiconductor laser 100 described above is that a light absorption coefficient adjustment region 60 is further integrated on the resonator end face side of the distributed reflector region 16. . The configuration of the other parts is the same as that of the distributed feedback semiconductor laser 100 shown in FIG.

  The two resonator end faces constituting the resonator of the distributed feedback semiconductor laser 300 are constituted by a left end face 46 of the light modulation region 10 and a resonator end face 66 on the light absorption coefficient adjustment region 60 side. Unlike the distributed feedback semiconductor laser 100, the resonator end surface 66 on the light absorption coefficient adjustment region 60 side can use a cleaved surface that is not coated with an antireflection coating.

  The CS optical pulse train generated by the distributed feedback semiconductor laser 300 is output from the resonator end face 46 on the optical modulation region 10 side. A reverse bias voltage is applied to the light absorption coefficient adjustment region 60 from the constant voltage source 64 via the p-side electrode 62 and the n-side common electrode 32.

  When the CS optical pulse train is generated by the distributed feedback semiconductor laser 100 described above, the wavelength dependence of the reflectance and penetration length of the distributed reflector region 16 of the distributed feedback semiconductor laser 100 is symmetric with respect to the Bragg wavelength. It is necessary to have In order to satisfy this condition strictly, the end face reflectivity at the resonator end face 66 on the distributed reflector region 16 side needs to be zero.

  In general, the reflectance of the cleavage plane constituting the cavity end face of the distributed feedback semiconductor laser is not zero but has a finite size. Even if the cleaved surface is coated, it is not easy to make the reflectance sufficiently small. When the reflectivity of the cavity end face of the distributed feedback semiconductor laser is not 0, the wavelength dependence of the reflectivity and the penetration length depends on the initial phase of the formed diffraction grating and the end face, as will be described below. It has a very large effect on reflectivity.

  The initial phase of the diffraction grating will be described with reference to FIG. FIG. 14 is a diagram for explaining the initial phase of the diffraction grating, and shows the distributed reflector region 16 in an enlarged manner. The initial phase of the diffraction grating is the phase of the diffraction grating at the light input end of the distributed reflector region 16 (left end of the distributed reflector region 16), as shown in FIG.

  The period of the diffraction grating is 240 nm, for example, in the case of an InP semiconductor laser with an oscillation wavelength of 1.55 μm. Therefore, in order to accurately set the initial phase of the diffraction grating, a manufacturing technique capable of reproducibly processing dimensions shorter than several tens of nanometers is necessary, which is very difficult at present. Therefore, it is currently impossible to process a distributed feedback semiconductor laser in which the initial phase of the diffraction grating is set according to the design value. That is, when processing a distributed feedback semiconductor laser, the initial phase of the diffraction grating cannot be used as a design parameter.

  On the other hand, the reflectance characteristics of the distributed reflector whose end face reflectance is not 0 are determined by interference between Bragg reflection by the diffraction grating and Fresnel reflection by the end face of the distributed reflector region 16. The phase characteristic of the reflected light due to Bragg reflection is a function of the initial phase of the diffraction grating. For this reason, the combined reflectance of both of the interference light between the Bragg reflected light and the Fresnel reflected light strongly depends on the initial phase and the end face reflectance of the diffraction grating.

The amplitude reflectivity r _DBR of a distributed reflector having a finite end-surface reflectivity is given by the following equations (9) to (13) (for example, “Semiconductor laser and optical integrated circuit” written by Ohharu Suematsu edited by Ohmsha 1st edition) reference).

Where n _DBR and L _DBR are the effective refractive index and region length of the distributed reflector region, λ _Bragg is the Bragg wavelength, α is the absorption coefficient, κ is the coupling coefficient of the diffraction grating, φ is the initial phase of the diffraction grating, r ₀ is the end face reflectance (amplitude reflectance).

Incidentally, r _DBR when the end face reflectance is 0 is given by the following equation (13).

With reference to FIGS. 15A to 15E, the energy reflectivity | r _DBR | ² of the distributed reflector calculated using the above-described equations (9) to (13) will be described. FIGS. 15 (A) to (E) are diagrams showing the energy reflectivity of the distributed reflector, and FIGS. 15 (A) to (E) are the reflectivity R ₀ of the cleavage plane and the energy reflectivity of the cleavage end face, respectively. Is shown for the case where the initial phase φ of the diffraction grating is changed as a parameter. In each figure, the horizontal axis indicates the wavelength in units of nm, and the vertical axis indicates the energy reflectivity in arbitrary units.

Here, the energy reflectivity shown in FIGS. 15 (A) to 15 (E) is calculated as n _DBR = 3.2, L _DBR = 50 μm, λ _Bragg = 1550 nm, α = 10 cm ⁻¹ , κ = 100 cm ⁻¹ It is. This calculation condition is a characteristic of a typical distributed feedback semiconductor laser.

The results shown in FIG. 15 (A) are the calculation results when the energy reflectivity R ₀ (| r ₀ | ² ) of the cleavage end face is 0, and the results shown in FIGS. 15 (B) to (E) are the cleavage results. When the energy reflectivity R ₀ (= | r ₀ | ² ) of the end face is 0.274 (| r ₀ | ² = 0.274), the initial phase φ of the diffraction grating changes from 0 to 1.5π every 0.5π. This is the calculation result when

When the energy reflectivity of the cleavage end face is 0, the energy reflectivity | r _DBR | ² of the distributed reflector does not change even if the initial phase φ of the diffraction grating is changed. Similarly, there is no change in the penetration length L _eff . The wavelength dependence of the energy reflectivity | r _DBR | ² and the penetration length L _eff of the distributed reflector is symmetric with respect to the Bragg wavelength as described with reference to FIG. According to the semiconductor laser, it is possible to generate a CS optical pulse train.

On the other hand, since the energy reflectivity R ₀ of the cleaved end face is not 0, the energy reflectivity | r _DBR | ² of the distributed reflector changes depending on the initial phase φ of the diffraction grating. In general, the reflectance and penetration length profile (wavelength dependence) of the distributed reflector region is asymmetric, and the wavelength at which the reflectance of the distributed reflector region is maximized and the penetration length at that wavelength also vary greatly. The wavelength dependence of the penetration length is also asymmetric, similar to the wavelength dependence of the reflectance of the distributed reflector region.

That is, the CS optical pulse train cannot be generated unless the energy reflectivity _R0 of the cleavage end face is sufficiently small. Therefore, in order to reduce the reflectance of the cleaved end face of the distributed feedback semiconductor laser (resonator end face 44 on the distributed reflector region 16 side), it is necessary to apply a low reflection coating. In this case, a problem in practical use is how much the energy reflectivity R ₀ of the cleavage end face can be reduced to a sufficiently small energy reflectivity. That is, the influence of Fresnel reflection at the resonator end face 44 on the distributed reflector region 16 side is sufficiently suppressed, and the reflection characteristics symmetric with respect to the Bragg wavelength as shown in FIGS. 8 and 15A are uncontrollable. It is a technical problem whether it can be obtained without depending on a certain parameter φ (initial phase of the diffraction grating).

With reference to FIGS. 16A to 16D, the initial phase φ dependency of the maximum reflectance, the peak wavelength shift amount, and the penetration length from the distributed reflector region will be described. FIGS. 16A to 16D show the maximum reflectivity from the distributed reflector region, the peak wavelength shift amount, and the cleaved surface reflectivity R ₀ in the distributed feedback semiconductor laser 100 when changed as a parameter. It is a figure which shows the initial phase (phi) dependence of penetration | invasion length.

  Fig. 16 (A) shows the maximum reflectance from the distributed reflector region, Fig. 16 (B) shows the peak wavelength shift amount, Fig. 16 (C) shows the penetration depth when the distributed reflector region length is 50μm. FIG. 16 (D) shows the peak wavelength shift amount when the length of the distributed reflector region is 50 μm. In each figure, the horizontal axis indicates the initial phase φ of the diffraction grating. In addition, the vertical axis of FIG.16 (A) indicates the maximum reflectance, and the vertical axis of FIG.16 (B) and FIG.16 (D) indicates the peak wavelength shift amount in units of GHz. The vertical axis of 16 (C) indicates the penetration length in units of μm, respectively.

The peak wavelength shift amount in FIG. 16 (B) is shown as a frequency difference (Δf _peak ). That is, the frequency difference (Δf _peak ) is given by the following equation (14), where the peak wavelength is λ _peak .

Here, c is the speed of light in vacuum.

As shown in FIGS. 16A to 16C, the reflection characteristics from the distributed reflector region when the energy reflectivity of the resonator end face 44 on the distributed reflector region 16 side is 0 (as 0% in the figure) In order to realize a reflection characteristic almost identical to that indicated by a dotted line), R ₀ of 0.1% is insufficient and must be reduced to 0.001% or less. Such small reflectivity is very difficult to achieve with a low reflection coating. That is, when the CS optical pulse train is generated using the distributed feedback semiconductor laser 100 having the end face of the distributed reflector shown in FIG. 3 as one end face of the resonator, the following problems are expected to occur.

The reflection characteristics of the distributed reflector region vary greatly depending on the reflected light generated when the energy reflectance R ₀ of the resonator end face 44 on the distributed reflector region 16 side is not 0 and the initial phase (φ) of the diffraction grating. As a result, the pulse characteristics of the generated CS optical pulse train vary greatly depending on the distributed feedback semiconductor laser element used, or the generation of the CS optical pulse train itself becomes difficult.

One means of suppressing the influence of the reflected light generated at the cleaved end face is to make the length L _DBR of the distributed reflector area sufficiently long so that the light input to the distributed reflector area reaches the end face of the distributed reflector. It is to make the structure which reflects with Bragg reflection before doing. FIG. 16 (D) shows the result of calculating the shift amount of the peak wavelength when the length L _DBR of the distributed reflector region is increased to 500 μm. Here, R ₀ is calculated as 27.4%. Despite the fact that one end face of the distributed reflector is a cleaved surface and the reflectivity is not reduced, the peak wavelength shift amount is almost 0 regardless of the initial phase of the diffraction grating, giving the maximum reflectivity. The state where the wavelength substantially matches the Bragg wavelength is maintained. That is, in this case, a reflection characteristic that substantially matches the reflection characteristic when the end face reflectance of the resonator end face 44 on the distributed reflector region 16 side is 0 is realized.

However, as is well known, when the region length L _DBR of the distributed reflector is increased, the bandwidth of the reflectance profile (the wavelength band in which reflection occurs) becomes narrower. This limits the number of longitudinal modes that can oscillate. As a result, the spectrum width when the mode synchronization operation occurs is limited. If the number of longitudinal modes capable of laser oscillation is limited due to the Fourier transform between the spectrum width and the pulse width, the shortest value of the optical pulse width obtained when the mode synchronization operation occurs is limited. Therefore, this results in a disadvantage that the variable width of the optical pulse width is limited.

  Therefore, the distributed feedback semiconductor laser 300 is capable of mode-locking operation that does not depend on the reflected light generated at the cleaved end face or the initial phase of the diffraction grating without increasing the length of the distributed reflector region. It is a semiconductor laser. The distributed feedback semiconductor laser 300 is connected to the resonator end face 44 on the distributed reflector region 16 side in the distributed feedback semiconductor laser 100, and is further provided with a light absorption coefficient adjustment region 60. Further, the resonator end face 66 on the light absorption coefficient adjustment region 60 side does not need to be coated with a low reflection film.

  A reverse bias voltage is applied from the constant voltage source 64 to the light absorption coefficient adjustment region 60 via the p-side electrode 62 and the n-side common electrode 32. By applying the reverse bias voltage, the band gap wavelength of the light absorption coefficient adjustment region 60 is shifted to the longer wavelength side, and as a result, the absorption coefficient of the light absorption coefficient adjustment region 60 is increased. The light that has passed through the distributed reflector region 16 from the left end to the right end passes through the light absorption coefficient adjustment region 60 and is reflected by the resonator end surface 66 on the light absorption coefficient adjustment region 60 side, and then again the light absorption coefficient. After passing through the adjustment region 60, it is input to the distributed reflector region 16 again. The intensity of light re-input to the distributed reflector region 16 is the product of the amount of light attenuation when reciprocating the light absorption coefficient adjustment region 60 and the end face reflectance of the resonator end surface 66 on the light absorption coefficient adjustment region 60 side. Determined by

  Therefore, if the amount of attenuation of light when passing through the light absorption coefficient adjustment region 60 is large, that is, the value of the light absorption coefficient of the light absorption coefficient adjustment region 60 is reflected on the resonator end surface 66 on the light absorption coefficient adjustment region side. By setting the size so that light does not reach the distributed reflector region 16, the intensity of the light re-input to the distributed reflector region 16 can be sufficiently reduced. This is equivalent to the case where the end face reflectance of the resonator end face 44 on the distributed reflector region 16 side is sufficiently lowered in the distributed feedback semiconductor laser 100. Therefore, if the amount of attenuation of light when passing through the light absorption coefficient adjustment region 60 is sufficiently large, even if the end face reflectance of the resonator end surface 66 on the light absorption coefficient adjustment region 60 side is large to some extent, in the distributed feedback semiconductor laser 100 Thus, it is possible to realize a reflectance characteristic that is very similar to the reflectance characteristic obtained when the end face reflectance of the resonator end face 44 on the distributed reflector region 16 side is zero. As a result, a CS optical pulse train can be generated from the distributed feedback semiconductor laser 300 even if the end face reflectivity of the resonator end face 66 is large to some extent.

Referring to FIGS. 17A to 17C, the reflected light from the distributed reflector region with respect to the diffraction grating initial phase dependence in the distributed feedback semiconductor laser 300 including the light absorption coefficient adjustment region shown in FIG. The maximum reflectance, peak wavelength shift amount, and penetration length dependency will be described. FIGS. 17A to _17C show the initial phase φ dependency of the maximum reflectance from the distributed reflector region, the peak wavelength shift amount, and the penetration _depth with the absorption coefficient α _EA of the light absorption coefficient adjustment region as a parameter. FIG. 17 (A) shows the maximum reflectance from the distributed reflector region, FIG. 17 (B) shows the peak wavelength shift amount, and FIG. 17 (C) shows the penetration length. In FIGS. 17A to 17C, the horizontal axis represents the initial phase φ of the diffraction grating. In addition, the vertical axis of FIG. 17 (A) shows the maximum reflectance, the vertical axis of FIG. 17 (B) shows the peak wavelength shift amount in GHz, and the vertical axis of FIG. Each length is shown in μm.

The calculation results shown in FIGS. 17A to 17C are obtained assuming the same parameters as in FIGS. 15A to 15E and FIGS. 16A to 16D. . That is, n _DBR = 3.2, L _DBR = 50 μm, λ _Bragg = 1550 nm, α = 10 cm ⁻¹ , and κ = 100 cm ⁻¹ are calculated. Further, the end face reflectivity of the resonator end surface 66 on the light absorption coefficient adjustment region 60 side was set to 27.4% assuming that the resonator end surface 66 is a cleavage plane. Further, the light absorption coefficient of the optical absorption coefficient adjustment region 60 (alpha _EA), respectively and 230 cm ^-1 and 345cm ^-1, the area length of the optical absorption coefficient adjustment region 60 was assumed to be 300 [mu] m.

When the absorption coefficient α _EA is converted from 230cm ^-1 and 345cm ^-1 to light attenuation per 100μm in length, it corresponds to -10dB and -15dB respectively, and has the same function as the function of the light absorption coefficient adjustment region 60 If it is a typical electroabsorption optical modulator, the value is sufficiently realizable. For comparison, the calculation results for the first distributed feedback semiconductor laser, which does not include the light absorption coefficient adjustment region 60 and the value of the end face reflectivity R ₀ of the resonator end face 44 is 0, are shown in the figure. Shown with a dotted line (r ₀ = 0).

As shown in FIGS. 17A to 17C, when the absorption coefficient α _EA of the light absorption coefficient adjustment region 60 is as large as 345 cm ⁻¹ , the maximum reflectance, the peak wavelength shift amount, and the penetration length are It can be seen that the amount is constant regardless of the initial phase. Further, they almost coincide with the characteristics of the distributed reflector of the ideal distributed feedback semiconductor laser 100 indicated by the dotted line (r ₀ = 0) and having the end face reflectance R ₀ of 0. The reflection characteristics of the distributed reflector region at this time are found to provide the desired reflection characteristics required for CS optical pulse train generation, in which the wavelength dependence of both reflectance and penetration length is symmetrical with respect to the Bragg wavelength. It was.

  According to the distributed feedback semiconductor laser 300 including the light absorption coefficient adjustment region 60 shown in FIG. 13, without depending on the initial phase of the diffraction grating in the distributed reflector region or the end face reflectivity of the resonator end face, As shown in FIG. 8, a distributed reflector having reflection characteristics symmetric with respect to the Bragg wavelength can be realized. As a result, variations in the reflection characteristics of the distributed reflector caused by the initial phase of the diffraction grating and the reflected light from the resonator end face 44 can be suppressed, and variations in characteristics between elements can be suppressed. In addition, according to the distributed feedback semiconductor laser including the light absorption coefficient adjustment region 60, the CS optical pulse train can be generated more reliably.

It is a schematic block diagram of a conventional multi-wavelength carrier-suppressed optical pulse signal generation device. 1 is a schematic block configuration diagram of a multiwavelength carrier-suppressed optical pulse signal generation device according to the present invention. FIG. 2 is a schematic configuration diagram of a first mode-locked semiconductor laser. It is a figure which shows the time waveform of CS optical pulse train. Is a diagram for explaining a frequency spectrum of the CS optical pulse train, (A) shows a frequency spectrum of the first optical pulse train repetition frequency of f _rep / 2, (B), the repetition frequency f _rep / 2 Shows the frequency spectrum of the second optical pulse train whose phase as a carrier wave is shifted by π with respect to the first optical pulse train, and (C) shows the optical pulse train obtained by synthesizing the first and second optical pulse trains. It is a figure which shows a frequency spectrum. It is a figure showing the frequency spectrum of the third and fourth optical pulse train and the combined optical pulse train, (A) is the frequency spectrum of the third optical pulse train, (B) is the frequency spectrum of the fourth optical pulse train, (C) is the combined light It is a figure which shows the frequency spectrum of a pulse train. It is a figure which shows the time waveform of the 3rd and 4th optical pulse train, and the combined optical pulse train, (A) is the time waveform of the 3rd optical pulse train, (B) is the time waveform of the 4th optical pulse train, (C) is the combined light It is a figure which shows the time waveform of a pulse train. It is a figure where it uses for description of the relationship between the reflectance spectrum of a distributed reflector area | region, penetration | invasion length, and a longitudinal mode. It is a figure used for explanation of the relation between the characteristic of the EA modulator and the optical gate waveform, (A) is a diagram showing the transmittance with respect to the voltage of the EA modulator, (B) is a DC voltage to be applied on the forward bias side (C) is a diagram for explaining the optical gate characteristics when the applied DC voltage is set to the reverse bias side and the modulation voltage is superimposed. FIG. It is a diagram for explaining the experimental results of CS optical pulse train generation of the first mode-locked semiconductor laser, (A) is a diagram showing the time waveform of the generated CS optical pulse train, (B) is a diagram showing the spectrum waveform It is. It is a figure which shows the modulator bias voltage and modulation voltage dependence of the optical pulse width of the CS optical pulse train output from the first MLLD. FIG. 4 is a diagram showing a CS optical pulse train with an optical pulse width of 3.3 ps output from the first MLLD, (A) shows a time waveform, and (B) shows a wavelength spectrum. FIG. 3 is a schematic configuration diagram of a second MLLD. It is a figure where it uses for description of the initial phase of a diffraction grating. It is a figure which shows the energy reflectivity of a distributed reflector. (A) to (E) show cases where the reflectance R ₀ of the cleavage plane and the initial phase φ of the diffraction grating are changed as parameters. In the first mode-locked semiconductor laser, when the reflectance R ₀ of the cleavage plane is changed as a parameter, the maximum reflectance from the distributed reflector region, the peak wavelength shift amount, and the penetration length depend on the initial phase φ. (A) shows the maximum reflectivity from the distributed reflector region, (B) shows the peak wavelength shift amount, (C) shows the penetration depth when the length of the distributed reflector region is 50 μm. (D) shows the peak wavelength shift when the length of the distributed reflector region is 50 μm. In the second mode-locked semiconductor laser, when the absorption coefficient α _EA of the light absorption coefficient adjustment region is changed as a parameter, the maximum reflectance from the distributed reflector region, the peak wavelength shift amount, and the penetration length depend on the initial phase φ (A) shows the maximum reflectance from the distributed reflector region, (B) shows the peak wavelength shift amount, and (C) shows the penetration length.

Explanation of symbols

10: Light modulation area
12: Gain region
14: Phase adjustment area
16: Distributed reflector area
18: Diffraction grating
20: p-side cladding layer
22: n-side cladding layer
24: p-side electrode in the light modulation region
26: p-side electrode in gain region
28: p-side electrode in phase adjustment region
30: p-side electrode in the distributed reflector region
32: n-side common electrode
34, 40, 42: Constant current source (or constant voltage source)
36: AC power supply
38: Constant current source
44: Resonator end face on the distributed reflector region side
46: Resonator end face on the light modulation region side
48: Insulating film
50, 54: Resistance heating film
52, 56: Constant current source
58: Coupler
60: Light absorption coefficient adjustment area
62: p-side electrode in light absorption coefficient adjustment region
64: Constant voltage source
66: Resonator end face on the light absorption coefficient adjustment region side
100, 300: Distributed feedback semiconductor laser
102: Reference clock signal generator
104: 2-branch electric distributor
106: 1/2 divider
108, 126: 1st electric distributor
110: Second electrical distributor
112-1, 112-2, ... 112-N, 124: Electrical amplifier
114-1, 114-2, ... 114-N: Electric signal generator
116-1, 116-2, ... 116-N: LN light intensity modulator
118-1, 118-2, ... 118-N: Continuous wave light source
120-1, 120-2, ... 120-N: Light intensity modulator
122: Multiwavelength optical multiplexer
128-1, 128-2, ... 128-N: MLLD
130, 134: Optical pulse train generator
132, 136: Optical pulse signal generator

Claims (4)

A reference clock signal generator for generating a reference clock signal;
A two-branch electrical distributor that inputs the reference clock signal and branches and outputs the first clock signal and the second clock signal;
An optical pulse train generation unit comprising a plurality of mode-locked lasers that generate and output carrier-suppressed optical pulse trains having different oscillation frequencies from each other;
An optical pulse signal generation unit including a plurality of optical intensity modulators that input each of the plurality of carrier-suppressed optical pulse trains and generate and output carrier-suppressed optical pulse signals having different oscillation frequencies.
A multi-wavelength carrier-suppressed optical pulse signal generation device comprising: a multi-wavelength optical multiplexer that combines and outputs the carrier-suppressed optical pulse signals having different oscillation frequencies. A reference clock signal generator for generating a reference clock signal;
A two-branch electrical distributor that inputs the reference clock signal and branches and outputs the first clock signal and the second clock signal;
An optical pulse train generator that generates and outputs carrier-suppressed optical pulse trains having different oscillation frequencies, and
An optical pulse signal generation unit that inputs each of the plurality of carrier-suppressed optical pulse trains, generates and outputs carrier-suppressed optical pulse signals having different oscillation frequencies, and
A multi-wavelength carrier-suppressed optical pulse signal generator comprising a multi-wavelength optical multiplexer that combines and outputs the carrier-suppressed optical pulse signals having different oscillation frequencies,
The optical pulse train generation unit inputs the first clock signal, and outputs an N-branch (N is an integer of 2 or more) for the (1-1) to (1-N) clock signals. A first electrical distributor;
The first to Nth N mode-locked semiconductor lasers having different oscillation frequencies are provided.
By inputting the (1-1) to (1-N) clock signals to the first to Nth mode-locked semiconductor lasers, respectively, the first to Nth mode-locked semiconductor lasers are mode-locked. Then, the first to Nth mode-locked semiconductor lasers generate and output the first to Nth carrier-suppressed optical pulse trains, respectively,
The optical pulse signal generation unit is configured to input the second clock signal, N branch to (2-1) to (2-N) clock signal and output the second electrical distributor,
First to Nth electrical signal generators that respectively input the (2-1) to (2-N) clock signals and output the first to Nth electrical signals;
1st to Nth N light intensity modulators,
The first to N-th carrier suppression optical pulse trains and the first to N-th electrical signals are respectively input to the first to N-th optical intensity modulators, and the first to N-th electrical signals are used to input the first to N-th electrical signals. Each of the N carrier-suppressed optical pulse trains is optically modulated, and the first to Nth N light intensity modulators respectively generate and output the first to Nth carrier suppressed optical pulse signals,
The multi-wavelength optical multiplexer receives the first to N-th carrier-suppressed optical pulse signals and combines the first to N-th carrier-suppressed optical pulse signals to generate a multi-wavelength carrier-suppressed optical pulse signal. A multi-wavelength carrier-suppressed optical pulse signal generator. The first to Nth mode-locked semiconductor lasers are
A light modulation region having a function of modulating the light intensity, the gain region population inversion is formed, and the phase control region effective refractive index is variable, different f _i (i Bragg frequency from each other from 1 to N And a distributed reflector region in which a diffraction grating is formed, and the light modulation region, the gain region, the phase adjustment region, and the distributed reflector region are connected in series. A distributed feedback semiconductor laser arranged and housed in a resonator, under the condition of outputting an optical pulse train having a repetition frequency of f _rep ,
The resonator length is adjusted so that each oscillation longitudinal mode of the first to Nth mode-locked semiconductor lasers is f _i ± q (f _rep / 2) (q is an odd number).
A distributed feedback semiconductor capable of mode-locking operation by modulating the transmittance of the light modulation region with a frequency f _rep and outputting an optical pulse train whose carrier is suppressed and whose repetition frequency is f _rep 3. The multi-wavelength carrier-suppressed optical pulse signal generation device according to claim 2, wherein the multi-wavelength carrier suppressed optical pulse signal generation device is a laser. The first to Nth mode-locked semiconductor lasers are
A light modulation region having a function of modulating the light intensity, the gain region population inversion is formed, and the phase control region effective refractive index is variable, different f _i (i Bragg frequency from each other from 1 to N A distributed reflector region in which a diffraction grating is formed, a light absorption coefficient adjustment region for adjusting a light absorption coefficient, the light modulation region, the gain region, A distributed feedback semiconductor laser in which the phase adjustment region, the distributed reflector region, and the light absorption coefficient adjustment region are arranged in series and housed in a resonator, and a repetition frequency is f _rep Under the condition of outputting an optical pulse train,
The value of the light absorption coefficient of the light absorption coefficient adjustment region is set to a sufficient size so that the reflected light reflected from the resonator end face on the light absorption coefficient adjustment region side does not reach the distributed reflector region,
The resonator length is adjusted so that each oscillation longitudinal mode of the first to Nth mode-locked semiconductor lasers is f _i ± q (f _rep / 2) (q is an odd number).
A distributed feedback semiconductor capable of mode-locking operation by modulating the transmittance of the light modulation region with a frequency f _rep and outputting an optical pulse train whose carrier is suppressed and whose repetition frequency is f _rep 3. The multi-wavelength carrier-suppressed optical pulse signal generation device according to claim 2, wherein the multi-wavelength carrier suppressed optical pulse signal generation device is a laser.

Images