METHOD FOR GENERATING GIGAHERTZ BURSTS OF PULSES AND LASER APPARATUS THEREOF

Abstract

A method for generating gigahertz bursts of laser pulses is provided, where: 1) time delay T2 of the delayed part with respect to the undelayed part of the input pulse is longer than a time period T1 between said input pulse and the next input pulse; 2) the bursts of output pulses have an incrementally increasing number of pulses; 3) intra-burst pulse separation inside the formed bursts is equal to T3=T2−T1 and corresponds to an ultra-high pulse repetition rate higher than 100 MHz. In another embodiment: 1) T2 is longer than M*T1, where M=2, 3, etc.; 2) output train of bursts is composed of bursts of pulses wherein M adjacent bursts have identical number of pulses; 3) T3 is equal to T3=T2−M*T1. The laser apparatus for implementing the method is provided.

Claims

1. A method for generating bursts of laser pulses from a primary pulse train, comprising delaying a part of an input pulse with respect to another part of the input pulse, wherein sets of undelayed and delayed parts of many input pulses form bursts of output pulses, wherein a time delay (T2) of the delayed part with respect to the undelayed part of the input pulse is longer than a time period (T1) between said input pulse and the next input pulse; said bursts of output pulses have an incrementally increasing number of pulses; an intra-burst pulse separation (T3) inside the formed bursts is equal to T3=T2−T1 and corresponds to an ultra-high pulse repetition rate higher than 100 MHz.

2. A method for generating bursts of laser pulses, which comprises forming delayed parts and undelayed parts of input pulses of a primary pulse train, wherein said delayed and undelayed parts of input pulses form an output train of bursts of pulses, wherein a time delay (T2) of the delayed part of any input pulse with respect to the undelayed part of that input pulse is longer than M*T1, where T1 is a time period between adjacent input pulses, M=2, 3, the output train of bursts is composed of bursts of pulses wherein M adjacent bursts have an identical number of pulses; an intra-burst pulse separation (T3) of the formed bursts is equal to T3=T2−M*T1 and corresponds to an ultra-high pulse repetition rate higher than 100 MHz.

3. The method according to claim 1, wherein a delay line for delaying parts of input pulses is incorporated into an optical chain with an element with two input ports and two output ports: the first output port is connected with a first input port via said delay line, the second input port receives input pulses of the primary pulse train, the second output port delivers output train of bursts; input pulses and their delayed parts received at the input ports are divided between the first and the second output ports with a certain splitting ratio; parts of pulses are amplified during propagation in said delay line; parts of pulses propagating in the delay line are blocked under control of a drive signal (D2).

4. The method according to claim 1, wherein the primary pulse train comprises a narrowband radiation and the input pulses are transform limited pulses with duration in the range between 10 ps and 10 ns.

5. The method according to claim 1, wherein the primary pulse train comprises a broadband radiation with a bandwidth corresponding to a transform limited pulse duration in the range below 10 ps, and dispersion management is accomplished during pulse propagation in the delay line; pulses of the output train of bursts have identical durations.

6. The method according to the claim 5, wherein pulses of the primary pulse train are transform limited.

7. A laser apparatus for implementing a method for generating bursts of laser pulses from a primary pulse train, comprising a seed source, a burst synthesizer and a power amplifier.

8. The laser apparatus according to the claim 7, further comprising: an optical switch positioned between the seed source and the burst synthesizer, and controlled by a drive signal (D1), an optical switch positioned between the burst synthesizer and the power amplifier and controlled by a drive signal (D3).

9. The laser apparatus according to claim 7, wherein the burst synthesizer is composed of an active fiber loop and a fiber coupler with a first and a second input ports and a first and a second output ports, wherein the active fiber loop characterized by an optical length (L.sub.opt) connects the first output port of the fiber coupler with its first input port and introduces a time delay (T2) of a part of the input pulse with respect to another part of said input pulse; and said active fiber loop contains: an isolator; a wavelength division multiplexer; a laser diode; a segment of a doped optical fiber; an optical switch controlled by a drive signal (D2); a segment of an undoped optical fiber; wherein the segment of the doped optical fiber is used for pulse amplification, while the optical switch is used to block or attenuate parts of pulses propagating in the active fiber loop.

10. The laser apparatus according to claim 9, wherein the active fiber loop also contains: a circulator and a side arm comprising a dispersion compensation element; wherein said dispersion compensation element is designed to compensate a chromatic dispersion of the active fiber loop at the end of a full round-trip.

11. The laser apparatus according to claim 9, wherein the active fiber loop also comprises: a circulator and a side arm comprising at least one optical router controlled by drive signals (D4, D5) and at least one additional segment of undoped optical fiber; wherein said at least one optical router is intended to direct pulses into a selected additional segment of undoped optical fiber; a total optical length (L.sub.opt) of the active fiber loop and the time delay (T2) depend on a rout which the pulse propagates inside the active fiber loop.

12. The laser apparatus according to the claim 11, wherein the active fiber loop also contains: at least one dispersion compensation element; wherein said at least one dispersion compensation element is designed to compensate a chromatic dispersion of the active fiber loop at the end of a full round-trip.

13. The laser apparatus according to claim 8, wherein the optical switch attenuates selected pulses of the primary pulse train.

14. The laser apparatus according to claim 8, wherein the optical switch blocks selected pulses of the primary pulse train.

15. The laser apparatus according to claim 8, wherein the optical switch blocks selected bursts of the output train of bursts.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0035] FIG. 1A-FIG. 1B. Schematic structure and operation principle of a fiber-optic master oscillator power amplifier (MOPA) (100) of this invention, comprising the following parts: [0036] seed laser (1); [0037] burst synthesizer (2); [0038] amplifier (3).

[0039] FIG. 1C. Schematic structure and operation principle of a practical realization of the MOPA (101) of this invention, comprising the following parts: [0040] seed laser (1); [0041] burst synthesizer (2); [0042] amplifier (3); [0043] optical switches (7, 8).

[0044] FIG. 2A-FIG. 2B. Burst synthesizer (2) of this invention, composed of an active fiber loop (2a) connected by a fiber coupler (70). The active fiber loop (2a) comprises the following parts: [0045] isolator (75); [0046] wavelength division multiplexer (76); [0047] laser diode (80); [0048] doped fiber (77); [0049] optical switch (78); [0050] undoped optical fiber (79).

[0051] FIG. 3—time frames illustrating an operation of the burst synthesizer of this invention.

[0052] FIG. 4A-FIG. 4B. Burst synthesizer (2) of this invention for ultrashort pulses, containing an element for dispersion compensation. A list of parts is as follows: [0053] fiber coupler (70); [0054] circulator (83); [0055] segment of undoped optical fiber (87); [0056] dispersion compensation element (81); [0057] wavelength division multiplexer (76) if necessary; [0058] laser diode (80); [0059] doped fiber (77); [0060] optical switch (78); [0061] segment of undoped optical fiber (79).\

[0062] FIG. 5A-FIG. 5C. Embodiments of the burst synthesizer with a selectable intra-burst pulse separation. A modified active fiber loop (2b) comprises elements of the active fiber loop (2a) and additional elements: [0063] multiple dispersion compensation elements (81a, 81b, 81c); [0064] multiple segments of undoped optical fibers (88a, 88b, 88c, 89a, 89b, 89c) of different lengths; [0065] optical router (90) with multiple output ports or several optical routers (90a, 90b) with two output ports.

[0066] FIG. 6. Burst synthesizer (2) of this invention, composed of a long active fiber loop (2c) connected by a fiber coupler (70). The active fiber loop (2c) comprises elements of the active fiber loop (2a) or the modified active fiber loop (2c), but introduces a larger delay.

DETAILED DESCRIPTION OF THE INVENTION

[0067] The goal of this invention is to generate bursts of short laser pulses with a selected pulse repetition rate, higher than that of the seed source, particularly in the range from 100 MHz up to 1 THz. It is another goal of the invention to generate bursts of ultrashort laser pulses, especially in the range of a few picoseconds or hundreds of femtoseconds. It is a special goal of the invention to obtain bursts of pulses with a controllable number of pulses, identical pulse duration and separation, and a desired shape of pulse burst amplitude envelope.

[0068] We are proposing a method for synthesizing bursts of pulses with the help of an active fiber loop comprising fiber-optic components for pulse delay, amplification and dispersion control. Burst termination element is also incorporated inside said active fiber loop. A preferred implementation of the method is used in a pulsed master oscillator power amplifier laser apparatus. Generation of GHz range pulse repetition rate bursts of a desired length and amplitude envelope is possible.

[0069] FIG. 1A-FIG. 1B illustrate very simplified block diagrams of this invention. The goal is to have a method for burst (or a train of bursts) formation from a primary pulse train. A burst synthesizer 2 incorporated between a seed laser 1 and an amplifier 3 makes a basis of a GHz range MOPA source 100 generating bursts of short or ultrashort laser pulses. The seed laser 1 may be a single unit (a fiber or solid-state pulsed oscillator, or a laser diode) or may be composed of several units (oscillator followed by a pre-amplifier or oscillator followed by a pre-amplifier and additional components such as pulse stretcher, etc.). The term “short” hereinafter refers to a pulse duration below 10 ns. The term “ultrashort” hereinafter refers to a pulse duration below 10 ps. The term “pulse” is understood as a discrete onset of electromagnetic radiation separated by an inter-pulse period when light is absent or negligible. The output from the seed laser 1 is understood herein as a “seed pulse” or a “train of seed pulses” (a primary pulse train 4 in FIG. 1B). A “train” of pulses refers to a periodic appearance of pulses; separation between adjacent pulses may be a long period of time compared to the time scale of the pulses. A “burst of pulses” refers to a series of any number of consecutive pulses with separation compared to the time scale of the pulses, while a “train of bursts” refers to a periodic or aperiodic series of bursts of pulses.

[0070] A mode-locked fiber oscillator is one of the choices for obtaining a primary train 4 of ultrashort optical pulses at high repetition rates in the range from tens to hundreds of MHz (pulse separation T1). The burst synthesizer 2 of this invention forms a train 5 of bursts wherein an intra burst pulse separation T3 is not necessarily a divisor of T1 and may be very small (ultra-high pulse repetition rate). The burst synthesizer 2 divides each incoming pulse into two parts and performs a delay for one of the parts. Also, additional control of amplitude and dispersion is performed. A delay T2, introduced by the burst synthesizer 2, is longer than the pulse separation T1 within a primary pulse train 4. Therefore, a delayed part 41.2 of the first incoming pulse 41 is outcoupled later than a non-delayed part 42.1 of the second incoming pulse 42. The first “burst” 51 is made of a single pulse (a non-delayed part 41.1 of the incoming pulse 41); the second burst 52 is a doublet of 42.1 and 41.2 parts of pulses. In the same manner consequent bursts 53, 54, 55 with an incremental number of pulses are formed. A detailed description of burst formation will be given below. It is possible to provide a constant amplitude of the pulses within a burst and among the whole train of bursts 51-55 of pulses. In a simplest embodiment, the amplitude of bursts 51-55 is equal to a half of the amplitude of the primary train 4 of pulses 41-45. The power amplifier 3 provides an amplified train 6 of bursts. The intra-burst pulse separation T3 is constant and is equal to T3=T2−T1. It is identical in all bursts, while the first pulse of each burst comes with a constant time period T1. Amplitude of pulses in the amplified train 6 of bursts depends on amplification conditions.

[0071] FIG. 1C illustrates a block diagram of a more sophisticated MOPA 101 of this invention and its time chart. This configuration is more practical when, for example, a constant number of pulses in all bursts is needed. Therefore, other bursts which contain a smaller number of pulses or higher number of pulses should be eliminated from the output train. The MOPA laser apparatus is supplemented with an optical switch 7 interposed between the seed source 1 and the burst synthesizer 2. Also, another optical switch 8 is interposed between the burst synthesizer 2 and the amplifier 3. Optical switches 7, 8 can operate as variable optical attenuators and/or as pulse pickers.

[0072] For example, only those bursts that contain four pulses are needed. In the given illustration, the optical switch 7 does nothing with a pulse repetition rate, therefore, a train 9 is identical to the train 4 of the seed source. Drive signals D1, D3 turn the optical switches 7, 8 into “ON” (highest transmission), “OFF” (zero transmission) or intermediate (partial transmission for attenuation, if necessary) states. The burst synthesizer 2 has its own optical switch inside which is controlled by a drive signal D2. The drive signal D2 governs a start and termination of “an evolution of a burst”: when the signal D2 is at the “ON” state, every consequent burst has a number of pulses increased by one; when the state is “OFF” evolution of a burst stops; when D2 is “ON” again, the next burst has one pulse again. The next burst having four pulses is formed after four cycles (time interval T4=B*T1, B—number of pulses within a burst). A train 10 at the output of the burst synthesizer 2 now has multiple subsets (sub-trains) 10.1, 10.2, etc. with an increased number of pulses; each subset lasts until the desired number of pulses within a burst is obtained. If only those bursts that contain four pulses are needed, the drive signal D3 turns the optical switch 8 into “ON” state only when required bursts 21-24 are formed. The train 11 of bursts 21-24 is then sent to the amplifier 3 where it is amplified; and an amplified train 12 of bursts is formed.

[0073] The burst synthesizer 2 of this invention has a property to form a desired shape of amplitude envelope of a burst. The amplitude of pulses is controlled by amplification conditions and at some extent by the optical switches. In order to form bursts 31-34 with a constant amplitude A.sub.out at the output of the whole system, bursts 21-24 with a growing amplitude of pulses should enter the power amplifier 3. The first pulse 21.1 of the burst 21 at the input of the power amplifier 3 has the lowest amplitude A.sub.1; the last pulse 21.4 of the burst 21 has the highest amplitude A.sub.4. The first pulse 21.1 enters the amplifier with an undepleted gain medium and is amplified better than consequent pulses 21.2-21.4. Amplification of the last pulse 21.4 is weakest due to gain saturation (depletion of the gain medium).

[0074] The mentioned burst amplitude shaping is achieved by adjusting amplification conditions. Amplitude A.sub.1 of the first pulse 21.1 is equal to a half of the amplitude A.sub.seed of the seed pulse 41, while other pulses 21.2-21.4 of the burst 21 have higher amplitudes since they are amplified during propagation in a delay line of the burst synthesizer 2. Pulses are depicted as infinitely short onsets of electromagnetic radiation in FIGS. 1A-1C. Of course, they have certain duration which is short or ultrashort. Whereas inter-pulse separation T3 corresponds to ultra-high, preferably gigahertz, pulse repetition rate.

[0075] An initial time moment t=0 is selected to be a moment when the first part 41.1 of the first pulse 41 crosses a reference plane. The optical switch of the burst synthesizer 2 may be switched on (drive signal D2) at any time between time moments t1 and t2 but obligatorily before a second part 41.2 of the first pulse 41 reaches said optical switch. The optical switch of the burst synthesizer 2 may be switched off at any time between time moments t3 and t4 but obligatorily before four delayed pulses circulating inside the burst synthesizer 2 reach the switch.

[0076] Different number of pulses within a burst can be synthesized in the same manner. Also, burst selection—with the optical switch 8 positioned downstream of the burst synthesizer 2—can be performed according to the needs of a custom application. Maximum repetition rate of identical bursts depends on a desired intra-burst number of pulses and initial pulse repetition rate. Another aspect is related to a duration of a burst. To avoid overlap of the generated burst with the next burst, burst duration have to be shorter than an inter-pulse separation of the incoming train to the burst synthesizer 2. In case of necessity of longer bursts, repetition rate must be reduced by the optical switch 7 positioned upstream of the burst synthesizer 2. The pulse train 9 of a modified pulse repetition rate would be formed.

[0077] The method given in FIG. 1C allows for amplitude control of individual pulses. The internal optical switch of the burst synthesizer driven by the signal D2, as well as the optical switch 7 driven by the signal D1 may have partial transmission and, therefore, may be used for amplitude control. It is easier to attenuate amplitudes of individual pulses with the optical switch 7, positioned upstream of the burst synthesizer 2, because every new input pulse comes with the MHz range pulse repetition rate which is a relatively low rate to change a transmission value for standard optical switches.

[0078] A burst synthesizer 2 according to this invention comprises (FIG. 2A): a 2×2 fiber coupler 70 and an “active fiber loop” 2a. The fiber coupler 70 has two input ports 71, 72 and two output ports 73, 74. The loop is formed by connecting the first output port 73 with the first input port 71. A primary pulse train 4 is injected into a second input port 72 of the fiber coupler 70, while an output radiation 5 is outcoupled through the second output port 74.

[0079] FIG. 2B demonstrates a detailed scheme of the burst synthesizer 2 depicting all elements of the active fiber loop and explaining the basic principle of operation. Each input pulse 41-43 is divided into two parts: one part is outcoupled through the second output port 74, while another part is delivered to the first output port 73 of the coupler and consequently to the active fiber loop 2a. Therefore, at the next cycle all radiation that enters input ports 71 and 72 is divided between output ports 73 and 74. An output train 5 of bursts 51-53 is formed from pulse parts that are coupled to the second output port 74. Other components of the active fiber loop are: an isolator 75, wavelength division multiplexer (WDM) 76, a segment of doped fiber 77, optical switch 78 and a segment of an undoped optical fiber 79. The isolator 75 prevents from backscattered light propagating back towards the fiber coupler 70. The WDM 76 is used for coupling the pump light from a laser diode 80 into the doped fiber 77. A core-pumping or a cladding-pumping may be used. Radiation propagating from input 77.1 towards output 77.2, is amplified in the doped fiber. Amplification is necessary for obtaining bursts of pulses with equal or controllable amplitude.

[0080] A delay T2 inside the active fiber loop 2a is determined from a pulse period T1 of the primary pulse train 4 and a desired intra-burst pulse separation T3 of the output train 5 of bursts as follows: T2=T1+T3. A required total optical path length L.sub.opt of the active fiber loop 2a is calculated from T2 or determined experimentally by measuring a time delay. In order to obtain GHz bursts, a physical length L of the loop is around 4 m for 50 MHz repetition rate of master oscillator. The segment of an undoped fiber 79 is spliced into a loop in order to achieve the required total length L. The elements 75 and 76 may be separate fiber-optic components, or a special hybrid wavelength division multiplexer-isolator (WIDM) component. The optical switch 78 is a fiber coupled acousto-optic modulator or electro-optic modulator, electronically controllable device that allows to vary amplitude and duration of a burst. Adjustment of an optical path length of the fiber loop 2a can be performed by heating or cooling the undoped fiber 79 or a part of it. That allows to fine tune the delay T2 in case of using GHz bursts with external resonators, e.g., for SOPO pumping. Alternatively, mechanical stretching of the undoped fiber 79 may be used for adjustment of the optical path length of the fiber loop 2a, hence, a delay T2. If all components are polarization-maintaining, the output laser radiation is linearly polarized.

[0081] In a first embodiment of this invention, the splitting ratio of the fiber coupler 70 is 50/50. That means that pulses entering the coupler's 70 input ports 71 and 72 are divided into equal parts and delivered to the output ports 73 and 74. Parts of pulses that are delivered to the port 73 propagate through the fiber loop and enter the port 71 again. They are divided and delivered to ports 73 and 74 again.

[0082] The whole operation is explained in FIG. 3 with the help of the following time frames: t=0, t=T1, t=2*T1. The simplest case is analyzed: bursts 51-53 of pulses of equal amplitudes are synthesized; the splitting ratio of the fiber coupler 70 is 50/50. Pulses are observed in the vicinity of reference planes A, B, C, D positioned on the input ports 71, 72 and output ports 73, 74 equally distant from the fiber coupler 70 (see FIG. 2B). In the reference plane A, radiation at the end of the fiber loop is observed. In the reference plane B, input radiation (pulse train 4 of seed pulses 41, 42, 43 is observed). In the reference plane C, a part of radiation that is outcoupled through the first output port 73 is observed. In the reference plane D, a part of radiation that is outcoupled through the second output port 74 is observed. All reference planes are now overlaid (as if the fiber coupler 70 has no size and does not introduce any delay).

[0083] The first time frame (t=0) illustrates that the first input pulse 41 (reference plane B) is divided into two equal parts. One part 41.2 is delivered into the active fiber loop (reference plane C). The other part 41.1 is delivered to the output (reference plane D); it is the first pulse of output radiation 5. There is no radiation at the end of the fiber loop (reference plane A) at this time moment.

[0084] The second time frame (t=T1) illustrates a moment when a second input pulse 42 reaches a reference plane B. A pulse 98 that propagates inside a fiber loop has not reached the reference plane A yet. The pulse 98 is obtained from the pulse part 41.2 by amplifying it (its amplitude is controlled by the doped fiber 77 and the optical switch 78; see FIG. 2B). Other characteristics of radiation may also be influenced by the fiber-optic components of the fiber loop, however, it is desired that their net action compensates each other and only amplitude is doubled. Radiation that is delivered to the loop again (reference plane C) is composed of a part 42.2 of the pulse 42 and a part 98.2 of the pulse 98. Radiation that is delivered to the output (reference plane D) is composed of a part 42.1 of the pulse 42 and a part 98.1 of the pulse 98. Pulse parts 98.2 and 98.1 are delayed with respect to the pulse parts 42.2 and 42.1. Pulses 42.1 and 98.1 form a burst 52 of pulses of output radiation 5. Amplitudes of 42.1 and 98.1 are equal.

[0085] The third time frame (t=2*T1) illustrates that when a third input pulse 43 reaches a reference plane B, pulses of radiation 99 that propagate inside a fiber has not reached the reference plane A yet. The radiation 99 at the end of the active fiber loop is formed from pulses 42.2, 98.2 by doubling their amplitudes. Radiation that is delivered to the loop again (reference plane C) is composed of a part 43.2 of the pulse 42 and a part 99.2 of radiation 99. Radiation that is delivered to the output (reference plane D) is composed of a part 43.1 of the pulse 43 and a part 99.1 of radiation 99; it is a burst 53 of pulses of output radiation 5. In the same manner other bursts of output radiation 5 are formed.

[0086] An embodiment of the burst synthesizer 2 depicted in FIG. 2B has no dispersion compensation elements. Therefore, it is suitable for narrowband radiation corresponding to transform-limited pulse duration longer or equal to 10 ps. If short bursts are needed, the burst synthesizer of FIG. 2B may be used for radiation with intermediate bandwidth. The necessity to compensate dispersion also depends on the length of the burst.

[0087] For example, in order to form a burst of 11 pulses, the part 41.2 of the first input pulse 41 circulates 10 round-trips inside the active fiber loop; the part 42.2 of the second input pulse 42 circulates 9 round-trips; etc.

[0088] A transform-limited 5 ps (bandwidth ˜0.31 nm) Gaussian-like pulse after 10 round-trips inside a 4 m fiber loop stretches up to 5.02 ps, after 30 round-trips—up to 5.17 ps, after 50 round-trips—up to 5.47 ps. In certain applications, it may still be allowable variation of the pulse width within a 50-pulse burst.

[0089] A transform-limited 1 ps (bandwidth ˜1.55 nm) pulse after 10 round-trips inside a 4 m fiber loop stretches up to 2.40 ps, but after 30 round-trips stretches up to 6.68 ps. Only short bursts of 1 ps pulses may be accomplished with the burst synthesizer of FIG. 2B without dispersion managing components.

[0090] FIG. 4A illustrates another embodiment of the burst synthesizer 2 of this invention. It is preferred for broadband radiation since contains a dispersion compensation element 81. Control of dispersion is necessary for broadband radiation (ultrashort transform-limited pulses or short chirped pulses). Pulses that are delayed travel a longer path inside dispersive media than those that are undelayed. Due to chromatic dispersion various spectral components undergo different shifts of their phases. As a result, duration of the delayed pulses may be changed and the pulse shape may be distorted. Every consequent pulse of any burst travels a longer path. A difference of spectral phases is accumulated. That means that a second pulse of the burst will differ from the first pulse, a third pulse will be different from the second and the first pulse, etc. In order to compensate the differences in spectral phases and to obtain bursts of pulses with identical durations, the element 81 with an opposite dispersion to that of fibers within the active fiber loop (elements 71-79) is needed. If the seed source is Yb-doped laser and radiates around 1 p wavelength, most optical fibers have positive (normal) dispersion at this spectral range. Dispersion of the element 81 must be negative (anomalous). For Tm- or Ho-doped laser sources generating around 2 pm, the burst synthesizer 2 must contain the dispersion compensation element 81 with positive (normal) dispersion. The most possible choice for dispersion compensation elements in fiber optical systems is a chirped fiber Bragg grating (CFBG), however, other types of dispersion compensation elements may be used (e.g. chirped volume Bragg grating (CVBG) or photonic crystal fiber). Since radiation power inside the active fiber loop is relatively low and nonlinear effects are negligible, a second order of chromatic dispersion (group velocity dispersion GVD) is mostly observed. CFBGs with any selected chirp parameter for compensation of the second order dispersion are available on the market. Therefore, phase shifts experienced during propagation inside the active fiber loop is totally compensated. CFBGs with compensation of higher orders of dispersion are also available. They should be used if non-linear processes inside a fiber loop, mainly in the doped fiber 77 or the consequent elements, cannot be ignored. The only case when compensation is unnecessary is a vicinity of zero-dispersion point.

[0091] The path inside the active fiber loop of FIG. 4A is as follows: radiation from a path 82 is diverted by an element 83 into a path 84, propagates inside the dispersion compensation element 81, is diverted to optical path 85 and path 86. If the dispersion compensation element 81 is a chirped fiber Bragg grating, it has a single fiber connector and operates as a mirror, i.e. reflects the light back to the same fiber (the reflected light is already dispersion-managed). Therefore, paths 84 and 85 are in the same optical fiber 87, and the element 83 is a fiber circulator. It also diverts the light from the path 85 to the path 86. The circulator 83 simultaneously works as an isolator, prevents from light travelling in a backward direction. The side arm (elements 83, 87, 81) of the active fiber loop 2a in FIG. 4A is located just after the fiber coupler 70. Therefore, dispersion control is done before the pulse propagates in the dispersive loop; i.e. dispersion management is done in advance (it is called pre-compensation). Duration of the pulses vary within a loop. At an ideal compensation case, pulse duration at the end of the loop is identical to that of the pulse part that is outcoupled through 74 without propagation through the fiber loop. It is also possible to insert the side arm (elements 83, 87, 81) in other places of the active fiber loop 2a but not recommended between the WDM 76 and the doped fiber 77. The dispersion of the chirped fiber Bragg grating 81 may be controlled by placing it into a heated mounting. It allows for a fine tuning of the dispersion and hence pulse duration.

[0092] There are several aspects when selecting a proper position for the side arm responsible for dispersion control. Let's analyze a case of FIG. 4A—the side arm is before the amplification system (elements 76, 77, 80). If the initial pulses (pulses of train 4) and their parts that are coupled into the active fiber loop are not chirped, then the side arm of the loop chirps them. This means that amplification in the doped fiber 77 is done for stretched pulses and non-linear effects are less observable. If the initial pulses are chirped, then the side arm can reduce their chirp and reduce their duration. Due to the increased peak power, non-linear effects could be more possible. If the side arm would be place in other locations, isolator is necessary, or WIDM instead of WDM.

[0093] FIG. 4B illustrates yet another embodiment of the burst synthesizer 2 with dispersion compensation. Herein, the segment of doped fiber 77 is inserted in a side arm of the active fiber loop. Hence, radiation propagates through the amplifying medium twice and the inversion is used more efficiently. Pumping radiation from the laser diode 80 is coupled through the dispersion compensation element 81, thus the WDM is unnecessary. The burst synthesizer of FIGS. 4A-4B are suitable for pulses for <1 ps. Other alternative configurations of the active fiber loop 2a of the burst synthesizer 2 may be obvious to the skilled in the art.

[0094] FIGS. 5A-5C illustrate embodiments of the burst synthesizer 2 with a selectable intra-burst pulse separation. An active fiber loop 2b of the burst synthesizer has a selectable length. There are several ways to accomplish the selectable length. The loop may have an additional side arm with optical length control, or an existing side arm may be supplemented with an optical router (a type of an optical switch which selectively directs optical signals from one channel to another).

[0095] FIG. 5B depicts a part of the active fiber loop 2b with the selectable length. The whole modified active fiber loop 2b comprises elements of the active fiber loop 2a and several additional elements. A side arm 87 of the loop connected to the main loop with the help of the circulator 83 ends by a segment 88a, 88b or 88c of an undoped optical fiber. An optical router 90 which has multiple output ports, directs radiation to one of alternative segments 88a-88c. Various lengths of the active fiber loop 2b are achieved by selecting lengths of multiple segments of fibers. A total length of the loop 2b when radiation propagates through segments 87 and 88a is equal to LI. A total length of the loop 2b when radiation propagates through segments 87 and 88b is equal to L2. A total length of the loop 2b when radiation propagates through segments 87 and 88c is L3. Different lengths of the active fiber loop 2b ensure a possibility to select one of alternative intra-burst pulse repetition rates. Selection is done by controlling the router 90 with a drive signal D4. For ultrashort pulses, it is important to compensate dispersion acquired during propagation in fibers. Therefore, different lengths of the active fiber loop 2b requires different dispersion compensation elements 81a, 81b, 81c. For short pulses, dispersion management is not so crucial (especially if LI, L2 and L3 do not differ significantly); identical dispersion compensation elements 81a, 81b, 81c or a single element may be used. Various types of optical routers having multiple output ports are suitable.

[0096] FIG. 5C depicts a more practical realization of the active fiber loop 2b with a selectable length. Several 1×2 optical routers (90a, 90b) are used. Alternative lengths of the active fiber loop 2b are achieved by selecting lengths of multiple segments of fibers: LI—when radiation propagates through segments 87 and 88a; L2—when radiation propagates through segments 87, 89a and 89b; L2—when radiation propagates through segments 87, 89a and 89c. In the same manner burst synthesizer which has many selectable intra-burst pulse repetition rates may be realized.

[0097] FIG. 6 depicts an alternative method to form bursts of pulses. The burst synthesizer X which has a long active fiber loop 2c such that ensures a delay of a pulse by more than two periods T1 can form a train of bursts wherein multiple consequent bursts contain the same number of pulses. If a length of the active fiber loop 2c satisfies the rule: T2=2*T1+T3, two consequent bursts have the same number of pulses (see FIG. 6). Bursts 91, 92 of the first series E of bursts, contain one pulse; bursts 93, 94 of the second series F of bursts, contain two pulses, etc. Two identical bursts appear with separation T1 equal to inter-pulse separation of the primary train. It may be an advantage in some applications where identical bursts repeating with small time interval are needed. However, the time for forming multiple bursts of many pulses increases. Nevertheless, if a certain application needs several bursts and can afford long periods between next series of several identical bursts, the burst synthesizer X is perfectly suitable. A time period T5 which is needed to wait for obtaining the first burst 93 containing two pulses is: T5=2*T1. A time period to obtain the first burst containing three pulses is equal to 3*T1. The rule for synthesizing M consequent bursts repeating at period T1 is to make a long active fiber loop 2c which satisfies the rule: T2=M*T1+T3. A time period which is needed for obtaining the first burst containing N pulses is equal to N*T1.

[0098] A distinctive feature and the advantage of the burst synthesizer and MOPA of this invention is that relatively slow amplitude modulators may be used. It is possible to control an amplitude of individual pulses in GHz burst with very high precision using the optical switch 7 since it must be faster than the full round-trip time of the fiber loop. For example, for 50 MHz repetition rate of a seed source, an opening and closing times of the optical switch have to be shorter than 20 ns. It is a moderate requirement for fiber-optic modulators. A slowly varying amplitude envelope of the burst may be successfully accomplished with the internal optical switch 78. Yet another advantage of this invention is that a number of pulses within a burst can be selected and changed by only driving said optical switches. There are no limitations for pulse repetition rate. The length of the active fiber loop is in the meters range while intra-burst pulse separation is in the range of gigahertz or even terahertz. The active fiber loop contains all the necessary components for pulse amplitude control and dispersion compensation. Bursts of pulses shorter than 1 ps can be generated with identical pulse duration and repetition rate. A variation of the method of this invention (given below) allows for synthesizing long bursts which are particularly attractive for certain applications. The only limitation in burst length is then a precision of dispersion compensation.

[0099] According to the preferred embodiment of this invention, pulses of each burst are not overlapped. That means that the minimum intra-burst pulse separation T3 is equal to pulse duration. For example, pulses with 10 ps duration can be packed without overlapping with 100 GHz repetition rate; 1 ps pulses allow 1 THz. If chirped pulses with 150 ps duration are used, 6 GHz burst may be generated.

[0100] Applications of pulsed lasers operating in burst-mode nowadays involve not just a variety of micromachining and material processing operations. Similar advantages are obtained in radiation interactions with biological objects: the amount of information collected per unit of time can be increased without exceeding the damage threshold of the biological object. Bursts of short pulses are used to generate packets of photoelectrons by illuminating photocathodes of electron accelerators. For such applications, the exact distance between light pulses is important, which is relatively easy to achieve using the method of this invention. High repetition rates open new possibilities in nonlinear frequency conversion in gas, liquids and crystals. Design and complexity of synchronously pumped OPOs or coherent pulse stacking systems is highly dependent on repetition rate precision of the pump radiation.

Other Embodiments

[0101] Long bursts which are longer than T1 may be formed as follows: 1) a first part of the burst is formed in the same manner as described above; 2) when a set of pulses that circulates inside a loop fills the entire length of the loop, the optical switch (7 in FIG. 1C) blocks the next input pulse. Pulses circulating inside the loop are outputted during the next cycle and join outputted at previous cycle thus making a continuous burst of pulses. By keeping the input pulses blocked a long burst can be formed. Since amplification is done in the active fiber loop, pulse amplitude does not decrease. Since dispersion management is accomplished in the active fiber loop, pulse duration is preserved.

[0102] For example, if intra-burst pulse separation is 0.5 ns (2 GHz), 41 pulses fill the 4 m-length fiber loop. A 80-pulse length burst can be formed within a time period 39*T1: 39 cycles incrementally grow a 40-length burst and then during the next cycle nothing is inputted to the active fiber loop, but the 40 pulses circulating inside the loop join previously outputted. If after the next T1 period, one more input pulse is blocked by the optical switch 7, the burst is elongated by additional 40 pulses from the loop.

[0103] Time interval needed for long burst formation is equal to a time that is required to fill the entire length of the loop with pulses. The only limitation is a closing time of the optical switch 78 of the active fiber loop. The switch 78 is used for termination of the burst. Since pulses inside the loop are closely packed in time (GHz pulse repetition rate) the switch's 78 closing time will shape a tail of the burst. E.g., if we generate a 120-pulse burst with 1 ns intra-burst period and use an acousto-optic modulator with 6 ns closing time, we obtain 114 pulses of the same amplitude and 6 pulses with decaying amplitude.

[0104] Another way to form long bursts is: 1) with the help of the optical switch 7 to form a train 9 with large pulse separation—larger than a duration of a desired long burst; 2) to use a long active fiber loop which introduced a time delay slightly longer than said large pulse separation. The idea is to form the whole long burst in the manner described in FIG. 1 A-FIG. 3: the first pulse of the burst is a part of an input pulse, other pulses of the burst come from the loop; they are parts of input pulses of previous cycles. An advantage of this way to synthesize long bursts is that amplitude of every single pulse is controlled and burst envelope of rectangular shape is obtained.

[0105] The method and laser apparatus of this invention can also be used for synthesizing a long rectangular pulse from several 10 ps-10 ns transform-limited pulses. If the length of the active fiber loop is such that a delayed part 41.2 of the input pulse 41 (see, for example, FIGS. 1B-1C) overlaps the undelayed part 42.1 of the next input pulse 42 a result is an elongated pulse. If T3 is less than pulse duration, one may obtain an M-shaped pulse or a rectangular-like pulse.

[0106] Application of the method of this invention in broadband laser sources has yet additional aspects. Broadband radiation of the seed source may be delivered to the burst synthesizer of this invention as a train of transform-limited pulses or as a train of chirped pulses. A variation of the MOPA of this invention comprises a pulse stretcher and compressor. The pulse stretcher inserted between the seed source 1 and the burst synthesizer 2 (in the MOPA 100 of Fig. IB) or between the seed source 1 and the optical switch 7 (in the MOPA 101 of FIG. 1C) chirps and, therefore, stretches pulses in time. If the time period T3 is less than duration of the chirped pulses, pulses overlap in the fiber coupler of the burst synthesizer 2. Due to interference of the delayed part of an input pulse and the next input pulse, spectrum of radiation is altered. Interference of many input pulses results in some spectrum modulation of radiation at the output of the burst synthesizer 2 of this invention. The pulse compressor added to the MOPA after the power amplifier 3 compresses radiation and results in ultrashort pulses. Output pulse envelope is modulated too.

[0107] There are many other alternative configurations of the active fiber loops 2a, 2b or 2c of the burst synthesizer 2 not illustrated in the given figures which does not depart from the main idea of this invention. In principle, the same ideas can be realized in a free-space equivalent of the loop, but there is a need for control of diffraction in order to avoid beam spreading while circulating multiple round-trips inside the loop. Also, there are many alternative configurations of the MOPA not illustrated in the given figures which does not depart from the main idea of this invention. The whole scope of the invention is defended by the appended claims.