WO2002093701A1

WO2002093701A1 - Controller calibration for small form factor sampled grating distributed bragg reflector laser

Info

Publication number: WO2002093701A1
Application number: PCT/US2002/015450
Authority: WO
Inventors: Larry A. Coldren; Michael C. Larson
Original assignee: Agility Communications, Inc.
Priority date: 2001-05-15
Filing date: 2002-05-15
Publication date: 2002-11-21
Also published as: WO2002093701A8

Abstract

Controller calibration methods for use with sampled grating distributed Bragg reflector SGDBR laser (102) is presented. An exemplary method includes conducting a two-dimensional mirror current scam of each front mirror current setting and back mirror current setting for a sampled grating distributed Bragg reflector SGBDR laser(102) to produce laser setting data corresponding to each front mirror current setting and back mirror current setting to generate a reference optical signal (114) of the SGDBR laser (102). A channel operating point is determined for each channel within the two-dimensional scan data. A fix up of the operating point to substantially minimize wavelength and power error can also be performed. A two-dimensional control surface is characterized at the channel operating point for each channel. A lookup table for controlling the SGDBR (102) laser is generated from the operating point currents, locker values and two-dimensional control surface data from each channel.

Description

CONTROLLER CALIBRATION FOR SMALL FORM FACTOR SAMPLED GRATING DISTRIBUTED BRAGG REFLECTOR LASER

CROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims the benefit under 35 U.S.C. §119(e) of the following co- pending and commonly-assigned U.S. patent applications which are incorporated by reference herein:

[0002] Provisional Application Serial No. 60/291,481, filed May 15, 2001, by Larry A. Coldren and Michael C. Larson, entitled "CONTROLLER CALIBRATION FOR SMALL FORM FACTOR SGDBR LASER," with attorneys' docket number 122.10-US- Pl; and

[0003] Provisional AppHcation Serial No. 60/291,375, filed May 15, 2001, by Larry A. Coldren and Paul F. Crowder, entitled "SGDBR LASER CONTROLLER," attorneys' docket number 122.11-US-P1.

[0004] This application is a continuation-in-part patent application of the following co- pending and commonly-assigned U.S. patent application which is incorporated by reference herein:

[0005] Utility AppHcation Serial No. 09/895,303, filed on June 29, 2001 , by Gregory A. Fish and Larry A. Coldren, entitled "GAIN VOLTAGE CONTROL OF SAMPLED GRATING DISTRIBUTED BRAGG REFLECTOR LASERS," attorneys' docket number 122.6-US-U1, which claims the benefit under 35 U.S.C. §119(e) of Provisional AppHcation Serial No. 60/215,742, filed June 29, 2000, by Paul F. Crowder and Larry A. Coldren, entitled "GAIN VOLTAGE CONTROL OF SGDBR LASERS," attorneys' docket number 122.6-US-P1.

[0006] This appHcation is related to the following co-pending and commonly-assigned U.S. patent appHcations which are incorporated by reference herein: [0007] Utility AppHcation Serial No. 09/895,848, filed June 29, 2001 , by Paul F. Crowder, entided "OPEN LOOP CONTROL OF SGDBR LASERS," attorneys' docket number 122.4-US-U1, which claims the benefit under 35 U.S.C. §119(e) of Provisional AppHcation Serial No. 60/215,739, filed June 29, 2000, by Paul F. Crowder, entided "OPEN LOOP CONTROL OF SGDBR LASERS," attorneys' docket number 122.4- US-P1; and

[0008] Utility AppHcation Serial No. 09/895,598, filed June 29, 2001, by Paul F. Crowder, entitled "POWER AND WAVELENGTH CONTROL OF SAMPLED GRATING DISTRIBUTED BRAGG REFLECTOR LASERS," attorneys' docket number 122.5-US-Ul, which claims die benefit under 35 U.S.C. §119(e) of Provisional AppHcation Serial No. 60/215,739, filed June 29, 2000, by Paul F. Crowder, entided "POWER AND WAVELENGTH CONTROL OF SGDBR LASERS," attorneys' docket number 122.5-US-P1.

BACKGROUND OF THE INVENTION 1. Field of the Invention.

[0009] The present invention relates to methods of caHbrating diode laser controHers. More particularly, the present invention relates to the caHbration of edge emitting diode laser controllers. And even more particularly, the present invention relates to the caHbration procedure for a smaH-form-factor ("SFF") SGDBR laser controHer ("controHer") and tunable laser assembly ("TLA").

2. Description of the Related Art.

[0010] Diode lasers are being used in such appHcations as optical communications, sensors and computer systems. In such appHcations, it is very useful to employ lasers that can be easily adjusted to output frequencies across a wide wavelength range. A diode laser which can be operated at selectably variable frequencies covering a wide wavelength range, i.e. a widely tunable laser, is an invaluable tool. The number of separate channels that can utilize a given wavelength range is exceedingly limited without such a laser. Accordingly, the number of individual communications paths that can exist simultaneously in a system employing such range-limited lasers is similarly very limited. Thus, while diode lasers have provided solutions to many problems in communications, sensors and computer system designs, they have not fulfilled their potential based on the available bandwidth afforded by Hght-based systems. It is important that the number of channels be increased in order for optical systems to be realized for many future appHcations.

[0011] For a variety of appHcations, it is necessary to have tunable single-frequency diode lasers which can select any of a wide range of wavelengths. Such appHcations include sources and local oscillators in coherent Hghtwave communications systems, sources for other multi-channel Hghtwave communication systems, and sources for use in frequency modulated sensor systems. Continuous tunabiHty is usuaUy needed over some range of wavelengths. Continuous tuning is important for wavelengtii locking or stabiHzation with respect to some other reference, and it is desirable in certain frequency shift keying modulation schemes.

[0012] In addition, widely tunable semiconductor lasers, such as the sampled-grating distributed-Bragg-reflector (SGDBR) laser, the grating-coupled sampled-reflector (GCSR) laser, and vertical-cavity lasers with micro-mechanical moveable mirrors (VCSEL-MEMs) generaUy must compromise their output power in order to achieve a large tuning range. The basic function and structure of SGDBR lasers is detailed in U.S. Patent 4,896,325, issued January 23, 1990, to Larry A. Coldren, and entided "MULTI- SECTION TUNABLE LASER WITH DIFFERING MULTI-ELEMENT MIRRORS", which patent is incorporated by reference herein. Designs that can provide over 40 nm of tuning range have not been able to provide much more than a couple of milliwatts of power out at the extrema of tiieir txining spectrum. However, current and future optical fiber communication systems as weU as spectroscopic appHcations require output powers in excess of 10 mW over the full tuning band. Current International Telecommunication Union (ITU) bands are about 40 nm wide near 1.55 μm, and it is desired to have a single component that can cover at least d is optical bandwidth. Systems that are to operate at higher bit rates wUl require more than 20 mW over the full ITU bands. Such powers are available from distributed feedback (DFB) lasers, but these can only be tuned by a couple of nanometers by adjusting their temperature. Thus, it is very desirable to have a source with both wide tuning range (> 40 nm) and high power (> 20 mW) without a significant increase in fabrication complexity over existing widely tunable designs. Furthermore, in addition to control of die output wavelength, control of the optical power output for a tunable laser is an equaHy important endeavor as optical power determines the potential range for die laser.

[0013] Fundamentany, maximizing the output power, while stabilizing the output wavelength and the maximizing the side mode suppression ratio are very desirable objectives in the control of SGDBR lasers. Thus, there is a need in the art for controUers which maximize the power output and stabilize the wavelength, particularly as the laser ages. Proper caHbration of the laser diode and controHer is key in meeting these goals. The present invention meets the foregoing objectives through a novel controHer caHbration technique.

SUMMARY OF THE INVENTION

[0014] The present invention involves the caHbration of a laser and an open loop controHer of the frequency (or alternatively "wavelength") output and power output of such a laser, which is preferably a Sampled Grating Distributed Bragg Reflector (SGDBR) semiconductor laser. The SGDBR laser generaUy includes at least four discrete sections: a gain section, a phase section, a first (or alternatively "front") mirror section and a second (or alternatively "rear") mirror section. AdditionaHy, a semiconductor optical ampHfier ("SOA") section may be included as weU as other discrete sections. The open loop controHer of such SGDBR devices provides stable SGDBR laser optical power and wavelength output by varying the control currents that ate appHed to each of the aforementioned sections.

[0015] The caHbration of the SGDBR laser and the open loop controHer in accordance with the present invention estabHshes a table of current or voltage settings to control the laser's optical output power and die output wavelength. Once the optical power and output wavelength are selected, the open loop controHer of the present invention selects a set of operating currents or voltages from the table corresponding to the selected output power and output wavelength. Further, the open loop controHer regulates the temperature of the SGDBR laser to a fixed, pre-selected value.

[0016] To generate the operating currents, each SGDBR laser is caHbrated using a caHbration routine, and each controHer is programmed with the values for the corresponding laser, which then controls the laser over its Hfetime.

[0017] A SGDBR laser generaUy includes a laser diode, a laser diode module ("LDM"), and a control hardware package; aU of which are housed in what shaU be referred to herein as a tunable laser assembly ("TLA"). The laser diode is housed within the laser diode module, which may be a butterfly package or some other smaU package weU known to those skiHed in the art for having mounted thereto and therewithin a laser diode.

[0018] The LDM is housed within and is a subcomponent of the TLA. AdditionaUy, the TLA houses the controHer, which comprises hardware and firmware loaded thereupon. The TLA also comprises connectors extending between the controHer and the LDM providing for communication therebetween.

[0019] To ensure proper functioning of the TLA, the controUer must be configured for each unique laser diode, and as such, each LDM. Set out hereinbelow are the steps associated with such a caHbration procedures.

[0020] The caHbration procedure of the present invention comprises caHbrating the controUer hardware and firmware of the TLA. The controHer hardware current sources and temperature sense circuits are caHbrated to facilitate long-term drift measurements. The laser optical output is caHbrated to the International Telecommunications Union ("ITU") channels. [0021] By properly choosing the operating currents, the current sources that deHver the currents to the SGDBR laser diode, and propedy regulating the temperature of the SGDBR laser, the open loop controUer of the present invention provides great stability of the optical output wavelength and power over the operating Hfetime, as weU as providing greater stabiUty over a wider range of ambient environmental conditions.

[0022] A exemplary method includes conducting a two-dimensional mirror current scan of each front mirror current setting and back mirror current setting for a sampled grating distributed Bragg reflector (SGDBR) laser to produce laser setting data corresponding to each front mirror current setting and back mirror current setting to generate a reference optical output of the SGDBR laser. A channel operating point is determined for each channel within the two-dimensional scan data. A fixup of the operating point to substantiaUy minimize wavelength and power error can also be performed. A two-dimensional control surface is characterized at the channel operating point for each channel. A lookup table for controlling the SGDBR laser is generated from the operating point currents, locker values, and two-dimensional control surface data from each channel.The lookup table can be programmed H to a controUer to operate the SGDBR laser to produce a caHbrated channel and power output.

[0023] Applying the method of the present invention, a typical apparatus for performing the SGDBR caHbration of the invention includes a computer for conducting a two-dimensional mirror current scan of each front mirror current setting and back mirror current setting for a sampled grating distributed Bragg reflector (SGDBR) laser, a reference locker to set a reference optical output of the SGDBR laser for the two- dimensional mirror scan to produce laser setting data corresponding to each front mirror current setting and back mirror current setting and a channel detector for determining a channel operating point for each channel within the laser setting data. The computer characterizes a two-dimensional control surface at d e channel operating point for each channel with the reference locker and generates a lookup table for controIHng the SGDBR laser from the two-dimensional control surface. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Referring now to the drawings in which like reference numbers represent corresponding parts throughout:

[0025] FIG. 1 is a laser system block diagram;

[0026] FIG. 2 depicts a block diagram of an open loop control in accordance with the present invention;

[0027] FIG. 3 is a two-dimensional Mirror Scan Block Diagram;

[0028] FIG. 4 is a block diagram of a channel detection method;

[0029] FIG. 5 is a block diagram of a two-dimensional control surface scan method;

[0030] FIG. 6 is a block diagram of the current sources in accordance with the present invention;

[0031] FIG. 7 is a block circuit diagram of a modified Howland current source circuit in accordance with the present invention;

[0032] FIG. 8 is a block diagram of a current mirror circuit in accordance with the present invention;

[0033] FIG. 9 is a block diagram of a power and wavelength control method in accordance with the present invention;

[0034] FIG. 10 is a block diagram of a DSP gain voltage control method for use in accordance with the present invention;

[0035] FIG. 11 is a block diagram of an analog gain voltage control method in accordance with d e present invention;

[0036] FIG. 12 is an analog phase lock circuit block diagram in accordance with the present invention; and [0037] FIG. 13 is a block diagram of a gain and phase current control method in accordance with the present invention.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0038] In the foUowing description, reference is made to die accompanying drawings which form a part hereof, and which is shown, by way of iUustration, an embodiment of the present invention. It is understood that other embodiments may be utilized and structural changes may be made without departing from the scope of the present invention.

1.0 Overview [0039] FIG. 1 is a laser system block diagram. The present invention is directed to the caHbration of an SGDBR laser controHer 100 (hereinafter referred to as a "controUer"). The controUer 100 monitors a multi-section, widely tunable SGDBR laser's 102 (hereinafter referred to as a "laser") gain section voltage 104, temperature 106, and wavelength locker 108 signals. The wavelength locker signal 108 is produced from an external reference 110 (a wavelength locker, alternatively referred to as an "FP etalon"). The laser 102 generaUy has a first or front mirror section (sometimes referred to herein as "FM"), a second or back mirror section (sometimes referred to herein as "BM"), a gain section for Hght generation (sometimes referred to herein as "Gn"), and a phase section provided to tune the output wavelengtii of the laser (sometimes referred to herein as "Ph") each controUed with current inputs 112. AdditionaUy, other sections may be incorporated onto the laser diode including, but not Hmited to a semiconductor optical ampHfier, a modulator, or some otiier weU-known component that may be fabricated on the same substrate as the laser.

[0040] As shown in FIG. 1, the controUer 100 adjusts each section's current (with inputs 102) and the laser's temperature to maintain a fixed optical output 114 power and wavelength. The laser's temperature is adjusted with a thermo-electric cooler 116 (or "TEC"), or some other weU known cooling mechanism or method. The laser 102 is controUed to generate optical output 114 at a substantially continuous power-level. [0041] The controUer 100 interfaces to a host (not shown) over a system interface 118, which is typicaUy a serial or paraUel interface. The host commands the operation of the controHer 100 and may be a personal computer, workstation, or some other weU-known device capable of sending commands to the controUer 100 through die system interface 118.

[0042] The controUer 100 regulates the laser's optical output 114 power and wavelength. The controUer 100 operates in one of the foHowing control modes, each of which shaU be described in more detaU hereinbelow:

A. Open loop control using fixed operating points,

B. Power and wavelength control using open loop control's fixed operating points as the initial operating points and regulating the optical power and wavelength to a reference,

C. Gain voltage control using open loop control's fixed operating points as the initial operating points and regulating the laser mirror aHgnment with the cavity mode, and

D. Power, wavelength, and gain voltage control using open loop control's fixed operating points as the initial operating po its.

1.1 Open Loop Control

[0043] As shown in FIG. 2, in an open loop control mode, the controUer 100 sets the laser optical output 114 power and wavelength by setting the laser section (BM, Ph, Gn, FM and SOA) currents 112 from values in a look up table. It regulates the laser's temperature to a fixed value by sending control code to the TEC 116. The look-up table values are generated by a caHbration routine. The values are fixed over the Hfetime of the laser 100. The choice of the operating currents 112, the current sources, and the temperature regulator guarantees maximum stability of the optical output 114 wavelength and power over the laser operating Hfetime and ambient environmental conditions. [0044] In some embodiments of the invention, the controHer can be implemented with "open loop" controUer hardware as described above, however feedback is provided (e.g. to control the mirror aHgnment). Thus, the controUer operates in a closed loop with respect one or more of the laser control parameters (e.g., mirrors, gain, or phase). Control loops for power and/ or wavelength control can also be appHed. In addition, temperature regulation also can be operated under a closed loop control. As such, there is often no clear distinction between open and closed loop operation of the controHer.

1.2 Operating Points

[0045] The laser operating points are typicaUy determined by one of three caHbration routines:

A. Incremental

B. Mirror reflectivity peak

C. Two-dimensional Mirror scan

2.0 Incremental CaHbration [0046] Incremental caHbration steps and locks the laser to each International

Telecommunications Union (ITU) wavelength channel using a caHbrated wavelength locker as a reference 110. See FIG. 1. It steps to the next channel by adjusting the phase current and locking the mirrors to the cavity mode with gain voltage control, which shaU be discussed in further detaU hereinbelow. Once at the channel, the laser wavelength is locked to the channel by adjusting the phase current using wavelength control and the laser power to a predetermined set point by adjusting the gain current with power control.

[0047] The process of incremental caHbration starts with the first and second mirrors aHgned at mirror reflectivity peak 0 and ti en steps to locate the next lower channel. At each cavity mode, the phase current is reset to its initial value and the search is continued. At die end of each mirror tuning range, the mirror currents are reset to the next mirror reflectivity peak. Once the wavelength wraps around, the process is repeated at mirror reflectivity peak 0 by searching for the next upper channel. The process is as foUows:

For each wavelength direction about mirror reflectivity peak 0 Until (wavelength wraps) Set gain current at nominal operational current

Set mirrors at next reflectivity peak Until (end of mirror tuning range)

Set phase current at minimum operational current; and Lock mirrors to cavity mode Until (passes cavity mode)

Lock power and wavelength at channel and aHgn mirrors

Record channel and currents; and

Step to next channel with mirrors locked to phase

3.0 Mirror Reflectivity Peak CaHbration [0048] Mirror reflectivity peak caHbration determines the mirror reflectivity peaks, generates the mirror tuning efficiency curves, and uses the curves to set the mirror currents for each channel. The process is as foUows:

Until (wavelength crosses mirror reflectivity peak 0)

Sweep mirror with cavity mode aHgned to mirror Locate the gain voltage minima, which is d e corresponding mirror reflectivity peak; and Record the currents Generate mirror tuning efficiency curve from reflectivity peaks Until (step through aU channels) Set mirrors to channel using mirror tuning efficiency curve

AHgn phase section to the mirrors Lock wavelength to channel using wavelength control Lock power to set point using power control Record the channel and current

4.0 Two-Dimensional Mirror Scan CaHbration

[0049] A two-dimensional mirror scan caHbration of the present invention (as may be employed for a smaU form factor TLA) determines the laser currents for operation at each ITU channel and the power and wavelength and mirror control surfaces and operating points at each ITU channel. The caHbration procedure for die s aU form factor TLA and laser involves the foUowing steps:

A. Conduct two-dimensional mirror current scan with power leveling and wavelength locking B. Channel operating region detection

C. Fixup of operating current values

D. two-dimensional control surface characterization; and

E. Generate lookup table

4.1 Two-Dimensional Mirror Current Scan [0050] AppHed to the TLA 300 as set out in FIG. 2 and FIG. 3, the two-dimensional Scan steps the mirror currents over their operating range whUe die controUer locks the gain current to a constant optical power and the phase current to an ITU channel. The controUer power and wavelength feedback is from a caHbrated locker 302 and coupler/ attenuator 304. The locker 302 caHbration value is fixed over the sweep. For example, the foUowing procedure can be appHed. A computer 306 (e.g, a PC) sends over the interface 118 (e.g., a paraUel interface) to the TLA 300, the front mirror and back mirror currents, stepping d em over their operating ranges. At each step, the TLA 300 attempts to lock the optical power and optical wavelength to their setpoints using the gain and phase currents. Also at each step, the TLA 300 responds over the interface 118 to the computer 306 with the status of the gain and phase current control, the gain and phase currents and the power 308 and wavelength 310 voltages (the locker reference and etalon signals). The resulting set of values is the two-dimensional scan data. 4.2 Channel Detection

[0051] As depicted in FIG. 4, the computer 306 uses the two-dimensional scan data to determine d e operating regions. GeneraUy, the operating regions are the front and back mirror current regions of the two-dimensional scan data where the optical power and wavelength remain locked to a channel. The computer 306 determines the center of an operating region, sets the TLA 300 at that point, and measures the channel number at that point from a channel detector 400. The computer 306 generates a table of channel numbers and operating currents and boundaries of the operating regions, representing a two-dimensional control surface for the laser.

[0052] However, in some embodiments of the invention operating regions are not simply detected by looking for boundaries where wavelength and power are not locked. For example, for some channels etalon and reference signals, as measured by the locker 302, will remain locked over the entire map except for those points where cavity mode hops (not supermode hops) occur. This occurs because the cavity mode spacing, detuned off the Bragg wavelength (i.e. off of perfect mirror aHgnment), is typicaUy sHghdy less than the locker channel spac g (e.g., 50 or 100 GHz). Instead, a more sophisticated channel detection technique can be employed which searches for "cusps" (regions of discontinuity, such as indicated by discontinuity in first derivative) of the phase or gain (or ampHfier) current surface or gain voltage surface of the two- dimensional scan data. This approach wiU detect both cavity-mode hops and supermode hops which completely bound the operating region of each channel.

4.3 Operating Pomt Fixup

[0053] Furthermore, embodiments of the invention can also include an operating point fixup process. Using this process, the computer 306 performs a more accurate recentering of the operating point within the mode after remeasuring mode boundaries with the locker etalon and reference target values that have been adjusted on a per channel basis to minimize wavelength and power error. 4.4 two-dimensional Control Surface

[0054] As shown in the block diagram of FIG. 5, using the two-dimensional control surface scan the computer 306 sets the TLA 300 at a channel, using the operating point currents as determined above. It sets the power and wavelength control setpoints for that channel and the TLA 300 relocks to the channel. The locker power and wavelength caHbration is known at each channel. It generates the gain and phase current control surfaces about the operating point. The TLA responds with the channel: status of control, gain, phase, and mirror currents and the control surfaces. The computer 306 records the caHbrated gain, phase, and mirror operating currents at the caHbrated optical power and wavelength setpoints and the center of the control surfaces. In other words, the control surface extremum is returned. In operation, however, the gain voltage control surface can be used to control the laser as described hereafter.

4.5 Lookup Table

[0055] The computer 306 then takes the control surface data for each channel and generates a corresponding lookup table. The lookup table can then be programmed into the controUer 100 of the TLA 300.

5.0 Current Sources

[0056] As shown in FIG. 6, the controUer 100 includes current sources 600 which drive each of the laser's phase, mirror, ampHfier, and gain sections. The current sources 600 are comprised of a voltage reference 602, individual 16-bit digital-to-analog converters 604 (DACs), and voltage-to-current (VT) converter 606. The DACs 604 connect to a digital signal processor (DSP) synchronous serial port (SSP) 608 through a programmable-logic device 610 (PLD). The PLD 610 provides a logic interface between the DSP SSP 608 and the DACs 600. Each VI converter 606 translates the corresponding DAC 604 voltage output to a proportional current that drives a corresponding laser section. 5.1 Voltage to Current Converter

[0057] As depicted in FIG. 7, a modified Howland current source (MHCS) can be used as the voltage-to-current converter 606. A current mirror 700, such as that shown in FIG. 8, is preferably added to the output stage of the ampHfier 702 to increase the drive current beyond that of the ampHfier 702 alone. A filter stage 704 was added at the load 706 to reduce noise. The current mirror 700 inverts the output of the ampHfier 702, which requires the source, V_in, at the inverting node of the ampHfier 702.

[0058] The current mirror 700 operates at a fixed gain that is determined, primarily, by the ratio of the resistors 800 in the emitter leads of the transistor 802. A resistor- capacitor (RC) compensation network 804 is added to insure stabiHty of the ampHfier 702 and current mirror 700. The gain of the current is variable up to a maximum ratio. The maximum ratio is determined by the additional drift introduced by heating of the transistor 802 and sense resistor 806 and the maximum tiiermal loss that can be sustained by the transistor 802 and sense resistor 806. If additional gain is required, an additional Q_mo & R_mo section can be added to the mirror 700.

6.0 Power and Wavelength Control

[0059] As shown in FIG. 9, the power and wavelength controHer 100 uses open loop control and feedback 900 from an external wavelength locker 902 (FP etalon) reference to lock die laser optical output power and wavelength to the reference. Power and wavelength control compensates for drift in the controUer current sources 600 and the laser 102 operating points over time and temperature. The power and wavelength controls may operate independentiy or interdependentiy.

6.1 Independent power and wavelength control

[0060] The least complex control algorithm is where the controls operate independentiy. Each control algorithm induces changes in one current or temperature independent of the other. The control algorithms are classical proportional, integral control routines. For example, the foUowing algorithm can be appHed: Optical power is adjusted by

Gain current (Ign), or

Current to a SOA (if integrated into the laser). Optical wavelength is adjusted by Phase current (Iph), or

Submount temperature Mirror currents are left fixed.

[0061] In most cases, gain current is used on four-section devices, and ampHfier current is used on five-section devices. Current to the semiconductor optical ampHfier (SOA) instead of current to the gain section can be used in all cases concerning power control or power leveling when an ampHfier section is present on the laser chip. Gain voltage control (See section 7) may be used in either case. However, when gain voltage control is combined with gain current-based power control, power control must be interrupted (i.e. gain current held constant) during acquisition of a gain voltage control surface.

6.2 Interdependent power and wavelength control

[0062] The independent control algorithm is slower and in its response to changes in the optical power output and optical wavelength. The mirrors and cavity mode become misaHgned as die control algorithm adjusts the gain and phase currents from their predefined values. The quaHty of the optical output may be reduced as a result of decreased side mode suppression ratio. AdditionaUy, the probability of a mode hop (wavelength shift) is increased as die mirrors and cavity mode become misaHgned.

[0063] The interdependent control algorithm induces primary changes in one current or temperature and corrects for secondary changes in die other currents with an adaptive filter or estimator. This compensates for wavelengdi sliifts or power changes and mirror misaHgnment induced when the control adjusts its primary variable. Using an interdependent power and wavelength control algorithm as foUows: Power is adjusted by the gain current (I_gJ,

Wavelength is stabilized by adjusting the phase current (IX) by an adaptive filter; and

Mirror currents are reaHgned by a fixed estimator, Wavelength is adjusted by the phase current (I_ph) or the carrier temperature

Power is stabilized by adjusting the gain current (L by an adaptive filter; and

Mirror currents are reaHgned by a fixed estimator.

The interdependent controls provide more robust, stable, and faster convergence of the power and wavelength to its reference value.

7.0 Gain Voltage Control

[0064] Gain voltage control uses feedback from the laser gain section voltage to keep the mirrors aHgned with the cavity mode. It aHgns the mirrors by minimizing the laser gain section voltage. The laser gain section voltage minimum is where the cavity loss is a minimum, roughly corresponding to maximum optical power output, wavelength stabiHty, and side mode suppression ratio. More specificaUy, the gain voltage minimum corresponds to the minimum loss condition when parasitic electrical effects are accounted for, but gain spectrum effects offset the minimum from mode center in a characteristic fashion. Additional output power may be achieved using certain techniques, such as by misaUgning the front mirror, however, in such a case, other characteristics may suffer, such as the side mode suppression ratio.

[0065] Gain voltage control can be implemented in the DSP using a numerical minima search or a least mean squares (LMS) quadratic estimator. Alternately, gain voltage control can be implemented in analog circuitry using a phase locker circuit (PL).

7.1 DSP Gain Voltage Control

[0066] A digital signal processor (alternatively referred to as a "DSP") may be used to implement the gain voltage control, as shown in FIG. 10. The DSP dithers the laser mirror currents 1000, 1002 and monitors the laser gain section voltage 1004. It uses a numerical algorithm to aHgn the mirrors by locating the mirtima of the laser gain section voltage.

7.2 DSP Minima Search Algorithm [0067] An example mkrima search algoriti m can be implemented as foUows:

Use three data points (mirror current, gain voltage) and estimate the slope of the gain voltage curve with respect to the mirror current, Step toward the gain voltage minima and calculate the next data point, Use the new data point and the two best points to re-estimate the slope of the gain voltage curve,

Continue the above step process, continuaUy searching for the gain voltage minima.

7.3 DSP LMS Estimator

[0068] The minima search algorithm may be susceptible to wandering around the gain voltage minima due to noise in the sampled gain voltage signal. The wandering is reflected as drift and noise on the optical signal. The LMS estimator reduces the wander and noise by using an array of data points to estimate the gain voltage surface, in effect, filtering the noise. The LMS estimator converges to the gain voltage minima faster and smoother than the minima search.

[0069] For fixed phase and gain section currents, the gain section voltage can be modeled using a causal Volterra series expansion over 2 input signals, the front mirror and back mirror currents. For ditiiering signals in the sub-lOOkHz regime, the analog circuitry and the device itself aUow a memoryless model, so a 5-tap adaptive quadratic filter model wiU suffice.

[0070] The LMS estimator can then be achieved using either of two classic adaptive filter update algorithms: a standard gradient descent adaptation (LMS or block LMS algorithm) or a recursive least squares adaptation (RLS algorithm — based on Newton's Method).

[0071] The RLS algorithm approach is used to achieve faster convergence of adaptive Hnear filters when the signals driving the system do not have sufficient spectral flatness to aUow a rapid gradient descent. However, in the case of adaptive Hnear filters, the gradient descent approach converges just as fast as the RLS approach when white noise can be used to drive the system. Recentiy pubHshed results indicate that comparable rates of convergence can be achieved with adaptive quadratic filters if a minor filter structure modification is used and (pseudo) Gaussian white noise can be used to drive the system.

[0072] There are two advantages of this LMS estimator approach. First, an initial tap- vector can be stored along with the four drive currents in the laser caHbration table in flash memory (resulting in much faster convergence). Second, the adaptation step size can be reduced as the system converges, reducing steady-state misadjustment in the mirror section currents.

[0073] Because of the aforementioned gain spectrum effects, the optimum setpoints for the mirror currents are actuaUy offset from the gain voltage minimum. Therefore, the objective is not to converge to the mitimium, but to use an LMS estimator to sense where the minimum would be based on the measured gain voltage surface in the vicinity of the operating point. The control system adjusts the mirror currents to operate at a caHbrated current offset from d e estimate of the minimum.

7.4 Exemplary LMS Estimator

[0074] An exemplary LMS estimator can use five independent data points to determine the surface. The LMS algorithm:

Dithers the mirror currents in a linearly independent fashion about the operating point where, a point Hes in each quadrant; and the step size is less tiian the power and wavelength accuracy; CoUects the gain and phase current at the mirror current when the power and wavelength are witi in control tolerance;

Runs the LMS estimator over the data set (at least five independent points); Resets the mirror operating point to the distance from the inflection points on die surface.

[0075] The LMS algorithm continuaHy operates in the background and the five- parameter fit to the quadratic control surface is:

2 2 r-| f + — + n- b + — + c ^{s m} simplify-*- r-f + s-f + n- ,b 2 + m-b + c 2ri I 2n 4-r 4-n

The parameters r and n define the surface curvature for the front and back mirror currents respectively. The parameters s and m define the offset of the surface extremum. The parameter c defines the offset of the surface. The independent variables f and b are the front mirror current and the back mirror current. The result maps the quadratic surface of the gain current or phase current. The extremums are at:

m f = — b = -

2r 2n

The LMS estimator that generates the surface parameters is:

r

where S denotes a summation over the data points of the terms multipHed together and z is the current of the surface. The distance is the df and db from the extremums. [0076] The above technique is preferably used with the gain voltage surface. In general, tiiere is a significant cross term (f * b) in the gain voltage surface, which goes to zero in the wavelength-locked case. In this case, therefore, a much simpler fit can be performed independentiy on the front and back mirror dither using three fitting parameters, and the resulting extremum is calculated.

7.5 Analog Gain Voltage Control

[0077] The digital algorithms implemented in the DSP are limited in speed and accuracy by the analog to digital converter (ADC) and digital to analog converter (DAC) as weU as the signal to noise ratio (SNR) of the circuit.

[0078] An analog gain voltage control is set out in FIG. 11. The analog phase locker's speed and accuracy is limited substantiaHy only by the SNR of the circuit. The analog phase locker (PL) is a high speed, analog-locking loop. It can be realized by a phase lock loop (PPL) or RF dither locker. The PL works with the open loop control circuit. The output of the PL adds to the output of die open loop control current sources. For example, the gain voltage 1100 can be appHed to separate PL circuits 1102A, 1102B of the controHer 100.

[0079] As shown in FIG. 12, an exemplary PL 1102 uses a high frequency narrowband stimulus 1200 to dither the mirror current. The PL 1102 measures the gain voltage ( _g) 1202 with a tuned, narrowband ampHfier 1204 and extracts the phase difference between stimulus and measured signal with a phase comparator 1206. The PL 1102 also drives an error ampHfier that adjusts die mirror current to the gain voltage minima and is sampled by an ADC 1208.

[0080] The PL error ampHfier output is measured by the DSP. The DSP adjusts the mirror current values in the open loop control lookup table to reduce the error to zero. The DSP effectively operates as an integrator function. [0081] FIG. 13 iUustrates the combined operation of analog gain voltage control circuits to correct the outputs to the two mirrors from the open loop digital controUer. The digital memory/DSP 1300 can set a first approximation current and voltage from a lookup table. The analog correction circuits 1102A, 1102B can provide feedback and correction signals to the device as described previously, and the digital controUer then monitors die correction signals 1302, 1304 and readjusts the currents and voltages to have the feedback currents from the analog correction portions approach zero. This aUows for correction of the laser controUer over the Hfe of the SGDBR laser, and accounts for changes in operating temperatures and conditions as weU as changes in the operation of the SGDBR laser internal components.

[0082] Gain and phase current control, such as that shown in FIG. 13, uses the extremum point (die maximum or minimum value of a function) of the gain voltage surface (as proxy for the gain and phase current surfaces) to keep die mirrors aHgned with the cavity mode. It aHgns the mirrors by operating the mirror currents at a fixed distance from the control surface extremums. The distance and extremums are determined during caHbration. The mirror operating point corresponds to best-cost function that maximizes the optical power output, wavelength and power stabiHty, and side mode suppression ratio. Gain and phase current control operates in conjunction with power and wavelength control.

[0083] Gain and phase current control can be implemented in die DSP using a least mean squares (LMS) quadratic surface estimator, such as that previously described. The DSP dithers the laser mirror currents while operating under power and wavelength control and records the gain and phase currents when the control loops are within tolerance. It can estimate a fit to the gain voltage surface as a function of the front and back mirror currents. Alternately, it can estimate a five-parameter fit to the quadratic control surface for the gain current and the phase current as a function of the front and back mirror currents. It sets the mirror currents at a distance from the surface extremums. [0084] The power, wavelength, and gain voltage controUer 100 operates the power and wavelength control and gain voltage control simultaneously.

8.0 Conclusion

[0085] The foregoing description of the preferred embodiment of the invention has been presented for the purposes of illustration and description. It is not intended to be exhaustive or to Hmit the invention to die precise form disclosed. Many modifications and variations are possible in Hght of the above teaching. It is not intended that the scope of the invention be limited by this detaUed description.

[0086] This concludes the description of the preferred embodiment of the present invention. The foregoing description of the embodiments of the invention has been presented for the purposes of illustration and description. It is not intended to be exhaustive or to Hmit the invention to the precise form disclosed. Many modifications and variations are possible in Hght of the above teaching. It is intended that the scope of the invention be limited not by this detaUed description, but rather by the claims appended hereto.

Claims

WHAT IS CLAIMED IS:

1. A method for laser caHbration comprising the steps of: conducting a two-dimensional mirror current scan of each front mirror current setting and back mirror current setting for a sampled grating distributed Bragg reflector (SGDBR) laser to produce laser setting data corresponding to each front mirror current setting and back mirror current setting to generate a reference optical output of the SGDBR laser; determining a channel operating point for each channel within the laser setting data; characterizing a two-dimensional control surface at the channel operating point for each channel; and generating a lookup table from the two-dimensional control surface for controUing the SGDBR laser.

2. The method of claim 1, furd er comprising performing fixup of the operating point to substantiaUy minimize wavelength and power error.

3. The method of claim 1, further comprising programming die lookup table into a controHer for the SGDBR laser.

4. The method of claim 1, wherein the two-dimensional mirror scan includes stepping the front mirror current and the back mirror current over their operating ranges.

5. The method of claim 1, wherein the two-dimensional mirror current scan is conducted with power leveling and wavelength locking to the reference optical output of the SGDBR laser.

6. The method of claim 5, wherein power leveling comprises locking a gain current to a constant optical power output at each front mirror current and each back mirror current.

7. The method of claim 5, wherein power leveling comprises locking an integrated semiconductor-optical ampHfier current to a constant optical power output at each front mirror current and each back mirror current.

8. The method of claim 5, wherein wavelength locking comprises locking a phase current to a substantiaUy constant wavelength output.

9. The method of claim 5, wherein power leveling and wavelength locking is performed relative to an external wavelength reference.

10. The method of claim 9, wherein the external wavelength reference comprises a Fabry-Perot interferometer.

11. The method of claim 1, wherein the two-dimensional control surface comprises a phase current surface, gain current surface or a gain voltage surface.

12. The method of claim 1, wherein characterizing the two-dimensional control surface comprises comprises applying the channel operating point and locking the channel to generate the two-dimensional control surface.

13. The method of claim 1, wherein determining the channel operating point includes determing boundaries of the operating region and a center of the operating region as the channel operating point.

14. The method of claim 13, wherein determing boundaries comprises identifying discontinuities within the laser setting data.

15. An apparatus for laser caHbration comprising: a computer for conducting a two-dimensional mirror current scan of each front mirror current setting and back mirror current setting for a sampled grating distributed Bragg reflector (SGDBR) laser; a reference locker to set a reference optical output of the SGDBR laser for the two-dimensional mirror scan to produce laser setting data corresponding to each front mirror current setting and back mirror current setting; and a channel detector for deterrrrining a channel operating point for each channel witiiin the laser setting data; wherein the computer characterizes a two-dimensional control surface at the channel operating point for each channel with the reference locker and generates a lookup table for controIHng the SGDBR laser from the two-dimensional control surface.

16. The apparatus of claim 1 , wherein a fixup of the operating point is performed to substantiaUy minimize wavelength and power error.

17. The apparatus of claim 1, wherein the computer programs the lookup table into a controUer for the SGDBR laser.

18. The apparatus of claim 1, wherein the two-dimensional mirror scan includes stepping the front mirror current and the back mirror current over their operating ranges.

19. The apparatus of claim 1, wherein the reference locker performs power leveling and wavelength locking to the reference optical output of the SGDBR laser.

20. The apparatus of claim 19, wherein power leveling comprises locking a gain current to a constant optical power output at each front mirror current and each back mirror current.

21. The apparatus of claim 19, wherein power leveling comprises locking an integrated semiconductor-optical-ampHfier current to a constant optical power output at each front mirror current and each back mirror current.

22. The apparatus of claim 19, wherein wavelength locking comprises locking a phase current to a substantiaUy constant wavelength output.

23. The apparatus of claim 1, wherein the reference locker comprises a Fabry-Perot reference.

24. The apparatus of claim 1, wherein the two-dimensional control surface comprises a phase current surface, gain current surface or a gain voltage surface.

25. The apparatus of claim 1, wherein characterizing the two-dimensional control surface comprises comprises applying the channel operating point and locking the channel to generate the two-dimensional control surface.

26. The apparatus of claim 1, wherein determining the channel operating point includes determing boundaries of the operating region and a center of the operating region as the channel operating point.

27. The apparatus of claim 13, wherein determining the channel boundaries comprises identifying discontinuities within the laser setting data.