Waveguide loss measurement method based on dual-channel detection

Abstract

The invention relates to a waveguide loss measurement method based on dual-channel detection, which is used for accurately acquiring a linear loss coefficient of a nonlinear waveguide to be measured. Firstly, pump light, probe light and auxiliary laser are combined and injected into a waveguide to be detected, and a four-wave mixing effect is triggered to generate idler frequency light; splitting the waveguide output light into two paths through a light splitting element, and extracting an idler frequency light signal and an auxiliary laser signal as a measurement channel and a reference channel; adjusting the waveguide temperature until the idle frequency optical power is stable, and fitting to obtain a linear interval between the probe light and the idle frequency optical power; and fixing the detection light power in the interval, generating a noise compensation factor by using the average power and the instantaneous power of the auxiliary laser, correcting the idle frequency light power in real time, and finally calculating the waveguide loss based on the corrected data. According to the method, real-time noise counteracting is achieved through two channels, parameter locking and linear interval screening are combined, the measurement stability and precision are greatly improved, and the loss detection requirements of various nonlinear waveguides are met.

Description

Technical Field

[0001] This application relates to the technical field of waveguide loss measurement, and in particular to a waveguide loss measurement method based on dual-channel detection. Background Technology

[0002] With the rapid development of optical communication, integrated optics, and optical quantum computing, waveguides, as core devices for optical signal transmission and processing, have seen their loss characteristics become a key indicator for evaluating device performance, directly affecting optical transmission efficiency, signal fidelity, and system integration. Currently, waveguide loss measurement techniques mainly include the truncation method, backscattering method, interferometry, and four-wave mixing (FWM) based on nonlinear optical effects. Among these, the truncation method, while simple to operate, requires the fabrication of waveguides of different lengths from the same batch, is destructive, and is difficult to adapt to miniaturized integrated devices; the backscattering method is non-destructive, but its spatial resolution is low, limiting its accuracy for low-loss waveguides; the interferometry method has high sensitivity, but is easily affected by environmental vibrations, temperature drift, and other factors, resulting in poor stability.

[0003] In recent years, measurement methods based on the four-wave mixing effect have gradually become a research hotspot for low-loss waveguide detection due to their advantages such as high sensitivity, non-destructive nature, and integrability. However, existing four-wave mixing measurement techniques still have significant shortcomings: First, environmental disturbances, such as mechanical vibration, airflow changes, and fluctuations in light source power, can simultaneously affect the transmission stability of both signal light and pump light, leading to random errors in idler power measurement. Existing single-channel detection schemes lack effective real-time noise compensation mechanisms, making it difficult to eliminate such interference. Second, the four-wave mixing conversion efficiency is significantly affected by parameters such as waveguide temperature and pump light wavelength. If key parameters are not precisely locked, fluctuations in conversion efficiency will occur, leading to systematic errors. Third, the linear relationship between probe power and idler power is a prerequisite for ensuring measurement accuracy, but existing technologies do not clearly define the screening method for the linear interval, making it easy to perform measurements in the nonlinear region, resulting in deviations in loss coefficient calculation.

[0004] The aforementioned problems have limited the application of the four-wave mixing method in high-precision waveguide loss measurement, and there is an urgent need for a measurement scheme that can achieve real-time noise compensation, stable parameter control, and accurate selection of linear intervals. Summary of the Invention

[0005] To address the shortcomings of existing technologies, the purpose of this application is to provide a waveguide loss measurement method based on dual-channel detection, which can improve the measurement accuracy and stability of the waveguide linear loss coefficient.

[0006] The above-mentioned objective of this application is achieved through the following technical solution:

[0007] A waveguide loss measurement method based on dual-channel detection includes the following steps: Step S1: Pump light, probe light, and auxiliary laser are combined and injected into a nonlinear waveguide under test, triggering a four-wave mixing effect to generate idler light; Step S2: The waveguide output light is split into two paths using a beam splitter, and the idler light signal is extracted from one path as a measurement channel; the auxiliary laser signal is extracted from the other path as a reference channel; Step S3: The waveguide temperature is adjusted until the idler light power in the measurement channel reaches a stable state, and the waveguide temperature and the wavelength of the pump light are locked at this point; Step S4: Maintain the state as described in Step S1. Step S3: The waveguide temperature and pump light wavelength are locked in step S3. The idler light power is recorded under different probe light powers, and the relationship between the two is fitted to obtain a probe light power range with a linear relationship. Step S5: Maintaining the waveguide temperature and pump light wavelength locked in step S3, and fixing the probe light power within the range obtained in step S4, a noise compensation factor is generated by measuring the average power and instantaneous power of the auxiliary laser. The noise compensation factor is used to correct the measured idler light power in real time. Step S6: Calculate the waveguide loss based on step S5. ;in, The four-wave mixing conversion efficiency; The pump light power; The operating power of the probe light; The corrected idler frequency optical power; The waveguide length is given. This is the linear loss coefficient.

[0008] As a preferred embodiment of the present invention, the following conditions must be met in step S1; rad / m; where, This is the wave vector mismatch. The wave vector of the pump light in the waveguide; The wave vector of the probe light in the waveguide, The wave vector of the idler frequency light in the waveguide is adjusted by regulating the temperature of the waveguide to ensure that the wave vector mismatch meets the requirements.

[0009] As a preferred embodiment of the present invention, in step S1, the pump light, the probe light, and the auxiliary laser have the same polarization state and are all matched with the main polarization direction of the waveguide to be tested, so as to maximize the four-wave mixing conversion efficiency.

[0010] As a preferred embodiment of the present invention, step S4 includes: Step S4.1: maintaining the temperature of the waveguide and the wavelength of the pump light locked in step S3; Step S4.2: setting the adjustment range of the probe light, starting from the lower limit of the adjustment range, adjusting the power of the probe light upwards in steps of 1mW~2mW, and holding for 3s~5s after each power point adjustment; Step S4.3: collecting 100 sets of idler light power data at the power point at a sampling frequency of 10Hz, removing the three maximum values ​​and three minimum values, and taking the arithmetic mean as the effective value of the idler light power corresponding to the probe light power; Step S4.4: importing all probe light power and idler light power effective value data into data processing software, performing linear regression fitting using the least squares method, calculating the goodness of fit R²; selecting continuous probe light power sub-intervals with R²≥0.998, which are the probe light power intervals with linear relationships described in step S4.

[0011] As a preferred embodiment of the present invention, the generation of the noise compensation factor in step S5 further includes outlier removal. When the absolute value of the deviation between the instantaneous power and the average power of a certain auxiliary laser is ≥10%, it is determined to be abnormal data, and the moving average of the first 3 normal data is used to replace the instantaneous power of the auxiliary laser to calculate the noise compensation factor.

[0012] As a preferred embodiment of the present invention, the noise compensation factor is calculated as follows; ;in, The noise compensation factor is mentioned above. The average power of the auxiliary laser; The instantaneous power of the auxiliary laser is given; the corrected idler light power is calculated as follows: ;in, The instantaneous power of the idler frequency light; This is for measuring the systematic error of the channel.

[0013] As a preferred embodiment of the present invention, the four-wave mixing conversion efficiency in step S6 needs to be corrected according to the actual operating temperature of the waveguide; ;in, The conversion efficiency is at the reference temperature; For reference temperature; Temperature coefficient; This represents the actual operating temperature of the waveguide.

[0014] As a preferred embodiment of the present invention, in step S6, when the four-wave mixing conversion efficiency is unknown, the linear loss coefficient is calculated using an iterative method.

[0015] As a preferred embodiment of the present invention, the iterative formula is as follows: ; ;in, The four-wave mixing conversion efficiency in the nth iteration; The linear loss coefficient for the nth iteration; The linear loss coefficient is the linear loss coefficient for the (n+1)th iteration; the assumed value of the linear loss coefficient for the initial iteration is set based on the waveguide material properties; the iteration termination condition is: dB / cm.

[0016] As a preferred embodiment of the present invention, the linear loss coefficient α calculated in step S6 needs to be verified by multiple measurements. The measurements are repeated 3 to 5 times under the same experimental conditions. If the relative standard deviation of the multiple measurement results is ≤2%, the average value is taken as the final result; otherwise, steps S3 to S6 are repeated to eliminate systematic errors.

[0017] In summary, the beneficial technical effects of this application are as follows:

[0018] 1. This application introduces an auxiliary laser in step S1 and uses a beam splitter in step S2 to split the waveguide output light into an idler light measurement channel and an auxiliary laser reference channel. Combined with the design in step S5 to measure the average power and instantaneous power of the auxiliary laser to generate a noise compensation factor and to correct the idler light power in real time, this design solves the problem that existing single-channel detection schemes cannot eliminate environmental disturbances. This design utilizes the characteristic that the auxiliary laser and idler light share the same transmission optical path and are subject to the same environmental interference. By using a compensation factor to cancel the noise influence in real time, it effectively reduces the random error of idler light power measurement and improves the reliability of the original data.

[0019] 2. This application solves the problem of fluctuations in four-wave mixing conversion efficiency caused by the lack of precise control of temperature and pump wavelength in the prior art by adjusting the waveguide temperature to stabilize the idler optical power in step S3 and locking the waveguide temperature and pump wavelength. The locking of temperature and pump wavelength ensures the stability of the waveguide phase matching state, avoids the system deviation introduced by the change of conversion efficiency with parameters, provides a stable basis for the subsequent calculation of the loss coefficient based on the four-wave mixing formula, and reduces system-level measurement errors.

[0020] 3. This application solves the problem that existing technologies are prone to measurement in nonlinear regions, leading to distortion of the correlation between idler light and probe light power. The linear interval is determined by recording the idler light power and fitting relationship under different probe light power in step S4, and the probe light power is fixed within the linear interval in step S5. The selection of the linear interval and the fixing of the power ensure that the premise of "the idler light power and the probe light power are linearly correlated" in the four-wave mixing effect is valid, avoiding the interference of nonlinear region data on loss calculation, and significantly improving the accuracy of linear loss coefficient based on formula back deduction. Attached Figure Description

[0021] Figure 1 This is a flowchart of a waveguide loss measurement method based on dual-channel detection.

[0022] Figure 2 A flowchart of a method for obtaining the linearly related probe optical power range. Detailed Implementation

[0023] The present application will be further described in detail below with reference to the accompanying drawings.

[0024] like Figure 1 As shown, a waveguide loss measurement method based on dual-channel detection is illustrated. First, multiple laser beams need to be combined and incident to trigger the target nonlinear effect. The pump beam, probe beam, and auxiliary laser involved here are preferably selected from the 1550nm band, the 1560nm band, and the 1310nm band, respectively. To achieve effective combining of the three beams, fiber optic couplers or spatial beam combiners from existing technologies can be used. By adjusting the optical path alignment accuracy of the beam combiner, it is ensured that the three beams are collinearly incident on the input end of the waveguide under test.

[0025] The waveguide under test must possess nonlinear optical characteristics. Commonly used nonlinear waveguides in existing technologies are selected, with materials including lithium niobate waveguides, chalcogenide glass waveguides, and quantum well waveguides based on semiconductor materials. When three beams of light are injected into the nonlinear waveguide under test at a set power, a four-wave mixing process is triggered within the waveguide due to the third-order nonlinear polarization effect, thereby generating a new frequency idler light. The pump light power is higher than that of the probe light and the auxiliary laser to ensure the significance of the four-wave mixing effect. The frequency of the idler light is determined by the frequencies of the pump light and the probe light, satisfying the energy conservation relationship, i.e., f4 = 2f1 - f2, where f4 is the idler light frequency; f1 is the pump light frequency; and f2 is the probe light frequency.

[0026] Furthermore, the mixed light output from the waveguide is split and the signal is extracted to construct a dual-channel detection structure. The beam splitting element used here is preferably a semi-transparent mirror or a fiber optic beam splitter, as is available in existing technologies. The splitting ratio can be set according to subsequent detection requirements. The two beams after splitting serve as the measurement and reference channels, respectively. For the measurement channel, the idler light signal needs to be extracted. This can be achieved by setting a bandpass filter in the optical path that matches the idler light wavelength. The center wavelength of this bandpass filter must be precisely aligned with the idler light wavelength. If the idler light wavelength is 1540nm, then the center wavelength of the filter is set to 1540nm, and the bandwidth is controlled within 1nm-2nm to filter out stray light interference from the pump light, probe light, and auxiliary laser. Ultimately, the measurement channel outputs only the idler light signal for subsequent power detection. For the reference channel, the auxiliary laser signal needs to be extracted. This is also achieved by setting a bandpass filter that matches the auxiliary laser wavelength. The purpose is to use the auxiliary laser as a reference standard to compensate for environmental noise during the measurement process, such as waveguide vibration, power supply fluctuations, and detector dark current changes. Since the auxiliary laser and the idler light are transmitted in the same waveguide and split by the same beam splitter, they are affected by environmental noise in a consistent manner. Therefore, the power fluctuation of the auxiliary laser can reflect the noise level.

[0027] Furthermore, the waveguide temperature and pump light wavelength are locked to ensure the stability of the measurement conditions. The waveguide temperature can be adjusted using a thermoelectric cooler (TEC) temperature control stage, as described in existing technology. The waveguide under test is fixed on the stage of the temperature control stage, and the waveguide temperature is monitored in real time through the feedback control module of the stage. The temperature detection accuracy can reach ±0.01℃, and the cooling / heating power is adjusted to change the waveguide temperature. During temperature adjustment, the change in idler optical power needs to be monitored in real time through the measurement channel. Preferably, a photodetector can be used to convert the optical power into an electrical signal, which is then recorded by a data acquisition card. When the fluctuation amplitude of the idler optical power is ≤±0.1% within a continuous monitoring period, such as 30 seconds, it is determined that the idler optical power has reached a stable state. At this point, the temperature control stage needs to be controlled to maintain the current temperature, thus locking the waveguide temperature. Simultaneously, the pump light wavelength needs to be locked, because even a small change in the pump light wavelength will directly affect the phase matching condition of the four-wave mixing, thus leading to fluctuations in the idler light power. Wavelength locking can be achieved by using a wavelength locking module based on a Fabry-Perot interferometer or a wavelength feedback control device based on a fiber optic grating. By monitoring the pump light wavelength in real time and adjusting the output wavelength of the pump laser, for example by controlling the laser temperature or injection current, the pump light wavelength can be stabilized at the wavelength value when the idler light power is at its maximum, thus achieving pump light wavelength locking.

[0028] After temperature and wavelength locking are completed, the linear response range of the probe light power needs to be determined to lay the foundation for subsequent accurate measurements. During this process, the locked waveguide temperature and pump light wavelength must be kept constant. This is achieved by adjusting the output power of the probe light, which can be done using a variable optical attenuator (VOA) or by directly controlling the injection current of the probe laser. The adjustment step size can be set to 1mW~2mW, with the specific step size determined based on the probe light power range and measurement accuracy requirements. Within the set probe light power range, for example, 0mW~20mW, different power points are selected sequentially. For each power point, a 3s~5s pause is required after adjustment to allow the probe light power to stabilize. Then, idler frequency light power data is collected through the measurement channel at a sampling frequency of 10Hz. Preferably, 100 sets of data are collected to reduce the influence of random noise. The 100 sets of data are then preprocessed to remove the three maximum and three minimum values ​​to eliminate sudden noise interference. The arithmetic mean of the remaining 94 sets of data is then calculated as the effective value of the idler frequency light power at that probe light power. All probe optical power values ​​and their corresponding effective idler optical power values ​​were compiled into data pairs and imported into data processing software such as Origin and Matlab. Linear regression fitting was performed using the least squares method, and the goodness of fit R² was calculated. The closer R² is to 1, the stronger the linear relationship. A continuous probe optical power sub-interval with R² ≥ 0.998 was selected. This interval is the linear response interval of the probe optical power. Within this interval, the idler optical power and the probe optical power have a strict linear relationship, which conforms to the theoretical assumptions of the subsequent waveguide loss calculation formula and can avoid calculation errors caused by the probe optical power exceeding the linear interval.

[0029] Furthermore, real-time noise compensation and correction of idler frequency optical power are achieved based on the reference channel. During this process, the locked waveguide temperature and pump light wavelength must remain constant, and the probe light power must be fixed within the aforementioned linear response range, for example, selecting the power value at the midpoint of the linear range to balance measurement sensitivity and stability. The average and instantaneous power of the auxiliary laser are measured through the reference channel. Power data of the auxiliary laser is acquired at a sampling frequency of 10Hz to 100Hz using the photodetector and data acquisition card of the reference channel. The arithmetic mean of all instantaneous power data within this time period is calculated, which is the average power of the auxiliary laser. Simultaneously, the instantaneous power of the auxiliary laser is acquired in real time. The sampling frequency is consistent with the idler frequency optical power acquisition frequency to ensure data synchronization. During instantaneous power acquisition, outlier removal is required; when a certain instantaneous power... With average power When the absolute value of the deviation is ≥10%, that is... If the data is determined to be an outlier, the moving average of the previous three normal instantaneous power data should be used to replace the outlier to ensure the reliability of the reference signal. Based on the processed auxiliary laser average power and instantaneous power, a noise compensation factor is generated. Its calculation formula is The physical meaning of this factor is that, by analyzing the instantaneous power fluctuations of the auxiliary laser, the noise level of the current measurement environment can be inverted, thereby compensating for the idler light power. Simultaneously, the instantaneous power of the idler light is acquired in real time through the measurement channel. The noise compensation factor is used to correct it in real time, and the corrected idler frequency optical power is... The calculation formula is: ,in The measurement channel's systematic error primarily originates from factors such as the dark current of the photodetector, the insertion loss of the bandpass filter, and the quantization error of the data acquisition card. This error is obtained through pre-calibration, for example, by measuring the power value corresponding to the detector's dark current under no-light conditions and measuring the power loss of the filtering and acquisition stages under illumination from a standard light source of known power. The results are then obtained by combining these factors. The specific value.

[0030] Finally, the linear loss coefficient of the waveguide is calculated based on the corrected idler power. The core calculation formula used here is: The meanings and acquisition methods of each parameter are as follows: The four-wave mixing conversion efficiency is related to the waveguide's nonlinear coefficient, pump wavelength, waveguide temperature, and mode matching degree. It can be calibrated experimentally beforehand, for example, by measuring different conditions in a standard waveguide with known loss. , , Substitute into the formula for inversion ; The actual incident power of the pump light can be directly measured by connecting a power meter in series in the pump light path; The fixed operating power of the probe light was also measured using a power meter. The corrected idler power obtained in the above steps; The physical length of the waveguide under test can be directly measured using vernier calipers or optical coherence tomography (OCT) technology. The linear loss coefficient of the waveguide to be determined is usually expressed in dB / cm or cm⁻¹.

[0031] By performing mathematical transformations on the above formula, we can derive... The calculation formula: By substituting the actual measured values ​​of each parameter into the formula, the linear loss coefficient of the waveguide under test can be calculated, thus completing the measurement of waveguide loss.

[0032] Furthermore, to ensure that the four-wave mixing effect triggered by the pump and probe beams within the waveguide under test has sufficient efficiency to generate stable and accurately detectable idler light, a specific wave vector mismatch condition must be strictly satisfied when the three beams are combined and injected into the nonlinear waveguide under test. Specifically, this condition is expressed as follows: The calculation is in rad / m, and the implementation methods of each physical quantity and condition need further explanation. The physical meaning and essence of each parameter in the formula should be clarified. The wave vector mismatch, at its core, describes the degree of phase matching between the three participating beams—pump beam, probe beam, and generated idler beam—as they propagate within the waveguide during four-wave mixing. Phase matching is a crucial prerequisite for the efficient occurrence of nonlinear optical effects like four-wave mixing, and the wave vector mismatch is the key indicator for quantifying this degree of matching. The formula... , , These correspond to the wave vectors of the pump light, probe light, and idler light in the waveguide under test, respectively. The wave vector, as a physical quantity characterizing the propagation properties of light in a medium, is directly related to the wavelength of the light and the refractive index of the waveguide material, specifically satisfying… ,in, For wave vectors; Let be the refractive index of the waveguide material under test for the corresponding wavelength of light. Let be the wavelength of the light in a vacuum. This means that when the same beam of light propagates in media with different refractive indices, the wave vector will change accordingly. Since the refractive index of the same medium differs for different wavelengths of light, the wave vectors of pump light, probe light, and idler light will vary due to their different wavelengths within the same waveguide under test. , , Different values ​​will be displayed.

[0033] Limit the wave vector mismatch to rad / m. From a physical perspective, during four-wave mixing, the energy and momentum of photons must be conserved simultaneously, and the wave vector mismatch... Essentially, it reflects the degree of deviation from the conservation of momentum, that is The smaller the value, the closer the momentum conservation is to the ideal state, the more efficiently the photons participating in the mixing can convert energy, the higher the efficiency of idler light generation, and the stronger the final output idler light power; if If the magnitude is too large, the momentum conservation deviation will cause most of the photon energy to fail to be converted into idler light, instead being dissipated as heat. This will not only cause a sharp drop in idler light power, but may even result in a signal that is too weak to be accurately captured by subsequent measurement channels, directly affecting the accuracy and reliability of waveguide loss measurement. Therefore, Controlled to ≤10⁻ 4Within the range of rad / m, this is a necessary condition to ensure the efficient occurrence of the four-wave mixing effect and to provide a stable signal source for subsequent measurements.

[0034] The refractive index of waveguide materials is temperature-dependent. Almost all materials used to fabricate nonlinear waveguides, such as lithium niobate and chalcogenide glasses, exhibit a slight but tunable change in refractive index with temperature. For example, the temperature coefficient of refractive index of lithium niobate is approximately 1 × 10⁻⁻⁻⁴. 5 / ℃, meaning that for every 1℃ change in temperature, the refractive index will increase by approximately 1×10⁻⁻. 5 The change in the wave vector. The relationship shows that when the waveguide temperature changes, the refractive index of the material... The changes will directly lead to , , The value changes accordingly, thereby altering... The result.

[0035] The specific adjustment process can be achieved using existing high-precision temperature control equipment, such as a thermoelectric cooler (TEC) temperature control station. The waveguide under test is tightly fixed on the stage of the temperature control station, and the actual temperature of the waveguide is monitored in real time through the temperature feedback module of the temperature control station. The monitoring accuracy can reach ±0.01℃. At the same time, the power signal of the idler light is collected in real time through the measurement channel. The idler light power is related to... Negative correlation The smaller the value, the larger the idler power; therefore, the idler power can be used as... Indirect monitoring indicators for whether standards are met. During adjustment, the set temperature of the temperature control station is slowly changed in small increments, such as 0.1℃ / min. After each temperature adjustment, it is held for 3 to 5 seconds to allow the waveguide temperature to stabilize. Then, the trend of the idler optical power is observed. As the temperature is adjusted, the idler optical power will gradually increase and reach a maximum value. At this point, the corresponding... Typically at its minimum; when the idler power stabilizes near its maximum value, and through theoretical calculations, combined with the refractive index data of the waveguide material at that temperature and the wavelengths of each beam, it is substituted into... calculate , , Then obtain ,confirm ≤10⁻ 4 When the temperature reaches rad / m, temperature adjustment is stopped, at which point the wave vector mismatch meets the requirements.

[0036] To further improve the conversion efficiency of the four-wave mixing effect and ensure that the generated idler light has sufficient intensity to support subsequent accurate measurements, specific requirements need to be set for the polarization states of the three laser beams and their compatibility with the waveguide. Specifically, the polarization states of the pump light, probe light, and auxiliary laser must be consistent, and the polarization states of these three beams must be precisely matched with the main polarization direction of the waveguide under test. Light is a transverse wave, and its electric field vibration direction is perpendicular to the propagation direction. The polarization state mentioned above is a physical quantity used to describe the distribution characteristics of the electric field vibration direction. Common polarization states include linear polarization, circular polarization, and elliptical polarization. In nonlinear optical effects, linearly polarized light is the most commonly used choice because its vibration direction is singular and easy to control, and it can be more stably adapted to the transmission characteristics of the waveguide. The requirement for the three beams to have the same polarization state is essentially to ensure that the electric fields of the pump and probe beams vibrate in the same direction within the waveguide. The essence of the four-wave mixing effect is a third-order nonlinear polarization process. In this process, the interaction strength of the light fields is directly related to the consistency of the electric field vibration direction. If there are differences in the polarization states of the three beams, the electric fields of the two beams cannot be effectively superimposed within the waveguide, and the intensity of the third-order nonlinear polarization will be greatly weakened, resulting in a significant reduction in the generation efficiency of the idler light. It may even be impossible for the signal to be captured by the subsequent measurement channel due to its weakness. Conversely, when the three beams have the same polarization state, the electric fields can be superimposed in the same direction to form a stronger light field interaction, providing a prerequisite for an efficient four-wave mixing process.

[0037] Due to their structural characteristics, waveguides under test exhibit polarization dependence when transmitting optical signals. This means that light with different polarization directions faces varying transmission losses, refractive indices, and mode stability when propagating within the waveguide. Among these, one or two specific polarization directions exhibit the lowest transmission loss, the most stable refractive index, and are less prone to polarization mode dispersion (PMD). This means that the signal broadening caused by the difference in transmission speed between different polarization modes occurs in these directions. This specific direction is called the waveguide's dominant polarization direction. For example, the common dominant polarization directions in single-mode optical fibers are horizontal and vertical polarization, corresponding to the two orthogonal polarization modes of the fiber's LP01 mode. However, in lithium niobate planar waveguides, due to the optical anisotropy of the crystal, the dominant polarization direction may be related to the optical axis of the crystal.

[0038] It is necessary to maintain the polarization state of the three beams matching the main polarization direction of the waveguide to avoid polarization loss and polarization distortion during light transmission in the waveguide. If the polarization state of the laser does not match the main polarization direction of the waveguide, for example, if the laser is 45° polarized while the main polarization direction of the waveguide is horizontal, the optical power of the laser after injection into the waveguide will be decomposed into components along the main polarization direction and components along non-main polarization directions. The components along non-main polarization directions will attenuate rapidly during transmission due to high loss. At the same time, the polarization mode coupling effect of the waveguide may also cause random distortion of the polarization state of the light during transmission. This will not only cause power loss of the three beams, especially the loss of pump light power, which will directly weaken the driving capability of four-wave mixing, but also cause the polarization state of the light field reaching the waveguide output to deviate from the initial setting, destroying the consistency of the polarization state of the three beams, and further reducing the conversion efficiency of four-wave mixing. When the polarization states of the three beams are precisely matched with the main polarization direction of the waveguide, the light propagates only along the main polarization direction when it is transmitted in the waveguide. This can minimize power loss and ensure that the polarization state remains stable during transmission. This ensures that the three beams maintain a consistent polarization state throughout the entire waveguide, providing a stable transmission environment for the efficient generation of idler light.

[0039] In practice, the above-mentioned polarization state control and matching can be achieved using existing mature equipment. For the polarization state consistency adjustment of the three laser beams, a polarization controller can be connected in series in the optical path of each laser beam. By adjusting the parameters of the polarization controllers, such as the position of the squeezed fiber and the rotation angle of the waveplate, the polarization state of the three beams is monitored in real time using a polarization analyzer until the analyzer shows that the polarization state parameters of the three beams are consistent, such as the vibration direction angle and degree of polarization of linear polarization. For matching with the main polarization direction of the waveguide, the main polarization direction of the waveguide to be tested needs to be determined first by polarization scanning method. Specifically, a laser beam with a known polarization state, such as a linearly polarized laser, is injected into the waveguide. By rotating the polarization direction of the laser, the laser power at the output end of the waveguide is monitored with a power meter. When the power meter shows the maximum power and the fluctuation is minimal, the polarization direction of the laser at this time is the main polarization direction of the waveguide. After determining the main polarization direction, the polarization states of the three beams are adjusted so that the polarization states of the three beams completely coincide with the main polarization direction, which satisfies the polarization state requirements and ultimately maximizes the four-wave mixing conversion efficiency.

[0040] like Figure 2 As shown, this illustrates the method for determining the range in which the probe optical power and the idler optical power have a linear relationship. This range is a key prerequisite for subsequently fixing the probe optical power and accurately calculating the waveguide loss. The specific operation process must strictly follow the logic below.

[0041] The first step is to maintain the stability of the measurement conditions, i.e., keep the locked waveguide temperature and pump wavelength constant. Adjusting the temperature stabilizes the idler power, essentially ensuring the waveguide operates efficiently and stably with the four-wave mixing effect. Locking the pump wavelength ensures constant phase matching conditions. If the waveguide temperature or pump wavelength shifts, it directly alters the four-wave mixing conversion efficiency, causing the relationship between idler power and probe power to deviate from the true trend, rendering the subsequently fitted linear range meaningless. In practice, a thermoelectric cooler (TEC) temperature control station continuously monitors the waveguide temperature. If the temperature deviates from the locked value, the station must immediately initiate feedback adjustment. Simultaneously, the pump wavelength locking module must monitor wavelength changes in real time, fine-tuning the laser injection current or temperature to ensure the pump wavelength remains stable at the locked value, providing a constant experimental background for subsequent probe power adjustment and data acquisition.

[0042] The next step involves adjusting the probe power and acquiring idler power data. The first step is to set a reasonable probe power adjustment range. This range must be determined by considering the nonlinear response capability of the waveguide under test and the upper limit of the probe laser's output power. If the adjustment range is too narrow, it may not cover the entire linear interval; if it is too wide, exceeding the laser's maximum output power or the waveguide's nonlinear saturation threshold, the idler power will no longer change linearly with the probe power. After setting the adjustment range, starting from the lower limit, the probe power is adjusted upwards in equal steps of 1mW to 2mW. This step size is chosen because too small a step size will result in too many data points and prolonged experimental time, while too large a step size may miss key changes within the linear interval. After adjusting to a new probe optical power point, it is necessary to pause for 3 to 5 seconds before data acquisition. This is because after the probe optical power is adjusted, whether the light intensity is changed by adjusting the adjustable optical attenuator or the output is changed by controlling the laser injection current, the optical power needs a short period of time to stabilize. If data is acquired immediately, the idle frequency optical power data will be deviated due to power fluctuations. The 3 to 5-second pause time ensures that the probe optical power is stable at the set value, thereby ensuring the authenticity of the idle frequency optical power data.

[0043] The data acquisition process requires high-frequency sampling combined with outlier removal and final mean calculation to process the idler frequency optical power signal, thereby reducing the impact of random noise. Specifically, a sampling frequency of 10Hz is used to collect idler frequency optical power data at the current detected optical power point, with a total of 100 sets collected. The 10Hz sampling frequency avoids data redundancy caused by excessively fast sampling and allows for the acquisition of a sufficient number of data points in a short time to reflect the statistical characteristics of the signal. The volume of 100 sets of data provides a basis for subsequent outlier removal, effectively reducing the impact of single burst interference, such as transient electromagnetic noise and detector dark current fluctuations, on the results. After data collection, the three maximum and three minimum values ​​in the 100 data sets need to be removed. This is because such extreme values ​​are usually caused by non-systematic sudden interference, such as brief vibrations in the laboratory environment or instantaneous fluctuations in power supply voltage. If these extreme values ​​are retained, they will raise or lower the overall data level, resulting in distorted results. After removing the extreme values, the arithmetic mean of the remaining 94 data sets is calculated. This mean is the effective value of the idler optical power corresponding to the current probe optical power. Compared with a single data point, the mean can more accurately reflect the true level of the idler optical power and further reduce the interference of random noise.

[0044] Finally, the fitting of the linear relationship and the selection of the linear interval are performed. All probe power points and their corresponding effective values ​​of idler power are organized into data pairs and imported into common data processing software such as Origin and Matlab. Linear regression fitting is performed using the least squares method. The core advantage of the least squares method is that it can obtain the fitting result closest to the true linear relationship by minimizing the sum of squared errors between the data points and the fitted line. It is the standard method for handling linear fitting problems in scientific experiments. After fitting, the goodness of fit R² needs to be calculated. This parameter is used to quantify the degree of fit between the data points and the fitted line: R² ranges from 0 to 1. The closer R² is to 1, the stronger the linear relationship between the idler power and the probe power; if R² is much less than 1, it indicates that there may be a nonlinear relationship or that the data contains significant interference. To ensure that the relationship between idler power and probe power strictly adheres to the linear assumption in subsequent measurements, a continuous probe power sub-interval with R² ≥ 0.998 needs to be selected. R² ≥ 0.998 indicates an extremely high degree of fit between the data points and the fitted straight line, suggesting a valid linear relationship. Continuity is also required because the linear relationship may only hold true in a certain intermediate segment during probe power adjustment. For example, nonlinearity may occur in the low-power or high-power segments due to waveguide nonlinearity saturation. Only a continuous sub-interval can guarantee that the linear relationship requirement will be met regardless of the selected point within the interval when the probe power is fixed. This final selected continuous sub-interval is the desired linear probe power interval.

[0045] Furthermore, the auxiliary laser setup facilitates the generation of noise compensation factors, which is a core step in achieving accurate idler power correction. Outlier removal is a crucial prerequisite for ensuring the reliability of these factors. If the instantaneous power of the auxiliary laser fluctuates abnormally due to sudden interference, unprocessed outlier data will directly cause the noise compensation factor to deviate from the true noise level, thus leading to deviations in the idler power correction results. Therefore, an additional outlier removal step is required during the noise compensation factor generation process. The specific criteria and processing methods are as follows: First, the criteria for outlier determination must be clarified. This determination is based on the average power of the auxiliary laser, combined with the fluctuation amplitude of the instantaneous power. In previous steps, multiple sets of power data for the auxiliary laser have been collected through the reference channel, and the average power has been calculated. This value reflects the baseline power level of the auxiliary laser under stable conditions and can be regarded as the true power when there is no sudden interference. The instantaneous power of the auxiliary laser acquired in real time afterwards... Need and Perform deviation comparison, when the power at a certain instant... With average power The absolute value of the deviation satisfies When the instantaneous power data is within a certain range, it can be determined that the instantaneous power data is an anomaly. The choice of 10% as the deviation threshold is based on the environmental interference characteristics commonly found in waveguide loss measurement experiments. Under normal experimental conditions, such as in a temperature-controlled, vibration-proof, and electromagnetically shielded optical laboratory, the instantaneous power fluctuation of the auxiliary laser is usually controlled within ±5%. The 10% threshold can effectively cover acceptable small fluctuations caused by brief airflow disturbances and slight electromagnetic radiation, while accurately identifying unacceptable abnormal fluctuations caused by sudden strong electromagnetic interference, such as the start-up and shutdown of high-power equipment near the laboratory, instantaneous detector failures, or sudden changes in dark current of photodetectors. This avoids misjudging normal fluctuations as abnormalities, while also ensuring that truly abnormal data is not missed.

[0046] For data deemed abnormal, the moving average of the previous three normal data points should be used to replace the abnormal instantaneous power. This data is then used to calculate the noise compensation factor. The core reason for choosing this processing method is to ensure the temporal continuity and authenticity of the data. The moving average can reflect the changing trend of the auxiliary laser power based on recent continuous normal data, avoiding data breaks caused by directly discarding outliers or replacing them with fixed values. If outliers are directly discarded, there will be no corresponding compensation factor at that moment, making it impossible to achieve real-time correction of idle frequency optical power; if the average power is used... Direct substitution ignores the slow drift of auxiliary laser power within the normal range, such as the slight attenuation of laser output power during long-term experiments, causing the compensation factor to fail to track the true noise level in real time. Calculating a moving average using the first three normal data points serves two purposes: firstly, the sheer size of the three data points smooths out minor errors caused by normal fluctuations and closely approximates the power state before the occurrence of abnormal data, avoiding the lag caused by introducing too much historical data; secondly, the characteristics of the moving average ensure that each outlier handling is based on the latest normal data, dynamically adapting to the slow changes in auxiliary laser power, making the substituted power value closer to the true instantaneous power when there is no abnormal interference.

[0047] Specifically, the moving average, also known as the sliding average, mentioned above is a data processing term. Its core principle is to smooth out short-term fluctuations in the original data by calculating the average of a continuous, fixed number of recent data points, thus more clearly reflecting the overall trend or true level of the data. As new data is generated, the data window for calculating the average moves forward, always based on the latest fixed number of data points, rather than using the entire dataset.

[0048] In waveguide loss measurement methods, the application of the moving average means is to replace the abnormal instantaneous power of the auxiliary laser with the moving average of the first three normal data. Specifically, when the instantaneous power of the auxiliary laser at a certain moment is determined to be abnormal, the abnormal value is not used directly. Instead, the three consecutive normal instantaneous power data before the occurrence of the abnormal value are selected. For example, if the 10th data is abnormal, the 7th, 8th and 9th normal data are taken, and the arithmetic mean of these three data is calculated. This average is the moving average.

[0049] The moving average is used to maintain data continuity, avoid data breaks caused by discarding outliers, ensure that there is a corresponding power value for calculating the noise compensation factor at each time point, and meet the needs of real-time correction. In addition, it is close to the real trend. The first three normal data can reflect the power change trend before the anomaly occurs. It can better reflect the real-time state than using a fixed average power. It not only eliminates the interference of sudden anomalies, but also continues the reasonable trend of the data, providing a reliable basis for the accurate calculation of the noise compensation factor.

[0050] The specific operation process needs to be seamlessly integrated with the instantaneous power acquisition and compensation factor calculation process. When acquiring the auxiliary laser instantaneous power at a set frequency through the reference channel... Then, first calculate the current... With average power The percentage of deviation; if the percentage of deviation is less than 10%, the data is considered normal, directly retained, and used to calculate the noise compensation factor at the current moment. If the deviation percentage is ≥10%, the data is marked as abnormal. Then, the three consecutive normal instantaneous power data points collected before the abnormal data are retrieved, and the moving average of these three data points is calculated. This average is then recorded as the instantaneous power. Finally, the values ​​are substituted into... The formula calculates the noise compensation factor at the current moment.

[0051] Through the above outlier removal and data substitution steps, sudden interference in the instantaneous power of the auxiliary laser can be effectively filtered out, ensuring that each noise compensation factor can accurately reflect the real noise level of the current experimental environment. This makes the subsequent idler frequency optical power correction based on the factor more accurate, providing a reliable optical power data basis for the accurate calculation of the waveguide loss coefficient.

[0052] Noise compensation factor The calculation formula is as follows: This involves two key parameters: one is the average power of the auxiliary laser. Secondly, the instantaneous power of the auxiliary laser. . This is the reference power value of the auxiliary laser under stable conditions. It is acquired by continuously collecting power data of the auxiliary laser at a set frequency using a photodetector and data acquisition card through a reference channel. The acquisition time typically needs to cover more than one minute to ensure sufficient data to reflect the stable state. Then, the arithmetic mean of all acquired valid data is calculated; this average value is the reference power. . The essence is the true power level of the auxiliary laser when there is no sudden noise interference, which can be used as a benchmark to judge whether the subsequent instantaneous power is affected by noise.

[0053] and This refers to the real-time acquisition of auxiliary laser power data. Its acquisition frequency must be completely consistent with the acquisition frequency of the instantaneous power of the idler light in the measurement channel to ensure that the idler light power at each moment corresponds to the auxiliary laser power at the same moment, thus achieving synchronous compensation. Fluctuations may occur due to environmental noise, such as slight waveguide vibrations, fluctuations in laboratory power supply voltage, and instantaneous changes in detector dark current. When noise causes a decrease in auxiliary laser power, It will be smaller than When noise causes the auxiliary laser power to increase, It will be greater than .

[0054] Noise compensation factor The physical significance lies in quantifying the degree of impact of noise on optical power at the current moment. When the noise level decreases, The calculated result will be greater than 1, meaning that the idler optical power of the measurement channel is likely also underestimated due to the same noise, and needs to be adjusted accordingly. Perform positive compensation; when When noise increases, The value will be less than 1, which may indicate that the idler frequency optical power is overestimated and needs to be checked. Perform reverse correction; if and Basically the same, then A value close to 1 indicates that no additional compensation is needed. This measurement method based on auxiliary laser essentially utilizes the fact that the auxiliary laser and idler light are transmitted in the same waveguide and split by the same beam splitter. Since both are affected by environmental noise in a consistent manner, the auxiliary laser acts as a noise probe, indirectly achieving noise correction for the idler light power.

[0055] Based on the above noise compensation factors This allows for the correction of the idler frequency optical power, and the corrected idler frequency optical power... The calculation formula is: This involves the instantaneous power of idler frequency optical signals. System error with measurement channel Two key parameters.

[0056] It is the raw power data of the idler frequency optical spectrum acquired in real time by the measurement channel, and its acquisition process is similar to... Synchronization involves converting the idler frequency optical signal into an electrical signal via a bandpass filter and photodetector in the measurement channel, which is then recorded as an instantaneous power value by the data acquisition card. However... It will be affected by two factors simultaneously: one is the environmental noise mentioned earlier, through... Compensation is needed for two reasons: first, the inherent systematic errors of the measurement channel itself; and second, the errors that need to be addressed through… Correction is needed, therefore multiplying by is required first. Cancel environmental noise, then introduce Correct system errors.

[0057] Here It is a systematic error in the measurement channel, which differs from the randomness of environmental noise. This is an inherent error determined by the hardware characteristics of the measurement channel, which is stable and can be pre-calibrated. Its main sources include three categories: First, the dark current error of the photodetector. Even in the absence of light, the detector will generate a weak current due to its semiconductor characteristics. This current will be misinterpreted as the power signal of the idler light, forming a fixed deviation. Second, the insertion loss error of the bandpass filter. While filtering out stray light, the filter will also cause a certain power attenuation of the idler light, and the attenuation amount is fixed. Third, the quantization error of the data acquisition card. When the acquisition card converts analog electrical signals into digital signals, it will generate a small error due to the limitation of the minimum quantization unit. This error shows a stable distribution in multiple measurements.

[0058] The specific values ​​need to be obtained through calibration experiments before the experiment. For example, for dark current error, the power value corresponding to the detector output current can be recorded when there is no idler light incident; this is the error component caused by dark current. For filter insertion loss, a standard idler light source with known power can be used to measure the power of the light source directly incident on the detector and the power incident on the detector after passing through the filter; the difference between the two is the insertion loss error component. After superimposing the various error components, the result can be obtained. The specific value needs to be determined. The symbol.

[0059] If the error causes the measured value to be too small, It should be a positive value; if it causes the measured value to be too large, It is a negative value. Therefore, in the formula... It needs to be determined based on the calibration results to ultimately ensure It can accurately reflect the true power level of idler light, which not only cancels out the random interference of environmental noise, but also corrects the inherent deviation of the hardware system, providing reliable idler light power data for the accurate calculation of waveguide loss coefficient.

[0060] Furthermore, in calculating the linear loss coefficient of waveguides... At that time, the four-wave mixing conversion efficiency It is one of the core parameters, and its value is not constant but fluctuates significantly with the actual operating temperature of the waveguide. If a fixed value is directly used... Values, such as initial measurements at room temperature, when substituted into the loss calculation formula, will be affected by... Deviations from actual operating conditions lead to significant errors in the final loss results. Therefore, it is necessary to consider the already determined actual operating temperature of the waveguide when calculating the loss. Targeted corrections are made, and the specific correction method is achieved through the following mathematical relationship. The physical meaning, acquisition method, and correction principle of each parameter can be explained in detail below.

[0061] This represents the four-wave mixing conversion efficiency at the reference temperature and is the baseline value for the correction calculation; For the corresponding reference temperature, a common stable temperature in the experimental environment is usually selected, such as standard room temperature of 25℃, i.e., 298.15K. This can also be adjusted according to the application scenario of the waveguide under test. For example, if the actual operating environment of the waveguide is 50℃, then... Set the temperature to 50℃ to reduce the correction range; The temperature coefficient of four-wave mixing conversion efficiency reflects the efficiency of a 1°C change in temperature. The key parameter for the change ratio is determined by the inherent properties of the waveguide material. This refers to the locked actual operating temperature of the waveguide. This temperature is the optimal temperature for triggering the efficient four-wave mixing effect, and it is also the true operating temperature of the waveguide throughout the entire measurement process. It must be compared with the temperature in the correction formula. Strict consistency is necessary to ensure the accuracy of the revised version. It conforms to actual working conditions.

[0062] Next, we need to explain the specific methods for obtaining each parameter, which is the practical basis for implementing the correction. Regarding the reference temperature... Corresponding conversion efficiency The temperature needs to be obtained through a pre-calibration experiment: Before the experiment begins, place the waveguide under test on a temperature control platform and adjust the platform to stabilize the waveguide temperature. Keeping the pump light wavelength, probe light power, and other conditions consistent with the subsequent formal measurements, the pump light power at this point is then measured. Detection of optical power Idle frequency optical power Combined with the known waveguide length Substitute into the basic formula of four-wave mixing At this point, the typical loss coefficient of the waveguide material can be used for approximate calculation, or the loss coefficient of a standard waveguide from the same batch with known loss can be used for calibration to deduce the loss coefficient. Below This process needs to be repeated 3 to 5 times, and the average value is taken as the final result. This avoids random errors in a single calibration.

[0063] temperature coefficient The acquisition of this data relies on the characteristic data of the waveguide material or specific measurements: different nonlinear waveguide materials Significant differences exist, for example, in lithium niobate (LiNbO3) waveguides. Approximately 1×10⁻³~5×10⁻³ / ℃, chalcogenide glass waveguides It is even smaller, about 1×10⁻ 4 ~5×10⁻ 4 / ℃, if the waveguide supplier provides a temperature coefficient handbook for the material, you can directly refer to the handbook. Value; for higher precision, experimental measurement can also be performed. Select multiple temperature points, such as 20℃, 25℃, 30℃, and 35℃, and label each temperature point accordingly. Then, with temperature as the x-axis, Plot a curve on the ordinate and calculate the slope of the curve using linear fitting. Slope = And then we can deduce =Slope / It should be noted that, The sign can be positive or negative; for most materials, the sign increases with temperature. It shows a slight upward trend, therefore It should be positive, but regardless of the sign, it must be completely preserved to ensure the accuracy of the correction formula.

[0064] The core principle of temperature correction needs to be understood by returning to the essence of the four-wave mixing effect: four-wave mixing conversion efficiency. The third-order nonlinear polarizability and phase-matching condition of the waveguide are directly related, and both of these factors are regulated by waveguide temperature. When the waveguide temperature changes, the refractive index *n* of the material changes accordingly. This change in refractive index not only affects the phase-matching condition, thus altering the degree of photon momentum conservation, but also indirectly affects the third-order nonlinear polarizability. The third-order nonlinear polarizability is related to the electron cloud distribution characteristics of the material; temperature changes alter the lattice vibration state, affecting the electron cloud response. The superposition of these two factors leads to… As temperature changes, for example, when the temperature increases, if the refractive index changes... If it is closer to 0, and the third-order nonlinear polarizability increases slightly, then... It will rise accordingly; conversely, if temperature changes disrupt phase matching, The temperature will then drop. Therefore, the actual operating temperature locked in step S3... Compared with reference temperature If differences exist It will inevitably deviate If not corrected, substituting it into the loss formula will directly lead to... The calculation deviation, for example, if the actual Higher than ,and Positive, uncorrected It will be smaller than the real thing. Calculated It will be too large because of the formula. and Positive correlation, If it's too small, you need a larger one. To satisfy the equation.

[0065] In actual correction calculations, the actual operating temperature of the waveguide must first be read and recorded by the temperature control station sensor. Then retrieve the pre-calibrated , and Value, substitute into the formula Calculate the corrected For example, if =25℃, =0.08 (i.e., 8%) =3×10⁻³ / ℃, locked =27.5℃, then the corrected =0.08×[1+3×10⁻³×(27.5-25)]=0.08×(1+0.0075)=0.0806, or 8.06%. This corrected value... Substitute into the waveguide loss calculation formula Only then can we ensure The calculation results are consistent with the actual loss state of the waveguide, avoiding the influence of temperature. Deviation introduces additional measurement errors, ultimately improving the overall accuracy of waveguide loss measurement.

[0066] In a preferred embodiment, When the numerical value cannot be predetermined, the iterative method for calculating the linear loss coefficient requires a clear iterative formula and convergence criteria to ensure the standardization of the calculation process and the reliability of the results. Specifically, the iterative process uses two core formulas to cyclically update the parameters and determines the iteration endpoint based on a preset termination condition. Meanwhile, the appropriate setting of the initial value provides a basis for iterative convergence. The details of each step are as follows.

[0067] The core formulas for iterative calculation include the iterative formula for the four-wave mixer conversion efficiency and the update formula for the linear loss coefficient. Specifically, the four-wave mixer conversion efficiency in the nth iteration... Through formula The calculation shows that this formula is derived from a transformation of the power relationship formula for four-wave mixing, with the core formula being... After rearranging the terms, we can obtain... The relationship with other parameters is shown here using the loss coefficient of the nth round. Substitute the values ​​to obtain the conversion efficiency for the corresponding round. The source of each parameter in the formula must be clearly stated. This is the effective value of the idler frequency optical power after noise compensation and system error correction; The actual incident power of the pump light is measured directly using a power meter. The operating power of the probe light within the defined linear range was also measured using a power meter; The physical length of the waveguide under test is obtained using high-precision measuring tools; The linear loss coefficient assumed in the nth iteration; This is an exponential term that reflects the effect of waveguide loss on optical power attenuation.

[0068] Based on the conversion efficiency of the nth round It can be done through the formula Calculate the linear loss coefficient for the (n+1)th round. The derivation of this formula stems from a logarithmic transformation of the core formula. Taking the natural logarithm of both sides and rearranging, we can separate the results. The expression, substituted into the nth round After that, the updated version was obtained. This process essentially involves using the conversion efficiency obtained from the previous iteration to deduce a loss coefficient that is closer to the true value, thereby achieving gradual optimization of the parameters.

[0069] Assumed linear loss coefficient for the initial round The initial values ​​need to be set appropriately based on the material properties of the waveguide under test; this is crucial to ensuring rapid convergence of the iterations. Different types of waveguide materials exhibit varying linear loss coefficients within typical ranges due to differences in fabrication processes and structural characteristics. For example, the linear loss of silicon-based photonic crystal waveguides is typically between 0.1 dB / cm and 2 dB / cm, while that of lithium niobate (LiNbO3) waveguides is mostly between 0.5 dB / cm and 5 dB / cm, and that of chalcogenide glass waveguides may be slightly higher, approximately between 1 dB / cm and 10 dB / cm. An intermediate value within the loss range of the corresponding material can be selected, such as 1 dB / cm for silicon-based waveguides and 2 dB / cm for lithium niobate waveguides. If historical measurement data of waveguides from the same batch and with the same structure are available, the historical data can also be used as a reference to further shorten the number of iterations. The initial value does not need to be absolutely precise; as long as it is within the typical loss range of the material, the iteration process can be guaranteed to converge to the true value.

[0070] The iterative process must terminate under a preset condition: when the absolute value of the difference between the linear loss coefficients obtained from two adjacent iterations reaches a certain threshold. When the value reaches dB / cm, the iteration stops. This is the final linear loss coefficient. Choose 10⁻ 5 dB / cm is used as the termination threshold, which is the result of comprehensively considering the measurement accuracy requirements and computational efficiency. The conventional accuracy requirement for waveguide loss measurement is on the order of 10⁻³dB / cm. 5 The dB / cm threshold is much higher than the actual requirements, which can ensure the reliability of the results. At the same time, this threshold can avoid the waste of computational resources caused by too many iterations. In practice, for most waveguides, 5 to 10 iterations are sufficient to meet this condition. The whole process can be completed automatically by programming tools such as Python and Matlab without manual intervention.

[0071] Through the synergistic effect of the above iterative formula, initial value setting principle and termination condition, even when the four-wave mixing conversion efficiency is unknown, the linear loss coefficient can be calculated systematically and efficiently, and the accuracy of the result can meet the stringent requirements of waveguide loss measurement, further improving the calculation logic.

[0072] In the waveguide loss measurement method based on dual-channel detection described above, the calculated linear loss coefficient... Multiple measurements are required to verify the results and eliminate any random or potential systematic errors that may exist in a single measurement, ensuring the reliability and accuracy of the final result. This verification process must strictly follow the logic of repeated measurements under the same conditions and statistical error analysis to determine the validity of the final result. The specific operation and judgment criteria are as follows.

[0073] First, the linear loss coefficient measurement procedure must be repeated under identical experimental conditions, with the number of repetitions set to 3 to 5. These identical experimental conditions are the core prerequisite for ensuring the comparability of multiple measurement data. Specifically, this includes: maintaining a constant waveguide temperature and pump light wavelength, with temperature fluctuations ≤ ±0.01℃ and wavelength fluctuations ≤ ±0.01nm; ensuring consistency between the determined linear range of the probe light power and the fixed probe light power value; keeping the acquisition parameters of the auxiliary laser and idler light, such as sampling frequency and outlier rejection rules, unchanged; and maintaining the initial state of the waveguide placement and optical path alignment, confirmed by observing the stability of the output light power. Choosing 3 to 5 repetitions is a result of comprehensively considering error statistics requirements and experimental efficiency. Too few repetitions cannot effectively reflect the dispersion of the data and make it difficult to distinguish between random and systematic errors; too many repetitions will significantly prolong the experimental time and have limited gain for error analysis. 3 to 5 repetitions satisfy the statistical significance requirements while also considering experimental efficiency.

[0074] After completing multiple measurements, the relative standard deviation (RSD) of these results needs to be calculated to quantify the stability and consistency of the data. The formula for calculating the relative standard deviation is RSD = (standard deviation / arithmetic mean) × 100%, where the standard deviation reflects the dispersion of each measurement from the average, and the arithmetic mean represents the central tendency of multiple measurements. The physical meaning of the relative standard deviation is that it expresses the relative fluctuation of the measurement results as a percentage, avoiding error evaluation bias caused by differences in absolute values. For example, if three measurements yield… The values ​​were 0.8 dB / cm, 0.82 dB / cm, and 0.78 dB / cm, respectively, with an average value of 0.8 dB / cm, a standard deviation of approximately 0.02 dB / cm, and a relative standard deviation of (0.02 / 0.8) × 100% = 2.5%.

[0075] Based on the measurement accuracy requirements, when the relative standard deviation of multiple measurement results is ≤2%, the measurement data is considered to have good repeatability and stability. In this case, the arithmetic mean of these 3 to 5 measurement results is taken as the final linear loss coefficient. This is because a relative standard deviation of ≤2% means that the fluctuations of each measurement around the average value are small, indicating that the experimental system is less affected by random errors and there are no significant systematic errors, such as slow optical path offset or detector sensitivity drift. The average value can be closer to the true value of waveguide loss. At the same time, the 2% threshold is a conventional accuracy requirement in the field of waveguide loss measurement, which can ensure the reliability of the results without causing reasonable data to be misjudged due to excessive stringency.

[0076] If the relative standard deviation of multiple measurements exceeds 2%, it indicates that the measurement system may have unresolved systematic errors. In this case, the current measurement process must be terminated, and all operations after the waveguide temperature determination step must be re-executed to systematically eliminate the source of error. Possible systematic errors include decreased stability of temperature or wavelength locking, deviations in the determined linear range, sudden interference in the calculation of the noise compensation factor, and changes in the optical path alignment. When re-executing the process, the above steps should be carefully checked. For example, the locking stability should be confirmed by extending the temperature monitoring time, the goodness of fit of the linear range should be re-verified, and the long-term power stability of the auxiliary laser should be checked until the relative standard deviation of multiple measurements is ≤2%, and then the average value should be taken as the final result.

[0077] Through the above-described multiple measurement verification and error analysis steps, random errors and systematic errors can be effectively distinguished, ensuring the linear loss coefficient of the final output. It accurately reflects the inherent characteristics of the waveguide under test, further improving the reliability and accuracy of the entire measurement method.

[0078] The embodiments described in this specific implementation are preferred embodiments of this application and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.

Claims

1. A waveguide loss measurement method based on dual-channel detection, characterized in that, Includes the following steps: Step S1: The pump light, probe light and auxiliary laser are combined and injected into the nonlinear waveguide under test to trigger the four-wave mixing effect to generate idler light; Step S2: The output light of the waveguide is split into two paths by a beam splitter. The signal of the idler light is extracted from one path as a measurement channel; the signal of the auxiliary laser is extracted from the other path as a reference channel. Step S3: Adjust the temperature of the waveguide until the idler light power of the measurement channel reaches a stable state, then lock the temperature of the waveguide and the wavelength of the pump light at this time; Step S4: Maintain the temperature of the waveguide and the wavelength of the pump light locked in step S3, record the idler light power under different probe light powers and fit the relationship between the two to obtain the probe light power range with a linear relationship. Step S5: Maintain the temperature of the waveguide and the wavelength of the pump light locked in step S3, and fix the power of the probe light and set it within the range obtained in step S4. Generate a noise compensation factor by measuring the average power and instantaneous power of the auxiliary laser, and use the noise compensation factor to correct the measured power of the idler light in real time. Step S6: Calculate the waveguide loss based on step S5; ; in, For four-wave mixing conversion efficiency; The pump light power; The operating power of the probe light; The corrected idler frequency optical power; The length of the waveguide; This is the linear loss coefficient.

2. The waveguide loss measurement method based on dual-channel detection according to claim 1, characterized in that, The following conditions must be met in step S1; rad / m; in, This is the wave vector mismatch. The wave vector of the pump light in the waveguide; The wave vector of the probe light in the waveguide, The wave vector of the idler frequency light in the waveguide is adjusted by regulating the temperature of the waveguide to ensure that the wave vector mismatch meets the requirements.

3. The waveguide loss measurement method based on dual-channel detection according to claim 1, characterized in that, In step S1, the pump light, the probe light, and the auxiliary laser have the same polarization state and are all matched with the main polarization direction of the waveguide to be tested, so as to maximize the four-wave mixing conversion efficiency.

4. The waveguide loss measurement method based on dual-channel detection according to claim 1, characterized in that, Step S4 includes, Step S4.1: Maintain the temperature of the waveguide and the wavelength of the pump light locked in step S3; Step S4.2: Set the detection light adjustment range. Starting from the lower limit of the adjustment range, adjust the power of the detection light upward in steps of 1mW to 2mW, and hold for 3s to 5s after each power adjustment. Step S4.2: Collect 100 sets of idler optical power data at the power point using a sampling frequency of 10Hz, remove the 3 maximum values ​​and 3 minimum values, and take the arithmetic mean as the effective value of the idler optical power corresponding to the probe optical power; Step S4.3: Import the effective values ​​of all probe light power and idler light power into the data processing software, perform linear regression fitting using the least squares method, and calculate the goodness of fit R²; select the continuous probe light power sub-interval with R² ≥ 0.998, which is the probe light power interval with a linear relationship mentioned in step S4.

5. The waveguide loss measurement method based on dual-channel detection according to claim 1, characterized in that, The generation of the noise compensation factor in step S5 also includes outlier removal. When the absolute value of the deviation between the instantaneous power and the average power of a certain auxiliary laser is ≥10%, it is determined to be abnormal data. The moving average of the first 3 normal data is used to replace the instantaneous power of the auxiliary laser to calculate the noise compensation factor.

6. The waveguide loss measurement method based on dual-channel detection according to claim 1, characterized in that, The noise compensation factor is calculated as follows; ； in, The noise compensation factor is mentioned above. The average power of the auxiliary laser; The instantaneous power of the auxiliary laser; The method for calculating the corrected idler frequency optical power is as follows; ； in, The instantaneous power of the idler frequency light; This is for measuring the systematic error of the channel.

7. The waveguide loss measurement method based on dual-channel detection according to claim 1, characterized in that, The four-wave mixing conversion efficiency in step S6 needs to be corrected according to the actual operating temperature of the waveguide. ; in, The conversion efficiency is at the reference temperature; For reference temperature; Temperature coefficient; This represents the actual operating temperature of the waveguide.

8. The waveguide loss measurement method based on dual-channel detection according to claim 1, characterized in that, In step S6, when the four-wave mixing conversion efficiency is unknown, the linear loss coefficient is calculated using an iterative method.

9. The waveguide loss measurement method based on dual-channel detection according to claim 8, characterized in that, The iterative formula is: ; ; in, The four-wave mixing conversion efficiency in the nth iteration; The linear loss coefficient for the nth iteration; The linear loss coefficient for the (n+1)th iteration; The assumed value of the linear loss coefficient in the initial round is set based on the waveguide material properties; The iteration termination condition is: dB / cm。 10. The waveguide loss measurement method based on dual-channel detection according to claim 1, characterized in that, The linear loss coefficient α calculated in step S6 needs to be verified by multiple measurements. The measurements should be repeated 3 to 5 times under the same experimental conditions. If the relative standard deviation of the multiple measurement results is ≤2%, the average value should be taken as the final result. Otherwise, repeat steps S3 to S6 to eliminate system errors.

Images