Method for tuning an optical transceiver and test / measurement apparatus

Abstract

To provide a method for quickly tuning an optical transceiver, a method for training a machine learning system, and a test and measurement device.SOLUTION: In a test measurement device 30 connected to an optical transceiver 32 and including one or more processors 38, the processors 38 are configured to: initially set operating parameters for the optical transceiver 32 to average parameters; acquire a waveform from the optical transceiver 32; measure the acquired waveform to determine if operation of the transceiver passes or fails; and send the waveform and the operating parameters to a machine learning system 42 to obtain estimated parameters. If the optical transceiver 32 fails, the processors 38 are configured to: adjust the operating parameters based on the estimated parameters; and repeat the acquiring, measuring, sending, and adjusting as needed until the transceiver 32 passes.SELECTED DRAWING: Figure 2

Description

TECHNICAL FIELD

[0001] The present disclosure relates to test and measurement systems, and more particularly, to systems and methods for tuning optical transceivers, for example, during manufacturing.

BACKGROUND ART

[0002] Manufacturers of optical transceivers may take up to 2 hours in the worst case to tune (adjust to the optimal state) an optical transceiver. Manufacturers typically achieve this by sweeping tuning parameters (varying values widely one after another). In some cases, up to 200 iterations may be required in the worst case, and 3 to 5 iterations may be required in the best case.

[0003] Currently, tuning of optical transceivers does not use any type of machine learning to reduce the time required for transceiver tuning and testing. The time required for machine learning training will increase the overall cost of the transceivers. Figure 1 shows a typical manufacturer's workflow. At 10, a user connects an optical transceiver (DUT: device under test) to a test and measurement apparatus. As used in the present application, the term "test and measurement apparatus" includes any apparatus capable of performing tests and measuring the resulting operation of an optical transceiver as a device under test.

[0004] At 12, the user sets the temperature and waits for the temperature to stabilize. Next, at 14, the user sets the voltage. At 16, the user sets the operating parameters of the DUT, operates the DUT, and then at 18, measures its operation. There are various forms of measurement, which will be explained in detail later. At 20, the system determines whether the measurement indicates that the component is operating within the desired range of measurements. At 20, if it is operating within the desired range, the parameters are measured again at 21. At 22, it is determined whether the parameters pass the criteria. If the parameters pass, at 24, the final temperature and voltage used in the test are saved and used for the next transceiver, and then at 26, the test of that component is completed.

[0005] If the DUT is not tuned at 20, or fails to meet the criteria at 22, it returns through 22, allowing the user to adjust the parameters at 16 and rerun the test. This process may be repeated multiple times until the DUT meets the criteria at 24. [Prior art documents] [Patent Documents]

[0006] [Patent Document 1] Special Publication No. 2015-528257 [Overview of the project] [Problems that the invention aims to solve]

[0007] Reducing the number of iterations required to tune the transceiver and thereby speeding up the tuning process will reduce both manufacturing time and, consequently, manufacturing costs.

[0008] The embodiments of the disclosed apparatus and method address the shortcomings of the prior art. [Means for solving the problem]

[0009] The process described above is time-consuming and consequently costly. Embodiments of this invention utilize machine learning to provide a more accurate starting point and to tune parameters more efficiently.

[0010] Using machine learning processes can reduce the time required for tuning, testing, and determining whether a component passes or fails. Figure 2 shows an embodiment of a test measurement system or device that uses a machine learning system to determine the operating parameters of a component. Output measurements may include TDECQ (transmitter and dispersion eye closure penalty quaternary), extinction ratio, average optical power, optical modulation amplitude (OMA), and level separation mismatch ratio, which can be tuned by the user and used to measure operation. Other tuning parameters for an optical transceiver that may be changed or adjusted may include bias current, modulation voltage, and others. [Brief explanation of the drawing]

[0011] [Figure 1] Figure 1 shows an example of a manufacturing workflow for tuning the parameters of an optical transceiver. [Figure 2] Figure 2 shows the test measurement equipment available for tuning the optical transceiver. [Figure 3] Figure 3 shows a flowchart of one embodiment of the process for training a machine learning system to tune an optical transceiver. [Figure 4] Figure 4 shows a flowchart of one embodiment of the process for tuning an optical transceiver using machine learning. [Figure 5]Figure 5 shows a flowchart of one embodiment of the process of tuning an optical transceiver using machine learning. [Figure 6] Figure 6 shows an example of a tensor image derived from a waveform used in a machine learning system. [Modes for carrying out the invention]

[0012] Figure 2 shows a test measurement device 30 that can be used for tuning an optical transceiver. The test measurement device 30 may have a wide variety of components. Figure 2 shows one possible embodiment and may have more or fewer components. The test measurement device in Figure 2 is connected to the DUT (device under test), in this case the optical transceiver 32. The test measurement device 30, such as an oscilloscope, may also have a photoelectric (O / E) conversion module 34, which is connected to the test measurement device 30 via a connector (36, etc.). The processor 38 represents one or more processors and will be configured to run a program including the manufacturer's test automation program 40 to cause the processor 38 to perform various tasks. Alternatively, the test automation program may be part of a separate computing device from the test measurement device 30 and interact with the processor 38. Figure 2 shows a machine learning system 42 as part of the test measurement device 30. Alternatively, the machine learning system may reside on a separate, independent device such as a computer and communicate with the test measurement device 30.

[0013] In this description, the term “test measurement device” as used in this application includes embodiments having an external computing device as part of the test measurement device. The memory 44 may reside on the test measurement device 30 or be distributed between the test measurement device 30 and the computing device. As will be described in more detail later, the memory 44 may include training data and runtime data used by a machine learning system. The test measurement device 30 may also have a user interface 46, which is a touch interface. It may have one or more of the following: a touchscreen, video screen, knobs, buttons, lights (light sources), keyboard, mouse, etc.

[0014] During operation, the optical transceiver 32 is connected to the test measurement device 30 via the O / E conversion unit 34 and the connection unit 36. The automated test process 40 transmits transceiver tuning parameters to the optical transceiver 32. These transceiver parameters may be based on average parameters received from a machine learning system 42 and generated through testing and training. The test measurement device 30 captures a waveform from the transceiver (DUT) and returns the waveform to the processor 38. The system measures the waveform and determines whether the transceiver 32 passes or fails with respect to its set of tuning parameters.

[0015] One of the measurements is the TDECQ mentioned above, which uses the PAM4 eye diagram. As mentioned above, this is one of several measurements taken during operation when the transceiver is tuned, others of which may include the extinction ratio (ER), average optical power (AOP), optical modulation amplitude (OMA), level separation mismatch (RLM) rate, and others.

[0016] The machine learning system 42 selects and adjusts tuning parameters at runtime based on its training. This training process builds a training set of tuning parameters and associated waveforms, which therefore provide information for selecting tuning parameters based on waveforms obtained from the DUT. Figure 3 shows the process of training the machine learning system, which also provides the starting parameters for the runtime testing process.

[0017] In the following description, several terms have specific meanings. In the training of a machine learning system, the process requires a set of training parameters and associated waveforms. As discussed in detail below, a user optimally tunes a large number of transceivers, resulting in a set of parameters referred to herein as "optimization parameters", and the large number of transceivers will hold the optimization parameters for each of these large number of transceivers. The term "average parameter" refers to a set of parameters, each of which represents the average of a specific parameter.

[0018] In the first iterative sub-process, at 50, a user connects an optical transceiver to a test and measurement apparatus. This process typically includes a procedure that allows the temperature of the transceiver to stabilize. At 52, the transceiver undergoes standard tuning, which refers to any current tuning procedure used prior to implementing the embodiments of the present application. Once a particular transceiver is tuned to reach a tuned (optimally adjusted) state, the parameters used in that state are the optimization parameters for that transceiver. The optimization parameters of each of these iterative processes are collected into a histogram or the like at 56, and an average is calculated at 58.

[0019] As will be described in more detail below with respect to FIG. 4 and the runtime environment, these averages serve as the initial starting parameters for the runtime process. This process continues at 54 until it is determined that a sufficient number of transceivers have undergone tuning. The determination of how many transmitters are sufficient will depend on the architecture of the machine learning system and how many accurate results the system needs to generate.

[0020] The second iterative sub-process begins at 60. The user connects the transceiver and stabilizes the temperature. At 62, the transceiver is tuned to a "next" set of parameters, and these parameter sets consist of the initial set of parameters in the first iteration. The system also records these parameters which are transmitted to the machine learning system. In each iteration, the parameters are swept. That is, the system adjusts each parameter to its next setting until all of the multiple settings for that parameter are completed.

[0021] At 64, the system captures the waveform for each respective set of parameters, and transmits the waveforms to the machine learning system. As part of the process of capturing the waveforms, the system may perform clock recovery and generate a partial eye diagram, also referred to as a short pattern overlay. In some embodiments, the partial eye diagram may include only 1-level transitions, where a 1-level transition means transitioning to another level with only one transition of low-to-high, high-to-low, etc., by one level, similar to the 1-level separated partial eye diagram described in U.S. Patent Application No. 17 / 592,437. The resulting overlay may constitute the waveform transmitted to the machine learning system, but in other embodiments, the waveform may be represented in other formats or vector spaces. In one embodiment, as part of the waveform capture process, the system also calculates a number of measurements such as TDECQ, OMA, ER, AOP, RLM, etc., as described above. At 66, the system transmits the collected information associated with each set of parameters to the machine learning system for processing, and this information includes histograms, averages, and information unique to each respective iteration, such as waveforms, short pattern overlays, and measurements. The machine learning system associates all of this information with the waveforms to train the machine learning system to identify the most accurate parameters in the next iteration. This process is repeated until it is confirmed at 68 that all iterations for one parameter have been swept.

[0022] As part of the machine learning training process, the system undergoes testing and validation. The system is supplied with test waveforms to the machine learning system, which generates predicted values ​​for the following parameter set. These parameters are then tested to verify that the machine learning system made accurate predictions. The training process continues until a sufficient number of transceivers have been screened and the machine learning system produces results with the desired accuracy (as described in step 70). Once this is achieved, the training process is completed (as described in step 72), and the system can move to runtime use. It should be noted that the system may return to the training process if necessary, such as if prediction accuracy decreases or if there are any changes in the properties of the transceiver under test.

[0023] Figure 4 shows one embodiment of a method for testing an optical transceiver in a runtime environment. In 80, the user connects the transceiver to the test measurement device and stabilizes the temperature. The temperature may be one of several adjustments to be performed, and may be various temperatures in multiple iterations. In 82, the system tunes the transceiver to a specific set of operating parameters. As used in this application, the term “operating parameters” means the parameters set for the transceiver. The operating parameters may change depending on their position in the process. Initially, the process sets the operating parameters to the average parameters created in the first part of the training process. In 84, the test measurement device acquires waveforms from the transceiver. Similar to the training process, this may include performing clock recovery and generating a partial eye diagram overlay containing only one level transition. In 86, the system may also perform measurements such as TDECQ, ER, AOP, OMA, and RLM. In 88, if the results indicate that the transceiver’s operation is within the desired range, the transceiver passes in 90.

[0024] In one embodiment of measurement and pass / fail determination shown in Figure 4, as shown in Figure 5, The process performs clock recovery 110, from which a tensor image is generated in 111. An example of a tensor image is shown in Figure 6. Figure 6 shows the input waveform on the left, with three channels representing the input waveform on red channel 120, blue channel 122, and green channel 124. Tensor 130 and the tensor at the same position spread across the figure correspond to the red channel, the position of tensor 132 corresponds to the blue channel, and the position of tensor 134 corresponds to the green channel. Short pattern images created from the acquired waveforms may be placed on all three color channels. Different code sequences may be placed on different color channels. Many combinations of patterns and channels are possible.

[0025] The tensor image is sent to the machine learning system as a waveform image for use by the machine learning system. This process then measures the waveform at 112 in Figure 5, and then checks at 114 whether the measurement result is within the range. The process then returns to 88 in Figure 4.

[0026] Referring again to Figure 3, in this embodiment, the waveform capture process would include the process of generating images and tensors to be used by the machine learning system. The machine learning system is trained on these tensor images. This allows the system to use the tensor images of the machine learning system at runtime.

[0027] If the results indicate that the transceiver does not operate within the desired range of measurements used, the transceiver is deemed unacceptable. The system may, in step 92, check how many times the transceiver has undergone the process. If the number is less than a predetermined number, the results are sent, in step 94, to the machine learning system for determining the next set of operating parameters. The machine learning system then provides estimated parameters for the next iteration. In one embodiment, the operating parameters of the transceiver in the next iteration consist of "adjusted parameters" obtained by subtracting the estimated parameters from the previous operating parameters to find the difference, and then adding this difference to the average parameters to generate new operating parameters.

[0028] Returning to 92, if the transceiver continues to fail despite repeated machine learning processes, the system uses a “standard” procedure for tuning at 96. This consists of some tuning procedure that was used prior to the implementation of the embodiments of the present invention. This process then continues until the transceiver’s operation meets the desired measurement at 98 and the transceiver passes at 90. However, if the standard tuning process fails to adjust the transceiver to operate within the desired range, the transceiver fails at 100.

[0029] In this way, the machine learning system can improve the efficiency of the tuning process. This allows the system to pass or fail transceivers more quickly, reducing both the time and cost of the manufacturing process.

[0030] Embodiments of the disclosed technology can operate on a specially programmed general-purpose computer, including specially created hardware, firmware, digital signal processors, or processors that operate according to programmed instructions. The terms “controller” or “processor” in this application mean microprocessors, microcomputers, ASICs, and dedicated hardware controllers, etc. Embodiments of the disclosed technology can be implemented by one or more computers (including monitoring modules) or other devices, using computer-readable data such as program modules and computer-executable instructions. Generally, program modules include routines, programs, objects, components, data structures, etc., which, when executed by a processor in a computer or other device, perform specific tasks or implement specific abstract data formats. Computer-executable instructions may be stored on computer-readable storage media such as hard disks, optical disks, removable storage media, solid-state memory, RAM, etc. As will be understood by those skilled in the art, the functions of the program modules may be combined or distributed as needed in various embodiments. Furthermore, these functions can be embodied in whole or in part in firmware or hardware equivalents such as integrated circuits or field-programmable gate arrays (FPGAs). One or more aspects of the disclosed technology can be more effectively implemented using specific data structures, such data structures are considered to be within the scope of computer-executable instructions and computer-usable data described herein.

[0031] The disclosed embodiments may, in some cases, be implemented in hardware, firmware, software, or any combination thereof. The disclosed embodiments may also be implemented as instructions carried or stored in one or more computer-readable media that can be read and executed by one or more processors. Such instructions may be referred to as computer program products. The computer-readable media described herein means any medium accessible by a computing device. For example, but not limited to, computer-readable media may include computer storage media and communication media.

[0032] Computer storage media means any medium that can be used to store computer-readable information. Examples of computer storage media include, but are not limited to, random-access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory and other memory technologies, compact disc read-only memory (CD-ROM), DVD (Digital Video Disc) and other optical disc storage devices, magnetic cassettes, magnetic tapes, magnetic disk storage devices and other magnetic storage devices, and any other volatile or non-volatile removable or non-removable media implemented by any technology. Computer storage media exclude signals themselves and temporary forms of signal transmission.

[0033] A communication medium means any medium that can be used for the communication of computer-readable information. Examples of communication mediums, though not limited to them, include coaxial cables, fiber optic cables, air, or any other medium suitable for the communication of electrical, optical, radio frequency (RF), infrared, sound, or other forms of signals.

[0034] In addition, the description of this application refers to certain features. It should be understood that the disclosures herein include all possible combinations of these particular features. Where a particular feature is disclosed in relation to a particular aspect or example, that feature may, to the extent possible, also be used in relation to other aspects and examples.

[0035] Furthermore, when this application refers to a method having two or more defined steps or processes, these defined steps or processes may be performed in any order or simultaneously, as long as the circumstances do not rule out such possibilities. Examples

[0036] The following examples are provided that are useful for understanding the technology disclosed herein. These embodiments may include one or more of the examples described below, or any combination thereof.

[0037] Embodiment 1 is a method for tuning an optical transceiver, comprising: connecting the transceiver to a test measurement device; setting the operating parameters of the transceiver to an average parameter set; acquiring a waveform from the transceiver; measuring the waveform to determine whether the transceiver is pass or fail; if the transceiver is failing, transmitting the waveform and operating parameters to a machine learning system; using the machine learning system to provide adjustment operating parameters; setting the operating parameters to the adjustment operating parameters; and repeating the acquisition process, the transmission process, the usage process, and the setting process until the transceiver passes.

[0038] Example 2 is the method of Example 1, further comprising the process of setting the temperature of the transceiver.

[0039] Example 3 is the method of Example 2, further comprising a process in which the above method is repeated multiple times at different temperatures.

[0040] Example 4 is a method of any of Examples 1 to 3, wherein the process of providing the adjustment operation parameters includes a process of subtracting the estimated parameters provided by the machine learning system from the operation parameters to find the difference, and a process of adding the difference to the operation parameters to generate new operation parameters.

[0041] Example 5 is a method of any of Examples 1 to 4, further comprising a step of completing the transceiver test if the transceiver passes.

[0042] Example 6 is a method according to any of Examples 1 to 5, wherein if the transceiver does not pass the process of repeating the process of acquiring data and the process of transmitting data, the process of repeating the process of acquiring data and the process of transmitting data is performed until a predetermined number of times is reached, and the process of testing the transceiver using another method.

[0043] Example 7 is one of the methods of Examples 1 to 6, wherein the waveform is represented as an eye diagram consisting of one-level transitions.

[0044] Example 8 is any of Examples 1 to 7, wherein the process of measuring the waveform includes a process of measuring the TDECQ (transmitter dispersion eye closure quaternary) value.

[0045] Embodiment 9 is a test and measurement device comprising a connection unit for connecting the test and measurement device to an optical transceiver, and one or more processors, wherein the one or more processors are configured to execute a program that causes the one or more processors to perform the following: first, set the operating parameters of the optical transceiver to average parameters; acquire a waveform from the optical transceiver; measure the acquired waveform; determine whether the operation of the optical transceiver is satisfactory or unsatisfactory; if the optical transceiver is unsatisfactory, send the waveform and operating parameters to a machine learning system to obtain estimated parameters; adjust the operating parameters based on the estimated parameters; and, if necessary, repeat the acquisition process, the measurement process and the adjustment process until the optical transceiver is satisfactory.

[0046] Example 10 is the apparatus of Example 9, wherein one or more processors are further configured to execute a program that causes one or more processors to set the temperature of the transceiver.

[0047] Example 11 is the apparatus of Example 10, wherein one or more processors are further configured to repeatedly execute the above program multiple times at different temperatures.

[0048] Example 12 is the apparatus of Example 10, and includes a program that causes one or more processors to adjust the operating parameters, which in turn causes one or more processors to subtract estimated parameters from the operating parameters to find the difference, and adds the difference to the average parameter to generate new operating parameters.

[0049] Example 13 is a method for training a machine learning system for determining the operating parameters of an optical transceiver, comprising the steps of: connecting the transceiver to a test measurement device; tuning the transceiver with a parameter set; capturing a waveform from the transceiver; transmitting the waveform and the parameter set to the machine learning system; and repeating the tuning, capturing, and transmitting processes until a sufficient number of samples have been collected.

[0050] Example 14 is the method of Example 13, further comprising the steps of setting the temperature of the transceiver, waiting until the temperature stabilizes, and recording the parameters.

[0051] Example 15 is the method of Example 14, further comprising the process of adding the above parameter set to a histogram of temperature parameters.

[0052] Example 16 is the method of Example 14, further comprising the process of adding the above parameter set to a histogram of temperature parameters.

[0053] Example 17 is the method of Example 14, further comprising a process of repeating the setting process, the waiting process, and the recording process until a sufficient number of samples have been collected.

[0054] Example 18 is a method of any of Examples 13 to 17, wherein the process of capturing the waveform further includes a process of generating an eye diagram overlay that includes only one level transition.

[0055] Example 19 is one of the methods from Examples 13 to 18, wherein the above parameters are changed and the above iterative process is performed so that each of the above parameters in the parameter set sweeps the range.

[0056] All functions disclosed in the specification, abstract, claims and drawings, and all steps in any method or process disclosed, may be combined in any combination, except where at least some of such functions or steps are mutually exclusive. Each of the functions disclosed in the specification, abstract, claims and drawings may be replaced by an alternative function that serves the same, equivalent, or similar purpose, unless otherwise specified.

[0057] For the sake of explanation, specific embodiments have been illustrated and described, but it should be understood that various modifications are possible without deviating from the gist and scope of the disclosed technology. Therefore, the present invention should not be limited to anything other than the appended claims. [Explanation of Symbols]

[0058] 30 Test and measurement equipment 32 Optical transceiver (DUT: Device under test) 34. Optical / Electric (O / E) Conversion Module 36 Connection part of the test and measurement device 38 processors 40. Test automation process (program) 42 Machine learning systems or neural networks 44 memory 46 User Interface

Claims

1. A method for tuning an optical transceiver, The process of connecting the optical transceiver to the test measurement device, The process of setting the operating parameters of the above optical transceiver to pre-determined initial start parameters, The process of acquiring the waveform from the above optical transceiver, The process involves measuring the above waveform to determine whether the above optical transceiver is pass or fail, If the above optical transceiver fails, the above waveform and the above operating parameters are sent to the machine learning system. The process of using the above machine learning system to provide adjustment operation parameters, The process of setting the above operating parameters to the above adjustment operating parameters, The above optical transceiver will repeat the above acquisition process, the above transmission process, the above usage process, and the above setting process until it passes the test. A tuning method for an optical transceiver equipped with [a specific feature / ability].

2. The process that provides the above adjustment operation parameters is, The process involves subtracting the estimated parameters provided by the above machine learning system from the above operating parameters to find the difference, The process of adding the above difference to the above operating parameters to generate new operating parameters. A method for tuning an optical transceiver, including the part of claim 1.

3. A test and measurement device, A connection section for enabling the test measurement device to be connected to an optical transceiver, One or more processors and Equipped with, First, the process of setting the operating parameters of the optical transceiver described above to pre-determined initial start parameters, The process of acquiring the waveform from the above optical transceiver, The process of measuring the captured waveform, A process to determine whether the operation of the optical transceiver is successful or unsuccessful based on the measurement results of the above waveform, If the above optical transceiver fails, the above waveform and the above operating parameters are sent to a machine learning system to obtain estimated parameters. A process to adjust the above operating parameters based on the above estimated parameters, The above optical transceiver is repeated as needed, including the acquisition process, the measurement process, and the adjustment process. A test and measurement apparatus configured such that one or more processors execute a program that causes one or more of the above processors to perform the above task.

4. The test measurement apparatus according to claim 3, wherein the program that causes one or more processors to perform the process of adjusting the above operating parameters includes a program that causes one or more processors to perform the process of subtracting the estimated parameters from the operating parameters to find the difference, and adding the difference to the initial start parameters to generate new operating parameters.

Images