Welcome. The latest radar systems employ phased-array antennas to provide a faster scan rate and more versatility in detecting multiple targets

Transcription

1 Welcome. The latest radar systems employ phased-array antennas to provide a faster scan rate and more versatility in detecting multiple targets across a wider special area. They include large numbers of transmit/receive, or T/R modules that precisely control radiated signals from each antenna element. In this paper, we will discuss the latest test methodologies and considerations for building efficient T/R module test systems from a variety of aspects. 1

2 First we will review phased-array radar systems and trends. Then, we ll go over T/R module test requirements and challenges. We will spend most of the time going over the following considerations for T/R test integration : - the latest measurement methodologies, - signal conditioning requirements and efficient signal routing, - idea for simplifying system calibration and a tool to maintain test accuracy, - synchronization of DUT control and automated tests, - and finally considerations for keeping the system uptime. 2

3 A phased array antenna is an array of antenna elements that control the relative amplitude and phase of signals from these elements to concentrate the energy in a desired direction. The attenuator and phase shifter in each T/R module quickly change their states, which enables a phased array antenna to scan the narrow beam very rapidly. Today s phased array antennas utilize 100s or even 1000s of antenna elements with T/R modules. Their energy can be directed and scanned to multiple positions in wide special areas, which makes them very flexible for use in radar, electric warfare, and communications applications. 3

4 Active electrically scanned array, or AESA antennas are the latest version of the phase array antenna. AESA can form a lot of antenna patterns. This provides versatility with wider a scan angle and multiple concentrated beams for detecting and tracking multiple targets simultaneously. Precise relative phase and amplitude control are crucial to the AESA capabilities. Shown here is a typical AESA antenna with 24 test ports and 6-bit phase and 6-bit attenuator control. It has 4096 settings per test port and is tested at 21 frequency points. That is over 2 million antenna patterns to be tested. 4

5 The chart shows three types of phased array antenna systems with their typical applications and characteristics. Type 1 radars are low power, and low cost arrays for communications, weather radars, perimeter detection, and automotive radars. The range of applications are rapidly growing. The phased arrays needed for future 5G systems will be in this category. Type 2 are medium power phased arrays used for military intelligence and surveillance which have a higher cost per element. This type of radar system is also growing rapidly as communication and situation awareness become more and more important. Type 3 are very high power per element and very expensive. They are used in applications for fire control radars, electronic warfare, and electronic countermeasures. GaN is the current semiconductor of choice for these high power applications. Applications for phased-array antenna are expanding and growing rapidly in aerospace and defense to commercial industries, with medium to low power but highly integrated form factor. 5

6 Digital beamforming in phased-array antennas is becoming popular in both military and commercial applications. It has a radar exciter, or signal generator and a receiver right behind each T/R module, and provides the most flexible beam control. This configuration minimizes power loss in the transmitter and increases receiver sensitivity. The tradeoff is that it increases the complexity and requires more components and space than analog beamforming. For these reasons, it was limited to large-scale and costly applications in the early 2000s. Today, higher integration with evolving semiconductor technologies enables more power in a smaller size, which makes the digital beamforming used in type 1 and 2, or medium to small-scale, more cost effective radar systems. The complexity of multi-receivers, signal generators, and high-power consumption with large data processing limits the size of the back-end RF and baseband sections of the radar. This becomes more challenging as the operating frequency increases. An alternative approach is a combination of analog and digital beamforming to optimize special and power requirements, as well as performance and versatility requirements. 6

7 Now, let s review the front-end T/R module and its trend. Here we see a typical T/R module. In one direction it s a pulsed power amplifier, in the other, it s a low noise amplifier. All the RF characteristics of these paths are dominated by these devices and circuit-designs around them. The unique part of a T/R module is the phase shifter and attenuation adjusters which allow each T/R module to have it s own settings. The attenuator and phase states can be re-programmed for each radar pulse. When T/R modules are tested, they are attached to a fixture with a coaxial interface or waveguide adapter for modules with built-in antenna. Then, the fixtures are de-embedded from the measurement results. 7

8 Past and current AESA assemblies use a brick construction, with T/R modules installed orthogonal to the antenna array. Newer AESAs use a planar configuration, with the antenna array on one side of the planar surface, and their corresponding T/R modules as an IC chip mounted on the other side of the planar. This topology enables smaller size, lighter weight, and lower cost, which expands the AESA applications with type 1 and 2 radars. There is a limitation to the newer configuration. Each radiating element has to be placed a half of wavelength apart in order for the beamforming to be controlled with the array of elements. Higher frequency operation requires much closer spacing of the radiating elements, which requires a much more tightly integrated T/R module chip. 8

9 Here is the summary of technological evolution over last three plus decades. The active phased array allowed lower RF power per element in the front-end. Lighter and smaller structures became alternative choices for bulky waveguide manifolds. This enabled more elements in a system, and allowed more sophisticated digital beamforming. Then with the planer structure, the whole system became smaller, lighter and cheaper, but required more components to be integrated into a smaller form factor. Now, as function blocks like front-end T/R modules, signal generators and receivers, baseband digital processors become available in IC chips and frequency range is extended, the phase array antenna systems will find applications in many industries. Then we begin to see needs of different scales and sizes, lower-cost structures, high-volume production, and maybe new performance requirements. Ultimately, front and back-end RF and digital sections will be fully-integrated and attached directly to the antenna array. It will drive full scalability in size, cost, power, frequency and bandwidth making it easier to adapt different requirements. 9

10 As the phased array antenna evolution continues, it may impact some of test methodologies in future. But in this section, we will focus on T/R modules most commonly used today and immediate future, and discuss their test requirements and challenges. 10

11 Let s start with the transmit path. The signal in the transmit path is typically pulse-modulated. It goes through an attenuator and a phase shifter then a power amplifier. In order to deliver amplitude- and phase-controlled signals to the antenna element, the input and output matches of the path, gain, and attenuation and phase offset deviations are measured at every possible setting over the operating frequency range. The attenuation step can be 0.5 or 1 db in 20 to 30 db range, and the phase offset is controlled in about 3-degree steps or less from -180 to +180 degrees. The number of combinations in this case varies from a few to several thousands. Each set of measurements are simple S-parameters, but there are a lot of amplitude and phase states to be tested. 11

12 Furthermore, the power amplifier characteristics are tested. This includes a 1 db compression point or P1dB, maximum output power at P1dB or at saturation, 2 nd and 3 rd harmonic distortions and output-referred IP3. It is quite simple to measure the gain, then find P1dB and the output power with a vector network analyzer. Notice they are all tested versus frequency. If it is tested at every 10 MHz in whole X-band, the measurements have to be repeated 401 times. Distortions are traditionally tested with signal generators with a combiner and a spectrum analyzer. This approach is simple to implement and very straightforward for fixed-input frequency, but it is time consuming for testing versus frequency. 12

13 Let s look at the receive path. The receive path has a power limiter at the front-end to protect the following lownoise amplifier from unexpected high-power signals. The phase shifter and attenuator are controlled in the same manner as the transmit path. The matches and gain are tested over the operating frequency and the various attenuator settings. They are simple measurements, but require a lot of steps. The input path is the front-end of the whole receiver chain where distortions are tested. They include 2 nd and 3 rd harmonics, input-ip2 and IP3 or IIP2 and IIP3. They are calculated with the main output power, 2 nd - and 3 rd -order IMD, and the gain. A spectrum analyzer is used for distortion measurements, and a VNA measures the gain. Testing IIP2 and IIP3 requires data from both instruments and some math. They are also tested versus frequency so it becomes even more complex. Noise figure testing is one of the most important characteristics of a receiver. In general, it is very slow with a lot of internal noise averaging. Therefore, noise figure is tested at a relatively coarse frequency resolution. This may hide true characteristics of the receiver front-end. The match of LNA input greatly affects the noise power at the output. These error sources must be carefully managed for better accuracy. 13

14 This is probably not a common in production, but T/R modules may be tested with transmitter and receiver subsystems in design characterization. The matches stay simple as S11 and S22, but transmissions become conversion gains between the input and the output at different frequencies. Amplitude measurements are relatively simple, although calibrations requires extra steps. Phase measurements need a slightly different measurement and calibration technique. Because the frequency converters in the paths, there are more opportunities for signals leaking through undesired paths. High-order mixing products are generated so there are more spurs. The mismatches in the paths also make the signal flow more complex. The screen shot above is an example of spectrum at each of the inputs, outputs and isolated ports on a simple transceiver module in transmitter mode. Harmonics and spurs on Tx and LO ports are shown in the upper windows and as you can see they cause a lot of spurs on the antenna port and leakages on the Rx port which are shown in the bottom windows. These spurs and leakages must be tested in transceiver subsystems. Additionally, wideband performance with modulated signals may be tested, such as noise-power ratio, transmitter noise density, and receiver error vector magnitude. Vector signal generators and signal analyzers may be used for testing these performances. They may not be needed in a T/R module test system, but it is better to be able to incorporate these additional capabilities 14

15 when needed. 14

16 We have just discussed the latest phased-array radar and T/R module test challenges. Now let s move onto considerations for building T/R module test systems for higher test throughput and better test quality. 15

17 To start, let s review what needs to be done to complete a T/R module test system. This is a conceptual building block for test system integration. Building a test system involves three key areas of challenges: - measurements, - switch matrices and signal conditioning, - and automation. (click) The measurement challenges are mostly addressed with tools and techniques that are available from instrument suppliers. Users need skills and knowledge to utilize them, but suppliers are typically able to assist. (click) The switch matrices and signal conditioning are unique to each test system and measurement requirement. It requires a good understanding of test specifications, RF and MW, and perhaps digital fundamentals. Suppliers for switches and signal conditioning devices will be able to identify the right products based on your requests, but you could spend great amount of effort for designing and optimizing the configuration. (click) 16

18 Automation is a huge task. Controlling T/R modules and sequencing the tests require a deep understanding of the module design, test plans, specifications, and expertize on hardware control and software integration. It also requires measurement knowledge to maintain the test quality so that the test systems perform as expected. The work associated to each area becomes larger as you go higher with the level of integration. 16

19 Next we ll cover the latest measurement techniques but first, let s review the traditional test approach. Traditional test systems are configured with many instruments, each of which was designed for a specific set of measurements. For example, a power meter is used for output power measurements and a VNA is used solely for S- parameter measurements. Separate signal generators and spectrum analyzers are used for distortions and spurious tests. A noise figure analyzer is added for the NF measurements on receive path. The test fixture is connected to different instruments through a large multiplexing switch matrices, which add large path loss and mismatches, causing drift and reduces system dynamic range. Software has to control each instrument with a slow interface and longer instrument response time. This becomes major problem for sequencing a lot of tests like T/R modules. The latest test systems often share hardware, such as receivers and signal generators, among different types of measurements. They sometime includes switching and signal conditioning for better accuracy. Then the internal software controls the hardware efficiently and provides a variety of measurement capabilities. 17

20 Here is an example of a VNA with the latest hardware architecture. This is PNA- X with 2-port configuration. It can have two internal sources with built-in harmonic filters with a signal combiner for two-tone IMD measurements. The sources are capable of pulse modulation, controlled by internal pulse generators, which is useful for testing devices in radar applications. There are signal path switches behind the reflectometer so that signals are routed to different hardware for additional measurements while maintaining S-parameter measurement accuracy. Additional hardware may be included, such as a lownoise receiver for noise figure measurements. Now, let s talk about the latest measurement methods for T/R module testing. 18

21 We ll start with gain, match, output power and compression measurements. These measurements are traditionally made with standard S-parameter measurements in a VNA as shown on the left. The compression, or P1dB, is found in swept-power S21 measurements then the output power can be found at the same output power as the P1dB. It needs many setups to cover the required frequency range which causes throughput and calibration challenges. (click) The latest VNA, like the PNA-X, offers a software option called Gain Compression Application or GCA. In a single setup, the GCA controls the stimulus frequency, power, linear gain, P1dB, and output power at P1dB versus frequency with fewer measurement points. Using the GCA, the setup and calibration are simpler and the measurement speed is significantly faster than the traditional swept-power S21 method. 19

22 Spurious tests are typically done using a spectrum analyzer with fixed-frequency input to the DUT. The input frequency is stepped to cover the frequency range. The search range becomes wider frequency and lower level making it time consuming. Repeating these measurements at different input frequencies can lead to hours of testing. The slow sweep speed is mostly due to the microwave pre-selector at the spectrum analyzer input in swept mode. Spectrum analyzers use it for measurements with highest level of accuracy. The latest spectrum analyzers also have a FFT mode for faster measurements, but it bypasses the pre-selector and loses the accuracy. (click) The PNA-X has software capability that turns the VNA into a spectrum analyzer. It enables fast spur search in wide frequency range with the stepped-fft method. It provides multi-channel spectrum analysis with superior accuracy at the DUT interface. Internal sources can be controlled at the same time so that spur tests are performed with various stimulus conditions. Moreover, it does not require a switch to change the signal path from the VNA to the spectrum analyzer. This improves the receiver s dynamic range by as wide as 10 db in high-frequency. 20

23 Let s move on to IP2 and IP3 measurements. They were tested with signal generators and a spectrum analyzer. When tested versus frequency, a user software was necessary to control all the instruments. You might use a spectrum analyzer s internal tracking generator capability, but it is still very slow. If you want to measure output IP3 at a specified DUT output power, it could be challenging. (click) Today we see a simpler way with swept-imd applications. The swept-imd controls the two-internal sources and switches to make two-tone IMD measurements. By simply setting the center frequency, tone-spacing, and the sweep range we can choose the test parameter from the pre-defined list. This is easier than setting up individual source frequencies, programming to sweep, finding peaks and calculating parameters using the traditional method. 21

24 One of the most important parameters of the receive path is noise figure, which is defined as signal-to-noise ratio degradation. The noise figure is traditionally measured with either a noise figure analyzer or a spectrum analyzer with a noise figure option using a technique called Y-factor method. The Y-factor method calculates noise figure from noise power measurements at the DUT output with hot and cold noise source states at the DUT input. This is very popular and works well when the noise source is connected at the DUT input. However, it assumes the DUT input is terminated with 50-ohm and calculates the noise figure. When configured with switch matrices in an ATE system, this is no longer true and the noise figure is greatly affected by the input mismatch. This method is also very slow, which limits the number of frequency points in practical use. (click) The noise figure application on PNA-X uses an optional low-noise receiver or standard VNA receivers that minimize the need for switching between the noise figure and other measurements. With industry-unique source-mismatch error correction, it delivers superior accuracy especially in an ATE environment. It calculates noise figure from the device gain and noise power at the DUT output with no signal at the DUT input, so it is called the cold-source method. This method enables typically 10 to 40 times faster measurement speed than traditional Y-factor method, enabling a finer frequency resolution with small 22

25 impact to the test throughput. 22

26 Classic ATE systems are configured with front-end switch matrices to connect multiple instruments to a DUT. The design is simple and it is easy to understand, but it is large, difficult to configure, less accurate and very slow. (click) The ATE systems with the latest approach introduce more software capabilities and utilize front-end receivers for multiple measurements. However, not all required measurements can be done with a single type of receiver using software. You may need a signal generator for higher output power or complex modulation, or a signal analyzer for wideband demodulation. Then you need to switch the paths to these instruments. And you would most likely need signal conditioning for some measurements and switch them in and out when changing measurement types. The best solution is to keep the front-end receivers as close as possible to the DUT interface so you can minimize the path loss and system drift. This is not always possible, but it is a good practice to optimize the path switches and maintain higher accuracy. 23

27 Let s talk about signal conditioning and routing in more detail. 24

28 We ll start with signal conditioning for high-power tests, which is often required for testing the transmit path. Here is the PNA-X block diagram that we reviewed earlier. (Click) And here is a set of external components for bi-directional high-power tests as our example. The maximum available output power from VNA internal sources is typically +10 to +20 dbm. A booster amplifier and a high-power reference coupler are used to provide higher power to the DUT. You will also need to make sure the signal level that is going into the instrument is attenuated low enough. (Click) The test port couplers can handle up to +43 dbm, but let s consider protections for up to +43 dbm output power in this example. If the coupling loss of this highpower reference coupler is 20 db, the power goes into the reference receiver can be as high as +23 dbm. To protect the reference receiver and avoid from compression, 20 db or more attenuation is recommended. (Click) The 43 dbm signal coming into the test port is attenuated by about 15 db in the coupling path then goes to the test receiver. An additional 15 db or more attenuation is required to keep the signal level below +15 dbm at the receiver 25

29 input. (Click) Lastly, the signal coming into the test port hits either the booster amplifier or internal source attenuator. This can cause unexpected reflection and errors or could cause damage. An isolator is recommended to keep the signal flow in onedirection but terminate in the other direction. 25

30 Adding appropriate signal conditionings is an essential for distortion measurements. Distortions can be caused by the test signals, DUT, and receivers. First, we want to make sure the test signals are clean. A low-pass filter is often used to reduce source harmonics. It is a simple and low-cost alternative to an high-end microwave signal generator with premium cost. The drawback is the low-pass filter limits the signal bandwidth and you may need a set of multiple filters to cover a wide-frequency range. Next, always make sure the receiver measures the signal with the linear range to avoid receivergenerated distortions. Use an attenuator if necessary. When two signals are combined for two-tone IMD measurements, they need to be well isolated so that they don t interact with each other. Isolators are recommended at the combiner inputs, but again, they are banded. Broadband attenuators are alternative if the path loss is not a concern. The latest instruments often have internal harmonic filters to maintain the signal purity for distortion measurements. PNA-X s two internal sources have two outputs, one is filtered and another is unfiltered. Consider these source port characteristics when setting up for harmonics and IMD measurements. The PNA-X s internal combiner has a good isolation between input ports and you will unlikely need an external isolation. When signals are combined externally, source isolation should not be ignored. 26

31 Even though PNA-X provides very accurate noise figure measurements with the internal low-noise receiver and source-mismatch correction, there are many cases the test setup causes large errors. Let s look at the input side of the DUT. First, if for some reason you are not able to use an impedance tuner, say due to the source match being close to 50-ohm, then consider adding 6 to 10 db of attenuation at the DUT input to improve the source match. Second, when an impedance tuner is used for the source-mismatch correction, it sets 4 to 7 different ECal states and measures the noise power and the source impedance at each state. If there is a large loss between the impedance tuner and the DUT, the source impedances collapses to closer to 50-ohm and it becomes very difficult to calculate the noise figure at 50-ohm source impedance. The path loss after the source impedance should be minimized. How about the output side of the DUT? It is common practice to use a low-noise amplifier at the output of a DUT, when the DUT has low-noise figure or small gain and does not output enough excess noise. This is still true for many DUTs if PNA-X s standard receivers are used for noise figure measurements. If the internal low-noise receiver is used, an LNA is not necessary. However it is good practice to minimize the loss between the 27

32 DUT output and the receiver so it maintains good sensitivity. In other cases, when measuring frequency converters, the receiver sometimes shows an overload warning. Try solving the problem by reducing the receiver gain setting. If that doesn t work, make sure there is no large signal leaking from the DUT and hitting the low-noise receiver. Use a low-pass or band-pass filter to block these leakages. 27

33 These signal conditioning devices often need to be switched in and out to optimize the performance of each measurement. A high-power setup for transmit path is completely different from the one for noise figure tests on the receive path. We discussed earlier that switching should be minimized between DUTs and receivers, but there are many cases you must switch them in front for better measurements, efficiency, or simplicity. If you can t avoid the switches, they must be linear, very low loss and good match. For these reasons, electromechanical switches are commonly used. There are a few different types of switches with unique capabilities. The first type of switch is a multiport switch. It has one input and typically from 2 to 6 output ports, and is used to expand the number of test ports for multiple DUTs, to route a signal generator to multiple outputs, or sharing a signal analyzer with multiple input ports. The second type of switch is a bypass switch. It is configured with a pair of oneby-two switches and used for switching a signal path from one to another. For example, it is used to switch in and out a booster amplifier and an isolator pair at the DUT input, or used in front of a receiver and switches between highsensitivity path with an LNA and high-power path with an attenuator. The third type of switch is a transfer switch. It is also configured with a pair of 28

34 one-by-two switches, but they switch to opposite states to switch between two inputs and two outputs, and reverse the signal direction. It can be configured as a single-pole, double-throw or bypass switch. It is often used as a bypass switch and for reversing the DUT connections between transmit and receive paths to simplify the test system configuration. 28

35 Here is a generalized example of switch matrices for a T/R module test system. It includes: - high-source output power with receiver protections, - source-impedance tuner and low-noise receiver for noise figure measurements, - high-dynamic range and auxiliary paths for additional signal conditioning or external instruments, - Tx and Rx reverse switches, - test port coupler reverse switches, - a source harmonic filter, - and multiport DUT switches. Most of them are switched in and out using transfer switches. Signal paths are terminated inside or outside switches to avoid unwanted signal reflections and interferences. 29

36 Once you have designed and built a test system with switch matrices, the next challenge is to design a calibration procedure that maintains the system s accuracy and minimizes user effort and errors. One approach is to integrate high-performance switches and calibration standards, such as an ECal module, a power sensor and perhaps a noise source and a thru path. This can be expanded to multiport DUT topology with multiport switches. All paths from calibration standards to the reference planes are measured and stored as user characterization files in the ECal module and in the PNA. This approach simplifies the calibration procedure and minimizes operator s errors. 30

37 Shown here is a configuration example with CalPods for measuring T/R modules in a thermal-vacuum chamber. The CalPod assemblies include multiple impedance states that are characterized over temperature range. They are left in-line during the calibration and measurements, and allow re-correcting the drift or any path characteristic changes after the initial calibration. This approach increases calibration intervals, and saves a lot of operation cost, especially for testing devices in thermalvacuum chamber. 31

38 Now let s move on to the final stage of building T/R module test systems. This includes DUT synchronization and test automation. Then last but not least, we ll cover the services to keep the system running smoothly. 32

39 This slide is all about the uniqueness of the T/R module attenuator and phase shifter and how we get the T/R module state to change at every pulse of the radar s PRF. You can see the trigger handshake interactions between the PNA-X and the DUT control FPGA. The PNA-X provides very flexible and low-latency triggering interfaces with DUT control FPGA, as well as other instruments in the system. The DUT control FPGA manages the timing of measurements, detects the measurement completions, and attenuator and phase shifter states of the T/R module over proprietary interface. 33

40 Here is a system design example with the latest methodologies and tools. It includes a PNA-X with either 2-port or 4-port configuration and switch matrices that are fully integrated with the PNA-X. It also includes spurious tests using the PNA-X SA option which can replace a stand alone spectrum analyzer. The system controller, removable data storage, and digital I/O with FPGA are based on the scalable PXIe architecture. The power supply unit can use expandable architecture with the N6700 modular power system. The oscilloscope at the top is used for timing verification. With recent PXIe expansion, the oscilloscope can be in the PXIe chassis as a Source/Measure Unit or SMU for DC voltage and current measurements. The system is very compact, scalable, and easy to customize compared to a classic ATE system. 34

41 Let s now take a look at software which is an important element of T/R module test systems. Keysight s Test Automation Platform or TAP is a Microsoft.NET-based framework, designed for speed and optimized execution. It provides a graphical user interface, which allows the user to quickly construct test plans with highly repetitive tasks. Customizable modular plug-ins are available for test steps, instrument/dut interfaces, and result storage. TAP s Connection Manager helps easily control the DUT, PNA-X and non- Keysight instruments, and a variety of measurements, including switch matrices. TAP s Command Line Interface enables integration with other manufacturing applications, and allows for various levels of customization. There are many plug-in tools for visualization and analysis as well. 35

42 For T/R module testing, we provide a custom PNA-X driver which can be scaled to execute a unique set of measurements and be synchronized to the DUT control required for phase shift and amplitude validation. You can easily enter the test and DUT control parameters. The results can be shown in test development mode or automatic test mode in graphical format. The timing analyzer shows each test duration and provides insight in where to optimize the test sequence for the maximum throughput. 36

43 You can use TAP s plug-ins to export the data into graphical or numerical format for easy analysis, test reports, and data archiving. The example here shows a graphical display on the TAP platform as you would see them on the PNA-X s display. All data can be saved with a variety of standard data formats. Keysight TAP is not just another programing language, it is a highly customizable and adoptable platform to speed up system integration for any skill level. It builds environments to archive plug-ins and re-use them to make future system integration simpler and quicker. 37

44 Reducing the cost of test is always a key requirement in almost any ATE system. The total test time, upfront capital costs, and ongoing maintenance costs have an anticipated effect on the total cost of test. On the other hand, system downtime, unless it occurs more often and for longer periods than expected, can be sometimes overlooked. As you can see in the support cycle, it is a complex, time consuming process to disassemble the system, box and ship to the manufacturer, diagnose and correct any issues, and then repeat the process in reverse in order to restore the test system. If an entire system goes down often a single instrument in the system must be removed for support, unless there is a spare available. During the down-time, there is revenue loss due to missed shipments and other commitments. The result is project goals are missed or the schedule becomes delayed. One indirect consequence is the loss of productivity of the test system operators and/or engineers maintaining the systems. Additionally, their confidence is reduced as repeated system shut-downs and increased schedules fall under increasing pressures. The higher the throughput required, the higher the impact of test system down-time. 38

45 Most companies try to mitigate system down-time with one or both of these approaches. First is to build backup stations as a way to mitigate down-time. It also improves throughput substantially if all stations are regularly utilized. However, it can create additional costs both upfront and during operation with additional engineers and floor space. The second approach is the most common. Purchasing spare instruments reduces the upfront cost from the first approach. It allows for a quick swap out when an instrument is removed from a system for repair or calibration. However, one of the main challenges is that the spare assets must be properly tracked, stored, calibrated and kept in full working condition. These are the currently accepted approaches to mitigate down-time. However, a new approach has begun to be adopted by the industry. 39

46 The new approach consists of partnering with the instrument provider to ensure maximum up-time and consistent system throughput. This enables the instrument provider to create a flexible, customized plan to meet the needs of the most demanding systems in terms of up-time support so that your support strategy can be planned upfront. For example, Keysight offers customized services within any required level of up-time and budget, including onsite support for repair, calibration, or both. This can be delivered using a mobile lab or by a dedicated onsite resident professional. For return-to-service center or factory, turnaround time can be guaranteed according to program requirements. Dedicated loaner services are an alternate approach to buying spares. The upfront capital expense is eliminated through an ongoing contract that provides the loaner instruments when they are needed. Keysight will account for and maintain enough loaners to meet the program requirements, which avoid the complexities of tracking and maintaining spares. Loaners are shipped when required, or can be kept onsite if desired. The response time and procedure is customized according to program needs and budget, ensuring that expectations are set upfront. These types of services ensure that an up-time is maximized. 40

47 Here are some reference documents for the techniques and tools discussed today. For any questions for your specific program, please discuss with your local Keysight representative and we are happy to assist you. 41

48 With that, I d like to take some questions. 42