电气专业(外文翻译)--使用agilent对快速射频信号发生转换器的生产改良(译文-英文)—-毕业论文设计.doc

Agilent Improving Throughputwith Fast RF Signal Generator Switching

Introduction
Today’s manufacturers of wireless components, devices, and systems face significant pressure to increase throughput and lower the cost of test. Improving the speed of automated test equipment (ATE) can be an important factor in reducing overall test times. This note describes techniques for optimizing RFsignal generators within ATEs to reduce test times and improve throughput. Although these concepts can be applied to any signal generator, the examples shown in this note apply to the Agilent N5181A MXG analog and N5182A MXG vector signal generators (250 kHz to 6 GHz). When equipped with the fast switching capability (Option UNZ), these signal generators can switch frequency, amplitude, or waveform in less than 1 millisecond in most cases.
Motivation for faster tests
More and more functionality is integrated into wireless systems, requiring more tests with more setups under more conditions. Connectivity methods are expanding to include not just voice and data throughGSM and CDMA, but a variety of data connections such as RFID, Bluetooth®,WiMAX, and UWB. New broadcast functions are also being supported, such as FM radio, Mobile TV, and Assisted GPS. These modes require testing and verification, over multiple channels, and different power levels, and with realistic waveforms. This can make the efforts to reduce test time and test cost even more challenging.
Understanding the Role of the Signal Generator
To optimize a signal generator, it is important to understand its role in the ATE system and how it is programmed in relation to other elements in the system.
Throughput versus switching speed
It represents a typical ATE system and the steps needed to make a typical measurement. Overall throughput is normally stated per measurement or per device, and is determined by many factors besides the signal generator settling shown in step 5. However, in highly repetitive test sequences, the signal generator may be stepped to hundreds or thousands of points becoming a significant factor in overall test time. In these cases, it is critical not only to choose a source with fast switching capabilities, but also to implement test programs that take advantage of this speed.
Some quick caculations can help determine the impact of source switching time on system throughput. For a given measurement, the total source switching time can be estimated by multiplying the point switching time by the number of source settings. Many measurements may require changes in frequency, amplitude, and waveform settings, and each type of change may have a different switching time specification.
Switching speed specifications
There are a variety of techniques for measuring and specifying switching speed. Important considerations are typically step size constraints; limitations on the size of the step, and the settling window; that is how close to the desired value the source is before it is considered finished switching.
It is important to consider the times for frequency, amplitude, and waveform switching, depending on the application. Agilent MXG frequency and amplitude settling times for LIST sweep and SCPI modes are shown in Table 1.
Optimizing nested loops
A test sequence may require that the DUT be tested with different frequency, amplitude, and waveform conditions. For example, a GSM/GPRS/EGPRS device calibration may require 10 channels in 4 bands (40 frequencies), 8 power levels, and 2 waveforms, for a total of 640 stimulus conditions. Applications like this require nested loops in the program, as represented in Figure 2. Planning how to nest the loops may have an important impact on throughput.
The switching interval of the inner loop, loop 3 in the diagram, is exercised most frequently, since the number is multiplied by the number of repetitions in loop 2 and then multiplied again by the number of repetitions in loop 1. For optimal time, the fastest switching characteristic should be placed in loop 3, and the slowest in loop 1. Note that LIST sweep mode, discussed later, allows up to 1601 points with arbitrary frequency, amplitude, and waveform settings, and fast switching less than 1 ms per point. Multiple loops can be embedded in a single LIST.
Traditionally for signal generators, waveform switching has been the slowest, and has therefore been in loop 1. When switching between waveforms in the internal baseband generator, the Agilent N5182A vector signal generator wave- form switching times are very close to the plitude and frequency switching times. This may enable the programmer to switch waveforms to loop 2 or loop 3, while maintaining or improving throughput, depending on the application
Downloading waveforms
When using the N5182A vector signal generator, waveforms can be stored within the internal 64 Megasample (MS) baseband generator or 100 MS of non-volatile memory. Downloading the waveforms from the computer or from the non-volatile memory can take seconds or even minutes; a typical transfer rate over LAN into the baseband generator is 270 kilosamples per second. This can therefore become a significant part of throughput for some applications. Waveform loading should therefore be done as infrequently as possible. This may be difficult when using very large waveforms such as video waveforms used in Satellite Digital Multimedia Broadcast (S-DMB).
Source switching speed background
RF signal generators have advanced to include many modern digital features. But switching characteristics and limitations are often inherent in the RF architecture. A typical signal generator block diagram is shown in Figure 3.
Whenever the signal generator is set to a new frequency, the frequency synthesis loop is used to phase lock the voltage controlled oscillator (VCO). Whenever a new power level is set, the automatic level control (ALC) loop is used to set the power versus a precise reference. The switching behavior depends on step size and is subject to inherent bandwidth limitations and settling times.
For digital waveforms the I/Q data must also be computed and downloaded into the playback memory. The sample rate and digital-to-analog converter (DAC) values must be computed and stored with the file for accurate playback. The N5182A vector signal generator has 64 Megasamples of playback memory.
Computer I/O considerations
The choice of computer I/O can also impact throughput when using SCPI programming. GPIB (IEEE-488-2) has the shortest latency; commands are transferred very quickly, up to 500 KBytes/sec for a 1-meter cable. USB and LAN have longer transfer periods, and a typical command may take up to 1.5 ms to transfer. However, USB and LAN provide faster data throughput and offer better performance for larger data transfers such as when downloading waveforms or LIST sweep parameters. GPIB offers better speed performance when program- ming every point via SCPI, and faster triggering for sweep mode using group execute trigger (GET). See the section, “Choosing the Best Triggering Method”
for more information on triggering the Agilent MXG signal analyzers over the bus.
Waveform sequencing
The signal generator’s on-board playback memory can be loaded with multiple waveform types. Individual playback segments can be repeated and sequenced in predefined patterns and the user can send SCPI commands to switch to different locations in the playback memory. An example sequence is shown in Figure 7.
Waveform sequencing offers advantages of very fast waveform switching and continuous transitions from one waveform to another. However, waveform sequencing requires that the frequency, amplitude, and sample rate be identical in all segments. ATE applications often require switching signals with different sample rates as well as frequency and amplitude settings. In this case, sweep mode in the Agilent MXG may be preferred. Sweep mode enables the source to jump to any pre-configured frequency, amplitude and waveform state, including different sample rates. A brief description of the Agilent MXG sweep modes is provided in the section, “Choosing the Signal Generator Control Method”.

使用Agilent对快速射频信号发生转换器的生产改良
介绍
今天的无线元件、装置和系统的制造商正在面对具体的来自增加生产而又要降低测试的费用的压力。在全面减少测试的时代, 改良自动化的测试设备 (ATE) 的速度可能是一个重要的因素。这个注解描述啦,为了优化射频信号,而使用自动化的测试设备去减少测试时间而改善生产的技术。虽然这些观念能被适用于任何的信号产生器, 但是,在这次的注解中被显示的例子适用于 Agilent N5181 A MXG 模拟器和 N5182 A MXG 矢量信号产生器(250 仟赫至 6个十亿赫兹)。当使用快速转变能力的设备时,这些信号发生器在大部份的情形下能在少于1毫秒中转变频率,幅度或波形。
对于使用快速测试的目的
越来越多的功能被整合在无线系统内,而这样就使在更多的条件下,要求更多的设备并且需要更多的测试。 通过GSM和CDMA,这种连接方法正在扩展到不只包括是声音和数据,还有多种数据连接,像是RFID,Bluetooth,WiMAX和UWB。 新的广播功能也正在被支持,像是FM收音机,移动电视,还有协助全球定位的测量站。在多个通道和不同的能量级上,这些模式都需要测试和和现实的波形进行确认。这能尽力减少测试时间,而且测试花费更低廉。
了解信号发生器的角色
为了要将一个信号发生器最优化,了解它在自动化的测试设备系统的角色和它是如何与系统其他部分联系很重要。
对变换速度的生产能力
表现一典型的自动化的测试设备 (ATE)系统和需要制造一个典型的测量步骤。整个的生产通常是通过每一测量或者每一装置表现的,而且除了包括在第 5 步骤显示的信号发生器以外还有很多的因素决定。然而，在高度地重复的测试序列中，信号发生器可能被应用成百上千次,这就使得信号发生器在整个的测试时间内变成一个具体的因素。在这些条件下,它是非常重要的,不只是因为有转变能力的来源,而且也是利用这种速度实现测试任务的工具。
一些快速的计算能决定在系统生产中转变时间来源的影响。对于一个给定的量，转变时间的总计来源能通过转变来源的种类来估计测量的时间。许多测量可能在频率、幅度和波形方面需要改变设定, 而且每种类型的变化可能有不同的转变时间叙述。

变换速度的叙述
有多种测量和叙述转变速度的技术。重要参考量典型地是步骤类型的限制;在步骤的大小方面的限制,和解决方案;在思考完成变换之前,最要考虑的是源的类型。 在转变的频率、幅度和波形种考虑时间很重要,要看使用的范围。为目录清除和 SCPI 模态设定时间, Agilent MXG 频率和幅度展示在表 1中。
佳化套入的环
测试序列可能需要 DUT 与不同的频率、幅度和波形一起测试。举例说,GSM/GPRS /EGPRS 装置校准可能需要4条基带(40 频率),8个能量级和 2个波形中需要 10个通道,即总数为640种的刺激条件。像这样需要在程序中嵌入的步骤所适用的范围将在如图 2中展现。如何在程序中嵌入步骤可能对生产产生重要的影响。
内部环的转变间隔,在图表中的环 3,时常被应用,因为数字是在环 2中以重复的数字相乘然后与环 1的数字再乘。最佳的时间,最快速的转变特性应该被放在内环 3中和最慢的应该被放在环 1中。注意：目录清除模态,稍后讨论,允许直到 1601 点为止以任意的频率幅度和波形设定,而且快速地转变得每点比1 ms少。多个环能在一个信号中应用。
传统来说对于信号发生器,波形的转变已经是最慢的,而因此应该被放在环 1中。当内部的基带发生器在波形之间转变的时候,Agilent N5182 A 矢量信号发生波转变时间与需求的幅度和频率非常接近。当维持或改良生产的时候，这可能使程序员能够在适用范围内,对环 2和环 3转变波形。
下载波形
当使用Agilent N5182 A 矢量信号发生波的时候,波形能在内部的64个MS基带发生器或固定存储的100个MS里面被储存。从计算机或固定存储中下载波形能使次数甚至时间一样;通过基带发生器的典型频率是每秒270,000个样点,因此,在一些应用中,这就变成具体的生产部分。因此,波形载入应该尽可能少地做。当使用大型波时,例如,在卫星数字多媒体广播中应用的视频波,这就十分困难。
转变速度的来源背景
射频信号发生器已经向包括许多现代的数字特征进步。但是,转变特性和限制时常嵌入射频领域中。一个典型的信号产生器区段图表在图 3中显示。
每当信号发生器设定新的频率时，频率综合环将用来锁象电压控制振荡器(VCO)。每当一个新的能量级被设定时,自动的水平控制 (ALC) 环用来比较叁考设定的能量级。依赖步骤大小和受制于固有的带宽限制的转变行即将解决。
对于数字波形,这个I/Q 数据也一定要被量化而下载到存储中。采样点将在数模变换器中数字化并存入存储中。这个Agilent N5182 A 矢量信号发生波有64个MS。
计算机输入／输出参考量
当使用SCPI程序时,计算机输入／输出的选择也将影响生产。GPIB(IEEE-488-2) 是有非常大的潜力的; 指令的转移业非常快,1公尺的电缆达500 KB/S。 USB和LAN有非常长的传输时间,一个典型的指令可能需要1.5ms转移。 然而， USB和LAN为较大的数据提供较快速的数据传输而且,当下载波形或者修改叁数时,提供较好传输性能。当程序在逐点执行SCPI和执行程序时快速的清除时,GPIB提供啦非常好的运行结果。看这部分区域 “选择最好的启动方式”,通过母线启动Agilent MXG信号分析的更多信息。
波形序列
波形序列提供从一个波形到另外一个波形非常快速转变和连续的输出。然而，波形序列需要频率、幅度和采样频率在所有的片段中是一样的。ATE的应用范围经常需要不同的频率,幅度和采样频率在波形变换中。在这种情况下,在Agilent MXG的清除模式将被优先应用。 清除模式将能够跳到任何一个需要的不同的频率,幅度和采样频率的状态下。