Originally, pulsed-RF test systems were needed to test devices opearting at power levels higher than those which could be continuously tolerated without destruction of the devices, such as occur in radar systems. Subsequently, pulsed measurements were used simply to measure device parameters at anticipated signal levels to assist in the design of circuits for pulsed operation. More recently, pulsed measurements have been used to characterise devices to assist in devising and subsequently fitting models for CAD. The ability of pulsed-bias measurements to obtain drain characteristics isothermally, and/or to obtain readings beyond device continuous Safe Operating Area (SOA), has proved to be valuable. This is especially the case with III-V semiconductors whose thermal effects are more pronounced and less theoretically well understood than those of Silicon, and which are preferably run at high current densities. Pulsed-bias measurements have resolved the wide discrepancy between GaAs device characteristics such as output conductance, as measured by high frequency, AC methods, and as extracted from traditional "curve-tracer" style families of drain-voltage and current curves that suffer from thermal droop. They also can proved data from which parameters of thermal models are straghtforwardly extracted. Different systems appearing in the literature have reflected the growing knowledge and continuing uncertainty of just what will be required of, and what system design will best provide, a pulsed-bias, pulsed-RF setup intended for device characterisation applications.

Much device characterisation today uses (DC) bias data and S-parameters alone. A criticism of this sheme is that there is no large-signal, high-frequency information available, to which to fit the model. This renders empirical models incapable of properly simulating this behaviour, and in particular, prediction of distortions such as amplifier third-order intercept may be grossly inaccurate. It is possible to extract parameters of a physically based model, using only DC and small-singal (S-parameter) data, and achieve simulations with some success. However, it would seem that some measurements spanning the timescale occuppied by device time constants must be required for complete model.

Richard: 前面做Ku的项目，对于脉冲模式工作的PA和测试有了一些很直观的认识。尽管大多数器件或者PA都是工作在CW模式，但是它们在Pulse模式下工作的特性还是可以揭示很多问题。从前没有做过模型，只是知道需要用到脉冲测试设备，现在知道用来干什么了。在成都的时候，还跟蛋蛋讨论过测试脉冲小信号参数的办法，也就是所谓“Pulse-bias and Pulse-S-parameters Test”。这个其实并不难，只是系统搭建起来比较复杂，各个单元的时序需要仔细设计，大概看了一些文章，已经有很多人实现了这种方案的测试。

***

我最最讨厌的中国女人，有两个：江青和柴玲。现在可以再加一个了：王千源。她们有很多共同点：首先是极其虚伪，然后是狼子野心，再而她们都醉心于政治却又都愚蠢至极。很不幸的是，她们都是山东人--当然这完全不是我讨厌她们的理由，只是师兄说道：真他妈给我们山东人丢脸，呵呵。

***

近期工作安排，比较务虚。

1. 文章+文章+文章...

2. 专利+专利

3. 某个申请参与

4. 解剖