低噪声放大器中共源结构与共栅结构对噪声的影响

请教各位高手，低噪声放大器如果采用共栅结构，在噪声方面与共源结构想比有何优劣，Thomas Lee书上说共栅在信号通路上存在沟道电阻，引起热噪声，但是在拉扎维书里计算MOS管噪声模型时，用的沟道热噪声，没有说明不同结构产生的不一样，也就是默认共源的也是存在这样的沟道热电阻，百思不得解，望高手解惑

求指导

求教自顶

我再自顶

这个还真不知道呢

noise source is one aspect and its transfer function is another.
Taking simple common source and common gate, the gm of common gate if fixed for impedance matching, and thus best achievable NF=3dB. While for common source, you have freedom to choose larger gm and thus better than 3dB NF could be obtained. Again, this only applies to basic structure. If you combine common source and common gate (like noise canceling), you can also obtain better than 3dB NF.

谢谢，从这个角度看，是明白了。小弟还有另一不明，我们采用共源共栅来替代单独的共源，我考虑的是2个主要原因，第一，共源的输出与输入隔离度差，共源共栅好很多，第二，共源共栅减小了共源管的增益，从而减小栅漏电容密勒效应，等效至输入端后，展宽工作频带。不知道这样理解对不对，请高手再次指教

I agree, and also higher gain

好像增益没有提高很多吧，输出端都是要匹配至50欧姆的，也就是共源增益是gm*(r0//50)，共源共栅是gm*(gm*ro1*r02//50)。这两种情况，gm后者要小一些，阻抗后者稍大一些，所以总的增益基本差不多吧。实际仿真结果基本也是这样，不知道对不对

I thought you are designing on-chip LNA

交流阻抗50 Ohm?

是在片的，也有COB的，做的感觉都是差不多的。现在总结设计的过程，感觉还是有很多问题。如果不是在片的，是不是级联时LNA后面的buffer阻抗较大，或者后级混频器的输入阻抗也较高，从而共源共栅增益高呢?

单级的LNA，一个独立的IP，输出匹配至50欧姆的

You need to make the difference between microwave IC and RFIC (nowadays, more and more microwave ICs become RFICs). If you design individual blocks (LNA,mixer,etc.), and then connect them on PCB, for sure you need impedance matching. However, if you are designing RFIC (LNA,mixer,etc. are put together on the single chip, as most SOC does), impedance matching is not necessary, and actually it just makes design more difficult.

明白了，非常感谢

说得很好，问的也很好

领教了:victory:

Have you ever simulated a common gate LNA with fixed gm?The best NF is 3dB?Books say so ,but simulation results prove itwrong.Actully, simulation results show better NF and NFmin can be less than 1dB.

tell me how you can get better than 3dB NF for simple CG LNA with input impedance matched. Unless for old large channel length, gamma is smaller than 1. BTW, I have never designed 'simple' CG LNA, since its theoretical limit is already bad.

I doubt the minimum NF of CG.So I design a simple CG LNA using 0.18um BiCMOS.I found NF is less than 3dB and NFmin was better than 1dB.

check impedance matching,what's your gm, what's connected to gate

犀利啊