dog1
|
Hello,
In order to answer this question we need to understand why we want to do impedance matching.
From what I heard and understand, there are two main reasons:
1. from transmission line point of view, the frequency response of transmission line and filter changes with the load (and the source) impedance. Also, when mismatch, there will be reflection and standing waves on the transmission line, which can cause problems (say, reflection into the free space through the antenna at the input of the receiver) and distributed V and I (deviation from standard circuit theory, thus making the circuit analysis invalid)
2. from circuit theory point of view, for maxim power transfer.
Note that the latter is based on circuit theory only and requires no distributed wave analysis.
when designing IC, say at the output of the LNA, if the mixer is place close to the LNA, the voltage and current in the interconnect is not much distributed over the line, and that the reflection in itself is not a problem. So reason 1 is not valid.
Additionally, reason 2 is valid only under the circumstance of drawing 1 in the picture attached, where you can have lossless matching network. However 1) on-chip inductor is lossy and costly. 2) in IC design we are very often interested in stable voltage and current, while by using matching network we are essentially creating an resonating network. Since in IC design, the load is very often purely capacitive, and the loss due to packaging and interconnect is small or non-exsiting, this often require large and almost purely inductive matching network. It is hard to build those inductors and the resonance gives large ringing on the line with pulsed input (long settling time and not suitable for digital and switching circuit). Say, in drawing (b), the input of the next stage is almost purely Cgs, thus the matching network will be almost just an L, and this LC network give a resonance and ringing.
|