A．Parts in a Matching Network

In a practical RF or RFIC design, a passive matching network built by one or no part exists scarcely. A passive matching network built by two parts is usually available in most designs though there is a topology limitation. A passive matching network built by three parts provides infinitive choices of the parts’ values. Empirically, in order to design a wide-band matching network, more than three parts might be necessary, which scarcely happens in most circuit designs. The passive parts in reality are not ideal so that they always attenuate the signal with their resistive components. Too many parts in a matching network might bring about a serious attenuation of signal.

Usually a matching network is built by two kinds of passive parts: capacitor and inductors. We would try to use capacitors as much as possible and try to use inductors as less as possible. The reasons are:

1. The cost of capacitor is much less than that of inductor if the circuit is built by discrete parts.

2. In IC design, the area of a capacitor is much less than that of an inductor. Therefore, the cost of a capacitor is cheaper than that of an inductor.

3. In IC design, The Q value of an inductor of an inductor is much lower than that of a capacitor.

B. Impedance Matching between Power Transportation Units



Nobody would doubt the importance of the impedance matching in an RF or RFIC design. However, there is a paradox about the impedance matching: is it necessary to do impedance matching from part to part? If yes, how to do it? If no, how come?

To answer this question, we should return to the goals of the impedance matching :

1. To reach the maximum of power transportation;

2. To eliminate the phase shift in the power transportation.

If a matching network consists of more than one part, any individual part cannot perform these goals independently. All of the parts in the matching network are actually cooperated together to maximize the power transportation and the elimination of phase shift In other words, a matching network is a basic entity or a “minimum cell” in the performance of the impedance matching goals. Impedance matching between the individual parts is meaningless. It is therefore concluded that the impedance matching is necessary between the power transportation units but not between the individual parts except when the matching network consists of only one part.

Very often impedance matching is not conducted between power transportation units in the actual circuit design. For example, in a cascode LNA with CE-CE or CS-CG configuration, the individual units, CE, CB, or CS, CG portion, are the power transportation units. The impedance matching is usually neglected between CE and CB portion or between the CS and CG portion. This is due to other considerations such as noise or the simplicity of the circuit. Otherwise, it is better to have the impedance matching between CE and CB, or between CS and CG if only the power transportation is taken into account.