Hi,


you might find this webpage useful.

http://vk1od.net/transmissionline/VSWR/VSWRMyths.htm

It seems that if you have a lossy line, the actual line "power loss" increases with the VSWR. However, according to the webpage, it may not be that much in practice. Note that "power loss" degrades the maximum achievable SNR. For a transmitter, it also reduces the maximum achievable efficiency.

Another thing to mention is that for very wide-band systems, a VSWR can actually distort the signal since different frequencies will see a different impedance.

Lastly, the conclusion of your analysis is kind of wrong. If the input impedance of the FET is treated as purely capacitive, then it is never possible to power match to the FET. Conventional matching schemes (inductive degeneration for example) deliberately add resistance to the FET input impedance in order to provide an impedance to match to. If we treat this resistance as infinite, then an ideal match to this infinite resistance will produce an infinite voltage at the FET input, and therefore, the matched design gets a larger signal voltage than the unmatched design.


regards,
Aaron

Hi,


I see that I posted that about 5 years ago, so I'm not really sure exactly what I was thinking back then. Anyway, what I can say is that the statement is more or less correct. As I pointed out in my first reply, power loss increases with VSWR, so that must be taken into account. Another thing is that having a matched impedance reduces uncertainty. For example, if I match to 50-ohm, then when I add a 50-ohm transmission line, it doesn't matter how long the line is - at the other end it will always be 50-ohm. However, if my load impedance is not 50-ohm, then the impedance seen from the other end of the line will depend on the line length.

Maybe you could say what you think is wrong with my original statement.


regards,
Aaron

I realized the power loss you are talking about is reflection (mismatched loss), not attenuation (matched loss).
I read Razavi's RF Microelectronics, he made several points about why we need impedance matching for LNA, which are kind of clear.

Conjugate impedance matching is only relevant for maximum power transfer from a real source to a real load.

An antenna is a relevant case because we want to transfer the maximum power from the antenna to the load. But the definition of the load becomes a bit ill defined when the load is the gate of a MOSFET transistor. As Aaron pointed out, there is always some resistance in the capacitive gate to match to. But, matching to the resistance in the gate would result in a fairly poor NF (3dB NF if the real part of the input impedance was strictly due to the resistive loss in the gate connection).

Achieving low NF with conjugate matching requires that a large portion of the real input impedance be contributed by negative feedback from the output to the input. It doesn't matter much if it is shunt or series feedback so long as the feedback that sets the real part of the input impedance is noiseless (i.e. inductive or capacitive).

Low NF can be achieved with poor S11 without feedback, but noiseless or noise cancelling feedback is required to achieve both a good match and low NF with a common source amplifier.