pcsnim techinque
and now met with the problem to choose a proper on-chip matching method.
pdk inductor itself is noiseless but the it takes much area and noise of its parasitic is troublesome.
but common gate lna show a higher noise and the gain (unlucky, input impedance of passive mixer is low) is too low, so it suffers a bad noise from the following blocks.
is there any good method to deal with this problem?
I think feedback amplifier is acceptable here. Look it in Gonzalez book (see attach).
match maybe first important in blocks that directly connect to board, such as LNA. matching methods are depending on your applications, wide-band or narrow band. If narrow band, L can be used because its tune facts and will also improve noise figure. As for wide-band, common-source with L degeneration may not be used easily, and common-gate or feedback topology is better choice.
to flamingo,
yes, you are right.
in my case , it is a narrow band lna, and I use on-chip L to match the input of Lna to 50ohm, the problem is, as you know the parasitic of on-chip L contributes too much noise...
I tried common gate configuration. theoretically speaking , it has intrinsic higher noise. it seems to be quite hard to make it as low as 4dB.
I wonder, normally how to make on-chip matching, since I notice that, for example the new version Bluetooth chip is on-chip matching, and the customer is easy to use.
Generally, the on-chip matching network is built exclusively by capacitors.
As for the input matching of CS LNA, you can try the PCSNIM (Power constrained simultaneous noise and input matching) optimization technique which is described well in the paper "CMOS Low-Noise Amplifier Design Optimization Techniques". I have tried this technique, it works ok for narrow band matching and NF<2dB meantime.