improve iip3

I have read several papers which say that the iip3 of the LNA can be improve by optimizing the input and output load. Like the figure below, by tuning the Z1(0), Z1(2ω),Z2(0),Z2(2ω)(ω is the operational frequence, Z1(0) refers to the load near the DC point, Z(2ω) refers to the load at the frequency 2ω), intermodulation distortion can be cancelled.
The problem is these papers did not give a detailed match network topology. Do someone have any advice on the match network which can set the Z(0) and Z(2ω) seperately while not impacting the match condition at the operational frequency?

Hi,

I think your problem can be solved by studying so-called diplexer circuits. They can be designed to show the needed low or high impedances at specific frequencies. See this URL for a useful discussion on diplexers:

http://www.qrp.pops.net/dip2.htm Be sure to go fully through on the text.

Now maybe the best link on your specific questions are here:

http://www.suptech.com/pdf/HTS%20Diplexer_%20LNA.pdf
http://www.dl5neg.de/9cm_lna/paper%2...9cm%20lna.html


rgds
unkarc

Well this is a great example of "Trade Offs". The IP3 is directly
related to the P1dB point IP3~ P1dB + 8dB. Now if you optimize
for a high P1db/IP3 your NF and Gain might get worse!

If it's realistic you can use a 90° hybrid coupler and put 2-LNA's
in Parallel and this will bring you P1dB up by ~ 3dB!

infact my LNA is designed in the form of three stages, and the third stage limits the total output 1dB compression point. So the increase of NF is not much concerned. A gain of 8dB for the third stage is possible satisfying. The structure with two amps may be too complex and wastes IC area.

http://www.digikey.com/Web%20Export/...WP_201087A.pdf

Maybe this article is useful.

You may look what methods others used to improve IP3:
2xbf998