LNA Noise Matching design

[megaknaller]

Hello guys,

I am designing an LNA and would like to match it for a nearly optimum noise condition. I´m running in trouble trying to design the input matching network, which should convert a 50 ohm source to my desired source impedance (let´s say 10 + j*10 Ohms).

So far the matching tools that I´ve been using (including the ADS Impedance matching wizard) basically match for maximum power transfer (given the source and load impedance), so the load impedance is transformed to the complex conjugate of the source impedance. What I´ve doing so far is using the same approach but in the opposite direction, being my "load" my source and entering the complex conjugate of my desired impedance as "source" so the wizard transforms the load (original source) to my desired impedance (for example, if I need the 10+j10 input impedance, i enter as "load" the 50 Ohms and as "source" 10-j10)

Somehow when I use the LC matching section, the calculated values are correct in that I get the desired input impedance, but when I use a transmission line/stub model (which I have to use in reality) the obtained values simply does not provide the required input impedance and it´s really driving me crazy because I´m afraid I´m not doing something correctly and I´ve already spent 2 or 3 days trying to figure it out.

If any of you would have a suggestion or method to obtain the required impedance transformation would be really appreciated. I can upload pictures of a basic schematic if you need it.

Thank you and best regards. Hope you can help me to sort this finally out.

MK

 

[BigBoss]

If the frequency of operation is not too hig, you can use L-C circuits to match.It's easy to do this in ADS using with Smith Chart utility..
Define your Zopt and take the complex conjugate this impedance and the other will simply be 50 Ohm and bingo..!!

 

[BradtheRad]

but when I use a transmission line/stub model (which I have to use in reality)

On the chance this is of interest...

Transmission lines are supported in Falstad's animated interactive simulator.

Here is a screenshot. The red and green bars move from left to right, uniformly...
Except for the bottom left segment, where the bars jump unevenly.



The menu has several schematics of transmission lines, showing what problems may occur, and solutions.

Free to download and use at:

https://www.falstad.com/circuit

Java needs to be installed on your computer.

 

[megaknaller]

Thanks guys for the interesting replies

@BigBoss: I have to use transmission lines because the frequency is quite high (180 GHz) and the lumped components are much more lossy. I've done it so far as you said but the smith chart utility wizard can only generate lumped elements matching structures. I haven't found any option to generate/select a transmission-line/stub model instead. Do you know if that's possible?

@BradtheRad: I'll give a look at the tool and see if I can work out something. Thanks!

Any other tool/method suggestion is highly welcome as well!

 

[volker@muehlhaus]

Have a look here:
https://www.rfcafe.com/references/electrical/lumped-distributed-components.htm

 

[BigBoss]

Thanks guys for the interesting replies

@BigBoss: I have to use transmission lines because the frequency is quite high (180 GHz) and the lumped components are much more lossy. I've done it so far as you said but the smith chart utility wizard can only generate lumped elements matching structures. I haven't found any option to generate/select a transmission-line/stub model instead. Do you know if that's possible?

No,you also can use Smith Chart utility to match the impedances with transmission lines components.

 

[ferdem]

My suggestion does not address your problem exactly but it may help you.
You may go step by step, in LC matching depending on the circuit type used(series cap first etc.), you place first component and adjust its value until the impedance/admittance intercepts the X circle. Then, you place the second component and adjust its value until the point reaches to the final point, matched condition. You either adjust component values or size of the distributed elements. In this manner, you can end up with a distributed matching network although you do not track lumped/distributed conversion. The necessity to study on lumped/distributed conversion is still exist.