Folded cascode LNA - tunning circuit to actual LNA

EDAboard.com Forum Index -> Analog Circuit Design -> Folded cascode LNA - tunning circuit to actual LNA

Author

Message

ansonyeap

Joined: 29 Oct 2009
Posts: 7

29 Oct 2009 10:12 Folded cascode LNA

hello. I'm a student working on the folded cascode LNA designed by my senior to work on 2.14GHz. the LNA have been fabricated and the chip is measured. the S-parameters measured shown some deviation from the desired spec. I am required to do optimization on the design based on the measurement result.

But i have not experience on interpreting the measured result and don't know what is the reasons that cause the S-parameters vary from the desired based on the result. can anyone help me? I have included the measured result. thank you

Sorry, but you need login in to view this attachment

deepak242003

Joined: 24 Dec 2008
Posts: 205
Helped: 10

29 Oct 2009 12:47 Re: Folded cascode LNA

Can you elaborate what the difference your are getting...???..
I assume it might be less measured gain....
You can work on your mathcing circuitry if it is the case....I coud not predict more until i KNow the differnce.... in results

ansonyeap

Joined: 29 Oct 2009
Posts: 7

29 Oct 2009 15:03 Re: Folded cascode LNA

ok. for my post-layout simulation the S-parameters is as follow

S-parameter S11 S12 S21 S22 NF NFmin
Value(dB) -9.63 -36.65 16.95 -14.96 1.17 1.11

on 2.14GHz operating frequency.

but the measurement result have shown that the peak of S21 have shifted to 2.06GHz instead of 2.14Ghz which is the desired one. The valley of the S11 and S22 also shifted to the 2.06GHz. the S12 is the worse. it suppose to have a peak(at -dB) at 2.14GHz but in the measurement it didn't have a peak.

the main difference is that all the S-parameters response measured is shifted to the left to 2.06GHz instead of 2.14GHz.

any suggestions on what is the causes for the shifting of the operating frequency to the left instead of 2.14GHz? for the S21(dB) is less than the simulation is due to the matching problem as I know. but what is the main cause that cause the responses to shifted? thank you.

snafflekid

Joined: 09 May 2007
Posts: 120
Helped: 16
Location: USA

29 Oct 2009 21:47 Re: Folded cascode LNA

I have found (did an LNA project also) that the parasitics are going to affect the center frequency. The parasitic models will get you close but a little tweak is usually necessary. You can try adjusting the length of bond wires by moving the die to different places in the package. I assume you are using inductive degeneration. Also do you really expect the LNA to work at 1V?

Google
AdSense
Google Adsense

29 Oct 2009 21:47 Ads

ansonyeap

Joined: 29 Oct 2009
Posts: 7

30 Oct 2009 3:33 Re: Folded cascode LNA

ya. the LNA is designed for low voltage supply and is a inductive degenerated folded cascode.

You mean that I have to work on the layout to solve the problems? if yes, than how could I know that which components to modified? thank you

snafflekid

Joined: 09 May 2007
Posts: 120
Helped: 16
Location: USA

30 Oct 2009 5:11 Re: Folded cascode LNA

When you designed the LNA did you include all the parasitic elements: Bond wire inductance, bonding pad capacitance, junction capacitance?

At this frequency 2.14 GHz I assume you are using the bond wires for inductance. You cannot change much because your chip is fabricated, but you are very close to the frequency you need. You can adjust the length of the bond wires to adjust the inductance.

The S11 is sensitive to the source inductance and it is a smallish value. Sometimes you have to parallel several bond wires to tweak the source inductor.

You should be thinking about noise figure also. Adjusting the source inductor will change NF also.

If you are using gain peaking on the output, you can adjust the length of the bond wire between the drain and supply to shift around the peak of S21 frequency.

I suggest you run some simulations of various adjustments to the bond wire lengths.