[SOLVED] LNA simulatipon results

[GSarris]

LNA simulation results

Hello,
I designed an LNA at 1.42GHz based on the NE3210s01. My BW is 100MHz and i get the following results. Nf=0.47dB, S11=-36 at my center freq and -13db worst over the band, a gain of 17dB approximately constant over the band and S22=-6dB at my center freq and -4dB worst. I wonder if my S22 is acceptable. My LNA is destined to be the first stage of a radio telescope. please provide me with your opinion.
Thanks in advance,
George

 

[sv1437]

Even S22 should be more than 10 dB in magnitude. You can improve S22 by matching the output impedance to 50 ohms. Remaining S parameters looks great.

-sv

 

[enjunear]

Agree with sv. A 10 dB return loss means that 10% of your power is being reflected back into the source. At a 4 dB return loss, this is even more significant... you aren't getting any signal out to your load, it's all being reflected back into the LNA (and might screw up the performance, or potentially burn it up).

Add a matching section to your output, but beware that you'll have to retune the input match (especially with such a poor initial output return loss... your input NFmin point will likely move quite a bit, unless that device has a LOT of isolation, which I doubt). You'll have to bounce back and forth from tuning the input match, to tuning the output match. There is no closed-form solution for simultaneous conjugate matches on a linear transistor, so you'll have to perform the tuning iteratively (welcome to RF engineering) :wink:

 

[vfone]

If S22 return loss is -6dB but you have flat 17dB gain on your LNA, maybe is not a big issue. Problems could appear when the LNA is connected to the following stage, which might require a good return loss at the input. Generally SAW filters needs good return loss at the input, otherwise get degradation on ripple or group delay.

 

[GSarris]

i redesigned the circuit and optimized it and now i get s11 and s22 under 10db (10dB worst). Do you think it's better this way? i see that in all the projects they tend to favor S11 at the expense S22 and vice versa. i don't know what to do because i don't have any specifications. this is a thesis where i try to achieve the lowest noise possible with acceptable other characteristics. I attach my simulation results. Please see them and tell me your opinion.
Thanks in advance
George

 

[vfone]

I think now you have too much variation of NF and Gain across bandwidth (which is about 7% from central frequency). For this relative narrow bandwidth you must have less variation of these parameters.

 

[BigBoss]

For an LNA, S11 is more important than S22.In your first design S11 looks good and maybe you might adjust the output stage components to make S22 better.
But Gain and NF look pretty good.

 

[RF_Jim]

For an LNA, S11 is more important than S22.In your first design S11 looks good and maybe you might adjust the output stage components to make S22 better.
But Gain and NF look pretty good.

At this point, it would be nice to know what his S12 (feedback) value is ... there could be a reason why the output was computed initially to be what it was ... as long as whatever he was driving was close to 50 Ohms (as the simulation would be set for) he should have operation as initially characterized ...

Jim

 

[enjunear]

i redesigned the circuit and optimized it and now i get s11 and s22 under 10db (10dB worst). Do you think it's better this way? i see that in all the projects they tend to favor S11 at the expense S22 and vice versa. i don't know what to do because i don't have any specifications. this is a thesis where i try to achieve the lowest noise possible with acceptable other characteristics. I attach my simulation results. Please see them and tell me your opinion.
Thanks in advance
George

Focus on your key parameters, first. Get your NF and gain in good shape, then tweak S22, but compensate your input match as you adjust the output matching. The minimum NF point at the input almost never coincides with the minimum return loss point, so expect a less-than-perfect S11.

 

[GSarris]

The particular transistor is a bit tricky in the design. It imposes values of Γs and Γl that are on the very right of the Smith Chart and close to the x axis. I can achieve good VSWRs in both output and input by choosing a Γs of bigger magnitude, but the matching networks that can be designed doesn't have Q low enough to give me the appropriate BW. I performed a bunch of calculations in Matlab and this is the best trade-off beetween S11 and S22 that can simultaneously give the BW i need. What would you do in my shoes? Would you choose one of these 2 designs or would you go for something else?
Thanks
George

UPDATE: I think i solved it. I achieved worst S22 of -9dB and worst S11 of -12dB across my band. My proffesor was very pleased. Thank you for your help.

 

[breasted]

Re: LNA simulation results

Hello,
I designed an LNA at 1.42GHz based on the NE3210s01. My BW is 100MHz and i get the following results. Nf=0.47dB, S11=-36 at my center freq and -13db worst over the band, a gain of 17dB approximately constant over the band and S22=-6dB at my center freq and -4dB worst. I wonder if my S22 is acceptable. My LNA is destined to be the first stage of a radio telescope. please provide me with your opinion.
Thanks in advance,
George

dear GSarris，
I also design LNA for radio telescope,but my boos want me to use SIGE HBT.so,could you provide me some books about broadband lna .
thank you !