Designer’s Guide Community

‹ Previous topic | Next topic ›

Pages: 1

Negative NF for LNA in Hspice simulation and 0dB of input and output matching (Read 3511 times)

nguks

New Member

Offline

Posts: 2

Negative NF for LNA in Hspice simulation and 0dB of input and output matching
May 11^th, 2015, 9:35am

Currently I am trying to simulate the noise figure, S11 and S22 parameters for LNA at 2.4GHz in HSPICE. Below is the code I used:
Calculate noise figure:
.ac dec 50 400MEG 3g
.NOISE v(vout) vin 1

Calculate S11 and S22:
.net v(vout) vin rin=50 rout=50
.print s11(db) s11(m) s11(p)
.print s22(db) s22(m) s22(p)

However, I get negative NF after calculate from the following:
**** resistor squared noise voltages (sq v/hz)

element 0:r3
rs 9.1186f
1/f 0.
total 9.1186f
rx 5.2622k

...

**** total output noise voltage = 433.9253f sq v/hz

NF = 10log(9.1186f/433.9253f) = -16.77dB

and get 0 value for S11 and S22:
freq s11 s11 s11
db mag phase
400.00000x 0. 1.0000 180.0000
418.85142x 0. 1.0000 180.0000
...
2.41024g 0. 1.0000 180.0000

freq s22 s22 s22
db mag phase
400.00000x 0. 1.0000 180.0000
...
2.41024g 0. 1.0000 180.0000

Can I know did I do wrongly? Or why these values appear?

Thank you.

IP Logged

Geoffrey_Coram Senior Fellow Offline Posts: 1999 Massachusetts, USA	Re: Negative NF for LNA in Hspice simulation and 0dB of input and output matching Reply #1 - May 12^th, 2015, 10:44am NF = 10log(9.1186f/433.9253f) Isn't NF the ratio of the total to the input resistor noise, rather than the other way around?
Back to top	If at first you do succeed, STOP, raise your standards, and stop wasting your time. IP Logged

nguks New Member Offline Posts: 2	Re: Negative NF for LNA in Hspice simulation and 0dB of input and output matching Reply #2 - May 13^th, 2015, 11:59am I think NF is 10log(SNRin/SNRout)
Back to top	IP Logged

Geoffrey_Coram

Senior Fellow

Offline

Posts: 1999
Massachusetts, USA

Re: Negative NF for LNA in Hspice simulation and 0dB of input and output matching
Reply #3 - May 15^th, 2015, 11:46am

But SNR should be the ratio of signal power to noise power, not the "total output noise voltage"

If you have (S/N)_in / (S/N)_out, then you do end up with something that has N_out / N_in, which is what I was trying to say, though I agree I didn't get the formula right.

If at first you do succeed, STOP, raise your standards, and stop wasting your time.

IP Logged

Pages: 1

‹ Previous topic | Next topic ›