Designer’s Guide Community

New Member

Offline

Posts: 3

RX-Chain noise figure calculation in Cadence
Oct 26^th, 2016, 5:37am

Hi all,

I am simulating a receiver chain in Cadence Spectre with the following components

PORT1(50 Ohms) --> SE LNA --> BALUN --> MIXER --> PORT2 (100 Ohms)

where SE stands for single-ended. The total Gain of the chain seems to agree quite well with my expectations, the noise figure nevertheless happens to be around 3-4 dB higher than expected if I plug in the Gain/NF values for each component in the Friis NF Equation. For the simulation I am running a QPSS combined with a QPNOISE analysis and then plotting the "Noise Figure" function under the QPNOISE results. For the noise analysis, I specified both output and input elements as "Probe", i.e. PORT2 and PORT1. The components are internally matched to 50 Ohm for the single ended, 100 Ohm for the differential ended respectively so I would discard a Noise Figure increase due to mismatching effects.

I was wondering if this 3-4 dB difference arises from the fact that the output noise sees a 100 ohm impedance while the input noise sees just a 50 Ohm one? This 1/2 factor would imply a -3 dB for the noise figure definition if I am correct?

Any advice or clarification is highly appreciated.

Thanks a lot in advance! :)

Back to top

IP Logged

dog1

Community Member

Offline

Posts: 52

Re: RX-Chain noise figure calculation in Cadence
Reply #1 - Nov 7^th, 2016, 6:08am

Hello,

It's been quite a while since I last did a RF design, so I may be wrong.

I don't think the port resistance gives you the problem. refering to the attached picture, the NF of the system is related to the ratio of Vn+In* (input resistance of the system) and R1, where Vn and In are the input referred noise contributed BY THE SYSTEM, and is intrinsic to the SYSTEM and not affected by R2 or port 2. R2 may indeed affect the (input resistance of the system), thus affecting the NF. But if your system is unilateral to a good extent (as most RF system of this kind can do), then R2 shouldn't affect (input resistance of the system) by a factor of 2.

An explanation that is more vague but is closer to the definition of NF is that when the port 2 is changed, (the gain of the signal) and (the noise from the system and the input port) scales altogether, and the output S/N stays more or less the same when the input S/N is not changed.

some ideas of where the 3dB comes from:
1. check the gain and NF (singnle-differentail?) of the balun.
2. check if the overall noise is SSB or DSB (double side band).

Hope you can find out where it is going wrong.

BR

CHEN

Back to top

inputref.jpeg

IP Logged

Ken Kundert Global Moderator Offline Posts: 2384 Silicon Valley	Re: RX-Chain noise figure calculation in Cadence Reply #2 - Nov 7^th, 2016, 4:55pm Why are you using QPSS/QPNoise? Seems like PSS/PNoise would be sufficient. -Ken
Back to top	IP Logged