Hello everyone,

I have a question about the Noise Figure (NF) of a direct conversion receiver that has an LNA,  followed by an  I/Q mixer. Each of the I and Q channels then goes through baseband channel select gain/filter chain followed by a sigma-delta ADC. Now due to some architectural reasons, the gain of the LNA/mixer front-end is quite low, so the baseband noise makes a significant difference to the overall receiver NF.

My question is how can the NF of the two baseband chains be correctly taken into account into the overall receiver NF given that we have an I/Q mixer here? I keep getting confused by the difference between double-sideband (DSB) NF and single-sideband (SSB) NF. What is the appropriate signal bandwidth to use for the baseband compared to the original signal bandwidth before I/Q downconversion?

What I am doing right now is to find the input NF of one of the baseband stages, and then taking 3 dB off in the level plan to account for the DSB vs. SSB NF. But this means that I somehow know that the noise in the I and Q paths is uncorrelated - but at least some part of it may be correlated due to common supply and biasing etc.

What is the recommended way to measure NF for a direct conversion I/Q receiver (any references)? And how can it be done in Cadence?

Thanks a lot!
mohta