When "sweeptype" is "absolute", noise_input frequencies of each sideband are calculated like following in Cadence Spectre's Pnoise.
Quote:noise_input_freq = absolute_noise_output_freq +/- n*fund , ( n=0, +/-1, +/-2, +/-3, ...., +/-maxsideband )
Max sideband frequency is "maxsideband*fund" in this case.
There is no confusion about this.
However when "sweeptype" is "relative", I have confusions about sideband's treatment in calculation of total noise
in Cadence Spectre's Pnoise.
For example, consider following netlist.
Quote:pss ( Vout 0 ) pss fund=1.0G harms=8 errpreset=moderate annotate=status
pnoise ( Vout 0 ) pnoise sweeptype=relative relharmnum=1
+ start=1 stop=10M dec=10 maxsideband=3 annotate=status
+ saveallsidebands=yes
Here noise_input frequencies of each sideband are calculated like following.
Quote:noise_input_freq = (fund+relative_noise_output_freq) +/- n*fund , ( n=0, +/-1, +/-2, +/-3 )
(1) noise_input_freq(sideband= 0) = +1GHz + relative_noise_output_freq ; This is a feedthrough noise. This contributes to USB.
(2) noise_input_freq(sideband=-1) = relative_noise_output_freq ; Up conversion noise from around DC. This contributes to USB.
(3) noise_input_freq(sideband=+1) = +2GHz + relative_noise_output_freq ; Down conversion noise from around 2*fund. This contributes to USB.
(4) noise_input_freq(sideband=-2) = -1GHz + relative_noise_output_freq ; This is a feedthrough noise. This contributes to LSB.(5) noise_input_freq(sideband=+2) = +3GHz + relative_noise_output_freq ; Down conversion noise from around 3*fund. This contributes to USB.
(6) noise_input_freq(sideband=-3) = -2GHz + relative_noise_output_freq ; Down conversion noise from around 2*fund. This contributes to LSB.(7) noise_input_freq(sideband=+3) = +4GHz + relative_noise_output_freq ; Down conversion noise from around 4*fund. This contributes to USB.
Here my questions are :
(a) What is an actual max sideband frequency in this case ?
Is it "(maxsideband+1)*fund=4*fund=4GHz" ?
(b) Numbers of evaluated noise_input frequencies of sideband are asymmetrical about DC.
For example, "-2GHz, -1GHz, 0Hz, +1GHz, +2GHz, +3GHz, +4GHz" in the above case.
Noise calculated by Cadence Spectre is a Single Sided PSD regardless of "freqaxis expression".
For example, noise voltage spectrum density of open circuit Resistor is "4*k*T*R [Vrms
2/Hz]" not "2*k*T*R [Vrms
2/Hz]" regardless of "freqaxis expression".
Actually Cadence Spectre's Pnoise don't use
(4) and (6) for calculation of total noise ?