Vivek Sharma
Guest
|
Hi,
I am using SpectreRF to check for the effect of aliased wideband noise in my delta-sigma modulator as it is quite hard to predict the increase in the noise level owing to sampling.
I am suprised to see that the simulated noise power is 100 times lower than what I would expect from handcalculations (even neglecting the effect of aliasing on thermal noise). I am quite sure that my calculations are correct, especially since the simulation results are way off the mark in practical terms, requiring miniscule amount of capacitance to realize a rather high level of accuracy.
I have not used the "pnoise" analysis option much, but I suppose that allowing 100 sidebands should be enough to take care of any amount of aliasing. Even using more or less sidebands does not change the result much.
Has anyone had a similar experience? Perhaps there is some special parameter that I have forgotten to set. Since I am trying to simulate kTC noise, I do not assign any PAC magnitude to any of the voltage sources in my circuit.
I have set "maxacfrequency" (PSS parameter) as required for 100 sidebands. The number of harmonics in PSS is set to 0.
If anyone can offer a suggestion, it would be great.
Thanks Vivek
|