Andrew Beckett
Senior Fellow
    
Offline

Life, don't talk to me about Life...
Posts: 1742
Bracknell, UK
|
Goodness me, you ARE picky. Since yunjung referred to the manual for SPECTRE, you might reasonably guess that he was talking about pnoise from spectre. Or is that completely unreasonable of me?
I've been away from the boards for a number of weeks, and am having to wade through all your rather harsh replies, particularly on this board. I need to be patient. If I see "EDA Tool Play" again, or that Cadence Spectre users know nothing about measurement, I shall probably scream :'(
Anyway, pancho_hideboo was quite correct in pointing out that if the PSS solution was found with harmonic balance, you must have computed sufficient harmonics - you cannot include noise contributions from harmonics you did not compute in the steady-state analysis. With shooting that is however possible, because there will be sufficient time domain data (usually at least 5 points per period of maxacfreq, which defaults to 4 times the highest harmonic requested in the PSS) to be able to find higher harmonics - the general rule is that maxsidebands shouldn't take you above the maxacfreq, although in practice you can often go up to twice the maxacfreq because you're still within Nyquist, and assuming that the noise contributions are small at those frequencies anyway, it's not normally critical.
So maxsidebands is what controls how many sidebands are contributing noise in the output, and in both cases your PSS solution needs to have enough information for it to be able to compute the transfer functions from those sidebands. Exactly what is enough for maxharms depends on whether you're using HB or Shooting.
Regards,
Andrew.
|