Ken Kundert
|
Jess, 1. I don't see any source for ripple either, though if it exists it should be easily visable in the supply voltage.
2. Harmonic balance results when viewed in the time domain often exhibit substantial ripple due to missing harmonics when the signal is abruptly discontinuous. However, it would not create ripple on an otherwise constant-valued signal.
The salesman was correct when he said the higher noise floor was due to the variable time step. In a sense, using a variable time step trades a lower total error for an error with a spectrum that is broadband. Thus it shows up as a noise floor. If the timesteps were uniform, the error would still exist, but it is somewhat concealed because it falls at the same frequencies where the signal already exists. In otherwords, with uniform time steps, the integration method appears to be LTI (linear, time invarient), but with nonuniform time steps it appars LTV (linear, time varying). An LTI system cannot move energy from one frequency to another, so the error is confined to fall at the same frequency as the signal. However, an LTV system can redistribute energy to other frequencies, and does, creating what appears as a noise floor.
-Ken
|