i guess the auto bias at Stage 1 -OR- that specific LM324's Spice model - is what does not get recognized here correctly (Shit happens)
Update 2019.06.03 :: LM324's Spice model has an input zero mismatch of +17µV -- this however does not represent an AC analysis error !!! in static DC bias --AND-- in DC amplifier configuration . . .
. . . there was an OpAmp Spice model that failed AC configuration ((a capacitive coupling)) analysis ... and it may be the same LM324 ...
... also the frequency plot can be achieved by multi-passing a transient analysis with stepped frequencies and integrating the DC offsets and by these integrating the AC amplitudes for the voltage points in interest (gonna be time consuming but doable) . . . if you want to waste a day or two for it you can further track the slopes in transient analysis and estimate by these a rough phase shifts (further by inserting a marker signal -- such as -- spike or missng or inverted period/cycle the greater than 2πN offsets can be determined) -- the fast part example ::
. . . and still not coz higher frequencies take time to integrate
it might be considered (by tiling 0-ohm impedance AC sources in series) -- to reach measure point by a lower fixed frequency and then swapping to SAMPLED ONE --!!!-- such causes a "frequency jump" that causes a bias disbalance and even when done in phase at zero-crossing it BANGS the offsets off their values - may take long to settle again . . . thus
it could occur that if we make our own frequency source that is stepped to higger smootly by ??
(-π to π) + 1 "sine" at desired time the BANG won't occur (never tested) . . . ← well in last case it's still possible to automate the entire process . . .
( . . . only it must be in prior be determined the smooth transition times for frequencies for the speciffic circuit , etc. . . . )
[Eop]
No comments:
Post a Comment