Transient Analysis Non-symmetrical Currents

[Just_Plain_Bob]

Running Transient Analysis on simple RC RL series/parallel circuits with sinusoidal voltage signal source.
Msim's calculated values for circuit current changes with total time for the sample rather than number of sampled points.

I increased number of steps to 20000 for a single cycle (1khz) and that had no effect on eliminating non-symmetrical +/- peaks in the current waveform. (eg, -5 mA, +4 mA).  It should be symmetrical about zero. (eg +/- 4.5mA)

I found that I needed a minimum time of at least 3 full cycles before final Msim currents were stable, reasonably accurate, and reasonably symmetrical. (eg 1khz stop time 0.003 minimum)

Msim apparently needs to average over several full cycles somewhat independent of number of time steps.
Maybe optimization for either speed or memory may have led to these problematic results.

In addition to the comments about number of cycles Msim needs to settle on values, the addition of inductor current seems to amplify the imbalance in circuit +/- current peaks which should be equal +/- from 0…

If there is a minimum time required to arrive at a balanced state, then there should be a way for Msim to calculate out 4 or 5 cycles and present those results as representative of the circuit even though the time parameters are set to display a single cycle. (eg 1khz stop time 0.001)

It's pretty painful to try and do zero crossing comparison between two vertically offset waveforms. I shouldn't have to determine correction values to add to the current waveform to make it symmetrical about 0.

This brings up another point. It would be nice if the Y axis could always include 0 and always present equal +/- scales on command rather than having to manually set the scales every time you zoom. I may be wrong, but toggling the Autoscale button didn't seem to prevent automatic re-scaling on every zoom.

Lastly, wish list items. You could use a phase meter and an analysis option in transient analysis that let's you choose a reference voltage or current and a second waveform for comparison. That of course would only work if Msim calculated everything out at least 4 cycles for a sinusoidal waveform to arrive at some stability and base the comparison out at that point in time, .....

Suggestions?

Bob

[_user32]

Just Plai… ,

Perhaps you could post the circuit you are working with to verify what you are seeing and we could propose a solution.  I can work with our R&D team to give you the 'real' answer, but here is my guess... I am thinking it has to do with either the time step (TMAX), Initial Conditions (IC) or the underlying SPICE3F5 simulation settings.

There are several things that may be going on here.  First of all, when the underlying SPICE engine starts in Multisim it is typically not using equivalent time steps to come up with a solution - it will generally settle on more equivalent time steps for continuous waveforms after a cycle or two.  You can adjust the time step (TMAX) very small and also possibly set the initial conditions on either the C or L side and thereby get a more accurate solution during the first cycle.   Also the default integration method is trapezoidal, which I understand does take several points in a transient condition to get an 'averaged' result during calculations, the gear method is a different solving mechanism that relies less on this past sample points for calculations.   From the menu take a look at: Simulation -> Interactive Simulation Settings.  TMAX is on the first tab and the more advanced sim settings are on the 'Analysis Options' tab - click on the 'Customize' button.   Multisim tries to pick the best settings for simulation accuracy and simulation speed, however tightening the SPICE settings will definitely improve the simulation accuracy.

As for the waveform comparison issues, are you using Virtual Instruments or the Grapher view to look at the waveforms?   Using the Virtual Instruments, you may be able to use AC coupling to get the waveforms to match (for current use a current probe + 2 channel scope for this type of comparison).

If possible, please post the circuit and the group can give better insight and suggestions.

Regards,

Patrick Noonan
Business Development Manager
National Instruments - Electronics Workbench Group
50 Market St. 1-A
S. Portland, ME 04106
Email: patrick.noonan@ni.com
Tel. (207) 892-9130
Fax. (512) 683-7754