For my simulation the first time I run I get a plot with an expected response. If I change nothing and hit run again I get a different response as if spice has calculated a new operating point to start with. If I close the plot window and hit run I will get the expected response plotted again. It’s almost like it is saving the final operating point from each run when the plot window is open … is that a new feature that I missed? Am I missing something here?
and on the second run (no change in circuit parameters)
you can see that vout is not 0v at t0 on the second run even though the ic=0 is set on C4 (vout)
isolated power supply with center ref.qsch (10.0 KB)
Perhaps can consider adding UIC by the end of .tran
1 Like
I am not with my laptop. If this can be replicated consistancely from your example, possibly you should send this schematic to Mike for review. Just running a simlution two times without changing anything should not yield different results.
1 Like
@jay314159265 I tested your .qsch multiple times and made sure that the .qraw file does not exist before opening your schematic. It consistently returned identical results. I conducted this test in both the 2-Feb and 4-Feb versions (most update).
1 Like
Thanks for looking at this issue. I re-booted my laptop and the problem went away (which was a bit of a surprise resolution to this issue). Now I’m wondering how this even happened. I have wanted the ability to start a simulation from the previous operating point for years and was actually excited thinking that might be what was happening. 
1 Like
interesting As i know it Spice is deterministic as in it should not be random, unless intentional.
Maybe you experienced a RAM bit error or similar.
Maybe V2 that is 0.0 is to blame since it is also tied to GND.
There is no current to measure there since the current does not close into GND there but into V1.
V2 can be 0 1 1k volts and it will not matter. This may be causing unexpected behavior.
In LTspice often i had to use 1 Meg or 1G resistor to GND in order to reference some node. Try replaceing V2 with 1Meg.
