Power-on transients aren't observed in a standard transient simulation,.....
Yes, that`s true. Therefore - when you try to be as realistic as possible you should use a power supply that is switched-on at the beginning of the TRAN simulation.
Such a switch-on is a very good help, for example, to allow self-start of an oscillating device.
- - - Updated - - -
Power-on transients aren't observed in a standard transient simulation, also for a real OP with finite bandwidth, because GBW isn't considered in initial transient solution. If you look at the waveforn of the positive feedback circuit falling into saturation, it starts at t=0 after initial transient solution.
The above described effect reminds me on a contribution in an electronic journal with excellent reputation:
A.S. Elwakil:
On the necessary and sufficient condition for latch-up in sinusoidal oscillators.(Int. Journal of Electronics, vol.89, No.3, 197-206)
In this article, it is prooved by Elwakil from a system point of view (using the pole distribution in the s-plane) why an IDEAL opamp with dominating positive dc feedback cannot work as an oscillator.
For this purpose, a modification of the classical WIEN bridge oscillator is analyzed that still fulfills Barkhausens oscillation condition - however with net positive dc feedback.
It is shown that (and why) the circuit starts oscillating and exhibits latch-up as soon as the amplitude reaches on of the supply rails.
(Not contained in that paper: It is interesting to note that a REAL opamp model shows another behaviour. It starts NOT oscillating but immediately goes into saturation due to a REAL pole in the RHP of the s-plane)
- - - Updated - - -