Quote:But when I simulate the Tran analysis, no matter how long it is, the reference voltage will be unstable in the final simulation time in 100ns period. But the value is very small.
Could you explain this in more detail as I do not completely follow it. Are you saying that bandgap is unstable in tran simulation, no matter what and the output of BGR is small?
In general, DC analysis tries to find an equilibrium point and it is not necessary that the equilibrium is a stable one. In a transient analysis, the simulator perturbs the system by moving it away from the equilibrium point. If the circuit is stable, it will return back to its equilibrium position. Going by what I understood from your post, it appears that the BGR is indeed unstable. I do not know which simulator was used to simulate this particular circuit, but I would be hard pressed to believe that it is a simulator issue.
For more details, you might be interested in taking a look at
http://kenkundert.com/docs/bctm98-MSsim.pdf pg 5 in particular.
Thanks and regards