Thanks, Pancho_hideboo.
For the dc analysis, it is not stable, the opamp behaves like a comparator.
The integrator only works at the trans-simulation, the input current source keeps absorbing current from the feedback capacitor, and the opamp will keep the voltage of the inverse node of the opamp to be equal to Vref, the output node of opamp voltage is Vout=Iin.*t/C.
For noise simulation, the negative loop seems not start to work, so the output voltage of the opamp is either vdd or 0. Then the simulation result is not right.
- - - Updated - - -
Thanks, FvM.
It is a good idea for setting the initial condition.
For adding a large resistor feedback, it might lead to output to be vdd or 0. Iin*Rf will form a voltage drop delta_V,
then the output voltage of opamp is Vref+delta_V or Vref-delta_V, which might be Vdd or 0 separately.