Thanks Analog Ground and LvW for your suggestion. I make a new schematic to run with cadence spectre (attached picture). I plot differential gain vesus frequency (close loop gain) , the result from spectre is not what i expect. From formular for close loop gain Acl = Aol/ (1 + B*Aol) B is feed back factor
From schematic B = C2 / (C1 + C2 + Cin) where Cin is total op amp input capacitor (Cin = 0.7pF in my simulation)
So the feedback factor B = 1 / (1 + 3 + 0.7) = 0.2
My open loop gain is 54dB, and from Acl = Aol/ (1 + B*Aol) ---> Acl = 1/B
---> Acl = 1/0.2 = 5
The spectre simulation shows Acl = 3 only
I thinking the open loop gain is a litle low but this is not the case because 54dB is OK.
Anyone has any idea to fix this problem please let me know. Thanks a lot
- - - Updated - - -
I make a mistake in caculate Aol:
Aol ~Cin/Cfb = 3/1 = 3. The simulation is correct