PSRR strange variability

Hello,

i'm designed a OTA, particulary a three stage. i'm using the reversed nested miller to create an adeguate frequency responce. it works well but the value of PSRR is very strange. I made the
corner analysis with cadence and for the Nominal case (TT) it has a strange value, instead in the others corners it seems quite good. i report some numerical value to undestand better:

Corner TT FF SS FS SF T=27°C
GAIN (dB) 58,2 57,7 58,49 58,18 58,01
PHASE(°c) 66,01 68,73 63,17 67,15 64,79
PSRR (dB) 95,43 47,74 48,82 50,52 50,85

Corner TT FF SS FS SF T=85°C
GAIN (dB) 55,96 55,11 56,45 55,63 55,98
PHASE 64,72 67,36 61,94 65,69 63,67
PSRR (dB) 66,17 44,31 48,99 47,12 51,2


Corner TT FF SS FS SF T=-10°C
GAIN (dB) 59,39 58,96 59,64 59,44 59,16
PHASE 66,94 69,7 64,07 68,15 65,66
PSRR (dB) 85,96 47,68 48,96 50,51 50,97


what happend? why it failure in the TT?
Thx

PS: for the simulation i use a dc voltage set to the analog ground and in the VCC i use a dc voltage + ac voltage and the PSRR=(db20(VF("/vout_ac")) - db20(VF("/vout_ps+")))

Sure, it is a good value of PSRR. But my OTA has a low CMRR (it is due to bulk strategy) and i suppose That Also the PSRR is not so hight.
i do not understand why in TT is so hight, instead in the other corners it is so low (but they are so close) Which are the reason of this difference? can i make something to reduce it?

for the Nominal case (TT) it has a strange value, instead in the others corners it seems quite good.

Click to expand...

??? The numbers show higher PSRR (better) for TT.

- - - Updated - - -

Higher than expected is a bit different from failure.

Everything is just simulation, it's easy to "look into" the circuit and find out why CMRR is different. My guess is that common mode coupling to first and second stage cancels by chance in the TT case. If so, you need to perform Monte Carlo analysis to get a realistic view of expectable CMRR for a device with non-ideal parameters.