?: LTC2400 with buffer, troubles, need help

[GigaJoe]

Hello, everyone

UPD: opamp faulty,  seems Chinese fake ... ( replace it , and all !!!! )

need some help in understanding:
U2 - LTC2400 , NOT 74

LTC2400 (U2) + buffer (U1) , input shorted. LTC Vref- 5.0000V , Vsup- 4.95V
Result: 
if U2 input disconnected , U1 output grounded over 1M resistor the output  2mkV 
removing resistor and connecting input LTC2400 (U2)  - R4, C3 - around 2.5V slowly discharging over voltmeter resistance.
shorting capacitor by 100K resistor, output very low , and i'm getting triangle impulses around 200 mV p-p.
LTC2400 digital output hang in the air.

?:  Where is my mistake .... ?  What else for debugging ?
or it opamp faulty ?

[Kleinstein]

The diodes for input protection are a little odd. The negative sid would allow up to -8 V and this too much for the LTC2400. This diode could go towards GND.
The positive side diode is only towards a divider, so with more current from the input the voltage could rise well over the set 4 V. The should be more like a lower impedance sink. So maby a divider to 3 V plus an PNP transistor to amplify the current.

[GigaJoe]

D2 - are correct in my real schematics ,  grounded to the neutral.

I'm not 100% certain about input diode protection , it just a prototype , to hook arduino over ADUM isolation.  So i have 2 independent transformers to run , for digital and analog parts , then debugging the code to get a data ,   them see any improvements ,  analog parts again ,  pcb construction etc ...  that a plan for now ...

thinking, for a next cycle need a different  model of LTC24XXX , with differential input and differential Vref,