Problem with LTSpice

Got a circuit to work very well using the downloaded model for the
AD8605.
Then i added one as a voltage follower to isolate a pot that had been
driving a resistor network.
Initially with the pot set "at" zero (tried 25mV and 2mV out of 5V to
be more specific), it worked fine (both cases).
Then i tried it with the pot set "at" 5V (actually 4.975V).
Refused to converge, even with ITL1 ans ITL6 set to 950.
Went back to my 25mV case, and it *still* refused to converge.
Would not work at any input voltage level!
Cut out that added buffer and all was OK.
Used "undo" sufficent times to restore it, and FRAP!

1) How did it go nuts?
2) How does one fix it?

 

Hello Robert,
you could either add the .asc file to your next reply here or
you send me all your design files(.asc, .sym, model) and I will
check it. The ladder may be the best idea.
I will then come back to the group with my solution.

I can tell you that most of the "old dog" SPICE tricks
regarding convergence don't make sense with LTspice.
The solver strategies are too different.

Best Regards,
Helmut

 

Are the DC earth returns OK? My usual error!

 

Grounds and supplys were OK; i gave up, tossed the file and started
from scratch.
That worked.