DLCTOL=1D-9 ORTTOL=1D-9 does not solve the problem.
>>How to solve the problem? By the way, are there failsafe options for normal relaxed PES scans (without bonds breaking, using simple black box methods) so that I can do something more useful than endlessly adjusting keywords and options?
>Black box methods are in Gaussian. Real quantum chemistry expects thinking and understanding rather than button pressing.
Black box methods are everywhere. You do know what I an talking about. And please, in this particular question, let me decide which approximation I need.