cry for help, at least suggest which direction to dig

a day ago all my 7 nvidia rigs worked fine, yesterday i have changed PSU on 3 rigs from cx750 to kcas 750, and now i got only 4 rig working

other 3 rigs, where psu were changed, dont work with any miner, tried ccminer many different versions, ewbf 034b, claymore zcash&dual. all miners giving some errors in a few minutes on any algo. claymore hangs rig, ccminer and ewbf any version just stops working, but rig continue working normaly.
have many bat files which are working correctly on exactly same rigs, but non of them works on these 3.
i have 1060 3gb ASus&palit, mobo msi z270 a-pro, b85 anniversary and h81.
nothing works even in stock. no bios were edited
i have tried
different driver versions 384.76/80 and the latest one.
reinstalled win 10
updated net framework
updated c++
increase from 16gb to 30gb
delete AB
copy many different vcruntime140.dll to syswow64 and system 32, after reinstalled again win 10

some error that i get on cc after few min,
Cuda error in func 'quark_blake512_cpu_setBlock_80_sp' at line 577 : unknown error.
Cuda error in func 'quark_blake512_cpu_setBlock_80' at line 547 : the launch timed out and was terminated.
GPU#1:invalid argument

honestly dont believe that the problem is just because of PSU? those PSU were working on other rigs with amd cards earlier.

any hints where to search for problem would be great