I have a rig with 13x RX480 highly tuned cards running SMOS where the rig crashes from time to time. There is nothing related in the Claymore miner logfile, also SMOS seems to not keep the Ubuntu logs in /var/log - any idea how I might isolate a potential card that is causing the problem without disassembling the rig and taking out cards? I don't have windows, only SMOS.

Strangely, this rig also do not accept any custom mem clocks via the SMOS console. Whatever numbers I put in, regardless if a single one or per card, it just takes the default mem clocks from the cards itself (which are optimized per card).

Any ideas? 

PS:
All cards were pre-tested under load prior building the rig on a testbench and the clocks of each cards were tuned indivdually so that they do not produce memory errors. The core clocks are all 1150 and core voltage is 887mV also pre-tested under load with each single card. Each cards temp in the rig do not get over 67°C. Running optimized Claymore-eth 10.6 or 11.6 where 10.6 seems to be more stable.