Another thing that probe was to cool the plate and the LED strip without result, so thermal failure is not ...
I pointed there at the beginning but the fault is very systematic, after 5 minutes it appears, it also seemed strange that a bad LED could modify both the power values ... I don't know what you think. Thank you for your input!
Thank you Victor for your contribution, I took the task of feeding the leds of the group that failed for 5 minutes one by one and the fault appeared in the last of the group, it was shortening after raising a little temperature and was sent to ground. It is incredible how the tension of the rest of the independent feeds pulls down. The electronics out there surprise you hahaha, comment below so I leave the solution, thank you very much, hug!