To my experience this should stay under 65 degrees celcius.
Whats the numbers under the "HW" column?
1. The miner is safe to run until 80c. It will not slow down when it reaches the thermal limit it will stop hashing altogether.
2. The HW errors column will always have some number of errors. You will see thousands of errors very rapidly if the hardware is having issues.
3. The 1.5t number you quote leads me to believe that you have 2 boards that are not hashing properly. Each board should run just over 1.5t on its own. i would try and figure out which boards are dropping then shuffle around the cables/connectors to see if the problem is based on the controller board plugs, the connecting data cable, or the hashing board itself.