My problems occured after the index was built successfully. 2-3 minutes later, during the build all was ok. No blocking etc.

The system is never completely quiet, but it was done in a quieter period.

Changing NON CL PK to Clustered can have a negative impact?

If so what is the best course of action for these heaps. I would insist all PKs are created on a table as clustered. But what is the best course of action to turn this db around. Testing this is not really an option. 1 single table can have between 30 - 100 dependant sprocs. Each sproc can have probably the same amount of parameters 10 - 30 maybe. NOT MY DESIGN. legacy of many years of bad design and administration.

To test one table would take a long time, there are about 1000 tables to sort out.

Real headache.