distinct count question

Hello Community, 

I have a situation where I need to count the distinct count of "problem codes" against their corresponding order.  

The table below perfectly illustrates the issue I am trying to solve.   

The count of RMAs is a distinct count.   I need the Header Problem code count to match that  (72).   So I need a measure that equates the RMA to the header problem code, and sums up the correct number of header problem codes  (which should match the distinct count of RMAs).     

In the table below, you can see if I just use the options  -  count or count distinct  -  I end up with the wrong values (for what I need) for the problem header codes.   The "distinct count" option just gives me the distinct count of the header code types in this data set  (3 in this case).  T3  T1  T5.      The "count" option just gives me the count at the line level.   

I need a "distinct count" of header codes, but I need it to be tied to the distinct count of the RMAs since each RMA will have a corresponding heading problem code.

Any guidance is appreciated!