>>You can try to use previous result as derived table
>
>Thanks
>
>It turned out that after having tested several approaches, I used the one from Borislav. I am sure I would have been able to have all proposed solutions to work OK with a little more time.

Did you see my latest reply (packed value)?

BTW, the links I gave you in my very first message contained all methods to get the desired result including the one showed by Boris.