Sort of worked, but parameters on spectrum don't match those in
handout.  Can this be polished to 
    - be consistant
    - give good results rather than proof-of-concept results ?