Ben,

I've got a friend that does our low tech PIC subsystems. He also rides in 150 miles bikes rides so I'd bring him in if I were doing this as a project. I don't mess with PICs personally, but I did show him how to calculate 2-D vectors in 18 lines of PIC code.

I won't be sharing that trick. :)

If I were to look at PICs, I'd start with the LCD interface requirements first. I'd try to use a 12xxx with a serial LCD module if possible because of the minimal circuit, minimal interconnects, but a 16xxx might be needed if serial LCDs were too expensive. We have no computational requirements to do this... just enough ROM spce for our table.

Of course, you need your spreadsheet to find the best crystal speed so you can judge whether it can be forced into 8 bit values only (simpler) but if you have more ROM space, doing table lookups at 16bits is not too hard.

My company does weird radio analysis systems. I sell a radio signal field strength meter subcomponent that takes 32 bit free floating point data, strips it down to 24 bits of significance, and uses the 3 byte values to look up its display classes just as the technique I mentioned. The address of its "found" class, is used to display a visualization output and the 32 bit version is passed onto the next subsystem for further analysis.

Its a great software mechanism. Learning about priority queues may reveal a neat trick to simplify the lookup process. :)

-aka Jay C. Box