I've yet to download your sample, but I just created a log using the MODLOG file (John Mark Osborne's site) as the basis for my work.
I've run into the FIELD NOT FOUND problem, though, in the CASE (name <> name OLD, name OLD & " -> " name, address <> address OLD, address OLD & " -> " address, etc) calculation for actually writing the field-changes to the log. I got, however, far further along than a mere 11 fields before I hit this problem (probably more like 100 fields in).
Does this sample of yours get around this by creating a new record for each change, so it's not writing to a log field associated with each record? How do I get around the limitation in my CASE?