Evidemment Google est ton amis... je me réponds à moi même :
élément de réponse trouvé sur Scuababoard :
Buhlmann has three versions of the ZHL16 algorithm, each of which
has slightly differing a/b coefficients. ZHL16a is for theoretical
use only. ZHL16b is for generating tables. ZHL16c has slightly
more conservatism built in, because it was intended for use in
dive computers (UWATEC's Aladin and many other late-model dive computers
use a variant of this). I originally used ZHL16c, but have added ZHL16b as
a user-configurable option. I first chose "c" because it is
supposedly better at funky multilevel profiles, but mostly because I have
a wrist-mounted version of this program for personal use. Generally, you
will notice little difference between the "b" an "c" options, except
for middle-length dives to middling depths, where the "b" version will
be slightly less conservative.http://zplan.com/download/README
Translated in layman's terms he says that the M values are different between the various models. "A" being useless for practical purposes. "B" being some kind of baseline and "C" being more conservative.
The actual differences in detail can be seen in this table:
A B C
106.4 same same
97.3 same same
83.2 same same
73.8 same same
66.8 same same
62.3 same 60.8
58.5 57.4 55.6
55.2 54.1 52.3
52.3 51.7 50.1
49.9 same 48.5
48.2 same 47.2
46.8 same 46.1
45.6 same 45.1
44.5 44.1 44.1
43.5 same 43.1
42.6 same 42.4
41.8 same same
source: Understanding M-values; Erik C. Baker, P.E.