ACARS tranmission was via satcom
Finally - i am glad, that avspook maybe closed the discussion how ACARS messages were transmitted.
I have 3-4 years of expierence woking with HF data trans. and 2-3 years on SATCOM.
So, VHF is limited in range - useful on ground.
HF - not an option. As some documents are stating, it is an option close to poles - where SAT signal is week and angles to sattelite close to 70 deg.
What is left - SATCOM.
Minimum SATCOM link bandwith is 16 kbps - all known ACARS data can be transmitted in 1 sec. SATCOM modem also can take responsibility to repeat data, if not send propertly, so only 2:14z events data can be incompleted\not fully transmitted.
Looking at AF447 antena location, i also can gues, that SATCOM equip. was high power-small antenna-wide beam, antenna can be pointed to geo satelite with 15-20 deg. accuracy and still works.
That also means that at 2:14z the plane very likely was in one piece, airframe intact, and antena pointing to SAT with 15-20 deg. accuracy. If pointing devices were working - pitch\roll angles likely not exceeding 60 deg. It hardly believable, that plane broke apart, but antena was not torn apart from pointig device, and still recieving data and power.
Now - very very speculative part:
SATCOM modem is allowed to transimt data only in its 'time frame'. (to avoid signal inteference with other ground transmitter at geo sattelite end). Time frame is very narrow and very precisely calculated, so if you move antena 100 meters close to the sattelite (because of altidude, air speed, round globe) it is detectable - signal arrives a little bit to early\late. (i was told that SATCOM operators can notice even if i move antena 100 meters on the ground)
If logged, delay data can be used in reverse ingineering - to calculate distance from sattelite to AF447 and can provide coordinates with 100 meters distance (hardly more than 1-3 kilometers on map) accuracy. But only if logged. Signal/noise ratio (S/NR) can give some idea how precisely antenna was pointed and indicate that a/c was not expierencing wild changes in pitch\roll.
Maybe this helps to close some speculiation.