2 LB 10-X AC have been fired multiple times from several distances, aiming for CT of an AWS. Details see Methodology section.
Test results:
Expected maximum damage: 20 (all pellets hit)

Note:
- I guess the main error / std deviation source is the % health granularity.
- average DMG is reduced as not all pellets hit the mech
- Long range of LB 10-X AC is 540 m, max range is 1620 m.
- At ranges <= 90 m, every pellet hits; at ranges <= 50 m, every pellet hits CT.
Methodology:
- Test vehicle: CTPLT-K2
- Test weapon setup: 2 LB 10-X AC, shooting both in one weapon group
- Target mech: AWS-8Q
- Aimed at middle of CT
- Map: River City (Night)
- Match was restarted when AWS was severly damaged.
- For each distance >= 180 m, more than 20 salvos (40 shot) have been fired (yes, 40 LBX shots don't kill the AWS). 90 m: 8 salvos, 135 m: 9 salvos
- dmg values calculated using % health
- 4 tests have been in more detail to include std deviation (of individual salvos vs. average salvo) as error bars.
Same map & mechs, 20 salvos from 200 m, aiming at CT. Then stripping remaining armor using ML and MG, counting ML shots and MG bullets to calculate remaining armor of a section. Error estimation is max error based on remaining health % (there's some dmg to internals). dmg % refers to how much of total dmg has been done to a section. Only one test has been conducted.

Edited by Phaesphoros, 23 March 2013 - 11:06 AM.