Day 1: 18 mb cap, 6 dB NM (no change from before)
Day 2: 18 mb cap, 3 dB NM
Day 3: 19 mb cap, 3 dB NM
Day 4: 19 mb cap, 6 dB NM (18.6 meg synch, effectively no synch cap)
Day 5: 18 mb cap, 3 dB NM
Day 6: 18 mb cap, 6 dB NM (end of DLM, current speed)
What's interesting is that there was no difference in noise margin between the 19 mb cap @ 6 dB NM and the 18 mb cap @ 6 dB, which supports the theory that output power is reduced when the cap is hit. This doesn't really make sense, since it's the extra noise margin on a line that adds stability, which seems to nullify the point of capping the speed (if stability is indeed Sky's reason for capping).
Since the line was stable throughout the test, I am disappointed that DLM chose to cap at 18 meg when the 19 meg profiles were perfectly stable. Yes, I'm aware that 18 meg speed is good, but I think this test highlights the case of Sky over-capping synch speeds when a line is perfectly capable of attaining a bit more, in this case resulting in 0.6 meg being lost.