@RealAleMadrid
Thank you for your reply. I recently changed my ISP, and they asked for the Technicolor DWA0120 to be sent back. I have used the DWA0120 for well over a year with very good results. This might be because both the Far and Near end were based on Broadcom chipset. With the DWA0120 out of the picture, I switched to a Vigor 130 which was very affordable on eBay. During the first evening, when I was streaming films, the connection started dropping with alarming regularity.
I managed to stabilise the connection by forcing a switch from the "fast" mode to the "interleaved" one. To achieve this, I disabled G.INP on the Vigor. It has been stable ever since. However, in the process I noticed a large number of errors on the upstream side. I think what you are saying about that coming from the cabinet, makes a good sense. That would definitely explain the large numbers.
With regards to FECS, I suspect they are seconds, after all. If I do "vdsl status more", it reports the unit of measurment as "seconds" next to FECS. It kind of makes sense, if you think of it. FECS is a proactive mechanism, and I suspect the sending end does not really know if the redundant data/checksums have been used by the receiving end or not. For that reason, it simply reports "I have been sending redundant data for X number of seconds".
> vdsl status more
vdsl status more
---------------------- ATU-R Info (hw: annex A, f/w: annex A/B/C) -----------
Near End Far End Note
<...>
FECS : 56655 170595 (seconds)
ES : 7 12254 (seconds)
SES : 0 21 (seconds)
LOSS : 0 0 (seconds)
UAS : 28 30350 (seconds)
<...>