@stavely
Interestingly, on my 2860 the last 2 re-syncs have shown 4 and 6 dB SNR immediately prior to disconnection - which may indicate something else going on.
But, yes, i have done most of your suggeastions already. the only thing i have not done is the alternative modem code.
Setting the SNR changes the starting value at sync, but has no effect on the gradual dropping to zero.

As far as the modem decreasing SNR to get the best speed - in theory it's an idea, but I do not see any increase in line speed as the SNR drops towards zero.

Unless your cab's DSLAMs support vectoring (not many do yet, but there are some trial sites) then the vectoring settings will have no effect.

Note to all: I am currently testing a new vector5 modem code from draytek support and initially it is ok (only been running about an hour so far). Time will tell whether it remains stable or goes into a re-sync loop like VT3 seems to. Fingers crossed. I'll try to post up the results when it has been running a bit longer.

jedi98 wrote: Note to all: I am currently testing a new vector5 modem code from draytek support and initially it is ok (only been running about an hour so far). Time will tell whether it remains stable or goes into a re-sync loop like VT3 seems to. Fingers crossed. I'll try to post up the results when it has been running a bit longer.

@jedi 98 - Any update on how you are getting on with the vector5 beta?

Many thanks.

So far...

I am testing two 2860n routers with v3.8.4.3 modem code 579F17 (that is the difference here, the main firmware is the same but the DSL modem code is new). One is on a good quality VDSL2 line which gives full speed and is only a short distance from it's cabinet. The other is on a marginal quality VDSL2 line which can only achieve 33/5Mbit and is quite a way from it's cabinet.

Router 1 (high speed line) has been up and connected now for 230 hours without retraining with no NE CRC errors and zero ES/SES. it is maintaining 80/20 with attainable rare reading about 85/29. This is comparable in reliability to VT1 on this line. VT2/VT3/BT versions would all go in to retraining loops after a day or so. STD version would give stable but at lower rate. The line is reporting that it does not have vectoring enabled (presumably dictated by the cabinet equipment.

Router 2 (lower speed/quality line) has been connected now for 56 hours without retraining. There are continual CRC errors cropping up and hence ES/SES are going up. However the code seems to be keeping on top of the errors using ReTX, G.INP etc. Rate is 33/3.6M with attainable showing as 34/5.9M. So far this is better than VT1 which was retraining on average every 12-24 hours. VT2/VT3/BT were again resyncing a lot wore frequently, sometimes several times per hour. The rate achieved is slightly higher than the 29M on VT1. The line used a lot of interleaving on VT1 but so far has not had to resort to that. The line is reporting that it does not have vectoring enabled.

Overall things seem to be going well, but it is early days for router 2. Router 1 has been rock solid so unless there is a disaster I will be keeping this version there.

Regarding: v3.8.4.3 modem code 579F17 (vector 5) testing on 'low' quality VDSL line

The firmware is producing a consistently faster speed and with no interleave. If is still re-syncing (re-training) about every 1-2 days, however, having analysed the stats from the VDSL I believe that the re-syncs are caused by the DLM and NOT the 2860 modem. This is likely due to the fact that the line is of very variable quality. There are unexplained cascades of CRC & FECS errors but I will have to monitor this for a while to get an idea of what is going on there.

So far on a good line this firmware is perfect and on a lower quality line it is better than the previous versions. I cannot say how it compares to a BT Hub on the same line or an Open Reach Modem since I would not be able to get enough stats off of those but I do not believe that they would do much better job.

Each time there is a re-sync (usually overnight) the FE LPR goes up by 2. The last re-sync has also zeroed ALL the FE stats, this does not normally happen - does that mean that the card in the cabinet was actually reset / power cycled? Anyone know the answer to this one?

@jedi98 - many thanks for the updates.

Not sure if this is relevant to your last query about possible power cycling of the cabinet, but when I use the Openreach modem, the line usually 'resets' every 14 days or so. I've always assumed that it was BT initiated, usually between the hours of midnight and 7am and never detrimental to the connection speed.