Plusnet 20CN Max Premium - dropping sync on incoming call

I've been struggling for some weeks to get a problem with sync dropping when the phone rings for an incoming call. Only if we answer the phone very quickly does the sync not drop, but then I see several hundred CRC errors logged in a matter of seconds, particularly on upstream. The loss of sync is usually so brief that PPP does not drop, but it is long enough that VoIP calls in progress when it happens are dropped and remote desktop over VPN drops out.

The OpenRetch engineer connected to my line inside the exchange and although they saw a lot of errors they didn't actually have the connection drop. He tried to tell me that it is because I am on Max Premium and that when they switched to standard Max it didn't happen. I've only recently moved to this house and have had this problem from the day we moved in. I've had Max Premium on two other exchanges (one in Sussex and one in Dorset) and didn't have this problem then so I think the engineers suggestion that this is a problem inherent to Max Premium is rubbish.

I have the router connected directly to the test socket. The problem occurs even if there is no phone of any kind plugged in to the filter.

Plusnet support is proving to be useless.

The last time I updated the open ticket it took nearly 48 hours before they picked it up and responded - that sort of massive delay is typical.

They keep saying they can't see any events that I am reporting and posting a graph from the radius logs. As there is no loss of PPP obviously the events don't show up. I've lost count of the number of times I've told them they need to stop looking at the radius logs and look at the sync events.

On the Plusnet forums it is obvious that there are lots of mistakes being made. Until their graphing broke a few weeks back it was obvious that they are seeing a significant increase in user numbers. My feeing is that their launch of the Unlimited was very successful - too successful and that their support is not able to cope with the extra demands. Very recently we've seen the news about O2/Be and Sky and that a lot of people are jumping ship to Plusnet which is putting even more demand on support and a lot of these migrations are going wrong.

Another bad sign is that they are fudging their support statistics here - specifically the Question Handling Average closure time. Normally we'd see that being updated regularly and the time would be a real figure in hours, minutes and seconds. Now it is only being updated infrequently and is obvious a made up figure such as 8 hours or 12 hours - goodness knows what the real figure is, but the last few tickets I've raised have usually been way longer than that before they are even looked at.

It's also worth pointing out that until early this year the target average was 8 hours. Then they sneaked out a change in the target to the 24 hours it shows now. In fact they kept the change so quiet that even their staff in the Digital Care team didn't know it was being changed!

Fundamentally Plusnet is (or was) a good ISP but it's obvious their support at the moment is badly stretched. For anyone considering Plusnet at the moment I'd advise wait and see if things improve.

Max Premium can push the upstream, so in a small number of cases there might be a higher chance of the upstream being the case for resyncs, as usual tracking noise margin will show if that is the issue.

Sounds most likely to be a HR fault and if engineer really did sync inside the exchange then I would have expected nothing else buy full sync. Unless there was a major issue affecting the line, again router stats tracked over time are the key

It's often a HR or battery fault. Is there any noise on a quiet line test?

Phone is absolutely fine.

You could use Routerstats to monitor the SNRM up and down, and see what happens when the phone rings.

I'm running RouterStatsand am seeing a remarkably stable noise margin. Downstream is typically 6db +/- 0.1 (sync usually 7000 to 7500 although with an attenuation of 21dB I might have expected better), upstream seems to be locked solid on 12db with full speed 832 sync. Evening noise margin might drop by about 0.2dB.

Yesterday I went 16 hours with no incoming phone calls and 16 hours with not a single CRC error logged up or down.

I am running RouterStats - see reply to MrSaffron.

Noise margin is a flat as a pancake until there's an incoming call and it bombs out.

What sampling interval have you got on RouterStats?

The default is only useful for background info about normal behaviour. For loss of connection diagnosis you need to drop it to the minimum, 5 seconds. Also make sure you are logging to file as well as saving the graphs. (On closing as well).

The combination of all that should give you some evidence you can submit to Plusnet.

If not an HR fault (you'd be nearing hiss on the line) then maybe an issue with the 21cn equipment in the exchange. Always difficult to prove this kind of thing if it IS an exchange issue, you need to apply an artificial 'load' to the circuit, and to do this you use attenuators, however sync will drop on an incoming call when using attenuators with or without a fault ....... Quality.

I seem to recall it is 1st gen Fujitsu 21cn cards that sometimes give the fault you describe.

Not round the Reading area are you ?

The version of RouterStats I'm running won't go below 10 seconds which is what I have set.

The evidence I've been giving Plusnet is from the router log. E.g.:

<173> Apr 13 11:46:31 xDSL linestate down
<173> Apr 13 11:46:45 xDSL linestate up (ITU-T G.992.1; downstream: 7584 kbit/s, upstream: 832 kbit/s; output Power Down: 19.8 dBm, Up: 12.3 dBm; line Attenuation Down: 21.0 dB, Up: 13.5 dB; snr Margin Down: 6.1 dB, Up: 12.0 dB)

They then look at the radius log, see no disconnect and call me a liar.