Please can an OCE check my line out; previously could sync at up to 5900 no problem, but recently can't connect at anything other than 4096 despite plenty of SNR in hand. No changes to wiring/hardware etc, so assume this is due to being moved to DLM? Perhaps the line training hasn't worked well? Also 4096 seems a 'coincidental' figure... is the line now being fixed/capped at 4Mb?
(all stated down/up)
Previous - typical
SNR 6 / 7 db
Line Attn 53.5 / 28.0 db
Data rate 5728 / 832 kbps
Power 19.5 / 12.5 dbm
SpdChk 4.85 / 0.73 Mbps
SNR 10.4 / 7 db (can see that down snr is now much higher than before)
Line Attn 51 / 27 db
Data rate 4096 / 880 kbps
Power - / 12.3 dbm
SpdChk 3.49 / 0.76 Mbps
I've completed a line test which hasn't detected a fault.
I can see that DLM has lowered the broadband speed due to a high number of errors on the line. Does your master socket have a test socket?
Are you experiencing any issues with the voice service such as noise on the line?
Thanks for that - good to know no fault was detected. I noticed that when errors do occur, it tends to be a large number in a sudden burst - might suggest an intermittent fault? Not sure if line test would detect that - unless it happened to go intermittent during the test? No problems with voice service - no discernable noise at all.
Yes I have an NTE5 single master socket with test port. Plugging my router directly into it requires some reorganisation because of the master socket's location, which is not particularly near to either a power outlet or my wired PC - I would have to run a power extension lead, and connect via wifi-only - but I can do this if necessary. The router has, for the last 10+ years, been connected via a permanent/fixed extension socket (only about 4-5m, in proper round telephone cable with a faceplate; not one of those crappy flat cable DIY kits). No changes have been made to the arrangement in a loooong time, but as a precaution I've just been round and checked & tightened all the terminals etc. Microfilters are correctly installed - one into the master, feeding a phone only; one into the extension socket, feeding router and a phone.
One other thing; according to router stats: significant numbers of errors (CRCs) are only appearing on the Up side, however it's only the Down side that appears to have been throttled by DLM? Up SNR and speed are as before; Down SNR is now 10-12 (usually 6), and speed ~30% lower.
If there was a fault at one point but which has now cleared, would DLM automatically try to train the line speed up again - having previously lowered it due to seeing errors? Or does retraining have to be manually invoked?
Thanks for the additional information. DLM may increase the sync speed if there are no errors on the line. Unfortunately we can no longer manually change the profile.
Would it be possible to connect the router at the test socket and we can try optimsing the connection to see how the sync speed and errored seconds compare?
Yes - as it happens, I connected into the test port about 10 minutes ago! Will leave it there until your instruction.
Thanks for confirming this. I've optimised the connection and the sync speed has increased to 6.1mb.
Would it be possible to monitor the connection at the test socket for 24-48hrs to see how the errored seconds compare following this?
Thanks - yes of course I will monitor it; it's an big immediate improvement, and looks rock-steady so far - I guess that may indicate that the internal wiring's not the best?
Presumably there's no problem with plugging a phone into the microfilter, which is now direct into the test port? Obvs else no landline....
Yes it would be fine to also connect the phone. If the errored seconds increase and DLM lowers the speed again at the test socket then the next step will be test with a different router to rule this out.
Sorry for delay - I monitored for a while, but then got busy with other stuff....
Anyway - it seems that there is still some kind of problem. From the router stats, I am not seeing many errors downstream; but upstream seems to have an intermittent issue. It'll be ok for a period, but them suddenly see a huge and rapid burst of upstream errors - the CRC count will go from 0 to 65500 almost instantaneously. Not recording any dips in Attn, and downstream is completely unaffected. Speed down to 4095 again, presumably DLM acting due to these errors.
Findings so far are:
1. Plugging directly into the test port does not make any difference - the problem still occurs. Would appear to rule out any issues with extension wiring/sockets.
2. Substituting the router does not make any difference either. Would appear to rule out router/hardware issues.
Any clues? There's no apparent issue with voice service, the line seems very low-noise; but I guess if there's an intermittent issue I probably wouldn't detect it. It doesn't look like my wiring or router is at fault - could there be some kind of wiring, hardware or interference/crosstalk issue somewhere between my line and the exchange?
Thanks for the additional information. Is your connection also unstable as I can see quite a few re-connections on the line?
When you tested with the alternative router did you also change the microfilter and adsl cable?
The connection is generally very stable - especially now that DLM seems to be aiming for 9dB or maybe even more, instead of the previous fixed 6dB profile. The disconnections yesterday were due to me trying to eliminate various possible causes of the upstream error bursts, ie by substituting router/cables/filters, and changing connection point (master socket, test port, internal extension sockets) to prove out my hardware and wiring.
Yes I did try with a different (TT-supplied) ADSL cable and microfilter. Even with these, and an alternative router straight into the test port, I was still seeing the same thing - ie stable for a short period (maybe 10-20mins), then a sudden and massive burst of upstream CRCs.
So as far as I can tell from that experimentation, all my hardware and wiring is fine since I get the same issue regardless of any configuration.
It doesn't appear related to data demand either - ie the error burst can happen when my router/network throughput is very low/zero, and is not triggered by high/100% or sustained bandwidth usuage eg during a download.
Also - did you just change something a few minutes ago? I saw a glitch, and despite router sync still 4095, the connection is *extremely* slow/laggy and keeps timing out? Also DNS server is now reported as 188.8.131.52 - I'm sure it was 62.24.something before?
ok will do.
Although it is hard to monitor any further after there has been an error burst, since the router's upstream CRC stat maxes out, and can't be reset/rezeroed without a restart. That entails a disconnection/reconnection - how will this affect DLM? Is there a way to restart without registering errors etc and causing DLM to think there's a problem?
I've re-checked your connection stats and although the sync speed hasn't increased the connection looks stable and the errored seconds are very low. Have you experienced any slow web pages since your last posts?
No slow pages now - all seems stable. Just not any closer to knowing where those intermittent error bursts were coming from... having ruled out, as far as I can, any internal potential causes - can only guess there's some external issue.
Worth retraining the line maybe?
Ok thanks for confirming this. I've optimised the connection again and the sync speed has increased, however this may reduce again if the errored seconds increase.
Let us know how your connection compares.
Ran great since you reoptimised, synced at 5200 and noticeably smoother than before. But this morning it's gone back to 4096; at some point there's been a disconnection during which the router stats get reset, so I can't see what happened or when. I did a manual resync just now to see if it'd go back to 5200, but no. Can you see what occurred from your logs?
Presuming there has been a burst of errors - it still seems to me that the source must be external to my line and wiring/hardware etc. Rubbish that DLM has such a drastic overreaction to a momentary externally-induced problem and cripples my line rate by 20%, even though it clearly can run without issue for days at a time.
Thanks for the update. I've re-checked the connection stats and it looks like DLM has lowered the broadband speed due to an increase in errors on the line. Are you experiencing slow web pages following this?
Would you like us to raise this over to our Network Team for investigation?