Kitz Forum
Broadband Related => ADSL Issues => Topic started by: aesmith on February 13, 2016, 06:18:22 PM
-
Hi,
We had a noisy phone line fixed recently, and although that's made a big difference I have to face the fact that the error rate is still too high for comfort, running from the high hundreds to a couple of thousand ES per day. I'm really looking for ideas as to how I can address this. The error rate goes through the roof in the evening, from around 20:00 at night until midnight or sometimes a couple of hours into the next day, and that's accompanied by a drop in noise margin. However I'm not completely sure that of that link, because for example sometimes like today there are regular error even when the nm is over 7dB and not really changing with a pretty big drop. See attached for example.
Any ideas as to anything I can do to track the issue down? I can't see me making any headway with the service provider.
Thanks, Tony S
-
> can't see me making any headway with the service provider.
Is changing ISP an option?
-
It is, but given the history of our line I would really want phone and DSL from the same provider. Unless I go for broke and get our phone number ported into SIP, meaning our phone line could just be for DSL with no calls.
-
Something to mull on…
-
I wonder if anyone could comment on the attached screenshots. This isn't an area I've handled myself to this sort of detail, but I wonder if there are any clues in these statistics which would point to where the errors originate.
Attached HLOG, Bits per Tone, QLN.
-
Also SNR per Tone. I think that Bit's Per Tone is dynamic, i.e. shows the current usage. However I'm not sure whether the others are recorded by the router only at time of connection. Last resynch was around 07:00 yesterday 16th Feb.
Thanks,
TonyS
-
Out of interest, what modem are you now using? I thought it wasn't possible to get the HLOG and QLN data from the 582n (well, it might be possible using some undocumented commands and the Broadcom DslDiag program, but that way is very inconvenient because it drops the DSL connection and puts it in a special diagnostic mode).
-
It's a Billion 7800DXL, and it certainly does report more info than the 582N.
-
I think that Bit's Per Tone is dynamic, i.e. shows the current usage. However I'm not sure whether the others are recorded by the router only at time of connection.
QLN and HLog are only recorded at the time of connection. All the other stats are dynamic.
-
Cheers. So would it be worth re-synching during the highest error period, to grab those readings at the worst time?
-
If you do that you'll almost certainly end up with a lower connection speed.
-
I was thinking to use the SNR target adjustment to correct for that. So let's say I connect during the day at 6dB target, then in the evening the NM drops to 4dB, I'd reconnect setting the target to 4 so I get more or less the same speed.
-
The dip in the bits per tone at about tone 160 (690 kHz) - possibly an AM radio station? (perhaps Radio 5 Live on 693 kHz)
I think the bits and SNR per tone graphs look fairly ordinary, for a fairly long line.
I suppose you could try comparing the bits per tone at a time when the errors are bad with the same graph that was taken at time with lower error rates. I don't think there's any great need to reconnect to get the other graphs, the bits per tone, SNR per tone, and QLN all show the same kind of thing really.
-
One odd thing, if the router resynchs the error rates are really high for the rest of that day, then the next day and subsequent days they are much less. I thought that might be because over the following 24 hours it gets a chance to move off the bad frequencies. However bit swaps per minute continue to follow the same rough pattern, with a lot overnight and fewer during the day.
-
Errors have still continued to reduce day by day, the four day ES graph shows the pattern clearly. However in contrast the noise margin has reduced day by day from over 6 during the first day after re-synch to now barely going above 5. Speculating, could it be that the modem is a bit optimistic when training up, then systematically swaps bits off the bad frequencies over the next couple of days?
Note resynch was at 07:04 on 16th, which is the first complete day on that graph.
-
Errors are still all over the place, I think that perceived reduction following a retrain was a red herring. Sometimes we have a good day, more often it's many hours of 20-50 ES per hour, giving daily totals of a few hundred to a couple of thousand. Current MTBE is 78 seconds for example.
What do people think? Is this just something I'll have to live with, or could this be a line fault? I'm conscious that the recent fault was only half fixed by moving us to a different pair within the 5 pair cable, since the engineer couldn't locate where the fault lay within that span. So I'm wondering whether there's cable damage that maybe degrade the other pairs, even though there's no audible noise on the line now.
Tony S