Kitz Forum
Broadband Related => ADSL Issues => Topic started by: Weaver on October 16, 2018, 03:26:06 PM
-
Thanks to our own johnson, who has been such a hero, I have obtained an SNRM-vs-time graph (https://www.dropbox.com/s/wgkqi7oo668nk9k/Snrm_graph_2018-10-16.jpeg?dl=0) directly from my ZyXEL VMG 1312-B10A modem using http. It now has an internal stats server feature, which is absolutely superb.
See this this image (https://www.dropbox.com/s/wgkqi7oo668nk9k/Snrm_graph_2018-10-16.jpeg?dl=0).
For some reason, there is a sizeable drop in the upstream SNRM - the upper trace on the graph - at 12:00. Very sudden. Does anyone have any ideas as to what might be the cause?
-
Mrs Weaver using the microwave oven? The bilge-pumps have auto-started? :-\
Something has been switched on.
-
You would think so. I shall ask if anything in particular has been switched on here.otherwise there are the two candidates of crosstalk and distant rf sources. Where would one look? (In terms of musing on the problem, I’m not going out roaming the moor with an x kHz receiver.)
It’s nice to now be able to see a steady decline in downstream SNRM starting at 17:00, from 3.05 dB down to 2.7dB with small occasional downward spikes down to 2.5 or 2.4dB. It is now 18:36.
-
It has all the hallmarks of the switching of something electrical. Has the SNRM recovered or is it still depressed?
-
It went down and stayed flat. I wonder if I should reverse the question and ask why it was so high before - perhaps the end of a special quiet period, for some reason, and we are back to normality. But that of course doesn’t help me.
Mrs Weaver swears she did not turn in any kit, not once-only anyway, at around 13:10 or what ever it is.
-
The other possibility is that the line was whacked by some one-off brief event, and it'll go back to normal on a retrain. I assume you don't have a line that's upset by incoming phone calls.
-
@ejs - does that fit in with the picture shown? Because the upstream SNRM went down and stayed down for good. That graph does not unfortunately show too much of a time period after 13:10.
I don’t have telephone service. Mind you, what would happen if someone tried to call the phone number anyway? I don’t know how a caller would find out that particular number.
( However, line #1, which is not the line shown, has a phone number that used to be well known. Indeed doctors and hospitals keep on ringing that number and then giving up, even eight years after it became defunct, so deeply is the now-useless info spread.) that is one great thing about going voip - no voice spam.
-
Here is a new graph image (https://www.dropbox.com/s/h1etw6rwxps1maq/SNRM-line%232-2018-10-17.jpg?dl=0). Things get even more interesting. It remains to be seen whether this pattern is repeated every day. It appears possible that some device somewhere is switching on at 11:20 roughly and off again at around 23:15, early twelve hours apart. So all this tells us something, but I am not sure exactly what. Some piece of large current electrical equipment that is on a timer? An electric motor ?
We have a borehole water pump which switches on and off. It is nowhere near the phone line. I suppose it could be spitting out a load of interference. If so, perhaps we could go to it and detect the rf? Could it easily be suppressed to fix it? This doesn’t seem to fit the pattern though, as the thing doesn’t run according to a twelve hour timer. It comes on when water storage is low and off again when it has filled the tank up again, presumably. That takes some random amount of time. All depends on water consumption patterns which are going to be reasonably variable, not absolutely rigidly fixed, although there will be a rough daily pattern. But people showering or drawing baths will be a bit random in some cases. There is no way a magic number like approx twelve hours, a very significant number, is going to come out of it, as far as I can see.
Things like heating systems are on daily timers, and that class of systems would be my proposed candidate. We do not have anything of that sort on timers. It would have to be some householder down in Harrapul, which is on the main road, who is on the Heasta road and is very near our village’s copper cable. (What is the proper term?)
-
I wonder if there is anyone who does rf interference surveys?
Googling such a service has been a total disaster - google really is utterly useless some times. [What I really want is a categorised hierarchical, organised database of suppliers of goods and services if various types, also with ‘tags’ and businesses often need to be placed in more than one category, and we need to have more than a single database ‘view’, more than a single tree.]
Calling someone out to locate such a thing might be a nightmare to organise and incredibly expensive because of the remote location. I don’t suppose, if such a true expert with real experience could be found, that they might liaise with and oversee a local electrician? In any case such a thing would be completely hopeless if specialist test kit were really needed, like a spectrum analyser, rather than basic stuff with just an ordinary radio of some appropriate type acquired for the job.
-
Are you monitoring every modem/line or just a single line?
If monitoring multiple lines do you see it on any other lines?
It also isn't necessarily something on or near your premises. With such a long, rural line with challenging terrain I'm sure your lines will pass a variety of automated pumps, generators, substations, properties, farms, businesses, etc.
The source of interference could be coming from a vast number of different possible sources.
With VDSL2 you can monitor bitswaps on specific tones which can help narrow down problem tones/frequencies.
I don't know if the ADSL2 stats are as accommodating.
Tracking down the source of this would involve walking about with an AM radio which I understand your circumstances don't allow.
I don't know of any firm who offer such a service apart from OpenReach themselves but I don't believe they would assist for a 0.8dB drop in SNRM that isn't service affecting.
Do you still have the ability to run a quiet line test with AAISP lines? Do you have a corded handset to run such a test?
-
It's not what I suggested earlier if it goes back up on its own without the line retraining.
As it only appears to affect the upstream, not even a LW radio would help because the upstream frequencies are lower than the LW band.
I advise ignoring it. It could well have been like that for years. This is the reason manufacturers / ISPs decide to make modems that don't give out much in the way of stats.
-
I advise ignoring it.
Or not to be too concerned about it.
It could well have been like that for years. This is the reason manufacturers / ISPs decide to make modems that don't give out much in the way of stats.
Very true. :D
-
It does tell me the right time to resynch my modem though. It indeed is not a problem for the upstream, just an oddity.
What is really really odd though is that it does not affect line #3 - I have just looked and have not got round to installing the graphing in the other modems yet.
You’re right, that is exactly why modem manufacturers do not want users to have this info.
-
What is really really odd though is that it does not affect line #3 - I have just looked . . .
Is that the most recently installed line (your fourth) or is it one of the original three? We know that the most recently installed line transits the cabinet NSBRD_P10, whilst the original three were DEL.
-
And that is indeed the most recently installed one. With the on-off noise.
-
Do you know where the route of the new line departs from the other three? Obviously you need to start thinking about any businesses from that point onwards.
Thinking about the times: a restaurant with a noisy extractor fan? Looking at Google maps at what could be the route: Claymore doesn't seem to open early enough, Creelers shuts too early. If you think of any likely candidates it mike be worth ringing them to ask if they turn something on/off at the right times.
-
Interesting! I wouldn't worry too much about the drop, once you have made certain there are no devices on the property that run on such a time frame there is likely little you cant do about it. If it was causing a resync then maybe further investigation would be worth it, but for now at least you know the best times to choose to resync.
Here is a new graph image (https://www.dropbox.com/s/h1etw6rwxps1maq/SNRM-line%232-2018-10-17.jpg?dl=0).
So combining both up and downstream, the best time is between 7AM and 11AM. There is a good 2db of wiggle room left at the worst points of the night on the downstream, so I guess a forced 1db reduction might be viable:
xdslctl configure --snr 80
I believe 80 gives just over a 1db reduction in target, so thats what I'd do in the 7-11 time period and monitor further.
-
Am I the only person who thinks that some supposed interference source that only affects the ADSL upstream seems quite unlikely?
-
You mean because anything putting out enough RF to cause this in the upstream should be having a similar impact on the down? I guess so.
But what are the other possibilities? I'd be worried there was some glitch with the graph if it caused exact round number changes or happened dead on hours, but to the best of my knowledge the graph is exactly what the modem is reporting.
Upstream on ADSL2+ is up to what ~ 140khz, right in the band for most SMPS... although I agree in my experience a PSU putting out noise will cause interference at many higher harmonics as well.
-
Given that it is only affecting lower frequencies, this could be a very narrow frequency source. Talking about the cable run, the Claymore restaurant might be near where the cable goes, I am not sure where it crosses the main road from the north to south side if it is not on the south side all the way, which it might very well be.
I don’t understand how there can be much difference in the placement of the newer line compared to the previous ones as they are all heading back from the same place to the one and only exchange.
It is very weird that it is not seen on the other line.
-
Today’s graph (https://www.dropbox.com/s/e25mibh72sslcmy/2018-10-19.jpg?dl=0) - not quite such a simple regular pattern
-
Thinking about businesses etc - Creelers’ restaurant is nit that near to the main road, it is a little way off, on a side road. There is the village hall, primary school on the south side of the main road and on the north side there is a large petrol station and a sizeable supermarket and car parking. The Supermarket and petrol station will have all sorts of electrical activity going on. I am wondering if some of the cables from the exchange run along the north side of the main road and some along the south side. We were told - see earlier threads - that a pair had been brought into service from some other bundle and it sounded as if there were enough pairs going from the main road to Heasta but not enough going along the main road from the exchange (west) towards the PCP very near (to the west of) the Claymore restaurant where the Heasta road meets the main road.
[Moderator edited to correct an "autocorrect generated typo".]
-
The plot thickens! Certainly seems like business hours for a restaurant or service station, will be interesting to see if it changes over the weekend.
And you are seeing no similar dips on the other lines?
-
Still no such activity on line #3. Yet to install monitoring on the final two modems but will do so ASAP.
-
Have you tried listening on a good old AM radio receiver to see if it picks up any electrical noise? If it does you can use it to direction find the source.
-
If someone were roaming around with a radio, they would have to somehow be close to line 2 but not to line 3.
-
And then it all changed, because the more recent total weirdness is now seen on line #3.