Had problems since before Xmas with Broadband connected but random no internet. Anyone help please?

[#1]

As title. Delayed doing too much due to health issues since early September. Currently used a new adsl filter into master socket for 5 days, but no improvement. I can lose internet 3 times a day or can be solid for 3 days. It is random.

I wonder if someone can check this out for me please? I dont think my usual routerstats are correct? It is the plusnet 2704N router. Noise margin downstream is always below 6.0 as is upstream. Line loss is pretty consistent around 21.8 / 40.0, regardless of noise margin. I read somewhere that downstream at 5.0 is too low?

When it is getting flaky, the internet led is flickering fast and occasionally the actual download speed exceeds the maximum speed!

If it drops out, either a router reset is required or simpler a power off and on again. Then it restores every time...until the next drop out. Occasionally I am aware (by uptime) that it has dropped and reconnected without my intervention.

3 examples of router stats from today:

8 March @ 0820

6 - Uptime:	0 days 10:44:57
7 - Data rate:	938/9525
8 - Maximum data rate:	1308/10888
9 - Noise margin:	5.6/4.4
10 - Line attenuation:	21.8/40.0
12 - Data sent/received:	27.1M/560.9M

8 March @ 1742

6 - Uptime:	0 days 08:12:58
7 - Data rate:	944/10177
8 - Maximum data rate:	1308/10704
9 - Noise margin:	5.3/2.4
10 - Line attenuation:	21.8/40.0
12 - Data sent/received:	87.9M/759.7M

8 March @ 2037 INTERNET LED flickering fast but still working ok

6 - Uptime:	0 days 11:08:06
7 - Data rate:	944/10177
8 - Maximum data rate:	1308/10024
9 - Noise margin:	4.4/1.3
10 - Line attenuation:	21.8/40.0
12 - Data sent/received:	166.7M/3.2G

In fact, it has stayed on even now at 2115.

Thanks, Robin

[#2]

@joseph1 Those figures, apart from the SNR values, look good for your line length.

Before I suggest following a slightly longer process, have you checked your phone line for any noise (dial 17070 option 2 from a preferably corded phone - the line should be silent). If that test passes, I suggest you connect to the test socket on your master phone socket, plugging in via a filter, and see if there is any improvement. Do this preferably at least one hour after sunrise.

[#3]

Just tried phone quiet test, but only with my panasonic dect, which is absolutely quiet. Will try to find my old corded phone tomorrow. I have been plugged into test socket direct about 5 days now and no difference. It is the new, unused filter came with my router.

I had a similar problem maybe 2 years ago and was told by pn that my router was too old (only last 2 or 3 years!!!) and they supplied a replacement for cost of post, but that made no difference. Openreach came out but nff and he did a few tests, again nff. However, from then on, I had very few drops.

Back then, i fitted ferrite clamps over cables near the router and pc, but it made no difference then and they are still there now. I also replaced the router smps with an old 12V linear one, again no difference. Tonight, I tried switching off house lights, again no difference.

In the last 9 months openreach have replaced a lot of cables in my area, hence the routine 10Mbps rather than the 8 or so previously.

My wife went to bed a while ago, but still no great change in stats. @2150

6 - Uptime:	0 days 12:20:54
7 - Data rate:	944/10177
8 - Maximum data rate:	1308/10264
9 - Noise margin:	4.4/1.6
10 - Line attenuation:	21.8/40.0

When the noise margin gets too low, thats when I know trouble might start.

Quite often in the early hours it drops out.

Thanks, Robin. bedtime now!

[#4]

My bedtime too, Robin. I'll have some more thinking time on this tomorrow morning.

[#5]

Here is an old router stat page:

6 April 2019 @ 1742

6 - Uptime:	0 days 12:41:50
7 - Data rate:	944/8219
8 - Maximum data rate:	1356/8488
9 - Noise margin:	6.4/6.3
10 - Line attenuation:	22.3/40.5
12 - Data sent/received:	44.2M/681.7M

[#6]

Robin - just a quick reply, Your D/L speed and noise margin in that old page look much more like what they should, TBH, given the apparent line length. I'm beginning to think your profile is too high, but will ponder more tomorrow - old brain is too tired now.

[#7]

@joseph1 Just had a re-read of this whole topic, and I think you may have at least two problems here, although they are probably related. (1) From a similar experience I had about 18 months back, which took four (at least) OR visits to resolve, I suspect your connection on the frame, or even the frame itself, at the exchange, is dodgy - my final solution was getting an SFI out, who spent about 4 hours checking my line all the way back to the exchange, where he identified a failing frame. He got that swapped out and after a short time whilst the DLM identified my 'correct' parameters, I have been at the same connection rate, with no fails - in fact my current 'up' time is around 190 days, and would be closer to 250 if I hadn't had a local electricity supply problem.

(2) Your SNR needs pinning to 6dB minimum - at your attenuation, anything much below that will guarantee line-drops as soon as night-time radio interference starts.

All this needs to be sorted via Plusnet, and relies on the individual you speak to there being experienced and not just reading from a crib-sheet.

I shall keep my eyes on this, but if you have questions, feel free to shout up.

[#8]

Thanks John,

Yes, "pinning" was something I had read about but forgotten.

Another was a Radius test from plusnet.

If applicable, are these done via opening a ticket or am I likely to get a response on this post to confirm them do you know?

I have now done the quiet line test and there is just a continual very faint but constant level white noise.

In the meantime I have used a MW portable, but although various noise issues around the house, as expected these days with smps everywhere, isolating each makes no difference.

My final test was to unplug everything pc wise that was mains connected, even the LAN cables from router. Leaving just the router running wifi and a laptop about 2M away on battery and wifi.

0.1dB difference better on main pc, although as I have seen from previous logs it varies every minute you take a reading.

Here are the 2 results:

9 March 1022 from laptop on wifi

- Uptime:	0 days 00:36:55
7 - Data rate:	944/10195
8 - Maximum data rate:	1352/11008
9 - Noise margin:	5.5/3.0
10 - Line attenuation:	21.8/40.0
12 - Data sent/received:	6.9M/191.5M

9 March 1028 back on main pc

6 - Uptime:	0 days 00:42:56
7 - Data rate:	944/10195
8 - Maximum data rate:	1352/11016
9 - Noise margin:	5.4/3.0
10 - Line attenuation:	21.8/40.0
12 - Data sent/received:	12.2M/323.6M

And another from a year ago.

8 March 2020 (time unknown)

6 - Uptime:	0 days 00:01:32
7 - Data rate:	952/8151
8 - Maximum data rate:	1364/8040
9 - Noise margin:	5.8/5.8
10 - Line attenuation:	22.2/40.5
12 - Data sent/received:	1.6M/6.5M

Thanks, Robin

[#9]

OK Robin - (1) + (2) These need requesting via either phone-call (although I will tag @Gandalf and see if he picks this up and reads the thread).

Your QLT does raise a small question - it should ideally be totally silent, but a very faint noise is not unknown.

Basically, as I see it, your upload is very good, so I'll ignore that, although it does appear slightly high. but your download SNR is too low. As above, that needs pinning at 6.0dB minimum.

Ignore the U/L noise margin difference, it is irrelevant to your current problem and as you say, it varies by a little bit anyway.

Final comment -for now - if you do succeed in getting your D/S SNR pinned to 6dB, your speed may reduce very slightly, but I am still of the opinion that there is some exchange-side problem, as noted previously.

[#10]

A further thought, Robin - which router are you using?

[#11]

Thanks John for both replies. I will try phoning.

My router is the PN Sagemcom 2704N. Running on a conventional (quiet) transformer psu.  Incoming dc has a few turns wrapped through a proper grade emc choke. Before I retired, I was a customer service engineer for an air traffic control communications manufacturer locally. My job was repairing and testing to component level the non warranty older tx and rx from worldwide, as well as RN & RAF. I have held a full radio amateur licence for 53 years. Hence my understanding of emc and good connections, but not how BT broadband works!

Robin

[#12]

Interesting job, Robin.

The reason I asked about your router was to see if would work with RouterStats, and the good news is that it will - at least I think the 'lite' version will. Might be an idea to down load and try it. Link to page: http://www.vwlowen.co.uk/internet/files.htm

Might give us a bit more to play with.

[#13]

Just spoke to CS and he is going to reset my margin level. Then for me to reset my router after four hours from now and try it again for a week. He was most helpful and understood what I was explaining.

Fingers crossed!

Robin

[#14]

Will try that.  (John?) Lowen developed quite a few variants for routers maybe at least 10 years ago. (He is a radio amateur as well.) I had one working with my old Belkin router. Back then I had a problem every night about 2100, loss of internet and his graph showed it as a massive drop in s/n. Took a few months back and forth BT/PN before a sfi was launched. The team arrived around 1930, tea and biscuits and observing my stats. As soon as it dropped, they went out with their MW radio. Not long after, they returned. Noise had been dfed to about 9 doors away. An ancient pc was switched on every evening, 2100ish. The psu was radiating tremendous interference. It had also affected a few other subs nearby as well. With a new psu, problem solved. It is back then when I detailed my experience on here how to trace interference affecting adsl. A couple of years later,  I had similar noise problems, confirmed by routerstats. This time it was my Belkin router smps under the desk!

Hence, my now always using a "proper" non smps.

Will try a download and see if I can remember how to drive it.

Robin

[#15]

Unfortunately F@ST 2704N is not listed. In router menu I tried 2704 Orange and 2504N Sky, but neither work.

Oh well, worth looking at.

Robin