Line Stats

Just over a year ago we had a fibre cabinet installed approximately 0.5 miles outside of the main village. Uptake has been good and it appears most people have now transitioned across to FTTC broadband.

A couple of weeks ago I was asked by a fellow resident which ISP I use, because he was looking to switch provider, mainly because he wasn't happy with the speed he was getting. When I asked a bit more, it began to sound like he was having issues with this line so I suggested we should monitor it for a while and see if that reveals anything.

He already had a Zyxel VMG8924-B10A router, so it was easy to hook up my Pi with DSLstats on it. While plugged in to the 5C master socket, testing to date has revealed three things:
1) The HLog plot has a big dip around tone 1670, see here.
2) The U2 band is not utilised (possibly related to item 1?).
3) Every so often the downstream SNRM would fluctuate severely, causing the broadband connection to drop.

I'll start with item 3. Following some troubleshooting, we think we have narrowed this down to Powerline adapters, but this is still to be confirmed. Essentially, each time he would stream content to a TV connected to one of these adapters, the fluctuations would start (mainly affecting the D3 band). We have now removed the adapters from the mains and all is good so far. These are the newer TP-Link adapters, so I'm planning to enable VDSL compatibility mode and see if that cures the issue. Unfortunately, in his property, the phone line comes in at the back of the house, and we don't know the cable routing to the master socket at the front of the house, so it could well be that the mains power cables run close to the phone cable, causing the interference. I'm hoping that with item 3 resolved, the DLM will soon start reducing the SNRM to get the download speed up (it's currently sitting at 8.9db 35Mbps, with an IP profile capped at 32Mbps, and interleaving is also high).

With regards to item 1, the HLog plot, from my limited research this seems to point to a bridge tap. Is that correct and is that the only thing it could be? Is it recognised by Openreach as a fault? If so, I guess it could be reported as such? The bitloading plot can be found here.

Thank you in advance for any informative feedback you are able to provide.

Yes, looks like a bridge tap to me. The Hlog is yuck, which would suggest that there is a major issue with the physical line. The dip is affecting the DS tones, so it shouldn't affect the U2 not being utilised, but it could be. Powerline adapters are known to cause issues. The VDSL compatibility mode should be on, so I would give that a try. Is the line rate exactly (give or take a few Kbps lower) 35 Mb? If so, then the line could be banded due to all the disconnections. It may be useful to paste some line stats from the router.

Edited by deleted (Tue 11-Dec-18 13:35:01)

Key question, was the hlog done with the powerline kit connected to the mains or not?

I'd have expected a bridge tap (which with some searching in the home can be eliminated or bypassed with the test socket) to affect a broader frequency range. The dig looks like a single noise source.

It may be there is noise issues and a bridge tap of course.

Thanks for the quick reply. The downstream speed is exactly 35000 kbps.... Let me know which line stats you'd like to see and I'll make them available.

I can confirm that the HLog plot was generated without the Powerline kit connected.

With regards to the house wiring, that is going to be challenge... The house is an old cottage which has been modified and extended on several occasions. The line comes in as an overhead wire at the back of the house, where there is one of those waterproof joints connecting the overhead wire to the house wire. At that point, it is a grey cable and disappears into the house. There is one other socket I found, which turned out to be upstream of the master socket, but this socket is not connected (gel crimps connecting the A and B wire). Both cables in that socket are white though, so there must be another joint upstream somewhere connecting the white cable to the grey cable coming into the house. Ideally, we'd perform a test from the waterproof joint outside the house, but I don't think it is a good idea for me to break into that joint .

Can you try connecting the router to the test socket, then repost the Hlog? That would eliminate if there's a noise source on/line fault with any internal wiring.

Just the line stats showing SNRM, attenuation, errors etc.

Tone 1670 is 7226 KHz I believe if you have a short wave radio tune in around that point and see if anything is being picked up (when the VDSL2 is turned off to avoid hearing the VDSL2 and same for powerline)

Shortwave radio stations have a much longer range than AM generally, no idea of ideal antenna length for that frequency but if phone wiring is not a twisted pair or has a bridge tap it will act just like the little wire antenna that many old clock radios had.

Connect to the test socket, sync, repost Hlog.

QLN and Bitloading would help.

I've never known powerlines to cause a bridge tap and fail to see how this would be possible without connecting the mains to the phone circuit.
Powerlines can cause the issues described in point 3.

I live in a brand new home and every new home I see is like mine.
They stick a power socket RIGHT NEXT TO the master socket.
Sensible idea as many people have modems and DECT phones and need power NEAR the phone socket.

I had such issues with powerlines. The live feed for this power socket ran parallel with the incoming OpenReach feed to the master socket.
Using powerlines sent FEC's from 100's/min to 100,000's/min and retx/LEFTERS would go crazy.

Removing this power socket and its wiring immediately cured my issues. The power socket is now about 1m from the OpenReach feed.

Powerlines can be nasty for VDSL2. This is even worse when the phone circuit runs parallel to the electrical circuit of the house.

Edited by j0hn83 (Tue 11-Dec-18 16:23:38)

Hmm, I wonder if most of the errors/SNRM drops my line experiences is related to the incoming power line running parallel to my master socket...

You'd think engineers who fit these sockets would realise that they're installing them next to a power line.

Edited by deleted (Tue 11-Dec-18 16:44:30)

Where others are talking about powerline kit they are referring to home plug adaptors e.g. https://amzn.to/2zSsyV3

Yeah, I know. Sorry, I wasn't clear. I do mean the incoming line, not powerline adapters.

The way to find out turn off the mains at fuse board run DSL kit and device to look at stats from another power source

Noise on the incoming power supply to a house is a long shot guess

Yeah, I just remembered that I did that and the FECs continued. The FECs/re-transmissions sometimes stop when nothing else turns off in the house. Weird.

Visited the property yesterday and generated the HLog, QLN and bitloading graphs from the test socket, see here. The Powerline adapters were not plugged in at that point, so the plots represent 'the best we can get'!

I then put the faceplate back on and repeated the plots, see here. As you can see, apart from the U2 section of the bitloading graph, they plots are very similar. This is as expected, as there are no sockets wired in downstream of the master socket.

We then reinstated the Powerline adapters and I started a big download while connected to the remote Powerline adapter. Even though I have enabled 'VDSL compatibility mode' on these adapters, the effect on the bitloading graph is obvious, see here.

As I said in one of my other posts, this is an old cottage which has been renovated and extended on several occasions. The Openreach cable comes in overhead at the back of the cottage, and the master socket is located at the front of the cottage, although we know this is unlikely to have been the original master socket location. From the test socket plots I think it is clear that the wiring isn't perfect in the first place and the Powerline testing suggests the wiring is too close to the mains electricity cabling.

As this is an issue with the wiring upstream of the master socket, would Openreach entertain the request of running a new cable from the joint outside the house to a new master socket close to this joint, i.e. minimise the cable length in the house? If not, it is just something we might have to do ourselves as the current setup is not fit for purpose .

Last but not least, line stats from mid-day yesterday:
Stats recorded 19 Dec 2018 12:00:21

DSLAM type / SW version: BDCM:0xa48c (164.140) / v0xa48c
Modem/router firmware: AnnexA version - A2pv6F039v.d26a
DSL mode: VDSL2 Profile 17a
Status: Showtime
Uptime: 3 days 16 hours 35 min 3 sec
Resyncs: 1 (since 14 Dec 2018 09:23:18)

Downstream Upstream
Line attenuation (dB): 20.6 0.0
Signal attenuation (dB): Not monitored
Connection speed (kbps): 34999 6209
SNR margin (dB): 8.4 5.4
Power (dBm): 12.6 5.1
Interleave depth: 16 1
INP: 44.00 0
G.INP: Enabled Not enabled
Vectoring status: 5 (VECT_UNCONFIGURED)

RSCorr/RS (%): 0.6518 0.4465
RSUnCorr/RS (%): 0.0000 0.0000
ES/hour: 0 9.90

That dip in your Hlog about tone 1600 should not be present, especially when connected to the test socket.

That's a job for OpenReach, as you've ruled out an internal wiring.

They charge to move the incoming feed, but if the issue is buried in the wall between the cable entering the property and the master socket they would be responsible for replacing this section.

They don't go ripping walls up, any defective cable is left in the wall and replaced.

Edited by j0hn83 (Thu 20-Dec-18 11:24:06)

Thanks for the feedback. How would we get this fault reported to Openreach? I suspect the ISP will simply say that the line is operating 'within tolerances'... FYI, the end user currently has the phone and broadband contract with John Lewis (Plusnet).

Call them up, ask for line tests.
Hopefully it fails.

Have you run a quiet line test with a corded telephone? Dial 17070 with the powerlines off.

The line appears to have bridge tap, it should fail the ISP tests and they should appoint an engineer from these tests failing.

If you are there when any engineer arrives, mention the Hlog (even better show them the graph).

I�d not bother mentioning the Hlog, it will mean sod all to most Openreach staff.

In reply to a post by Zarjaz:

I�d not bother mentioning the Hlog, it will mean sod all to most Openreach staff.

Perhaps those who install fibre may not.

Any decent engineer who does regular pair quality tests on copper circuits should know what an Hlog is.

If it's mentioned then it won't be left as is.

In reply to a post by j0hn83:

Have you run a quiet line test with a corded telephone? Dial 17070 with the powerlines off.

Yes, we did that initially and the line sounds fine (unfortunately I guess )

Any decent engineer who does regular pair quality tests on copper circuits should know what an Hlog is.

I disagree. Where within a PQ test does Hlog come into its results ?