Slow broadband question

I was round at some elderly friends last night. Their broadband has never been brilliant, around 4Mbps.

Last night I could see it was slow and did some TBB speed tests which showed download of around 1Mbps.

Looking at the router stats... (down / up)
Connection Speed: 1151Kbps / 819Kbps
Line attenuation: 50.1 / 28.6
Noise margin: 6.1dB / 9.8 dB

For an attenuation of 50dB Kitz gives 4832 to 5152Kbps depending on precise ADSL variant. Anything in this range would be perfectly satisfactory.

I rebooted the router, no difference.

They, like all(?) domestic Eclipse customers get unlimited BB so no "excess use slowdown".

Any thoughts on what may be the problem?

Could their ISP (Eclipse) have somehow locked them to a lower CS?

If not what?

WBC ADSL2+ service that has been banded by the DLM?

Andrew

With 6db noise margin, unlikely banding.

OP - have you done a quiet line test?

Looks like a very poorly performing line.

Are all phones / sky boxes etc filtered?

Are they using the master phone socket?

Did you try the test socket?

Have they acquired any new or replacement electrical or electronic kit since you last visited. Perhaps a smart TV or a Sky upgrade? The fact the downstream margin is in the 6dB area suggests a constant noise source.

To me the previous 4Mbps or so on that attenuation suggests a ring wire still connected at the master, but that of course isn't the major problem. Possibly a failed filter somewhere? Or something been added without one.

What is the router, in case we can get further stats. Such as error rates. The sync speeds suggest ADSL2/2+. Though interestingly the sync of 1151Kbps is remarkably close to the 1152Kbps af a fixed speed 1Mbps connection. Almost as though it has been banded at that. I don't know what effect that has on ADSL2+ noise margins though. I doubt if it is that, simply worth pointing out, as surely the downstream margin would be sky-high if it were?

Reply to all posts

In reply to a post by MrSaffron:

WBC ADSL2+ service that has been banded by the DLM?

Possible, I'll contact the ISP.

In reply to a post by lee111s:

With 6db noise margin, unlikely banding.

OP - have you done a quiet line test?

Not yet, it was getting late, and I didn't have a corded phone with me.

In reply to a post by ukhardy07:

Looks like a very poorly performing line.

Are all phones / sky boxes etc filtered?

Are they using the master phone socket?

Only 1 phone socket in the apartment. No Sky or any other devices apart from the same Panasonic DECT phone..

In reply to a post by RobertoS:

Did you try the test socket?

No, no time.

In reply to a post by RobertoS:

Have they acquired any new or replacement electrical or electronic kit since you last visited. Perhaps a smart TV or a Sky upgrade? The fact the downstream margin is in the 6dB area suggests a constant noise source.

No new kit. No Sky. Definitely no new (or Smart) TV.

In reply to a post by RobertoS:

To me the previous 4Mbps or so on that attenuation suggests a ring wire still connected at the master, but that of course isn't the major problem. Possibly a failed filter somewhere? Or something been added without one.

No new devices, a failed filter is a possibility. They use PLAs but have given fine service since I set them up for them last year. And they were getting ~4Mbps then.

In reply to a post by RobertoS:

What is the router, in case we can get further stats. Such as error rates. The sync speeds suggest ADSL2/2+. Though interestingly the sync of 1151Kbps is remarkably close to the 1152Kbps af a fixed speed 1Mbps connection. Almost as though it has been banded at that. I don't know what effect that has on ADSL2+ noise margins though. I doubt if it is that, simply worth pointing out, as surely the downstream margin would be sky-high if it were?

Router is Netgear DGN1000

Just a thought, when using the test socket obviously the PLAs will not be in use, but I would also look out for a difference in stats between them being still powered up and off.

I wonder also about the router PSU. And of course a router factory reset is needed as part of the diagnostic process. I can't easily look up accessing full stats at the moment but would expect telnet to work. Possibly needing the initial debug command in the CLI as given on this page, as well as the stats command. (Last two sections).

Just come off the phone to Eclipse (as usual got straight through to a real human being with good technical knowledge, no script jockey).

He checked the line and found it was ADSL2 and banded, though no apparent reason. He reckons that with their line length - 3-4km, they'd be better on ADSL2 (?), he's initiating the change so we'll see. He said the router should automatically pickup the changes and, hopefully, start running a bit quicker.

I've let the elderly couple know and they'll keep an eye on things to see if it does seem to be going quicker.

As ever Eclipse are so easy to contact and to talk to.

At that attenuation, almost the same as mine was, then I agree ADSL2 may be a bit better that ADSL2+. Possibly in the low hundreds of bps. Equally, it may be a fraction worse. Upstream is unaffected.

My preference if the DGN1000 works as per the 834 series would be to leave the BTW setting as ADSL2+ and force it down through telnet as on my previously linked page. On the grounds that if it is worse on ADSL2 then easy to revert. You can't force it up if BTW have it set to ADSL2.

I'm a bit chuffed I queried the 1151Kbps as suspect . (No other replies were present when I started that post).

Edit - But can we telnet into the DGN1000! Best done through Eclipse I think in the circs.

Edited by RobertoS (Tue 07-Feb-17 11:20:46)

Strange that the SNR is 6 though while banded though!

Unless of course the target margin is 3, but even then, that wouldn't free up 3meg.

But not impossible

See also this post. A lot there that could contribute to the banding and the 6.1dB combination.

Router factory reset would be an early step for me. Whatever Eclipse are doing.

Pretty sure DGN100 don't Telnet. It's not an 834.

An update.

I've just had an email from the couple. They've run 2 speed tests using the instructions I left with them last June . They show download of 4.54 and 4.46 Mbps; upload 0.62 and 0.61 Mbps. And their ping is 27 and 36. All so much better than last Monday.

I have to say that although the reason for their banding is impossible to determine, the ease with which I could contact Eclipse and get the issue resolved makes me appreciate the benefits of a more expensive ISP. I so rarely need to contact Eclipse - either for myself or others - but when I do it's a good experience.

I so rarely need to contact Eclipse - either for myself or others - but when I do it's a good experience.

I can more or less say the same for Plusnet but of course there are others who can't, don't know if I am in the minority or not.