Very poor download speed - 0.1Mbps bursting to 0.2Mbps

I'm regularly helping an elderly couple with their assorted computer problems. Was over there today to fix a browser issue - he'd unwittingly zoomed the font size to its minimum.

While there I had the opportunity for a quick look at their very slow broadband. They run 2 desktop PCs and occasionally a laptop. The desktops are connected via PowerLine adapters, the laptop via wireless.

I need to take my laptop over and connect it directly to the Netgear router to see if there's any significant difference in performance but the numbers from a desktop PC are not very good - to say the least.

TBB speedtest gives:
D/L: 0.1Mbps, tbb x 1, bursting to 0.2 Mbps
U/l: 0.56Mbps bursting to 0.63Mbps

Wow!

The connection data from the router (CS/LA/NM)
Down: 286 kbps / 50.7 db / 7.3 db
Up: 808 kbps / 29.5 db / 9.4 db

With a Line Attenuation of 50.7 Kitz suggests a DSLMax of 4832 kbps, even the worst figure in the Kitz results is 4000 kbps, around 14 times better than they're getting.

The PLAs may be losing a small amount of speed but given their connection speed any PLA loss is irrelevant.

Clearly there's something seriously wrong and I need to go there on a working weekday so I can talk to Eclipse, their ISP. I'm well aware that there are issues from at least some domestic users that Eclipse is limiting speeds, but I'm with Eclipse and get 7-8Mbps with LA of 41.5 and no real complaints.

Could it be that somehow their line is capped? They live in a block of retirement apartments built within the last 10 years or so.

Thoughts on what the possible issues might be would be welcome.

I think it's all just conjecture until you see the router stats.

You should also investigate and then describe the set up of the line and extension wiring.

Take a small cross headed screwdriver with you, that way, once you have noted initial stats 'as is', then you can test again with the router in the test jack of the NTE.

I've posted the router stats. It's connected to the only socket in the apartment, no extension wiring, and the only other connected device is a DECT phone.

In reply to a post by cheshire_man:

I've posted the router stats. It's connected to the only socket in the apartment, no extension wiring, and the only other connected device is a DECT phone.

Sorry, yes you have, Bank Holiday afternoon innit.
At a guess DLM has capped the sync rate down, highway you need to establish is why ? Is the line noise free ?
If it is capped I would have expected to see a higher downstream SNR.

Haven't done a quiet line test with a corded phone - will take one with when I next go over - I can hear no noise when I ring them, but that's by no means conclusive.

I clearly need to talk to Eclipse to see if the line is capped in some way.

Those stats look very wrong. The upload speed is about what you'd expect (perhaps a tad low), but the download is dreadful. It is not consistent with the attenuation. A few possibilities come to mind

1) there is still something wrong with the domestic wiring. I know you've said it's a single socket with no extensions, but it's as well to absolutely double check everything and trace it back to where the phone line enters the premises. Downloads speeds are much more vulnerable to bad wiring that upload.

2) The capping thing has been known (at line rates). There is a fixed one that matches what was the old fixed 256kbps. I seem to recall it was 288kbps, but could be wrong. It's a possibility, but I think the SNR margin is to low to explain a discrepancy of approaching 4mbps. I'd have expected the downstream SNR margin to be possibly into double figures.

3) There's a lot of noise on the line. Is the voice clear? Is the line stable, does it drop out a lot (there may be a log file showing resyncs - if it does resync, is it consistently bad?).

4) Possibly the router/modem is faulty, or the cable or microfilter. If you've got access to a different ADSL modem/router, then it's a good idea to get that to sync up and see if it gets the same results. Also, whilst you're at it, make sure the router firmware is up-to-date if that's user updateable.

5) there could be an issue in the OR network, possibly a bridge tap (which OR shouldn't have) or other issue.

6) it could be a linecard fault (no easy way to check that)

Also, has it always been this slow? If it's deteriorated suddenly then that suggests an emerging fault.

If everything else fails, then it would have to be reported to the ISP. If it's that far from expectations (and I'd check the expected rate from the BT Wholesale checker if that's possible), it need an engineer visit.

I have seen similar oddities on a sky adsl line, and that was the modem (netgear,) failing *

*failed capacitors

In reply to a post by TheEulerID:

Those stats look very wrong. The upload speed is about what you'd expect (perhaps a tad low), but the download is dreadful. It is not consistent with the attenuation. A few possibilities come to mind

1) there is still something wrong with the domestic wiring. I know you've said it's a single socket with no extensions, but it's as well to absolutely double check everything and trace it back to where the phone line enters the premises. Downloads speeds are much more vulnerable to bad wiring that upload.

As it's a block of retirement apartments goodness knows where it enters the building. I can only go by what I can see and measure in my friends' apartment.

In reply to a post by TheEulerID:

2) The capping thing has been known (at line rates). There is a fixed one that matches what was the old fixed 256kbps. I seem to recall it was 288kbps, but could be wrong. It's a possibility, but I think the SNR margin is to low to explain a discrepancy of approaching 4mbps. I'd have expected the downstream SNR margin to be possibly into double figures.

3) There's a lot of noise on the line. Is the voice clear? Is the line stable, does it drop out a lot (there may be a log file showing resyncs - if it does resync, is it consistently bad?).

I've never heard them mention drop outs or similar. When I looked at the stats today while I didn't explicitly note the uptime I saw it and I have no recollection of it being a low figure. I've never heard noise on the line but will do a quiet line test with a corded phone when I can.

In reply to a post by TheEulerID:

4) Possibly the router/modem is faulty, or the cable or microfilter. If you've got access to a different ADSL modem/router, then it's a good idea to get that to sync up and see if it gets the same results. Also, whilst you're at it, make sure the router firmware is up-to-date if that's user updateable.

I can try a different microfilter and cable, and also an old non-wireless Netgear router that I keep as a standby and that performs well on my line.

In reply to a post by TheEulerID:

5) there could be an issue in the OR network, possibly a bridge tap (which OR shouldn't have) or other issue.

6) it could be a linecard fault (no easy way to check that)

Also, has it always been this slow? If it's deteriorated suddenly then that suggests an emerging fault.

They only moved into the apartment in January 2015 so can't really say for certain one way or another. I have recollection of talking to Eclipse about their line speed and I think it was from their new home rather than their old one, and Eclipse tweaking something that gave a speed increase; but to be honest I can't be absolutely certain on this. The ISP, router and hardware (not the PLAs) are the same as at their old home, less than 1 mile away.

In reply to a post by TheEulerID:

If everything else fails, then it would have to be reported to the ISP. If it's that far from expectations (and I'd check the expected rate from the BT Wholesale checker if that's possible), it need an engineer visit.

I've emailed them to see if they can run the BT Wholesale checker but it may be too much for them.

To be clear, by BT Wholesale Checker I'm talking about the availability checker which gives a speed estimate (although it doesn't seem to be working just at the moment). You should be able to run that yourself.

https://www.btwholesale.com/includes/adsl/main.html

possibly a bridge tap

A bridge tap is going to be more of a problem on an FTTC service ...... and in it's true sense as a fault, I agree, you don't get them on Openreach services.

Ah, I'll let them run the one I pointed them to, a BT Wholesale Speed Test, if they're able. The results from the one you pointed to are, in the ADSL products section (all showing available) are:

Product         Down         Up        Down Range
-------         ----         --        ----------
WBC ADSL 2+     Up to 2.5    -         1 to 4
ADSL Max        Up to 2      -         1 to 3.5
WBC Fixed Rate  1            -         -
Fixed Rate      1            -         -

I wonder where Cab 18 is on the Glossop (MRGLO) exchange.

In reply to a post by cheshire_man:

I wonder where Cab 18 is on the Glossop (MRGLO) exchange.

Type in the property postcode here and bobs yer uncle

https://windows.mouselike.org/be/index.asp?DoAction=...

Thanks for that. They're about 0.3 miles from the cab. It's the cab to exchange distance that's the killer. Fibre - which is available - would be a real benefit to them but I doubt they could be persuaded.

What are the FTTC estimates?

FTTC may not sort out the problem anyway. It depends where the obvious fault lies. I say there is obviously a fault because of the attenuation and the Fixed Rate figure of 1Mbps. In effect that is guaranteed, as it either sync's at 1152kbps or not at all. In the circumstances, probably not at all.

I forget, when was the router last rebooted?

Distance from the exchange is not the killer. At 50dB they should connect above 4Mbps. I forget the approximate figure and not got easy access to the graphs right now.

Edited by RobertoS (Tue 31-May-16 09:49:31)

In reply to a post by RobertoS:

What are the FTTC estimates?

As requested

I see they changed the checker to refer to VDSL (and G.Fast) instead of FTTC.

Looks like VDSL would solve the problem

It'll get the data to Cab 18 more cleanly. Depends on where the loss is really occurring between the theoretical 4+Mbps the LA indicates, and the 0.1-0.2Mbps they actually get.

To be honest, if they were able to get 3-4Mbps I think they'd be delighted, after all that's ~20 times faster than at present!

It'll probably be a week or two before there's a mutually convenient date to go there to progress.

An update - replying to myself for convenience.

I'm planning to go over next Wednesday armed with various bits of kit and the collective wisdom from here and elsewhere to progress.

This afternoon I decided to phone Eclipse Technical Support to see if they were able (and willing from me, though a customer, not the one with the issue) to do any investigation.

Got through in 5-6 minutes and detailed the issue and currently available information, especially Connection Speed and Line Attenuation, and wondered if he was able to see if, for example, there was some sort of cap on the line. He said he'd get it checked and got back in a minute or so to say it was capped and that they'd requested a reset. I'll get the couple to do another speed test if possible in a couple of hours to see if there's any difference. Give the chap his due, he was very helpful and, I hope, effective.

I'll probably still go over next week and do more checks. It may be there's noise on the line, the router's near end of life, or the filter or cable, whatever is dodgy.

Maybe there's some progress.

I'm at their place now. Have got the D/L CS to just over 1100Kbps, just about usable, though not right for LA of 49dB. But certainly much better.

I've changed the dangly filter and, currently, running off the test socket in the NTE5 wall box. I've tried a spare router of mine but that made no difference.

Looking at the faceplate connections there are 3 wires: 2-blue, 3 orange, 4 not connected, 5 white/blue. One of these is, I presume, the bell wire and needn't be connected. There's no connections visible to any other sockets, indeed there are no other sockets in the apartment.

I'll leave it 30 minutes or so to avoid upsetting the DLM and then replace the face plate to see if it slows down.

Would it be worth removing the bell wire? If so which wire?

ring wire is pin 3

T3 is the bell wire, orange. Best to remove it as it can't do any harm, unless!

Some very old phones used at extensions need it to make them ring, and a very few poor design modern ones. However, if removing it stops anything ringing, then it either isn't on a filter; or the filter has failed; or it's a cheapo filter that doesn't trigger the ring tone. Nearly all filters do trigger the ringing.

Do they happen to have a burglar alarm?

Just reconnected the NTE5 face plate and the D/L CS has halved.

What on earth is happening?

Their only phone is a Panasonic DECT system less than 2 years old.

No alarm, no Sky, nothing but that DECT phone.

They have an entryphone to the apartment block, I doubt that uses the phone wiring, or could it?

You said there were no visible connections to any other sockets but that there are connections on the back of the plate so there must be an extension somewhere. If the extension isn't in use then you could remove all of the wires from the front part of the socket and see if that improves things. It sounds like there could be something unfiltered at the other end of that wire.

If I understand you correctly, the wires on T2,3,5 can only be to feed some extension. But there's no sign of an extension in the apartment. And certainly nothing in use.

Presumably therefore the single socket on the front of the face plate is directly connected to the test socket via the internal plug. Those 3 wires then being a feed to elsewhere.

Or have I misunderstood it?

Edit: The 3 wires are connected via IDC connectors, not screw ones.

Edited by cheshire_man (Wed 08-Jun-16 13:02:45)

You said there were no visible connections to any other sockets but that there are connections on the back of the plate so there must be an extension somewhere. If the extension isn't in use then you could remove all of the wires from the front part of the socket and see if that improves things. It sounds like there could be something unfiltered at the other end of that wire.

Agreed.

So is this a 'normal" NTE, with a detachable lower front portion ? If so you should have dial tone in the (now) exposed test jack and all the extension wiring should be terminated on the frontplate you have removed.

You mention an entry phone system, ask the owner if it rings through to the landline when buzzed downstairs. It's highly unlikely, but I have seen it a couple of times over the years.

Any chance of seeing the full stats being displayed by the router ? SNR error counts and the like . I wonder if this is REIN, ?

In reply to a post by Zarjaz:

So is this a 'normal" NTE, with a detachable lower front portion ? If so you should have dial tone in the (now) exposed test jack and all the extension wiring should be terminated on the frontplate you have removed.

Correct

In reply to a post by Zarjaz:

You mention an entry phone system, ask the owner if it rings through to the landline when buzzed downstairs. It's highly unlikely, but I have seen it a couple of times over the years.

It does not. It rings a small wall mounted unit which has a door release button.

In reply to a post by Zarjaz:

Any chance of seeing the full stats being displayed by the router ? SNR error counts and the like . I wonder if this is REIN, ?

Chuckled at that, it's tipping it down currently and the thunder has just started

It's a Netgear DGN1000, I'm trying to find out how get those stats.

Mind you, the CS using the test socket is now 1240Mbps, a fourfold improvement, and it's usable at that speed.

In reply to a post by cheshire_man:

Just reconnected the NTE5 face plate and the D/L CS has halved.

What on earth is happening?

Either the connections on the plug part of the faceplate are slightly corroded, or those on the inside of its corresponding socket.

There's also a chance, if its the plug part, that the contacts inside the test socket aren't free of corrosion. Accounting for not getting full speed with a dangly. Is there any sign of damp around the master socket or perhaps on the other side of the wall it is mounted on?

I suggest not trying to clean anything by conventional means, but trying a few insertions and removal of plugs into the two sockets. That can clear contamination. Without the router connected while you do that!!

Edited by RobertoS (Wed 08-Jun-16 13:55:39)

Avoid ten disconnects/reconnects in an hour.

If the wiring on the removable small part of the faceplate is still there then that is the issue.

i.e. no need for it if there are no other extensions (probably is one somewhere but hidden behind a blank socket or furniture.

A photo to confirm everyone is talking about the same components will help too.

I've now come back home as they needed to get on with other things.

I've left it with the dangly connected to the test socket and the lower face plate gently taped on top for tidiness. The socket is amongst a wall of book shelves and is quite safe from unwanted attentions.

They're now talking about contacting BT to install an extension socket in the second bedroom/study. If so they'll get the engineer to sort out that wiring. And - hopefully - to install a socket with built-in filter. If they decide to go for that I've offered to be there when the engineer's due to be there.

When they bought the apartment February 2015 and moved in it was completely empty. They'd have seen another socket if there was, though it's possible there's a blanking plate somewhere.

Whatever the situation their phone, PCs, are all working very well. The entryphone also works fine.

There's still a question as to why with the DS LA they have the DL CS is only 1240Kbps, even though that's some 4 times the earlier figure.

The replies today while I was there were very helpful, thank you. And no, I was careful not to disrupt the router more than 3-4 times in any hour.

The sync speed may be banded i.e. to avoid it flapping around too often so if all is stable now given a long period of good it may go up further.

Sounds good. That wire on the back of the faceplate as you have described it is going somewhere and however it is terminated is likely to be what has been causing the issues (at least that would be my guess).

In reply to a post by MrSaffron:

The sync speed may be banded i.e. to avoid it flapping around too often so if all is stable now given a long period of good it may go up further.

I wondered if that might be the case and have asked them to do a speed test every couple of days and see if it improves.

In reply to a post by ian72:

Sounds good. That wire on the back of the faceplate as you have described it is going somewhere and however it is terminated is likely to be what has been causing the issues (at least that would be my guess).

The apartments are 10 years old so who knows what previous owners may have done...

Certainly by comparison with the situation at the start it's like an E-type compared to a bubble car. Their work is not bandwidth intensive, emails, browsing, that sort of thing. So 1-1.2Mbps is fine, even if it's shared between 2 PCs.

BTW, anyone know the BT charge for fitting an extension socket?

Edited by cheshire_man (Wed 08-Jun-16 16:44:46)

BTW, anyone know the BT charge for fitting an extension socket?

Expensive, is the answer. Why do they need an extension anyway ?

Thanks for the comment .

To run the computers in their study. The current arrangement using PLAs works but there is a real mess of mains & LV cables, flexes, as well as phone & data cables, in one small corner. Eliminating the data cabling there would help to tidy things up. They have a lot of mains powered devices and far too few sockets, consequently at least 3 or 4 4-way extension boards in both the lounge & study. The apartment's lease is very clear on wall disruption so fitting additional sockets is not feasible.

Using WiFi isn't an option, went from WiFi to PLAs because of WiFi problems and neither I nor they wish to go through that lot again!

I may have to go over again and unplug all their devices and see what can be done to re-route things. They do want all the extension sockets individually switched as they turn all things off when not being used, monitors, lights, etc. Even devices that power down to an almost nothing standby. Nothing I've said about standby and power cycling wear and tear has been able to change that practice.

They do want all the extension sockets individually switched as they turn all things off when not being used, monitors, lights, etc.

Do they regularly turn off the power to their router ? This can sometimes* be interpreted by DLM as an issue and sync rates capped as a result.

Why not run straight CAT5 extensions, no power consumption needed. Ease of use too.




*as in very, very occasionally.

No the router is left one.

I've tried to get them to go Cat5 but to no avail.

I've tried to get them to go Cat5 but to no avail.

There's no helping some folk [sigh]

Then they don't want an extension socket and BT are not the right people.

What they need is an Ethernet cable installed with a socket (which doesn't have to be wall mounted) where they have the computer (ideally run several cables as it is always good to have a spare or two). A good sparky that has Ethernet experience could run this (but do check they know what they are doing as they should properly test the cable after install to make sure all connections are good).

As they aren't using the telephone extension then those wires to terminals 2 and 5 on the master socket can be removed as well - that may even be enough to fix the issue they have been getting (and is essentially what you have done by using the test socket).