Different speedtests

Hello,

New to this forum; apologies if this post is in the wrong place. As you will see, I don't understand very much about these things.

I live in a part of West Sussex that suffers very low broadband speeds. If I try to view a 2 minute video clip on the BT news website, it generally stops to reload every 15 seconds. If I run the TB speed test, I usually get 0.2-0.7 Mbps; very occasionally up to 1.4. BT exchange is non-LLU. I buy all my phone services from 24Talk.

Ran TB speed test last night, got 0.4. BT speedtest produced 1.3Mbps. Am I comparing like with like? If so, how do I interpret the difference? If not, what am I to understand?

Would be grateful for any answers, in simple non-tech language!

thanks
veryslow

First step is to ignore the speed tests and check what speed your ADSL hardware is connecting at. This gives you the maximum technical speed you will ever get, and helps you to judge speedtests.

As things stand you could be connecting at 4 Mbps, so all those results are poor.

As a rule the BT Speedtest bypasses a chunk of the providers network, and thus avoids some congestion giving better results sometimes.

That's a bit of a difficult question, as some of us get very similar results from the two, others get big differences like you describe.

What might help, and might also help us improve your connection a bit, would be if you can give us a bit more information, as below:-

Please run a test using www.speedtest.net and post the link that gives to share the result. That will tell us what underlying ISP 24Talk are using, as they may just be a reseller.

And see if your router is listed here. If it is, try to follow the instructions and copy/paste the stats you get out. But if that looks difficult, or it isn't listed, tell us what it is and we'll see if there's an easier way.

Have you got a phone socket that has a split front, where the bottom half looks like it will come of if two screws are removed? So it would look the one on the left here? If you have, is that the one you use for the broadband, without a plug-in extension cable? Just the ADSL cable that came with your router.

My router is a Netgear.

ADSL Link Downstream Upstream
Connection Speed 1984 kbps 448 kbps
Line Attenuation 63.5 db 31.5 db
Noise Margin 9.6 db 10.0 db

How do I relate this to speedtests?

Speedtest.net result =
http://www.speedtest.net/result/2423746915.png

Router stats in post above.

Phone sockets are not of the split front type. Master looks like LJ 2/1A as in pic. I'm connected to an extension.

Great - thanks for that bit.

To relate that to speed tests, the first thing is to read about IP Profiles. You can see from the table there that yours will be 1500kbps, so a speed test result of 1.3Mbps is about right.

In my previous post I said about us trying to improve your connection. At the level you are at we don't need to get you much to get a big percentage jump in the real throughput.

See if you can do the other things I asked.

Got to go for a bit, but there are two things we can look at.

Even though you don't have the modern master socket, as long as you are careful, even though you say you are non-techie, it is very likely a ten-minute job at each of your sockets could add several hundred kbps to your connection speed. By several hundred, I'm meaning almost certainly 500, and anywhere from that to the exceptionally high over 2Mbps we have seen. I'l get back to you.

This could also make you line a bit more stable, as that 9.6dB noise margin points to a bit of instabilty. That's maybe inevitable with your long line though. The normal value is around 6dB, and that alone accounts for 500-750kbps.

So watch this space . In the meantime have a look at the insides pictures here of the LJ 2/1A and LJ 2/3A.

Thanks for all the comments so far. I don't know if my phone circuitry is non-standard or whether this might make a difference? Brief description:

BT metal box outside has BT cable coming in. From there 2 cables:
1) cable (via outside of house) to inside upstairs socket (the one connected to router); and on to an extension from this one.

2) wiring to a junction box inside. From there 2 more cables:
2a) wiring to a socket - BT master? has a T on the front
2b) wiring to a second socket upstairs.

Well it wouldn't be done that way for a new installation these days, but nothing to suggest a problem so far as I know.

Have a read about the ring wire. That's what we need to deal with, just that the instructions there are for the modern NTE5 master.

What I would do is disconnect absolutely everything from the phone line for a few minutes and go round each of those sockets in turn, open them up, and gently remove the wire from T3. (It should be numbered). If you have a pair of small-nose pliers they should lift the wire out easily. It is best not to cut them. Make sure you don't connect two terminals together accidentally with the pliers.

Coil the spare bit a little to keep things tidy, and put the faceplates back on. make a note of which is the master - see the cylindrical capacitor at the top in the picture. Though it may not be at the top.

Then connect only the broadband filter and router at the usual socket and read/post the stats straight after. Note that whatever the stats say - hopefully a decent improvement, your download speed won't change for a few days. Then connect whatever you have back on, one at a time! Re-read the stats after each one is added, but no need to post them. Don't add them all at one go. We need to see if anything hits the noise margin, so that's what to look for in the successive sets.

if the box outside splits the wiring into a T circuit with things on each end then you're going to get less than full speed, ideally you want the incoming line to go right to where your router connects and anything else off there to be filtered.

From your description I think it goes

Text
1 23	incoming --- junction box  -- (1) ------ router socket ----- extension socket                   |------------(2)------- junction box ------ socket with T                                           |------- second socket upstairs

if these all have dial tone and are on the same line you have a chance to improve it.

yarwell, yes, you have described it correctly.

RobertoS, thanks, I will do as you suggest; will be tomorrow morning.

Just switched on after several hours away - thinkbroadband speedtest = 61 Kbps (yes, really!), BT produced 490 Kbps. Don't think I'll be doing much surfing tonight.

OK, I did as RobertoS suggested.

TB speedtest result before doing anything = 0.7 Mbps. It's usually better morning than evening.
Router =
ADSL Link Downstream Upstream
Connection Speed 1792 kbps 448 kbps
Line Attenuation 63.5 db 31.5 db
Noise Margin 10.4 db 10.0 db

I unplugged everything.

Discovered I had forgotten one socket from the list posted above. It's first in line on (1) so this bit should read
1) cable (via outside of house) to downstairs socket; then to upstairs socket (the one connected to router); and on to an extension from this one.
This wiring was installed for me a few years ago by a friendly telecomms engineer (not employed by BT). All three sockets had no T3 connection. All three have a small cylindrical object inside at the top (as in my master but smaller).

The wiring to the junction box, master and extension (2 above) is older; I guess original (house built 1982). 2a does look like a master. Both had T3 connected; I have now removed these.

Rebooted router, reconnected ADSL cable + filter, restarted pc.
Router =
ADSL Link Downstream Upstream
Connection Speed 2368 kbps 448 kbps
Line Attenuation 63.5 db 31.5 db
Noise Margin 9.3 db 11.0 db

TB = 1.43 Mbps. I have see this level just twice before in the last 6 weeks, both morning.
BT = 1.48

Plugged 3 things (2 sky boxes + 1 phone with 3 digital cordless handsets) back in and took a router reading after each one. Router stats exactly the same after each one.

What do I conclude?

I would conclude that the wiring was dropping the speed from

Connection Speed 2368 kbps 448 kbps

to

Connection Speed 1792 kbps 448 kbps

The former will give you a 2M IP profile. Did you restart the router or its ADSL modem after each change - you only see the full effect of a wiring change if the modem reconnects to the DSLAM under the new conditions.

ETA: I wasn't clear if you removed the excess master sockets and wiring or just took the wire off T3 ?

Edited by yarwell (Thu 10-Jan-13 11:09:36)

In reply to a post by veryslow:

This wiring was installed for me a few years ago by a friendly telecomms engineer (not employed by BT). All three sockets had no T3 connection. All three have a small cylindrical object inside at the top (as in my master but smaller).

What do I conclude?

I will leave Roberto to comment on the stats &c but have one comment to make.

COWBOY ...

In reply to a post by MHC:

In reply to a post by veryslow:

This wiring was installed for me a few years ago by a friendly telecomms engineer (not employed by BT). All three sockets had no T3 connection. All three have a small cylindrical object inside at the top (as in my master but smaller).

What do I conclude?

I will leave Roberto to comment on the stats &c but have one comment to make.

COWBOY ...

RobertoS hasn't time to think about the stats right now, though wouldn't disagree with yarwell's post.

Your final word there is exactly what I would say about the chap who did the wiring. There's only one way to go with that, and that's replace the lot. I wouldn't even guarantee the wires themselves are suitable.

Dare we ask for the colours of the wire and which ones are connected to which terminals!

Well we could ask veryslow .

OK, back for a minute.

I misread your latest post at first. I think we have a good result there.

Re my approval of yarwell's post and my previous one, I don't think I do quite agree about rebooting the router after each item was reconnected. That might trigger a noise margin rise by the DLM unless the reboots had a decent time gap between them. IT might have been best when you did it, but I wouldn't bother redoing it, seeing as the noise margin didn't change.

The wiring certainly is a mess, but how much improvement you could get by having it all redone is debatable. Probably not enough to justify the expense for the moment.

I think we should leave it as is for now and see if that noise margin is lowered by the system in 10-14 days time. So unless anything goes downhill before then I recommend leaving the router on until the end of that.

If the DLM does lower it, it normally causes a re-sync anyway, often overnight.

If you find that has happened, a single daytime reboot with stats taken immediately before and after like you just did is what to do. (No need to disconnect anything).

I would be wary of the Sky boxes. Particularly if they are old ones. They can cause trouble. A few recommendations.

Do a quiet line test now. 17070, option 2. It should be effectively silent. If not, unplug the Sky boxes and try again. (I'm ssuming they are powered up).

If they are off, check the stats before and after switching them on for the first time after now.

Thanks again for all the helpful comments.

On the wiring, it was/is not practical to run the new cabling from the BT master socket. May I ask, out of ignorance again, what causes the concern? RobertoS said that the T3 connection was redundant nowadays, so T3 would not have been connected? I will happily identify colours and terminal connections tomorrow and post again.

RobertoS said

I think we should leave it as is for now and see if that noise margin is lowered by the system in 10-14 days time. So unless anything goes downhill before then I recommend leaving the router on until the end of that.

I will certainly do that and let you know.

I don't understand

Do a quiet line test now. 17070, option 2. It should be effectively silent.

. The sky boxes are 14 months and 2 months old respectively. And they were on at the time of testing.

I've just done speedtests. TB = 0.23. BT = 1.05. 2 minute video clip on BBC news website stops to reload every 15 seconds - so no improvement in real performance.

In reply to a post by veryslow:

On the wiring, it was/is not practical to run the new cabling from the BT master socket. May I ask, out of ignorance again, what causes the concern? RobertoS said that the T3 connection was redundant nowadays, so T3 would not have been connected? I will happily identify colours and terminal connections tomorrow and post again.

From your description of capacitors in every location, it suggests that they are all masters. Every master has a 1.8 uF capacitor and 470k resistor in series across A to B legs. That can put additional loading on both voice and ADSL/VDSL signals. If the installer did not understand the need to have just one master then it is likely he did not understand the need to use the right cable. Telephony cable is made up of 2 or 3 twisted pairs (with different twists) and that help in the rejection of electrical noise. Unfortunately some installers use Burglar Alarm cable because it is cheaper - not twisted and also finer. That results in possible intermittent connections and noise pick up. The noise may not be an issue on the voice side (but it can be) however it will almost certainly have an effect on ADSL/VDSL. Just by the colours it is possible to tell the type of cable and if it is telephony grade that a pair has been used.

Previously and separately I had contacted 24Talk.They have just come back to me again. This is a summary of the exchanges.

20/12/12 I reported low TB speedtest results for the last month.

27/12/12 No response yet. I reported more low TB speedtest results.

07/01/13 No response yet. I chased.

08/01/13 Response: "Thank you for your email our engineers believe they have resolved this issue. Please can you reboot your router and pc and then once they are logged back on to the internet run 5 speed tests using www.speedtest.btwholesale.com."

08/01/13 I gave BT results 1.49Mbps, 1.32, 1.19, 1.38, 1.19 and TB 0.4. I said "It still feels slow. Watching a 2 minute video clip on BBC news site, it stopped to reload every 14 seconds."

These different speedtest results prompted me to start the current thread here - see very first post.

10/01/13 24Talk responded
Our carrier has asked for the following checks to be completed as they can find no fault:
1. Check the speed after disabling Firewall or Antivirus if any, installed on the system.
2. Test at main BT socket, remove extension cables if any, swap filters and check the entire setup.
3. Optimize browser, delete cookies, Temporary Internet file, and scan with anti-virus for virus and spyware removal.
And also check speed at http://www.speedtester.bt.com at three different times and get back to us with the details.
Note:--> According to BT Wholesale, your phone line should be able to support a 1Mbps or greater ADSL connection"
The speed tests you have already returned are on or near 1Mbps.
Please let us know the results within 72 hours to avoid the ticket being closed by our carrier.

10/01/13 I replied:
1. Sorry, I don't know how to disable Firewall and antivirus.
2. Difficult given location of pc, router and master socket. I will try to do this Friday.
3. I already do this frequently.
http://www.speedtester.bt.com results:
08/01/13 2000 = 1.3 Mbps
09/01/13 1830 = 0.49
10/01/13 1030 = 1.48
10/01/13 2050 = 1.05
Right now, 2115, and every evening, a 2 minute video clip on BBC news website stops to reload every 15 seconds. 1Mbps should be better than this? This is what I'm experiencing, in practice, every day.

I will dismantle my system Friday evening and test performance directly via the BT master socket - unless you guys advise me otherwise.

MHC has explained about the poor wiring. There should be only one master, but the Openreach engineers who post here don't get too upset if there are two. I imagine if any of them see your description they will not be impressed .

As MHC says, the actual cabling used is just as important.

Re real performance increases, do you remember about IP Profiles, see the link in this earlier post of mine. That explains that a real speed increase could take a few days, and the table there gives you an idea of how big the increase could be.

Re the Quiet line test, whilst audible noise doesn't affect broadband signals, if it is present there is very often broadband frequency noise as well. Sky boxes, even new ones (though the older ones were much worse), can emit broadband and audible noise, as can anything - such as a DECT phone with a faulty power supply. (The bulky power plug).

This is why I asked you to add items back one at a time and check the stats after each. I hope you checked all the stats, not just the connection speed, as that would not change. The noise margin was the key thing when doing that.

In reply to a post by veryslow:

I will dismantle my system Friday evening and test performance directly via the BT master socket - unless you guys advise me otherwise.

I'm not sure it's worth it, but probably best to do what they say. Just so you can say you've done that. Given the fairly useless responses from them you just posted, (my post above was waiting to go for half an hour as a friend rang, so I hadn't seen this one), mainly stock and occcasionally wrong responses from a script, plus the blatant cop-out 1Mbps estimate quote, I'd be thinking about leaving on principle, never mind practicality.

On the other hand, all might now be a bit better in a few days. So might become irrelevant.

I forget, have we asked how much you are paying for phone and broadband, and what usage allowance you have?

Edited by RobertoS (Thu 10-Jan-13 22:20:05)

if you don't retrain the modem you won't see the benefit of wiring changes - frequencies that could not be used initially that become available can't be used without a retrain and the noise margin only picks up frequencies that are in use.

May I ask, out of ignorance again, what causes the concern?

Each master socket provides an AC signal path from t3 of attached wiring or plugged in equipment onto t2, via its capacitor. So having multiple masters provides a set of antennae to pick up interference like AM radio and feed it into your wiring.

Having wiring that is in a T configuration with the modem on a branch is less than optimal, allows reflections or signal loss to the unused (for ADSL) parts. Google for "bridge tap" which describes what this does when it occurs in the telephone network (usually US) - it degrades ADSL performance.

So I would re-state that the best solution for any line, especially a marginal one, is to terminate the incoming line at a socket that feeds the modem and anything else is connected via a filter to that socket. In this way no unnecessary wiring is exposed to the ADSL signal or vice versa.

Answers to some of the questions posed earlier:

Yarwell said "I wasn't clear if you removed the excess master sockets and wiring or just took the wire off T3 ?" - I just took the wire off T3.

MHC asked "Dare we ask for the colours of the wire and which ones are connected to which terminals!"
T2 = white/blue
T5 = blue/white
Not connected:
orange
white/orange
green/white
white/green
brown/white
white/brown

RobertS, on the Sky boxes, I didn't understand how/where to do the quite line test "Do a quiet line test now. 17070, option 2" - is this via the sky box menu?

RobertoS said "Re real performance increases, do you remember about IP Profiles, see the link in this earlier post of mine. That explains that a real speed increase could take a few days, and the table there gives you an idea of how big the increase could be." - I will be patient!

RobertoS said "This is why I asked you to add items back one at a time and check the stats after each. I hope you checked all the stats, not just the connection speed, as that would not change. The noise margin was the key thing when doing that." - Yes, I did. Line attenuation and noise margin were each the same after every reconnection.

RobertoS asked "I forget, have we asked how much you are paying for phone and broadband, and what usage allowance you have?"
Last year I was paying monthly (all incl vat)
Line rental = £7.30
Broadband = £13.00
10GB limit
Prices have just increased but it's not completely clear from latest bill what the new charges are.
I also have calls with 24Talk.
In November I agreed a further 18 month commitment.


Yarwell said "if you don't retrain the modem you won't see the benefit of wiring changes" - is that the same as "reboot"?

RobertS, on the Sky boxes, I didn't understand how/where to do the quite line test "Do a quiet line test now. 17070, option 2" - is this via the sky box menu?

No, your phone.

Rebooting the router is one way to retrain the modem on the line, you may also have options on the graphical interface to it.

In reply to a post by yarwell:

if you don't retrain the modem you won't see the benefit of wiring changes - frequencies that could not be used initially that become available can't be used without a retrain and the noise margin only picks up frequencies that are in use.

Errrm - yes. But if the modem is connected and noisy items are added, particularly Sky boxes, noise margin is likely to fall. No retrain necessary in that scenario, which is what the OP was doing.

Well at least the wiring might be OK!

The right twisted pair has been used - although it should be Blue/Wh on 2 and Wh/Blue on 5 but that will not matter in this context.

Well, I did as 24Talk requested. Results at 2010 before moving anything:
ADSL Link Downstream Upstream
Connection Speed 2016 kbps 448 kbps
Line Attenuation 63.5 db 31.5 db
Noise Margin 9.9 db 10.0 db

TB = 0.61
BT = 1.33
2 min video = reload every 15 secs

Dismantle system, move downstairs, plug router into BT master socket. At 2030:
ADSL Link Downstream Upstream
Connection Speed 1984 kbps 448 kbps
Line Attenuation 63.5 db 31.5 db
Noise Margin 9.6 db 10.0 db

TB = 0.16
BT = 0.73
2 min video = reload every 10 secs

it';s the wiring that needs changing, plugging into different places will only affect it marginally.