FTTC speed dropped suddenly by 10 MBits

Before
http://i.imgur.com/RczQGJ8.png


After
http://i.imgur.com/lhQBmAs.png

What would have caused this?

Heating just turned for autumn?

No I haven't turned on the heating yet.

I noticed a big chunk lost, what sort of things would cause interference in that range?

Noise or a problem with the line.

Can you get the QLN and Hlog graphs? Those might distinguish.

In reply to a post by WWWombat:

Noise or a problem with the line.

Can you get the QLN and Hlog graphs? Those might distinguish.

Before
http://i.imgur.com/RPuJ1OD.png

After
http://i.imgur.com/5PBRICT.png

This is the first time in 2 years this has occurred whilst on FTTC.

Edited by deleted (Thu 15-Oct-15 20:00:12)

In reply to a post by MrSaffron:

Heating just turned for autumn?

That's really weirdo! As for ADSL2+ can be affect by xmas tree light. But FTTC affect by central heating on? That's odd

Central heating has always been a potential problem. Mainly the pump. Possibly a dodgy diverter box. (Whatever it's called).

I requested a link check online and the results are:

Results of line test
We cannot detect a problem with this line

Edited by deleted (Thu 15-Oct-15 20:58:58)

Rf noisy electrical devices don't just emit in the adsl2+ band

In reply to a post by MrSaffron:

Rf noisy electrical devices don't just emit in the adsl2+ band

Any advice please on what to do further?

First instance is turn off stuff and see if things improve

Power the modem off for 10 seconds then on again.

In reply to a post by BatBoy:

Power the modem off for 10 seconds then on again.

Best way to do so without tripping DLM?

In reply to a post by BatBoy:

Power the modem off for 10 seconds then on again.

Should be 30 minutes really to avoid DLM.

That is the best way, unless you want to leave it off for 30 minutes. But do not repeat within 24 hours.

Power cycled both router and modem for 30 seconds and still the same.

You mean this is still the same? http://i.imgur.com/lhQBmAs.png

That looks to be caused by some RF interference. It could be a failing port in the cabinet

In reply to a post by BatBoy:

You mean this is still the same? http://i.imgur.com/lhQBmAs.png

That looks to be caused by some RF interference. It could be a failing port in the cabinet

Yes it looks like that still.

I'll upload it anyway
http://i.imgur.com/Fc3Njh5.png

The SNR before was 4.1 @ 49mb

The SNR is now 6.1 @ 39mb

I think your line had resync and now the target SNR is connecting at 6.1db and dropped the speed down due to the loss of the 2dB SNR. Crosstalk is my guess to be honest.

Edited by lockyatlrg (Thu 15-Oct-15 22:41:49)

He power-cycled the modem. One re-sync .

In reply to a post by smsmasters:

In reply to a post by WWWombat:

Noise or a problem with the line.

Can you get the QLN and Hlog graphs? Those might distinguish.

Before
http://i.imgur.com/RPuJ1OD.png

After
http://i.imgur.com/5PBRICT.png

This is the first time in 2 years this has occurred whilst on FTTC.

Clearly the Hlog graph hasn't changed, ruling out certain issues with physical connectivity.

However, the QLN graph has changed significantly, particularly
- In D1, from tone 100 to tone 512
- In D2 from tone 1250 to tone 1750

The noise is mainly in a) 400kHz - 2.2MHz, and b) 5.4MHz - 7.5MHz

It is noticeable that those ranges don't include the whole of D1 and D2. In fact, the way the D1 noise stops at the ADSL2+ boundary is uncanny

For some of D1, at tones 300-512 - the increase in noise, seen in QLN, amounts to 25dB or more ... and the consequent SNR drops to zero; hence the lack of bitloading here. The worst-affected parts of D2 seem to have an extra 5-10dB more noise, and a consequent SNR reduction of the same - but only knocking 2-3 bits off.

Where the noise changes, it is very peaky, rather than a gentle spread across the spectrum. To me, that suggests this isn't crosstalk.

I'll have to leave it to others to suggest reasons why we could be seeing these signs.

I also note that the line configuration has FEC activated both upstream and downstream, but without interleaving. Between the 'before' and 'after', FEC has been ramped up on the downstream, changing from around 1% of overhead to 6% of overhead - which accounts for a significant fraction of the speed drop (maybe one third).

It is hard to see why the modem or DSLAM have chosen to do this.

I connected directly to the test socket using a micro filter and my sync has dropped even more slightly. http://i.imgur.com/6Q3FqCy.png

Looks like the noise problem is external to my abode.

Well the ISP did some "checks" on the line and they said "I'm not finding a fault with your service, sync speed is 37.0 Mbps, the speeds you were offered at the point of sale were between 36 Mbps and 50 Mbps".

I think I should switch to Virgin Media.

Edited by deleted (Fri 16-Oct-15 10:50:58)

What about any events before?

In reply to a post by lockyatlrg:

What about any events before?

This is the first time in over two years a speed decrease of 10 Mbps has occurred so suddenly.

Previously the speed decrease was gradual, i.e. from 55 Mbps to 53 to 50, etc...

Well I noticed some road works going on nearby, perhaps this could be a source of interference?

Edited by deleted (Fri 16-Oct-15 11:23:10)

Are the roadworks close to the cabinet or maybe above where the underground cables ore?

Do they have a generators running? Or are there traffic lights? The lights used for roadworks will have a master and then radio links to all of the others operating with it - they could be of sufficient power to kill a whole band of DSL. There will also be some very basic motion sensors too - that could be another source although around 6MHz is low (microwave sensors tend to be 6 or 10GHz).

They could have dislodged some cables yes.

In reply to a post by MHC:

Are the roadworks close to the cabinet or maybe above where the underground cables ore?

Do they have a generators running? Or are there traffic lights? The lights used for roadworks will have a master and then radio links to all of the others operating with it - they could be of sufficient power to kill a whole band of DSL. There will also be some very basic motion sensors too - that could be another source although around 6MHz is low (microwave sensors tend to be 6 or 10GHz).

The cabinet is about 200m I would say from the roadworks. All cables are underground. I noticed there are temporary traffic lights erected but I'm not sure if they are being powered with generators.

Does this correlate with the loss of bits as show in the graphs at certain tones?

Edited by deleted (Fri 16-Oct-15 11:39:33)

Even if they are not generator powered and the noise you are seeing is probably not from a "motor" (generator dynamo, alternator, compressor pump &c) there will still be the other radio and detectors to be considered which are more likely to be up in those frequency bands.

In reply to a post by MHC:

Even if they are not generator powered and the noise you are seeing is probably not from a "motor" (generator dynamo, alternator, compressor pump &c) there will still be the other radio and detectors to be considered which are more likely to be up in those frequency bands.

Okay, well, I'm not sure what i can do in this situation. My ISP won't take action unless sync falls below 36 Mbps.

Even if it does drop to below 36Mbps and the roadworks are to blame, what will your ISP actually do?

How long will the roadworks be there for? Can you wait and see if it recovers?

From your graphs the suggestion is that there are two specific noise sources - relatively high powered which is why there is a wide band of interference, rather than a failing card, cross talk, bad joint, worn cable &c.

Can you ask a neighbour if you could look at their line - use your modem if necessary and see if they are being hit at the same frequencies.

In reply to a post by MHC:

Even if it does drop to below 36Mbps and the roadworks are to blame, what will your ISP actually do?

How long will the roadworks be there for? Can you wait and see if it recovers?

From your graphs the suggestion is that there are two specific noise sources - relatively high powered which is why there is a wide band of interference, rather than a failing card, cross talk, bad joint, worn cable &c.

Can you ask a neighbour if you could look at their line - use your modem if necessary and see if they are being hit at the same frequencies.

If it drops below 36 Mbps then I guess there's not much my ISP can do if roadworks are to blame.

I really have no clue how long the roadworks will take but it seems to take 2 weeks per section and they move further up along the road and repeat the process.

Not sure about neighbours, not really close to them

Anyway I've reported this to my ISP and I'll see what they can do.

increase in QLN suggests to me crosstalk, shoot the neighbour who had FTTC installed

just read wombats reply.

I agree with him the QLN change is a bit unusual in that the adsl range is also affected, but in D2 and rest of D1 its typical crosstalk behaviour.

A possible issue here could be the power masking is messed up, your QLN should dip on the adsl power cut back range, and the new graph it doesnt do that, yet the bitloading is lower. That suggests you have increased noise in the adsl2 frequency which shouldnt be happening.

Edited by Chrysalis (Fri 16-Oct-15 14:49:26)

My ISP keeps saying there is no problem with the line, but surely a sudden speed decrease from 49 Mbps to 39 Mbps indicates a problem? They keep fobbing me off now with "Your promised speed is between 26mbps to 35.5mbps".

In reply to a post by smsmasters:

My ISP keeps saying there is no problem with the line, but surely a sudden speed decrease from 49 Mbps to 39 Mbps indicates a problem? They keep fobbing me off now with "Your promised speed is between 26mbps to 35.5mbps".

Old criteria (for Openreach to agree to an engineer) used to be
a) Automatic tests show a fault on the line; or
b) The speed has dropped, gradually, to somewhere below the 10th percentile; or
c) The speed drops more than 20% in one hit (within, say, a week)

I don't know if (c) remains as valid criteria, but it would certainly apply to your line.


Note 1) For (b), note that the range shown in the checker is the 20th percentile; the criteria nowadays is the 10th percentile - a much lower speed.

Note 2) The BTW dsl checker provides two ranges, one for "clean", which should apply for an engineer install, and one for "impacted" which should apply for a self-install.

What ranges does the checker report for your line? What range did the ISP quote initially? How was your FTTC connection installed in the first place?

Surely you've worded that badly?

As I see it, on a self-install there is nothing to prevent attaining or even exceeding the Clean estimate if the home wiring is fine and the master socket is used?

In reply to a post by WWWombat:

In reply to a post by smsmasters:

My ISP keeps saying there is no problem with the line, but surely a sudden speed decrease from 49 Mbps to 39 Mbps indicates a problem? They keep fobbing me off now with "Your promised speed is between 26mbps to 35.5mbps".

Old criteria (for Openreach to agree to an engineer) used to be
a) Automatic tests show a fault on the line; or
b) The speed has dropped, gradually, to somewhere below the 10th percentile; or
c) The speed drops more than 20% in one hit (within, say, a week)

I don't know if (c) remains as valid criteria, but it would certainly apply to your line.


Note 1) For (b), note that the range shown in the checker is the 20th percentile; the criteria nowadays is the 10th percentile - a much lower speed.

Note 2) The BTW dsl checker provides two ranges, one for "clean", which should apply for an engineer install, and one for "impacted" which should apply for a self-install.

What ranges does the checker report for your line? What range did the ISP quote initially? How was your FTTC connection installed in the first place?

An "engineer" (I think it was a contractor) replaced the faceplate and installed everything for me.

Back in 2013 I don't recall a range specified by my ISP when I ordered.

Back in 2013 I did initially sync at around 55 Mbits, even as high as 60 Mbits at one point.


The checker reports:
FTTC Range A (Clean) 55.3 35.7 13.9 8.3
FTTC Range B (Impacted) 46.5 27.4 13.7 6.1

Edited by deleted (Fri 16-Oct-15 22:28:39)

Perhaps a silly question but have you noticed any difference in noise during telephone calls since this happened?

I had a similar issue seeing my attainable sync drop from 120Mbps to 90Mbps over the course of a few months due to crosstalk from other customers being added to the DSLAM. Since then, the line has had a very faint buzzing sound on it. Despite having a half witted engineer out to say it was just "Normal VDSL noise" (Despite any noise being a cause for a fault), it still remains to this day unsolved.

What may have happened is an engineer connected 1 or more customers to the DSLAM who use the same D-side cable as you under the pavement which has caused a noticeable drop in sync due to crosstalk.

On the plus note if your connection drops a few more Mbps, you'll have the ability according to the figures given to you by your provider to get an engineer out to have a look at it.

In reply to a post by RobertoS:

Surely you've worded that badly?

Maybe, maybe not. It reads OK to me, but I'm tired.

As I see it, on a self-install there is nothing to prevent attaining or even exceeding the Clean estimate if the home wiring is fine and the master socket is used?

Well, given that the top of both ranges is still only the 80th percentile, there is nothing to stop speeds exceeding either range.

The real point is that a <proper> engineer installation should result in a provable clean line; the engineer should do enough work to ensure that the line qualifies as "clean". With Openreach having made sure of this (through a suitably qualified and skilled employee or contractor), they are effectively telling the ISP that they can sell the A range speed to the end-user, and are willing to stand by such estimates.

Oh ... and a "clean" line, to me, implies a lot about the quality of the line outside the house - not just the socket and internal wiring. It is the end-end quality that matters. I've had three FTTC lines installed by engineers (two Openreach, one Kellys), and all of them have taken the requisite care over this - though the latest one made use of more automated tests than the previous ones.

A self-install, by a knowledgeable and skilled end-user, can result in a line that is every bit as good as an engineer installation. However, Openreach don't know that, so are only willing to stand by a B range estimate. You won't find it easier to get an engineer to attend - and to me, that makes a self-install something of a false economy.

In reply to a post by smsmasters:

An "engineer" (I think it was a contractor) replaced the faceplate and installed everything for me.

Back in 2013 I don't recall a range specified by my ISP when I ordered.

Back in the early days, there was only the option of an engineer installation, so there was no need for two ranges - every installation done back then should classify as "clean".

Of course, we know that a good proportion of installations weren't carried out competently, but the important point is that an engineer installation means Openreach should be using thresholds based on the A range.

Nowadays, there is the option to self-install, but this is inherently riskier. BTW says this in their FTTC handbook:

When customers order Self Install they should understand that there is a higher risk of the line being impacted with wiring issues and End Users getting lower line and throughput rates. We recommend that CPs use Range B Broadband Availability Checker (BBAC) values when providing estimated speeds to their customers.

In reply to a post by smsmasters:

The checker reports:
FTTC Range A (Clean) 55.3 35.7 13.9 8.3
FTTC Range B (Impacted) 46.5 27.4 13.7 6.1

You've put a couple of confusing message in this thread. In some places you say the ISP tells you that you were sold a speed between 36 and 50, but in one place you say they've said between 27 and 35.

From my point of view, the ISP can indeed tell you (under criteria (b)) that they won't report anything until the speed drops below 36, as a range A threshold.

If they try to use the 27Mbps figure (range B), your response would be that your line was an engineer installation, so range A should apply.

However, criteria (b) is moot in your case ... because you have suffered an extreme speed drop in a short period; you want them to assess your line on criteria (c), and you need to persuade the ISP on the merits of this alone. This will be harder to get past low-paid, low-trained first-line support ... and I can't say for sure that it still applies.

The most recent version of BTW's FTTC handbook says this:

For all services, if the line rate (synch rate) falls below 2 Mbit/s, a fault may be reported to BTW which will be investigated. Additionally, if the line rate has decreased by more than 25% over the previous14 days, then a fault can be reported to BTW.

BTW also have a "90 day rule", by which you can cancel within the first 90 days if speeds are not up to scratch. The "90 day rule" itself doesn't apply to you now, of course, but the fault criteria are interesting:

This deals with performance issues relating to a service that has been in service for less than 90 days and cannot be adequately fixed by Openreach.
The service must have been investigated by Openreach for under-performance since its Service Activation date (i.e. the line will not synchronise at a speed above the 10th Percentile point of sale predicted rate, (please see section 16.2), or the line rate has dropped by more than 25% over a 14 day continuous period).

The section about the 10th percentile has this to say ...

When ordering GEA-FTTC, you may cancel within 90 calendar days of installation if the line speed degrades to the point of being significantly lower than the speed estimate provided at the point of sale

...

We define �significantly lower� as not meeting the 10th percentile speed for lines with the same line characteristics (dB loss when calculated at 300kHz) when the line is operating on the Dynamic Line Management (DLM) �Speed� policy. The 10th percentile speeds for each line are published on the Range of Values file on the BT Wholesale.com website.

That tells us BT's policy is to consider 10% of its lines as faulty!

whats the new criteria?

a BT bod on kitz said the system has added a colour code status to engineer tools, but I was curious if the old GEA tests still applied such as a large drop which you put in (c).

UK support said if speed drops below 36 they would investigate.

Indian support later told me only if speed drops below 27.

It is very hard getting first line support to do anything at all in my case.

Edited by deleted (Sat 17-Oct-15 07:24:52)

In reply to a post by Chrysalis:

whats the new criteria?

I don't know that there are any newer criteria. I'm being cautious

The most recent handbook I have is quite old, going back to when BT first changed (b) from the 20th percentile (ie the bottom of the estimated range) to the 10th percentile... to a touch of controversy, IIRC.

It is quite plausible there have been changes since then...

Have your speeds dropped by 25% in 14 days? If i read the posts correctly, you've gone from 50.6 to 36.9, which I make to be 27%.

If so, have you talked to support about this fact? Do they agree that this is an alternative criteria? Does their supervisor/manager agree?

To get anywhere right now, you need to get them to agree that there is another way of looking at the issue, beyond just looking at the estimate.

In reply to a post by WWWombat:

Have your speeds dropped by 25% in 14 days? If i read the posts correctly, you've gone from 50.6 to 36.9, which I make to be 27%.

If so, have you talked to support about this fact? Do they agree that this is an alternative criteria? Does their supervisor/manager agree?

To get anywhere right now, you need to get them to agree that there is another way of looking at the issue, beyond just looking at the estimate.

Yes that's how much of a decrease it is.

I'm not getting anywhere with support, it seems they don't really care to be honest.



I've been asked for:
Numerous speed tests
Estimated speeds given

They're not actually caring about finding the root cause of this.

Anyhow I'll keep trying...

EDIT: Finally reached a support agent who will be raising this issue with the "Network Fault Specialist Team".

I asked "Is this a problem in the FTTC cabinet?"
Their reply: "No in the network"

I don't understand how this can be?

Edited by deleted (Sat 17-Oct-15 22:23:08)

I suggest he asks for a GEA test.

if that 25% criteria still exists then that test will fail, but he needs to get a move on because it has to be a recent drop, no use doing the test months after the fact.

ask them to run a GEA test and to paste the results to you.