cancel
Showing results for 
Search instead for 
Did you mean: 

Lost 8.4Mbps downstream but "nothing is wrong"

dinsdale
Hooked
Posts: 8
Registered: ‎15-05-2016

Lost 8.4Mbps downstream but "nothing is wrong"

Hi, I used to have a very reliable fibre connection with G.INP enabled and 56Mbps sync. Towards the end of February I noticed that my downstream SNRM had dropped from around 6dB to 3dB and my attainable sync rate was being reported as 48Mbps. This made no difference at all to performance but I was concerned about what might happen if there was a power outage or I ever needed to reboot my Billion 8800NL router. This was put to the test in early April and my sync dropped to 48Mbps. 

 

Plusnet tech support told me nothing was wrong as the BT speed estimate for my line was 27Mbps. As luck would have it the next day downstream speed had dropped to 9Mbps with 0.12Mbps upstream so an engineer was sent. He performed a reset which made a big improvement on that low point but sync was only 38Mbps with 6dB SNRM. I let DLM do its thing for ten days before rebooting the router but sync wouldn't rise above 47Mbps. If the forum lets me attach screenshots from the BT speedtester you'll see the best down/up rates of 51.31/6.31Mbps, worst of 6.83/0.14Mbps and current of 42.91/5.63Mbps.

I'm around 500m from the cabinet but was told recently by Plusnet that the network sees 1000m because the return distance is taken into account. This was news to me and my original sync rate was roughly in line with the chart published at http://www.increasebroadbandspeed.co.uk/2013/chart-bt-fttc-vdsl2-speed-against-distance . I've tested also with the Plusnet Hub One router and a Fritzbox 7390 but can't improve on 47Mbps sync. 

 

I appreciate that throughput will diminish as demand on the cabinet increases (my node is still green) but this doesn't seem to match my experience as the initial performance drop was overnight. It seems more as though a limiting profile has been applied and it may be pure coincidence that the currently advertised BT Infinity product offers speeds of 'up to 52Mb'. The speeds are still pretty good but I'd like to know what might have caused the performance drop. Downstream line attenuation with fibre has always been 26dB so I don't think it's a line fault. Tests done at the inner part of the master socket show no measurable difference to those done with my normal setup of 4M Cat5e cable with RJ-11 plugs routed to floor above.. 

11 REPLIES
dinsdale
Hooked
Posts: 8
Registered: ‎15-05-2016

Is the GEA line length test usually accurate?

My sync speed has become even lower than described in my other thread https://community.plus.net/t5/Fibre-Broadband/Lost-8-4Mbps-downstream-but-quot-nothing-is-wrong-quot... .

 

Until a couple of months ago I had a solid 56Mbps sync with nominal errors which dropped to around 47Mbps and is now exactly 40Mbps even after trying three different routers. At one stage recently my downstream speed was 21Mbps as my profile had mysteriously been temporarily downgraded to ADSL. The current persistent 40Mbps sync is apparently due to Crosstalk, according to Plusnet tech support. Yeah, right!

 

One odd fact which emerged from a GEA test done last Thursday is that my line length is 1045.8m. My house is about 500m from the street cabinet and I've been told previously by Plusnet that the physical distance of the cable run is effectively doubled because the distance both to and from the cabinet is taken into consideration. Hmmm. A distance to speed graph estimates that at 500m my original 56Mbps sync was about right but at 1000m I'd probably get nearer 27Mbps. Even now I'm 50% above that which seems to question the 1045.8m reading so what should I make of the figures?

 

Is the graph wrong? If it's not and my line *appears* to be twice as long as it probably is, does that suggest there's a high resistance fault somewhere? The run from the cabinet is all by overhead cable passing through trees. Downstream line attenuation has always been around 26dB though. Beyond that I'm guessing that I've been put on the old 40/20 profile although Plusnet say no.

 

 

 

 

Community Veteran
Posts: 5,187
Thanks: 486
Fixes: 21
Registered: ‎10-06-2010

Re: Is the GEA line length test usually accurate?


dinsdale wrote: 

I've been told previously by Plusnet that the physical distance of the cable run is effectively doubled because the distance both to and from the cabinet is taken into consideration. Hmmm.

I suspect that is nonsense. The signals are only going in one direction, the downstream signals are transmitted from the FTTC cabinet and received at your modem, the upstream signals are transmitted by your modem and received at the FTTC cabinet.

Why do you say "Yeah, right!" to the suggestion that it could be due to crosstalk? Different lines will add different amounts of crosstalk noise to your line, it's not unheard of for a single line having FTTC enabled to hit yours with enough crosstalk noise for you to lose several Mb.

dinsdale
Hooked
Posts: 8
Registered: ‎15-05-2016

Re: Is the GEA line length test usually accurate?

 

I'm not disputing the existence of Crosstalk and its negative potential but doubt it's the reason why my router always now reboots to a sync of exactly 40000Kbps. In recent times my default SNRM was reduced from 6dB to 3dB (OR engineer later said he had put me back on 'a 6dB profile') and my account was on an ADSL profile for a couple of days so I'm suspicious that some other restriction is still at work to cause the apparent glass ceiling at 40000Kbps with 48100 the 'attainable rate'.

 

Over the last couple of weeks I've noted that upstream G.INP has been enabled so wonder if that brought something else with it. The symbol and Interleave rates now vary a lot. During the almost three month long period of stability at 56Mbps sync the downstream G.INP was a steady 48 with 8 Interleave but now it varies between 44-54 with Interleave 4-16; it's currently 45/16 with upstream 40/4. It may well be a dynamic protocol but if it's reacting to a newly changing condition it may be a clue as to what might be wrong.     

 

If the effective distance for my line length is nearer 500m than 1045.8m, what is causing the GEA test to report a higher value? I think OR use this figure to estimate line speed and my upper estimate for a clean line is 34.2Mbps so Plusnet/OR regard anything over this level as an acceptable no-fault condition.

Community Veteran
Posts: 5,187
Thanks: 486
Fixes: 21
Registered: ‎10-06-2010

Re: Is the GEA line length test usually accurate?

The crosstalk or noise source or line issue may have also affected the error rate, which would have caused the DLM to make adjustments, and eventually the DLM will impose a speed cap with a "banded profile".

It didn't sound like your target SNRM changed from 6 dB to 3 dB, if that had happened your line speed would have increased.

Community Veteran
Posts: 19,102
Thanks: 443
Fixes: 21
Registered: ‎31-08-2007

Re: Is the GEA line length test usually accurate?

Hi dinsdale, welcome to the forum by the way.

It might have been better to have started this in your previous thread as although no-one replied to it at the time, they might in the future and things could get rather disjointed (I'll ask the mods to merge the two threads keeping the original title as that seems the most relevant or lock the other in favour of this one).

This distance "thing being double" is as ejs has said  - complete nonsense. If you know the route that the cable takes to the Cab you can check the actual distance yourself.
Also, there is no 3dB Target SNRM on FTTC, it is always 6dB.There is no 10 day training period on FTTC either. Connections are usually unrestrained for the first 24hrs and then DLM imposes changes to suit the line conditions. Thereafter it continually assesses the line (as with ADSL) but any changes it feels the line needs are usually done in the early hours (unlike ADSL which will have them applied fairly immediately).

If your stats were showing a steadyish near ~3dB SNRM when your sync was still at ~56Mbps then that suggests that either there was some interference present (and possibly still is) or there is now significant crosstalk for some reason - were you one of the first to take up FTTC from your Cab?

I take it you are on a 80/20 product? Do you have any audible noise on your phone line? Apart from your Billion 8800NL, what other Modem(s)(/Routers) do you have?

What does the BTw test show your current profiles as?

Moderator
Moderator
Posts: 27,209
Thanks: 1,850
Fixes: 160
Registered: ‎14-04-2007

Re: Is the GEA line length test usually accurate?

Moderators Note

Topics merged.

Customer and Forum Moderator. Windows 10 Firefox 63.0 (64-bit)

dinsdale
Hooked
Posts: 8
Registered: ‎15-05-2016

Re: Is the GEA line length test usually accurate?

I'm on the 80/20 product. I first noticed the reported 3dB margin in late February but my main concern was that it was accompanied by a lower attainable rate of 48Mbps. There was no resync at that time, sync of 56.122Mbps was maintained and a BTW speedtest done on 1st March showed 51.31Mbps downstream and 6.31Mbps. I was advised by Plusnet that SNRM was irrelevant to fibre and not to worry. As performance was not affected and the router is normally on 24/7 all was well for three weeks until sync was lost and I had to reboot.

 

The new sync speed was just under 48Mbps with lower performance but by the next day throughput had dropped to 9Mbps downstream and 0.12Mbps upstream. The Openreach engineer's equipment confirmed the 3dB reported SNRM. He commented that the 3dB margin doesn't work for everyone so he was putting me back on a 6dB profile, then reset my connection which resulted in a 38Mbps sync with 6dB.

 

I try not to upset DLM so rebooted after about ten days and got a sync just below 47Mbps which appeared to be the new upper limit whether the product of contention, crosstalk or an undiscovered fault. Vectoring is supposed to be coming here in the summer but there is no confirmed date. The recent introduction of upstream G.INP may be nothing more than coincidence but it was then that the current 40Mbps sync with 9.4dB SNRM was noticed after a network initiated resync.

 

I use a Billion 8800NL as my everyday router but also have an AVM Fritzbox 7390 with latest 6.30 OS and the Plusnet Hub One. There is the very occasional crackle audible on the phone line but it's inconsistent. The old 17070 number doesn't work any more so I haven't tried a quiet line test as such. I was the first FTTC customer on my cab (SWVLD cab 7) but changes have occurred very quickly rather than the expected more progressive losses as uptake increases and my problems pre-date the mass BT 38>>52 bandwidth increase on 2nd May. My current BTW profile is 38.67Mbps    

Andrue
Pro
Posts: 775
Thanks: 90
Fixes: 1
Registered: ‎12-01-2015

Re: Is the GEA line length test usually accurate?


dinsdale wrote:

The recent introduction of upstream G.INP may be nothing more than coincidence but it was then that the current 40Mbps sync was noticed after a network initiated resync.

 


It sounds like your line was already borderline at 3db but you were getting away with it. My guess is that the error count was within DLM's tolerance so it saw no need to act. When you rebooted your modem it went through the normal negotiation process to get you a 6db SNRM which resulted in the reduced connection speed, much as you yourself expected.

 

G.INP is probably just be a symptom that your line has deteriorated even further. As to why - that's hard to say. A fault engineer might be able tell you but first you have to persuade PN to get BT to send one out.

 

And yes, the introduction of BT's new package is a coincidence. I speculated here that it might impact other cabinet users but the luminaries on Thinkbroadband have quashed that suggestion.

dinsdale
Hooked
Posts: 8
Registered: ‎15-05-2016

Re: Is the GEA line length test usually accurate?

The 3dB margin was only there for about three weeks. The sync went from its normal 56Mbps/6dB to 56/3 then 47/3 and the SNRM only reverted to 6dB after the Openreach engineer's visit. It then stayed at 6dB until the present 40Mbps sync limit since when it has been 9.4dB. 

 

I'm connected to gateway pcl-ag03, located in City Lifeline House.

Community Veteran
Posts: 19,102
Thanks: 443
Fixes: 21
Registered: ‎31-08-2007

Re: Is the GEA line length test usually accurate?

Out of curiosity do you keep a check on your Current Line Speed ? Not that it will have any effect on you sync speed in any case - just worth keeping an eye on if you have any issues with throughput speeds. I assume from the way your post is worded that you've tried a resync whilst the SNRM is at ~9.4dB?
Just a word on that, DLM does not like more than one resync in any two successive 15 minute intervals, and indeed may still see reasons to act in conjunction with high errors even with one resync/drop. My advice is always to login to the router and drop the PPP session first (minimises the risk of a stale session issue) and then power off the modem(/router), (generally in daytime) for at least 30 minutes and preferably for at least 60 minutes - it's said other counters get reset.

Going back to what that OR engineer "said", either he didn't express what he meant very well, or was getting confused with ADSL2+, as I mentioned there is no 3dB Target Margin. A resync would probably  have resulted in the same speed as a "reset". The Modem's "attainable rate" estimate (and that's all it is) is based on current sync speeds and SNRM (probably assumes no change in usable tones) determined by it's own internal algorithm, so different devices might give slightly different values.

dinsdale wrote:
Is the graph wrong? If it's not and my line *appears* to be twice as long as it probably is, does that suggest there's a high resistance fault somewhere? The run from the cabinet is all by overhead cable passing through trees.
dinsdale wrote:
There is the very occasional crackle audible on the phone line but it's inconsistent. The old 17070 number doesn't work any more so I haven't tried a quiet line test as such.

Audible crackle indicates all is not well with the line - a phone line fault, especially an HR joint, but other faults may cause similar problems like Earth or Battery Faults, will certainly affect the BB performance, so it's possible that under such circumstances the GEA test may not estimate the line length correctly.
I notice that the GEA test results did not include a Copper Line Test - in fact it said "Unable to run".
Is your Line Rental with Plusnet or any other BT reseller, or is it with an LLU provider by any chance?

I also note that the GEA test result does show you on a 80/20 profile, but the sync speed limit and SNRM does suspiciously suggest a "banded" connection. This really needs a knowledgeable person from the Plusnet Help Team to look at this.

I also note that the test identified some supposed Interference at Customer Premises with a longest occurrence 1945-2345 - would that fit with anything you may do/equipment you may use - eg. a not so new Sky Box?

My guess is that either a problem with the line or perhaps some interference (which might even be cross-talk resulting from a line fault), or a combination of both is at the root cause.
A Copper Line Test (CLT) certainly needs to be carried out in case there is a detectable persistent fault, but it's possible it's intermittent and may not show on a given test. (Ideally it's best done when the crackling on the line is present - do you notice if it appears during adverse weather - wind/rain etc especially as the cable passes through trees ?).
Also you need to give careful consideration to whether anything electrical/electronic that you use might be generating interference for continuous periods.

dinsdale
Hooked
Posts: 8
Registered: ‎15-05-2016

Re: Is the GEA line length test usually accurate?

I've tried numerous resyncs over the last several weeks but I always leave the router switched off for at least 30 minutes. I'll add the dropping of the PPP session to my procedure Smiley   Since the 40Mbps sync limit with 9.4 SNRM appeared I'd see the same figures after every resync so I try hard to be patient and leave it for a few days to appease DLM.

 

Line speed has always more or less been a consistent fraction of sync speed. I do keep an eye on it pretty much out of geekiness but have always been impressed by the great performance. The Openreach engineer seemed very knowledgeable within the scope of his training but I agree he was probably thinking in ADSL terms as fibre is still a new technology here.

 

I haven't checked for EMI since ADSL days so I'll test for that again if the latest report from Plusnet doesn't do the trick. They emailed me today with another test which found "Impairment in copper joint detected most likely in local network. Please continue to submit a trouble report". This will be referred to their supplier so hopefully Openreach will respond during this week. The line is rented from Plusnet. 

 

What little line crackling there has been usually occurs just as I pick up the handset but doesn't seem weather related. If the problem persists beyond this week I'll pay closer attention although the discovery of the copper joint issue sounds very promising.