Kitz Forum

Broadband Related => FTTC and FTTP Issues => Topic started by: 22over7 on April 17, 2017, 09:22:03 PM

Title: Sync drops by a third
Post by: 22over7 on April 17, 2017, 09:22:03 PM
Something fairly catastrophic happened to my long-time stable line before 7:50pm Friday giving DS speedtests of just of 60Mbps; afterwards, 40Mbps.  The upload (17Mbps-ish) took a small hit too. As far as I could tell, by error-rates were fairly normal.

The event looked very bizarre on my SNRM dslwebstats (I'm SandReck). The graph had a huge spike, and even went negative (?!).  I don't know what to think.  I had the dslwebstats feed off for a day or so, then when I realised, clicked the record button. All hell broke loose. Total coincidence?  Nobody seemed to have been fiddling with by Huawei cabinet.  I use an unlocked HG612.  I powered all my equipment off, waited a bit, then on again.  No joy.

Before, I was 16/1 interleaved. Now I'm 1/8 interleaved. My SNRM margins were about 7.4 db down, 13.6 up; now they're almost identical, about 6.3 or 6.4. with a pronounced "wiggle" -- about 3 waves in 10 minutes.  Maybe the US SNRM is drifting a bit higher.

I just wondered, if any of the network gurus on this form can see my (SandReck's) dslwebstats data, and enlighten me, if possible, as to what-the-hell-might-have happened to provoke such a dismaying, sudden, deterioration?  Internet access seems subjectively rather  crummy now, though I can appreciate many people have it lots worse.

I'm with Plusnet, for almost 3 years. They seem to dislike having people on 80/20, and very often people on such a package find themselves misconfigured on 40/2 or something like that (I'm something consistent with 40/20).  I'm hoping it's some misconfiguration their end. I've raised a fault, but have no idea when/if they'll take a look. Anyway, here, I'm just after a little enlightenment. I can dig out current data quite easily (I've attached some post-apocalyptic modem stats), but my line had been so stable before, that I'm not sure I can remember much about its former state (except via historical dslwebstats data).

Title: Re: Sync drops by a third
Post by: kitz on April 18, 2017, 01:52:37 AM
If I had to guess the symptoms look kind of crosstalk related.  However that is one heck of a hit to take all at once.

Your downstream SNRm went from 7.4dB to -2.8   (10.2)
Your upstream SNRM went from 13.2dB to 6.3dB 

The upward spike is interesting, thats the sort of thing you could expect to see if a whole lot of users suddenly got knocked offline and had to resync.  Some of the BCM chipsets such as those used by the HG612/Zyxel seem to come back up much faster than say the HomeHub5a. 
However the corresponding spike in your max attainable doesnt quite give the headroom I would have expected if your downstream SNRm really was at 14dB so Im not quite sure exactly what happened there.   

The spikes indicate that what ever affected you, also had an impact on some of your neighbouring lines.

19:38 is a bit of an odd time for a new disturber to appear, but it is possible that someone got FTTC swapped over during the day but it was a self install and they didnt switch their modem on until that time when they got home from work.

What could have happened if your modem is happy doing its thing,  neighbour turns on their modem.  Bam crosstalk which immediately makes your SNR drop to minus 2.8dB.   The line cant hold at such a negative figure so attempts to resync at the target of 6dB which accounts for your loss of speed.

Although its not unusual for crosstalk to suddenly knock 10Mb from a lines max attainable, I dont think Ive seen what looks like one single disturber take 20Mbps off a line that was previously syncing at around 60Mbps.

This isnt anything to do with Plusnet configuration, they have no control over your SNRm or sync speed (other than provision type ie 80/20,40/2).  Such a drop does however need investigating so you will have to go through the fault process :(

----
PS note for others.  Ive not looked at all stats or params.  Only concentrated on main figures and not any config params. 
 
QLN is a bit messy... some noise... and also upstream crosstalk.  Slight signs of crosstalk bump at tone 3500 onwards.   Noise in D1 & D2 makes it a bit impossible to spot any obvious crosstalk in those carriers.
Hlog looks fine indicating physical line ok.  Slight bump also around 3500 but not anything Im too concerned about.
Attaching zoom of SNRM

Im only able to type one handed atm so its taken me a while to make this post and I need to get to bed myself.   
Title: Re: Sync drops by a third
Post by: ktz392837 on April 18, 2017, 08:03:51 AM
Quote from: 22over7 on April 17, 2017, 09:22:03 PM
Something fairly catastrophic happened to my long-time stable line before 7:50pm Friday giving DS speedtests of just of 60Mbps; afterwards, 40Mbps.
I have a (perhaps two) major crosstalk disturbers that has a similar impact 6-7db each.  Overall I have lost 40% due to crosstalk and as I am on ECI cabinet I do not know if it can worse - I hope not but not convinced unless they get GINP/3db working for at least a little compensation.
Title: Re: Sync drops by a third
Post by: 22over7 on April 18, 2017, 01:41:52 PM
Many thanks for the replies, especially for pointing me at things that may be most relevant.  I'll try to assimilate some of the (copious) relevant information on this site. 

My cabinet (serves 427 according to www.telecom-tariffs.co.uk/codelook.htm) and PCP are about 200 metres away as the crow flies, but maybe 600 away electrically,
going literally round the houses and ending with a long cable run along back-garden walls, ending in a pole serving at least 8 properties.  I guess crosstalk can
occur anywhere in/beyond the DSLAM though.  I'll try to get calamity-anecdotes about last Thursday from the neighbours I'm still speaking to. And wait for Plusnet's fault-team. And investigate what Virgin can offer.

Just for lolz, one of the properties passed was a cannabis farm, when my line was at its best (just south of 70Mbps on speedtests).

Title: Re: Sync drops by a third
Post by: NewtronStar on April 18, 2017, 04:57:24 PM
Your Bandplan Downstream and Upstream are still the same but the bits per tone have lowered this will cause a lower sync rate than you were used to, going to show your bit tones before and after

 
Title: Re: Sync drops by a third
Post by: 22over7 on April 18, 2017, 06:03:09 PM
Tyvm NewtronStar. Before understanding fully understanding what I'm looking at, it's evident my line has been hammered.

(Some nonsense deleted)
Title: Re: Sync drops by a third
Post by: WWWombat on April 20, 2017, 07:36:02 AM
I can't post pictures at the moment, but comparing some of the historic (as opposed to current) "quiet line" (QLN) graphs tells us this is likely to be crosstalk.

I looked at QLN for 3 dates -
1st February - clean, and low noise in D2 and D3, and above the ADSL2+ frequencies in D1.
13th April - generally the same but spikier. Happens sometimes.
17th April - much noisier now, but unchanged in ADSL2+ spectrum. Enough to explain a 10dB+ change in SNR

The QLN data only gets measured and updated (in the modem) at a resync, so we don't see live changes. However, we can see the impact of the noise increases in the "SNR per tone" graph.

Openreach should accept reports of a drop in speed of 25% in a 14 day period, so Plusnet ought to be able to get an appointment for you. There isn't much they can do about crosstalk though. Even using a different pair may not improve things.
Title: Re: Sync drops by a third
Post by: 22over7 on April 20, 2017, 08:37:42 AM
That fits, WWWombat.  The diagnosis/suggestion of cross talk is far beyond me, but in the past couple of months I had noticed that my "max attainable", which used to be well above 70Mbps (on 3Dec2016 it was 75,208Kbps), had been drifting lower and lower. Earlier this month it was now and then just below my actual sync rate. (Which puzzled and alarmed me.)

I cannot remember/recover the date now (early March?), but there was some work on my cabinet -- something to do with "ducting", perhaps, or resituating the cables to the PCP on the other side of the road.  Great, I thought; things will improve.  Fat chance. However, this is just empty speculation.  I guess it is simply that several new lines were crammed into the cabinet, as BT's assets get increasingly sweaty.
Title: Re: Sync drops by a third
Post by: snadge on April 28, 2017, 11:48:33 PM
Quote from: 22over7 on April 20, 2017, 08:37:42 AM
That fits, WWWombat.  The diagnosis/suggestion of cross talk is far beyond me, but in the past couple of months I had noticed that my "max attainable", which used to be well above 70Mbps (on 3Dec2016 it was 75,208Kbps), had been drifting lower and lower. Earlier this month it was now and then just below my actual sync rate. (Which puzzled and alarmed me.)

I cannot remember/recover the date now (early March?), but there was some work on my cabinet -- something to do with "ducting", perhaps, or resituating the cables to the PCP on the other side of the road.  Great, I thought; things will improve.  Fat chance. However, this is just empty speculation.  I guess it is simply that several new lines were crammed into the cabinet, as BT's assets get increasingly sweaty.

Hi 22over7  :)

My max attain slowly dropped by 6Mbps in just a few months and I live in a quiet village where ADSL is sub 2Mbps, so fibre take up is likely high, crosstalk (in this sense) is signals from other VDSL customers lines bleeding into neighbouring lines within the cable bunch, this causes a reduction in attainable sync rate.

This happens because Fibre (VDSL) uses high frequencies (up to 30Mhz in some cases but usually 17Mhz as is your case) so in laymen's terms is "louder" than regular ADSL (which uses up to 2.2Mhz) and its the 'cross-talk' of these signals cancelling each other out.

BT are trialling a technology called Vectoring which is to cancel out X-talk, but it could be a while before thats Nationwide and the 'norm', as WWWombat says, even switching pairs may have a negate effect, it could better OR worsen the situation.

trying can do no harm, the BTo engineer can always put you back on your original pair

I see BTo at my cab every week now
Title: Re: Sync drops by a third
Post by: Ronski on April 29, 2017, 08:03:47 AM
Your cabinet could have been at capacity,  so it's quite possible that the work on your cabinet was to install more tie pairs to increase capacity. This could have resulted in a sudden surge of orders when it was able to accept more.

As far as I'm aware BT or more to the point Openreach are not trialing vectoring, it is a mature technology, but unfortunately as far as we can tell it is not going to be rolled out to all cabinets. ECI cabs don't support it very well without some costly modifications and it's only being used on some Huawei cabs to reach targets,  mainly for BDUK.
Title: Re: Sync drops by a third
Post by: 22over7 on April 29, 2017, 09:51:56 AM
Quote
...Fibre (VDSL) uses high frequencies (up to 30Mhz in some cases but usually 17Mhz as is your case)
(from snadge)

Ah! It really pays to read every bit of these mydslstats reports carefully! Thank you.  I'm a little amazed that wires originally intended to carry audio up to c. 8Khz can support this.  Some pennies are dropping. (A long time ago I taught a (un!)advanced course on basic error-correction techniques for various kinds of media. Thanks heavens I never considered phone lines: it seems to be a lifetime's study. )

As for the work on my cabinet, I think I found some relevant paperwork: https://citydev-portal.edinburgh.gov.uk/idoxpa-web/applicationDetails.do?activeTab=documents&keyVal=OB2P9OEWHFB00 (https://citydev-portal.edinburgh.gov.uk/idoxpa-web/applicationDetails.do?activeTab=documents&keyVal=OB2P9OEWHFB00)

It seems to be about installing another cabinet and PCP, next to the one(s) I'm connected to; what I've observed so far is likely some preliminary work to prepare for this.  Mind you, the plans and drawings seem remarkably inaccurate.  From local knowledge, I suspect this is for delivering FTTC to a conspicuous fibre not-spot in the other direction from my neck-of-the woods. I'm not crossing fingers or counting chickens that anything will improve in my direction.  Oh well, too bad.
Title: Re: Sync drops by a third
Post by: Ronski on April 29, 2017, 10:36:04 AM
It is PCP27 on Mornigside exchange, which appears to be a new PCP and a 128 port Huawei for some exchange only lines - there's quite a lot on Morningside exchange

https://www.telecom-tariffs.co.uk/codelook.htm?xid=401815&cabinets=13522

I note permission was granted on Wed 14 Sep 2016, so it shows just how long these things take.
Title: Re: Sync drops by a third
Post by: burakkucat on April 29, 2017, 04:10:38 PM
Quote from: 22over7 on April 29, 2017, 09:51:56 AM
I'm a little amazed that wires originally intended to carry audio up to c. 8Khz can support this.

The PSTN actually uses less than half that bandwidth, 300 - 4000 3400 Hz, to be precise.

[Edited to correct the nonsense -- see below.]
Title: Re: Sync drops by a third
Post by: WWWombat on April 29, 2017, 07:24:41 PM
Quote from: burakkucat on April 29, 2017, 04:10:38 PM
The PSTN actually uses less than half that bandwidth, 300 - 4000 Hz, to be precise.

Even less...

When I wrote code for carrying traffic over digital transmission systems (DASS2, Q931, ISDN etc), you had to specify what kind of bearer circuit was being carried on the 64kbps digital path ... and the regular telephony one was known as "3.1kHz", representing 300Hz - 3.4kHz.

Those numbers have stuck in my brain for nearly 30 years.
Title: Re: Sync drops by a third
Post by: burakkucat on April 29, 2017, 07:36:50 PM
Quote from: WWWombat on April 29, 2017, 07:24:41 PM
Even less...

When I wrote code for carrying traffic over digital transmission systems (DASS2, Q931, ISDN etc), you had to specify what kind of bearer circuit was being carried on the 64kbps digital path ... and the regular telephony one was known as "3.1kHz", representing 300Hz - 3.4kHz.

Those numbers have stuck in my brain for nearly 30 years.

b*cat bows and goes to find his brown-paper bag as he realised he had typed a nonsense, above.  :paperbag:
Title: Re: Sync drops by a third
Post by: 22over7 on May 02, 2017, 05:45:35 PM
The BT engineer just left. He was here from 1pm to gone 4:30, up poles, in PCP, head scratching, running several tests.  Noisy copper. Line swap, DLM reset.  Left, saying (1) DS now stable at 67, (2) he'd keep an eye on it as he wasn't 100% happy about where the noise was coming from.  Supernaturally diligent and pleasant guy.  Explained everything he was doing. Can I send Openreach some kind of unsolicited glowing testimonial?  Sadly, I've forgotten already his second name. Ross someone, from Yorkshire.

I started uploading my stats again. At the moment (17:13), my-idea-of-the-basics are:

SNRM                    5.9     / 7.1
DS/US attainable:  79904/22572
DS/US actual:        69594/19999
PWR:                     13.8  / 7.4
ATTN(DS)               18.0
INT                        1363 (??)/ OFF

Of course, TBB speedtests are still hard-clamped at around 40/16. The engineer watched it run, and said it must be Plusnet doing that.

My Plusnet ticket is on hold till tomorrow 9:00am, when I'm away all day, mostly inaccessible.  I think I might add a note (P'net highly discourage this) saying more-or-less
the above, in the hope of triggering whatever administrative thing they have to do to allow me to get the service I'm paying for. Or perhaps I'd better wait for them to get the engineer's report.

 
Title: Re: Sync drops by a third
Post by: burakkucat on May 02, 2017, 06:09:06 PM
Quote from: 22over7 on May 02, 2017, 05:45:35 PM
The BT engineer just left. He was here from 1pm to gone 4:30, up poles, in PCP, head scratching, running several tests.  Noisy copper. Line swap, DLM reset.  Left, saying (1) DS now stable at 67, (2) he'd keep an eye on it as he wasn't 100% happy about where the noise was coming from.  Supernaturally diligent and pleasant guy.  Explained everything he was doing. Can I send Openreach some kind of unsolicited glowing testimonial?  Sadly, I've forgotten already his second name. Ross someone, from Yorkshire.

Yes, certainly.  :)

Please visit this Openreach page (https://www.openreach.co.uk/orpg/home/contactus/tellussomething/tellus.do), scroll down to the "Let us know how we did" heading and then take the "send them a thank you message" link.
Title: Re: Sync drops by a third
Post by: WWWombat on May 02, 2017, 06:17:04 PM
Plusnet might merely need you to stop & start the PPP connection to get the speed limit back. Maybe, but maybe not.

Looking at the graphs, and comparing, I can see
- That the Hlog graph has barely changed with the new line, so both seem to be OK
- The QLN is much better than it has been recently.
- The QLN now if back to being similar to how the old line was at higher frequency - above ADSL2+. In the lower frequencies, ADSL2+ and below, it is getting much LESS noise.

However, while you have synced at a high speed, you are getting a high rate of errors, even though (old-style) FEC and interleaving is turned on. I'd say there was a chance that DLM might even need to turn you onto a higher interleaving setting.

If you have the right modem, it might be worth making changes to set a lower maximum sync speed, to keep the error rate down. Others can guide you on that method.

Are you on a Huawei cab? If so, then the best aim would be to get DLM to change you over to G.INP, which would need you to keep error rates under control.
Title: Re: Sync drops by a third
Post by: 22over7 on May 02, 2017, 10:20:54 PM
I think this got lost (reduced):

to burakkat: thanks much for the engineer big-up link. I hope I did it effectively. Stellar individual.

to WWWombat: thanks advice. To anyone: very grateful for any links and advice about reducing sync speed to avoid new problems now (on HG612/Huawei).

Pi.
Title: Re: Sync drops by a third
Post by: banger on May 03, 2017, 01:11:20 AM
Quote from: 22over7 on May 02, 2017, 10:20:54 PM
I think this got lost (reduced):

to burakkat: thanks much for the engineer big-up link. I hope I did it effectively. Stellar individual.

to WWWombat: thanks advice. To anyone: very grateful for any links and advice about reducing sync speed to avoid new problems now (on HG612/Huawei).

Pi.

There is a sticky post at the top of this forum if you have a read on how to cap speed on the HG612 to reduce sync speed and errors at the same time and it also may reduce your interleaving with less errors thus delay.

I found that G.INP when enabled really gives you a boost in speed and reduces errors alot but if you are not on a Hueawi cabinet its unlikely you will see it.

Edit: I see you are on MDWS and G.INP is not active, you may get full sync with low errors if G.INP kicks in over the next couple of days. At present you are on high interleaving so its wait and see without rebooting the router/modem.
Title: Problem resolved (for the time being)
Post by: 22over7 on May 04, 2017, 03:42:12 PM
My problem of my broadband speed tests being stuck at 40/16 has gone, though (1) (possibly) Plusnet resetting some-profile-or-other, and (2) (sheepishly) me, some hours later, logging of from Plusnet and then back in again. (At the prompting of a singularly helpful CS help bod in one of their forums.)

A TBB speed test now runs at 74/16.  :)
I do appreciate though that the line is noisy and sync is likely to deteriorate, some.   
 

Title: Re: Sync drops by a third
Post by: WWWombat on May 04, 2017, 10:17:02 PM
I can see that G.INP has been turned on, and the error rate has gone down, even with a jump in speed of 10Mbps - banger's prediction was right. Looks OK at the moment, but of course the dreaded crosstalk can cut in at any time...
Title: Re: Sync drops by a third
Post by: 22over7 on August 09, 2017, 07:26:43 AM
I doubt it is of general interest, but I have been experimenting with capping my sync rates on my modem. 

It is (to me) a dark art, but it seems to work!   After DLM reset, my line went to "INT 8" after a short while, then "INT 16" after a longer while.  I was able to pull it back to "INT 8" with a downstream cap of 72Mb/s and upstream of 16Mb/s (at least I think that is what did it). With these caps I get speed-tests of about 66Mbps and 13Mbps. My attainable is roughly 89/22. 

There's quite a difference in responsiveness between "INT 16" and "INT 8"; its only when downloading an operating-system or a movie that throughput matters, and then only a little.

Could I get rid of "INT 8" by setting my sync-caps even lower?

What are the relative effects of upstream and downstream caps?  I do copy a few 10's of MB up sometimes, but the time
it takes is seconds at most.

What might happen if I left the down sync cap at 72, and lowered the up to 10? 
(Most errored seconds on my line seem to relate to the upstream.)
Title: Re: Sync drops by a third
Post by: Dray on August 09, 2017, 08:28:15 AM
Probably because there is no G.inp on the upstream
Title: Re: Sync drops by a third
Post by: tbailey2 on August 09, 2017, 08:55:26 AM
Wrong. There are at least eight active users on MDWS with G.INP set on the upstream...
Title: Re: Sync drops by a third
Post by: Dray on August 09, 2017, 08:59:42 AM
That sounds interesting. When I got G.inp on upstream it was because something was failing inthe cabinet resulting in huge instability which G.inp coped very well with
Title: Re: Sync drops by a third
Post by: ejs on August 09, 2017, 09:01:54 PM
Quote from: 22over7 on August 09, 2017, 07:26:43 AM
There's quite a difference in responsiveness between "INT 16" and "INT 8"; its only when downloading an operating-system or a movie that throughput matters, and then only a little.

There really shouldn't be, because the interleaving depth is really not a significant value. The interleaving delay will be the same, and the interleaving depth tends to vary according to how much data can be shuffled up in the interleaving delay interval. So when you have less bandwidth, there'll be less data in the same time interval.

With G.INP though, the interleaving depth is the size of the DTU (the smallest chunk of data that can be retransmitted). By capping the rate, you've decreased the size of the DTU, and the change in interleaving depth is secondary to that.
Title: Re: Sync drops by a third
Post by: Chrysalis on August 09, 2017, 09:23:13 PM
yeah latency is usually more important than overall speed.  Which is why I consider banding preferable to interleaving, but there is an obsession of speed by a lot of people. :)
Title: Re: Sync drops by a third
Post by: 22over7 on August 21, 2017, 09:27:23 PM
It's maybe interesting to see what a difference sync-capping makes, by looking at what happens to a BQM when one forgets to apply it, on installing a new modem.
https://www.thinkbroadband.com/broadband/monitoring/quality/share/6bcea9e0ffe0448345d6001ae25d4cf44662bc80-21-08-2017 (https://www.thinkbroadband.com/broadband/monitoring/quality/share/6bcea9e0ffe0448345d6001ae25d4cf44662bc80-21-08-2017)
My errored seconds at (80/20 uncapped) went through the roof for a day, latency was horrible, and speed tests weren't great.

The second red band is me setting
caps back on (in my 30 mins offline), and having the bright idea of turning wan pings off -- till I remembered an hour later that's how TBB's BQM works.

BTW, the new modem is a VMG132-B10D; old was HG612. It's nice to be able to ssh in without weird flags. At least.
Title: Re: Sync drops by a third
Post by: banger on August 21, 2017, 09:58:55 PM
Are you on MWDS? If so what username?
Title: Re: Sync drops by a third
Post by: 22over7 on August 22, 2017, 08:41:51 AM
Yes banger, I'm on MDSW, as SandReck.