Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[burakkucat]

The "silly period", by consideration of the US SNRM, can be seen to be between 0600 - 2000 hours. It does not show any of the really violent swings that are characteristic of a HR or semi-conductive joint.

We are uncertain as to the degree of audible noise that may be heard when the telephone is used.

Ideally we need to see the graphical results for a time period when an incoming call is being signalled on that circuit. Ideally make use of the faultsman's ring-back, initiate the backwards call and unplug the telephone. Leave it for at least one minute or longer. (I can't recall if there is a time-out on the facility.) Assuming that the "ringing" voltage has been applied to the circuit for at least one minute then any potential effect on the SNRM should be visible by viewing the relevant plot.

[d2d4j]

Hi

I hope you don't mind, but there were 1 post by William which to me, is a possible reason and one I have come across before.

I'm sure he posted that the overhead wire passes by trees, or through branches of trees.

I am not sure of the degree of ingress of the branches to the overhead wire, but if the branches are touching the wire, or touch the wire during breezes or windy conditions, I have experienced conditions similar which caused issues on dsl connections, which disappeared when the branches were cut down so they no longer were able to touch the wire

It's just a thought for a possible explanation of the issues

I stress this may not increase any speed beyond what the connection is capable off, but would likely stabilise the connection

I hope that helps but sorry if I'm wrong

Many thanks

John

[burakkucat]

Thank you John. Yes, now that you have mentioned it I, too, can recall that post . . .

I know what Black Sheep would do if he was tasked to fault-find on that circuit -- he would get his drop-cable measuring rods and use them to hit the cable whilst someone else monitored the line.

It could well be that abrasion by the tree branches have damaged the cable's insulation.

However, I am having a problem in attempting to rationalise why the "silly period" in the US SNRM only appears during the hours of daylight. 

[Chrysalis]

I would say swings of this nature are unusual on the US, I couldn't find any other MDWS user's with such swings (admittedly I didn't check many).  Although at the same time this swing will be within normal operating parameters as far as openreach are concerned, they only care about service affecting problems, so e.g. if the line drops out or if the speed is significantly below estimates. It is the reason why we have lines with a 6db margin not 0db, because of swings.

I also would never consider a passed JDSU test as decent evidence of a fault free line, that test by design will only detect very severe problems. But that is the agreement CPs have with openreach, that test only needs to pass and assuming GEA tests also pass, openreach are free to declare a line as fault free.

Ironically my current pair throws up an alert on the JDSU, my engineer spent a few hours doing a pair swap and gave me the pair with the best sync, and then proceeded to tell me in a gutted manner the JDSU didnt like the pair, I told him to not worry about it.

[kitz]

I think most those of us whom continue to monitor on MDWS are the geeks and therefore by nature are most likely to have optimised their lines as best we can.

I too havent gone through MDWS too far, but if you look at just the A's then aesmith, alecR and andy265 all have a line that doesnt look perfect and Im not counting those that swing by 1dB. 

Swings are accepted event by DSL technology world-wide, its not just an Openreach. I'm certain I read a tech journal many years (10yrs?) ago (and if you look back on some really old posts of mine Im sure I will have said this many times) that 3dB was considered quite within normal parameters for swings.
As you say its why we have 6dB SNRm, some overseas ISPs use 9dB and didnt TT at one time use 9dB as the norm? 
Several years ago before Sky started using G.INP on their adsl2+ lines by default they had a very high level of INP set by default which also took SNRM to 9dB and a high level of Interleave that sometimes they would tone down.

What I fear here is that William may have been led to believe that something drastic is wrong with his line, when it is still operating within normal params.  Yes he could call out an engineer, but if there is a slim chance the engineer is going to see the daily swing, then what do we expect that engineer to do?   This is a downside of adsl over copper.

---

BTW if you go look at my own upstream SNRM for the past 180 days, note my upstream variance regularly going between 13.5dB and even down to 11.5dB - the straight line in the middle was no monitoring because I was testing another router. So I will snapshot a zoom in from that period showing swings between 13.3dB and 11.5dB.

One thing I find interesting and Ive been meaning to tell, because I thought it may also interest others.
In March I happened to notice an Openreach van parked on my neighbours drive.  After he went I noticed that MY downstream SNRm jumped up by 0.5dB.   

My downstream generally sat at about 6.8dB, but note how on Mar 10 it jumps to 7.3dB.  At first I thought perhaps it was a crosstalker, but no its continued to stay around there.  I asked my neighbour about the visit and she said "the wires in her grey box had disintegrated" - basically exactly the same corrosion of the metal terminals in the BT66 that I had had 2 years ago when I had a fault.   I know its not a big deal, but I found it odd that she'd had a line fault and immediately after that was fixed, I gained 0.5dB. I wonder how common place this is?       

[mlmclaren]

I know this.   Same engineer also tried to tell me that CRCs didn't matter and that it was FEC's that were important.   Because my line didnt have any FEC's then it was perfectly fine.    Think about it for a moment.   The line at that particular time wasn't interleaved, of course it wouldnt have any FEC.
I should add this was not a Openreach Engineer gone through the ranks, but one of those BT took on at the time when they needed staff quickly and were recruiting ex military.   He was ex-navy and then proceeded to argue the case about FEC with me.
Seriously he chose the wrong person to try that one on, but can you imagine if that had been anyone else... such as William... or joe bloggs.

I had this same problem, both times... the 2nd time wasn't so bad though it was after I explained I understood the technolgy more than he may of thought...

The reality seems that Openreach don't want us to see or understand DSL.... maybe it will the next big Watchdog investigation where somebody goes in undercover and catches them dishing out the lessons on talking .....

Unfortunatly when it comes to talking about DSL stats or even comparing them, everyone should just ask one question which is "Does it work.... and what does BT Wholesale say you'll get"

I knew something was fishy with my address when the lines got put into the system properly.... they way the ranges of speed where changed on BT Wholesales checker....

The ranges really widened up after I ordered, specially on the 'Impacted' side of things.... I mean 67 - 35 is a bit of a joke...

and that leads me to my next question which is "what specifies a clean line to an impacted line?" is my line now classed as impacted? or does the system of "an engineer tested line is clean and self install is impacted?"

These issue we all keep seeing on VDSL such as crosstalk are only getting worse, the offers from CP's are getting very good at making people upgrade and the increase in bandwdith requirement in the 'basic' household is also grown pretty fast.

My grandparents currently have a Sky LLU connection at around 10/1... this is sufficient enough for them to do browsing and streaming but due to them streaming alot more and also now owning smartphone's and them looking at content heavy webpages its starting to have a drag on the connection, there's also the case that the Skype call quality at 1Mb/s is rubbish and it can take an age for them to backup photo's to Onedrive or upload a video from an event or holiday....

So at some point in the next 6 - 12 months I can emagine they will be getting popped onto VDSL... I can't see them being the only people in the boat, surely there is many more.

[niemand]

The reality seems that Openreach don't want us to see or understand DSL.... maybe it will the next big Watchdog investigation where somebody goes in undercover and catches them dishing out the lessons on talking .....

This can be said for every large operator, though. All of them want to simply present a black box you plug into and the Interwebs come out. For 99% of the population this works just fine, too. The big boys have to cater to the lowest common denominator

[mlmclaren]

This can be said for every large operator, though. All of them want to simply present a black box you plug into and the Interwebs come out. For 99% of the population this works just fine, too. The big boys have to cater to the lowest common denominator

True...

[gt94sss2]

I have small HR issue on my line but it's not service effecting and no way am I going to get 5 or 6 engineers out to trace a possible small fault which they won't find, and HR faults need to really bad and service effecting before an OR Engineer can find the fault.

It also concerns me that some BT boffins are aware of how easy it is to get hooked on line stats and that some will go into acute state of worry for what is normal parameters.     Ringing any bells?   Is it any wonder why certain ISPs lock their routers so that such info unnecessarily panics their users

I have to say that I have seen a number of posts where users are concerned about very minor variations in their stats for lines which otherwise appear to operating fine, objecting to the fact that due to DLM their ping times are higher than they used to be or having unrealistic expectations about the impact of crosstalk etc.

What one has to realise is that if Openreach do come out and touch the physical line, there is a possibility that a line which may not have had a major problem before could then be more likely to go wrong (as another recent thread demonstrates).

Like NS I suspect I have a HR fault and some noise on my line but its going to have to get worse before Openreach can find it..

[Chrysalis]

My favourite one is when I had an aggressive engineer tell me my estimated speed, is the speed at the cabinet and then it declines further to my property

[WWWombat]

Wow. It's taken me a while to work through this thread... I'll make a few separate posts on the things I noted of interest.

On the physical reality:
I agree that swings are part and parcel of DSL of all types - it is part of the price to pay for re-purposing an old voice network, and the variety of installations that comes with such a network, into a high-speed data network.

I was always of the opinion that under 1dB was a good line, and that up to 3dB could be expected. From this, I thought the reasoning for a 6dB target was to allow the swing to go no lower than 3dB ... because the 3dB margin was where the errors really started to bite.

My belief, therefore, that use of margins below 6dB is something to be restricted until the variance of a line could be shown to be well below 3dB.

Crosstalk is a spanner in the works of this thinking. Especially for VDSL2. And it makes a bigger spanner if the disturber has a habit of turning their modem off.

[WWWombat]

On the technology of "impulse noise protection":
My weakest point on understanding the aspects of DSL is indeed "coding gain" - especially when you include the hidden parts, such as trellis coding, and then start combining them as inner/outer pairs. (@kitz; I might like hoovering up knowledge, but something like this makes even my eyes glaze over)

I noted @ejs made a comment about whether a G.INP-enabled modem could make predictions about the improvements in coding gain, and attempt to make an allowance for extra speed. That is an interesting avenue to explore, alongside the relative efficiency of different methods.

However, I tend to think that we are seeing the "allowance for extra speed" come out in a different way; and that BT are recognising this added efficiency through this different way: By allowing the target margin to drop below 6dB. By allowing the target margin to drop below 6dB, they are inviting more (raw) bit errors into the stream, under the belief that the new system can cope with, and fix, more errors.

Of course, we still have some amount of FEC and interleaving on G.INP-activated lines, but at a much lighter level. I wonder what the different coding gains are for different levels of FEC/interleaving? I assume it can vary...

Incidentally, I do believe it is possible for the sync speed to improve with G.INP on lines that had no previous FEC/interleaving intervention (i.e. INP=0). I don't know why - the standard framing parameters didn't explain it - so my assumption was that it was either in the re-framing that came about with G.INP or it was in the coding. The gain would be highest for those who would swap from lines with INP=3+, but modest improvements seemed to happen.

This graph seems to back that up:

If most lines (those with INP=0) would stay the same, shouldn't there be a lot of dots placed along the unity line?

An old ISPreview article (describing the approach being taken for G.INP mk.II) includes this quote:

For the majority of lines, we have noticed a small (1-2Mbit/s) increase in line speeds, as a result of retransmission enabling the DLM to increase the headline rates. … Note that in some instances the available memory in the modem can limit the maximum headline rate, by a few Mbit/s, however this is quite rare

I'm not sure what to make of part of that quote. "Enabling DLM to increase the headline rates" Surely (at the time) DLM hadn't intervened on "the majority of lines"? Not in the sense we knew it in early 2015, anyway. Where DLM had intervened old-style, then the improvement would likely be much more than 1-2Mbps - eg around 4mbps on a 40Mbps line would be more normal.

The ISP Forum in April 2015 also stated that:

Generally has positive impact on experience and headline rates
– 6 fold improvement in error performance
– Majority lines seen small increase in speed 1-2Mbps

[WWWombat]

On William's line:
From MDWS, the behaviour looks entirely like we would have expected 18 months ago, prior to the introduction of G.INP in any form.

- Before the DLM reset, DLM had intervened (INP=3, delay=8, banded)
- Before the DLM reset, the line was seeing bit errors from 7pm through to 7am, peaking before midnight. This generated many FECs, but almost no CRCs nor ESs.
- At the reset, an open profile is put in place (INP=0, unbanded)
- The sync and attainable speeds roughly matched, in the region of 40-41Mbps.
- After the reset, the line continued seeing bit errors in the same 7pm-7am period. These now appeared as many CRCs, with a high level of ESs, and quite a lot of SESs.
- The ES rate, around 3,000 per 24 hours, looked to be enough to trigger DLM intervention
- DLM intervened, with "low interleaving" (INP=3, delay=8, unbanded)
- After intervention, actual sync speed dropped 10%ish to around 37Mbps, while attainable rose 10%ish to around 45Mbps.
- Since DLM intervention, the line continues seeing bit errors in the same 7pm-7am period. These have reverted to FECs with almost no CRCs, ESs or SESs.
- The volume of FECs appears, at first glance, to be higher than before the DLM reset. Perhaps because the line is running slightly faster, at a slightly lower SNRM.

I know of no good reason why William's line seems to have gone back to 2014 behaviour though.
- Why didn't the reset put the line onto a "low interleaving" setting?
- Why, on intervention, didn't it turn on re-transmission?

The main issue right now is, perhaps, to figure out the cause of the biterrors. There's a bit of me that wonders about streetlights - especially if some of them turn off around midnight.

As it stands, the difference between sync speed (37Mbps) and attainable speed (45Mbps) seems to be explained by the standard artifact of the presence of FEC and interleaving. If DLM de-intervened, both speeds would like settle at the intermediate value of 41Mbps again. If G.INP was applied to the open profile, then the best guess would match BT's statements: perhaps an extra 1-2Mbps could be gained.

[WWWombat]

and that leads me to my next question which is "what specifies a clean line to an impacted line?" is my line now classed as impacted? or does the system of "an engineer tested line is clean and self install is impacted?"

I split it out with two factors, the line itself, and the installation:

- The line itself is either clean or impacted. If clean, it has no faults. If impacted, it has faults that can cause reduced speeds.
It is perfectly possible for a self-installed FTTC product to be a clean line with no faults.

- Openreach really care whether a line can be assured to be clean and fault-free, or whether it must be assumed to be impacted. This assurance only comes from an engineer installation, as the engineer runs tests; without the assurance, with a self-install, no-one knows for sure.

- With the assurance of an engineer install, Openreach are willing to commit to speeds in the A range. Without the assurance, Openreach are only willing to commit to speeds in the B range.

- The assurance is only felt when you wish to report a fault where the only symptom is low speed. If you bought self-install, then you will be ignored until the speed has dropped well below the B range. If you bought an engineer install, then you will only be ignored until you fall well below the A range. If you want better service in the event of a fault, you maybe should choose the engineer installation in the first place. Its almost like buying an extra warranty.

You'd think that, if you started with a self-install, and then had an engineer around to check for faults, you'd get promoted from the B range to the assurance of the A range. I wonder if this happens.

[Chrysalis]

wow thats news, so all these new self installs openreach will now only look at the impacted range?