Saturday, 20 October 2012

When is a fault not a fault?

When it is too costly for BT to fix, maybe?

Having a fun one with BT (well, "fun" is not the word) where an unfortunate customer has lost just over 25% of the 63Mb/s speed they had on FTTC when their neighbour also got FTTC.

It is right on the borderline, but (we think) the wrong side (for BT) in that it did drop just more than 25%. Not many people realise that an FTTC line should not drop by 25% over a 14 day period. It is there in the handbook, but even BT do not seem to understand this. All they seem to care about is that the availability checker shows the estimated speed which matches what the line gets (well, it does now, as it seems that they changed the checker!).

BT seem totally happy to ignore their own rules if they are not convenient. What is worse is they know the cause of the problem (a short length of Aluminium cable) and the fix (to replace it with Copper), but that is not something they want to do - even though they are the ones that made this 25% drop rule.

We'll see how it goes - if they actually agree it has dropped more than 25% but still refuse to fix the fault, then this will be more of a story. I'll post more when we get to the bottom of it.

From our point of view we don't guarantee any speed, so whilst we are not in breach of contract with our customer, and we charge the same for 46M as we do for 63M, it is not exactly very nice on the customer to lose the speed like this. As a result we are doing everything possible to try and force BT to follow their own rules and fix the problem.

Update: Unfortunately, analysis of the sync rates and changes suggests that for this customer the sync rate dropped just too slowly. i.e. more than 25% overall, but over more slightly than 14 days. The rate drop within 14 days being just below 25%. Asa  result BT have refused to take any action to restore the previous speeds. In the long term vectoring may provide a means to reduce the crosstalk affecting this customer, but this could be some time off.

10 comments:

  1. Especially with BT, to stay within their quotas?
    "there is no fault"
    10 minutes later it miraculously starts working again...

    ReplyDelete
  2. Don't put any efforts into getting the aluminium replaced with copper, firstly its a blatant mis-diagnosis, further still the entire concept of aluminium limiting service is a myth - evidently a myth believed by some within BT as well.

    ReplyDelete
    Replies
    1. Not convinced it is a myth, but I will do some research.

      Delete
    2. I used to work for an LLU who primarily dealt with MPF's (not shared with BT), we consistently saw poorer performance and a higher failure rate on Al lines. Doing noise and return lost tests for the full ADSL2+ spectrum showed poorer quality results on Al lines. As such sync speeds where always less.

      If it is not the actual element in the cable which matters, perhaps the cabled used by BT during the Al era is significantly poorer quality then the Cu variant. Perhaps less twists, thinner conductor, etc.

      The support call RevK is describing above sounds shockingly like the call I logged on Friday, although the story differs as mine has lost 25% over six months, marked drops of speed when it looks like other customers have purchased FTTC.

      Here's hoping AA's BT bashing can get them to sort it out for this customer, and me!

      Delete
    3. Corrosion on the Ali cable can also be a big problem, in some cases the engineer only has to knock it slightly and it breaks in multiple places due to having corroded as a result moisture ingress over time.

      Delete
    4. During the Aluminium era, BT weren't oblivious and used a greater wire gauge to compensate and would therefore doubt that BT would specify the use of fewer twists. The only way to know what you have in the network is to have someone from the field (or where records exist) tell you what is written on the plastic sleeve. Sure, you can see guage changes and joints in test results but you cant differentiate what those are.

      I'll concede that the joints are potentially more volatile, I couldn't draw on experience to comment on this but it would make sense given the methodologies of the day.

      It sounds like crosstalk in the basic form of neighbour gets FTTC, original customer's rate drops by 25%. Unless the bundle has become rapidly saturated by more than one FTTC customer, a drop of 25% would suggest something is wrong and you want to have an Openreach tech to check the longitudinal / AC balance is OK (tested at the customer end ideally)

      Delete
  3. Seems to me that the second pair is interfering with the first - probably because the twisting (to reduce/eliminate the field generated by each pair) is so loose that it's not working... the fact that the new copper pair probably has tighter twists than the antique aluminium pair is convenient and the script jockeys at BT will know no better (they could of course replace it with aluminium cable with the same twists per inch and get a similar improvement - or indeed sufficiently spatially separate the cables - which is probably unlikely within the 2" customer endpoint duct)

    ReplyDelete
  4. Is BT using vectoring on their VDSL2? This sounds like the kind of case it was invented for.

    ReplyDelete
  5. Aluminium is an excellent conductor. But, as another poster has said, it readily corrodes leading to very high impedance where it is joined, if not actually falling apart.

    ReplyDelete
  6. Revk I find it interesting you posted ali was the cause of the severe instant crosstalk effect?

    On my own line I initially had a 110/36 attainable, I thought it was so high I would be immune to any crosstalk. Also I was first on the cabinet, I will also point out now my cabinet has only been live 2 months to this date.

    2 weeks after the install it dropped in a split second to 90/36, it was not gradual it just happened mid morning on a working day (installation). At this point I got worried as a sudden 20mbit loss of attainable sync I considered huge. If it happened once it can happen again but what grounds could I report this as a fault? As I still had a 80/20 sync and I am not even supposed to unlock the modem for the stats.

    Indeed the inevitable happened, a week after the above I had another instant sharp drop again around 10am on a working day, this time taking me below the 80mbit sync. Attainable went down to 73/24 so also this time my upstream was impacted. So within a month of install I went from 110/36 to 73/24. I am still on 6db fast path as ironically the snrm is extremely stable outside of these 2 events.

    So here is the dillema as far as fault reporting goes.

    My estimated was 65.9/20 - My line still syncs above this rate. I believe BT estimate at worst case scenarios so the fact I am now close the estimate shows things arent good.
    My actual sync speed is a loss of 9mbit down from 80 to 71 so not even 10%, upstream sync is still full speed although the error rate has increased significantly due to the lower margin and power masking (more on this below). In reality BT arent going to see that as a fault, even tho my attainable sync has lost over 30% in both directions which is huge.

    Obviously we all know crosstalk is an issue but most people seem to lose 1-2mbit here and there when someone gets connected, on 2 seperate occasions like your customer I lost huge chunks.

    What I do know is this.

    My upstream power was fiddled with, I stayed in touch with my install engineer, and he did some digging for me and the 2 dates matched to installs and that he said the 2nd one was a longer line where they adjusted my signal to improve their service. The graphs I will post below show my lower upstream frequency the power has been cutback which is making the weaker higher upstream frequency take over the upstream work as well as reducing the snrm. Sadly he cant touch my line such as swapping the pair without a fault order been raised.

    Next my QLN across D1 D2 D3 and U2 frequencies went up a LOT, on D3 it went up less severe and U1 had no increase at all (showing the power on it been cutback as it has much lower bitloading now). I am posting below data for when I had a 90/36 attainable and the 2nd link is the 73/24 attainable. I didnt make graphs for 110/36 as I only had that sync for less than 24 hours after I unlocked the modem. DLM has never touched the line.

    It does show that vectoring in my view is badly needed on BT's network, some parts of it are clearly of poor standard, and isp's in holland and germany are rolling it out yet it seems openreach arent too interested.

    90/36 http://www.chrysalisnet.org/hg612/line_stats-P-20121214-1443.png
    73/24 http://www.chrysalisnet.org/hg612/line_stats-P-20121222-1621.png

    ReplyDelete