Anyone understand BT's G.Fast/Halo 1 FTTC Broadband?

[jeremyc]

I have a recurring problem with my 'Fibre 250 with BT Halo 1' broadband: every night it renegotiates the downstream sync speed until way below the guaranteed minimum bandwidth (and sometimes until it gives up with a connection altogether).

I've had multiple visits from engineers who have confirmed the copper connection to the cabinet is OK, "reset" something in the back end and restored my 200Mpbs+ downstream bandwidth. However, over the next 10 or so days this gets renegotiated down to below the 150Mbps minimum.

So to my questions:

1. Is it possible there is something in the configuration of the infrastructure that is causing this? Could it be an over-congested DSLAM or fibre link between the exchange and cabinet?
2. Would crosstalk interference between copper cables in common bundle between my property and the cabinet cause this?
3. How can I escalate the fault resolution with BT to get to someone who understand the configuration/management side of the infrastructure, rather than the engineers who are great at understanding the copper cables but have no idea about the 'black magic' between the copper and the exchange?

It seems bizarre to me that it can perform well for a week and then suddenly rapidly degrade.

[bimsb6]

Gfast is very dependent on a good copper connection from the cabinet and needs to be a very short cable run ( within a few hundred metres of cable) .

[jeremyc]

bimsb6 said:

Gfast is very dependent on a good copper connection from the cabinet and needs to be a very short cable run ( within a few hundred metres of cable) .

I'm within the distance guidelines for the copper run - about 300m from the cabinet I believe.

I can understand a poor copper connection giving a poor performance, but don't understand why it would give great performance for a week before being degraded. It's almost as if there are traffic management/shaping algorithms that say "let it run for a week and then wind it down every night".

From my end-point evidence, my whole physical infrastructure would appear to happily support 200Mbps+. I'm trying to understand what might be going on to limit performance in the system.

[randlemarcus]

Possibly because there's an optimistic noise or something monitor that goes (days 1-4) "ooh, lets just ignore these high readings for a bit, see if they go away", then days 5 onwards, goes - "nope, there's an actual problem, see if reducing speed makes it go away"

https://helpforum.sky.com/t5/Broadband/Ultrafast-s...

[KingofKong]

I had similar with my Infinity, it took 5 different engineers over the course of 3 months to find why my connection was dropping from 68Mbps down to 5ish. The first couple of engineers were contractors, not actual BT staff, the next couple just checked what the first lot had already checked and were keen to blame my internal wiring as the issue.

Finally, the last visit was from a senior engineer with lots of test equipment who found a loose cable in the green cabinet, why the others didn’t think to check there I really don’t know.

Somewhere inside the cabinet my connection was only just hanging in there with one or two strands and this was causing my disconnections and ultimately throttling my connection.

My connection is now at 67Mbps and has stayed connected for 45 days, 5 hours and 47 minutes.

BT are so frustrating, but keep onto them until they send out a better engineer?

I also got a credit for 3 months line rental by way of apology, but I just wanted the connection fixed.

[bimsb6]

jeremyc said:

bimsb6 said:

Gfast is very dependent on a good copper connection from the cabinet and needs to be a very short cable run ( within a few hundred metres of cable) .

I'm within the distance guidelines for the copper run - about 300m from the cabinet I believe.

I can understand a poor copper connection giving a poor performance, but don't understand why it would give great performance for a week before being degraded. It's almost as if there are traffic management/shaping algorithms that say "let it run for a week and then wind it down every night".

From my end-point evidence, my whole physical infrastructure would appear to happily support 200Mbps+. I'm trying to understand what might be going on to limit performance in the system.

Is the circuit being reset at each visit , this could then degrade as the circuit settles down .

[TuonoPants]

KingofKong said:

I had similar with my Infinity, it took 5 different engineers over the course of 3 months to find why my connection was dropping from 68Mbps down to 5ish. The first couple of engineers were contractors, not actual BT staff, the next couple just checked what the first lot had already checked and were keen to blame my internal wiring as the issue.

Finally, the last visit was from a senior engineer with lots of test equipment who found a loose cable in the green cabinet, why the others didn’t think to check there I really don’t know.

Somewhere inside the cabinet my connection was only just hanging in there with one or two strands and this was causing my disconnections and ultimately throttling my connection.

My connection is now at 67Mbps and has stayed connected for 45 days, 5 hours and 47 minutes.

BT are so frustrating, but keep onto them until they send out a better engineer?

I also got a credit for 3 months line rental by way of apology, but I just wanted the connection fixed.

This sounds very typical, simply testing the line is not enough. The connections all need to be checked visually and also reseated, the copper connections in the cabinets and elsewhere are probably of the punch-down type (Krone or similar) and need to be given a good push with a punch-down tool to make sure that they are making good contact.

Repeatedly connecting test tools and expecting different results, or the answer to magically appear, is all too common.

[jeremyc]

Thanks for all the suggestions. Digging through the event log in the router I can see that the 'down' SNR margin is degrading over time (whereas the 'up' margin is steady) - figures are at time of nightly renegotiation:

date	down sync speed	down SNR margin
25 Jan	229	3.7
28 Jan	221	3.9
29 Jan	226	3.9
30 Jan	204	6.4
31 Jan	212	6.3
1 Feb	187	7.9
2 Feb	187	8.3
3 Feb	93	12.1

If I look at the 'live' down SNR margin it is currently 8.9dB. All 'up' SNR margins have been 3.1-3.3dB

So presumably there's still some fault causing a fluctuating SNR on the down line. Why would the SNR margin degrade over time?

I'm just coming up for my fifth engineer visit for these problems - hopefully I've now got more data to give him a clue.

[KingofKong]

TuonoPants said:

This sounds very typical, simply testing the line is not enough. The connections all need to be checked visually and also reseated, the copper connections in the cabinets and elsewhere are probably of the punch-down type (Krone or similar) and need to be given a good push with a punch-down tool to make sure that they are making good contact.

Repeatedly connecting test tools and expecting different results, or the answer to magically appear, is all too common.

It must cost BT a fortune to send out a succession of drones when the first ‘proper’ engineer discovered and resolved the fault in 15 minutes.

The other engineers were obsessed with blaming my internal wiring, but although it’s a huge house there’s only two extensions and they were already disconnected.

[jeremyc]

The saga continues. My fifth Openreach engineer visit confirmed that all of their infrastructure is fine: the copper from cabinet to my premises has no faults and can support bandwidths of 230Mbps+, and the fibre link from the cabinet supports 330Mbps. Their only suggestion was that it might be a problem on the BT side. So DLM reset, everything back to normal (low SNR margin, 238Mbps downlink) and the cycle continues again ...

Yet there seems no way to escalate the problem to BT (rather than Openreach) engineers - I spent 45 minutes in a call to their support line which did eventually result in me being connected to a 'FTTC support' person, who then succeeded only in cutting me off.

I've now tried raising a complaint with BT via their website, sending an email to their CEO and tweeting them.

Anyone have any other ideas how I can get to talk to a BT engineer who understands G.Fast/FTTC?

[KingofKong]

I suspect you’ll get a reply from the Exec Team if you’ve emailed the CEO. I think based on the history of your issues they’ll get it resolved, it’s the people who email the CEO and get all shouty because they got disconnected once in a game of CoD that they are likely to ignore.

I know when I requested a senior engineer I had to wait a few weeks longer than the usual plug/unplug first line guys you often get.

[CharlesElliott]

The SNR is dependent on your sync speed so can only be compared at the same sync. If your sync speed drops the SNR will go up all things being even

I hope you get it sorted.

BT usually have 6dB as the minimum sync SNR so if it drops below that, renegotiation is likely.

[CharlesElliott]

Higher SNR margin is better by the way.

[jeremyc]

CharlesElliott said:

Higher SNR margin is better by the way.

Yes, but if the target SNR margin the system is trying to achieve is higher than the actual then the line speed is decreased to try and get there.

Currently my sync speed is being renegotiated every night, except for the first two nights after DLM reset/engineer visit.

The uplink SNR margin is rock steady at around 3.1dB; the downlink SNR margin starts out at around 3.1dB and increases to 13dB at which point I call out the engineers and we start the cycle again.

[jeremyc]

CharlesElliott said:

BT usually have 6dB as the minimum sync SNR so if it drops below that, renegotiation is likely.

Not what I'm seeing: renegotiation is happening every night, even with SNR margins at >6dB.

[jeremyc]

I'm in danger of ranting to myself here.

www.increasebroadbandspeed.co.uk said:

When your broadband service is first switched on, the target SNR margin is set by the BT exchange equipment at 6 dB. On a good line, BT generally considers that a 6 dB SNR margin is appropriate to maintain a connection without a significant number of drops or errors.

As part of the Digital Line Management process, the exchange equipment continuously monitors your broadband connection, for example recording the number of times that it drops and the number of errors it suffers. If the line experiences any issues (such as bursts of errors at a particular time), the exchange equipment can increase the target SNR margin – usually in steps of 3 dB. This action – aimed at stabilising the line – reduces the connection speed.

BT’s DLM system, or BT’s manual intervention, can increase the target SNR margin to a maximum value of 18 dB.

Once the BT exchange equipment has increased the target SNR margin, it can take days for the target SNR margin to be reduced and, in some cases, it is never reduced. In the worst case scenario, you may be saddled with very slow downlink speeds caused by an 18 dB target SNR margin when your line could work perfectly with a 3 dB (or less) SNR margin.

I wonder if somehow the DLM has a crazy target SNR margin set which would explain the behaviour I'm seeing.

website also said:

The exchange equipment attempts to set the downlink connection speed such that there is a sufficiently high SNR margin to guarantee a reliable connection (without regular drops and/or an unacceptable number of errors). This specific margin that the exchange tries to achieve is called the target SNR margin. The higher the target SNR margin, the lower your connection speed.

[bimsb6]

jeremyc said:

CharlesElliott said:

BT usually have 6dB as the minimum sync SNR so if it drops below that, renegotiation is likely.

Not what I'm seeing: renegotiation is happening every night, even with SNR margins at <4dB.

Too low an snr will give errors and cause dropouts , if the line keeps erroring the snr will rise to try and make the connection stable this results in a drop in speed , thats why i asked earlier if the line was being reset as this gives you the roller coaster of high speeds after reset then snr going up and speeds dropping as the errors kick in .

[jeremyc]

bimsb6 said:

Too low an snr will give errors and cause dropouts , if the line keeps erroring the snr will rise to try and make the connection stable this results in a drop in speed , thats why i asked earlier if the line was being reset as this gives you the roller coaster of high speeds after reset then snr going up and speeds dropping as the errors kick in .

Understand this, and this is exactly what is happening.

DLM is reset, speeds then constantly decrease (the blue line on my graph) as SNR margin increases (the orange line on my graph).

However, could it be that the target SNR margin in the DLM is set incorrectly meaning that the line speed is constantly decreased in order to try and get the SNR margin to the unrealistic target level?

Left unchecked, my connection ultimately fails completely with speeds dropping too low.

[bimsb6]

jeremyc said:

bimsb6 said:

Too low an snr will give errors and cause dropouts , if the line keeps erroring the snr will rise to try and make the connection stable this results in a drop in speed , thats why i asked earlier if the line was being reset as this gives you the roller coaster of high speeds after reset then snr going up and speeds dropping as the errors kick in .

Understand this, and this is exactly what is happening.

DLM is reset, speeds then constantly decrease (the blue line on my graph) as SNR margin increases (the orange line on my graph).

However, could it be that the target SNR margin in the DLM is set incorrectly meaning that the line speed is constantly decreased in order to try and get the SNR margin to the unrealistic target level?

Left unchecked, my connection ultimately fails completely with speeds dropping too low.

Not heard that theory before , i don’t think it works like that, the dynamic line management changes the snr to stabilise the connection speed not the other way round .

[jeremyc]

bimsb6 said:

Not heard that theory before , i don’t think it works like that, the dynamic line management changes the snr to stabilise the connection speed not the other way round .

Surely the signal and noise levels are what they are for a given connection speed (and line).

What the DLM can do is wind down the connection speed so the signal level increases with respect to the noise level (less errors) and hence the SNR margin increases.

So the DLM changes the connection speed in order to improve the SNR margin in order to meet the target it is aiming for to provide a stable connection.