VDSL attenuation and SLOW speeds

Looking for a bit of advice here.

Just got a VDSL2 service at an address that the checker for the company (Phone Coop) predicted would have a speed between 35-40 mbits. Max speed 70ish.

Service was activated just after Christmas and the modem reports a speed of 28 MbPS - poor, that�s also the fastest it�s done since it was put in, normally it�s about 24. Sustained download speeds are worse, around 21-23 MbPS.

The router reports line attenuation as 6 numbers. I don�t really understand what all the different numbers mean. I noticed most people get one figure for upload, one for download? I have 3 for each. *shoulder shrug*

They are (line attenuation):

Download:
18.1, 49.3, 76.2 DB

Upload:
4.2, 40.6, 58.2 DB

Noise margin:

Upload:
6.1 DB

Download:
6.2 DB

I found an attenuation to sync speed calculator for VDSL2, but due to my 3 figures for upload/download, I can�t work out how/where to put all these numbers in, to calculate the sync speed based on these attenuation figures. Apparently they are called D1,D2, and D3, and are something to do with an 8C profile. I couldn�t find a coherent answer on turning that into one number to do a sync speed calculation.

Screenshots of technical info from my router:

https://imgur.com/a/sdKMLsK

Secondly:

I plan on replacing the NTE5 box for a new one with a mk4 VDSL filter built in, and I�m hoping I can dredge some dregs of extra speed from a new twisted pair RJ45 to modem connector wire.

Unless a RJ11 twisted pair cable would be better? I know the VDSL port on the new Mk4 can take a RJ45. I�ve seen cables that are designed for this on amazon. Not sure where the biggest speed boost is to be had.

Any ideas?

Edited by deleted (Mon 11-Feb-19 21:12:49)

What does the BT wholesale checker say for your line, in terms of the expected max/min sync speeds? (Check with phone number, not postcode). https://btwholesale.com/adslchecker/

Sustained download speeds will always be about 5% below sync speed, because of TCP/IP overhead.

A twisted pair is better than straight, but if the router is 1-2 metres away from the socket it's unlikely to make any measurable difference.

RJ11 vs RJ45 won't make any difference. You're just using 2 pins within a 6-pin housing versus 2 pins within an 8-pin housing.

I read somewhere that the RJ45 makes a difference because it activates the grounding shield on the twisted pair cable, whereas the RJ11 has no grounding pin. This makes sense.

I will be buying a 5 metre cable, so it will be further away. It�s currently 5 metres away on flat cable.

As for the speed test, I cannot run it because it requires flash and we do not own a PC/Mac. Very old tech, not for me.

Okay, so I installed the MK4 VDSL face plate and Mk2 back box from BT (latest edition). Noise margin has gone from 6.1 DB up and down to 6.2 up and 6.4 down, apparently a higher Noise Margin (which is apparently SNR) is better, so I�m happy to have gotten some improvements on the line. Downloads are consistently faster, and upload speeds are noticeably improved. Not bad for £10. We�re now on 29-30 meg download. So getting there.

You can read lots of odd things on the internet, and rj45 does not possess a magic grounding ping

RJ45 is just the definition of a plug and socket pair, of which RJ11 is a member of the same family

At this point, as you say the key is what the BT wholesale checker says - its not unknown for ISP to get this wrong, also not unknown for the wholesale checker to be wrong, but first need to see the figures

Hope this helps:

https://imgur.com/a/e3hPZT3

I had confused the wholesale thing with the speed checker which needs flash. I can click the wholesale thing. Here it is!

What's the router manufacturer and model? Are you running latest firmware? I doubt it's your router - as 30mbit is reasonable. I'd expect lower speeds and packet loss/drops/disconnects if it was your modem.. 90% sure.

The 3 outputs of attenuation are the three bands of the VDSL frequency range in use. D1, D2 ,D3
I do not have a good example of what good levels are..

Ultimately, you're on the lower end of your estimated speed - which I'd find frustrating to see is the ISP was saying 70'ish.

Doing some research online, I see that D1 band falls over the ADSL2+ frequency, and those attenuation levels I am more familiar with. So, I'd recon D1 is a good indicator to use if your over all copper line is decent quality. Its about average.

Another thing I've learnt from personal experience and evidence.... DO NOT under estimate an ISP using [censored] old legacy equipment in the DSLAM/Exchange which is causing slower speeds and poor service even though your copper line is somewhat clean.

What's my evidence?
I work for an ISP and at one stage had a direct comparison of two service providers to my home.
Both FTTC, same annex, same NTE and settings etc..
Both services ran down the same cable (two twisted pairs inside).

One circuit would never sync above 51Mbit.. other circuit 78Mbit.

ref: http://www.atnf.csiro.au/people/atzioumi/jdsu/xDSL%2...

It�s a:
Technicolor TG588V V2.

I think the firmware is up to date but I�ll take a look. It�s not a great router, but realistically, it�s not causing the speed issue ha. I�ll upgrade the cable to twisted pair and I�ll demand a DLM reset, hopefully that�ll push me up to 35. If I get it up to 35, I�ll live with it, until I can get a better ISP. I might also demand an engineer come out. It�s at the bottom of, maybe just below the bottom end of the impacted range, I�ve read it�s reasonable to get an engineer out to look at it for that.

The speeds you are seeing then are within the range expected, though towards the lower end of things.

So looks like the provider was perhaps a bit optimistic in its original estimates.

IMPORTANT

The line connection speed is not going to magically improve if you swap provider, since the physical line and cabinet hardware will remain the same.

Not really, because my upload speed is: 2.88 mbps and the lowest estimate for the impacted upload speed is 5.7 mbps, which is significantly below lowest expectations.

The download speed is currently 28.53 mbps maximum in the last 24 hours, that�s the fastest it�s been (I know I said it did 29-30, but I was overegging it), and the lowest estimate for impacted speed is 29.4 mbps. It�s looking like it�s got a tendency to drift closer to the handback threshold rather than acceptable impacted low end estimates.

Edited by deleted (Tue 12-Feb-19 16:19:20)

You said things improved after a bit of work - it is possible the DLM has intervened to reduce errors prior to this and thus speeds may go up in time.

Also its possible the BT Wholesale checker is being optimistic and what you are seeing is the most likely reality. If you private message your postcode I can have a look at the local area and see how likely going a lot faster looks based on the distances involved.

I strongly believe you are not going to see 35 Mbps, at least on a regular basis; possibly after a powercut before crosstalk becomes high but could be fairly unstable until it resyncs, plus could encourage DLM to take negative action.

As someone has already mentioned, your line is about right for the anticipated downstream sync speed based on the stats provided.

The upstream sync speed appears a bit enemic, however (but this may be normal for your line, regardless of what the BTW checker says).

As Saffron said, it's possible the BTW checker is being optimistic. It does happen, and there have been cases where it has been worse.

I would pass this back to the ISP if you are unable to achieve the estimates provided by them (and because you are below the threshold for a "clean" line). They could do a few things, which includes sending an engineer out or waiving contract terms such as contract duration (thus allowing you to leave early).

Before you agree to an engineer visit, just be aware that the costs may be passed onto you if no fault is found or if it is found that your hardware/wiring is at fault. Though with the former there is a clear lack of ability to achieve the estimates so no fees should be passed on, and with the latter it seems you have tried a different faceplate (and, presumably, tried the test socket) so not a great risk there.

Just a post to thank everyone who has replied. Now in fault finding with the ISP and you�ve all educated me which I really appreciate.