FTTC - discrepancy between the reported Max Rate, and the actual Data rate

[g67575]

Thought this might be of interest:

So, around a 27mbps difference between the actual Downstream Data / Sync rate and the reported 'Max Rate' on an Asus router.

I have SNR Margin set to 3DB and 'Rx AGC GAIN Adjustment' set to 'High Performance' in the DSL settings menu.

Interleaving was enabled about 30 mins ago, which seems to prevent most CRC errors (when it decides to turn itself on), at the cost of a little bandwidth.

Another thing I notice about interleaving, is that on the SNR graph displayed by the router, the lowest points on the graph appear higher (all over 10dB), suggesting an improved SNR.

 

[Caged]

Asus routers have a reputation for having very poor VDSL modems, do you have anything else you can try?

 

[g67575]

Yes, I've had a TP-Link router, Talktalk router and 2 Technicolour routers. All synced at a max of ~51mbps downstream. The SNR was only sometimes at 3DB, which often meant that it synced lower than 50mbps (due to the SNR Margin defaulting to 6DB).

The reported 'max rates' on these routers were generally lower.

The good thing about most Asus VDSL modems, is that they can force on an SNR margin of 3DB, to boost the downstream sync speed.

As for the connection, it's solid with a DSL-N16 router /modem.

 

[Orcvader]

What do you get in speed tests?

It could also be on the ISP/Openreach side, with DLM potentially limiting it. You could try calling your ISP and see if they can reset the DLM profile.

 

[g67575]

Download speed is 54/55mbps when downloading files.

 

[Caged]

I think your line is banded or there's a fault somewhere. The TP Link should be able to work as just a dumb modem and then you can ask your ISP to try a profile reset. I have 22dB of line attenuation on my downstream and sync at 68Mbps.

 

[g67575]

Thanks for your help.

SNR is the main factor that determines sync rate, as far as I'm aware. Here the SNR graph for the line shown by my router:

It's not too bad really, it looks like the SNR is above 10 DB even on the D3 frequency band (highest VDSL frequencies). But 20 DB would be more ideal. But it looks like this part of the signal is quite prone to noise.

In general, I think ISPs can only reset your DLM after sending a BT engineer to check your wiring and equipment.

I've changed ISPs before, and received similar line stats.

 

[g67575]

SNR is how much of the signal is actually received reliably, isn't it?

A lower figure means both a noisier line and lower overall signal strength.

The sync rate on FTTC seems to be determined by the SNR, minus the Line attenuation of each VDSL2 frequency band. So, both are important factors.

The 'SNR margin' adjusts the sync rate accordingly, to ensure an adequate 'safety margin' between the SNR (received signal strength) and line attenuation (signal degradation over distance). Typically, this is configured by Openreach to be 3DB or 6DB on FTTC.

I assume this margin is necessary to cope with fluctuating line conditions caused by things like electrical noise and weather.

 

[g67575]

TLDR

My line's main weakness appears to be high 'Signal Attenuation' on the higher frequency 'D2' and 'D3' VDSL2 downstream bands, as shown here:

 

[VersionMonkey]

SNR is how much of the signal is actually received reliably, isn't it?

Kinda.

A lower figure means both a noisier line and lower overall signal strength.

Not necessarily. In isolation a lower SNR tells you nothing about the absolute values of either the received signal or the level of noise. On the other hand if you know the level of the signal is unchanged, then a lower SNR would tell you that the noise must have increased.

The SNR is the difference between the level of the received signal and the level of the noise.

The sync rate on FTTC seems to be determined by the SNR, minus the Line attenuation of each VDSL2 frequency band. So, both are important factors.

No, that's not how I understand it at all - if you have a reference for that it would be interesting to read.

The line attenuation is a measure of how much the signal is reduced over the length of the line. This will influence the measured SNR - so for example picking numbers completely at random out of thin air: Say the signal is originally 20dB and you have 15dB of attenuation, then the received signal level would be 5dB. If the noise is at 2dB, then your SNR would be 3dB.

The sync rate is established solely on the SNR, it makes no sense to include the line attenuation in that calculation because attenuation is already inherent in the SNR.

The 'SNR margin' adjusts the sync rate accordingly, to ensure an adequate 'safety margin' between the SNR (received signal strength) and line attenuation (signal degradation over distance). Typically, this is configured by Openreach to be 3DB or 6DB on FTTC.

The margin is a target SNR - during the sync, the modems (yours and the one in the cabinet) will effectively test the line to see what it will support while sticking to a minimum SNR. I believe that DLM can also go to a 9dB margin.

I assume this margin is necessary to cope with fluctuating line conditions caused by things like electrical noise and weather.

Correct. Noise on the line will naturally vary over the day - in particular radio signals are propagated differently at different times of the day, so the line will pick up varying amounts of RF. Plus local noise producers may not be on at the time of the sync.

Attenuation is an inherent property of the line - it shouldn't vary by much (although I imagine some faults can cause it to change). Plus the original signal power will be constant, so the received level of the signal will most likely be more or less the same - it's the variation in the noise level that can cause the received signal to be swamped leading to errors and ultimately resyncs. It's that that DLM is trying to prevent by setting an initial minimum SNR.

But that's just what I've picked up and inferred over the years, so would be interested in any alternative references that you may have.

 

[Caged]

You're probably best off signing up at the Kitz forum and posting your DSL stats, people there can figure out what's going on.

 

[g67575]

The sync rate is established solely on the SNR, it makes no sense to include the line attenuation in that calculation because attenuation is already inherent in the SNR.

Thanks for taking the time to answer.

I think you are correct about the importance of the SNR to sync rate.

I read about the formula for ADSL here:
https://networkengineering.stackexchange.com/a/43794

But, maybe it is a bit different for VDSL? I did simply it a bit, as in his example, the poster assumes a 'desired downstream in Kilobits per second: 8096 (ADSL)'. His calculation is SNR - Attenuation = SNR Margin.

Maybe the only correct way to interpret this, is that a high 'SNR Margin' (or separate SNR margin figures for each band) is actually the best prediction of a high sync rate, assuming the 'SNR Margin' hasn't been set manually by the user.

I'm curious, do you agree with my conclusion that the limitations on my FTTC connection sync speed, are the result of high attenuation on the D2 and D3 downstream frequency bands?

But, a D1 downstream of ~10DBs isn't particularly great either, according to this website:
https://www.speedguide.net/dsl_speed_calc.php

An attenuation of 5DBs on D1 would apparently provide a downstream sync rate of 78311kbps, even without the other VDSL2 bands, in most cases. 4DB should allow a sync rate of 80,000kbps, where a sync rate limit is imposed.

 

[VersionMonkey]

Thanks for taking the time to answer.

I think you are correct about the importance of the SNR to sync rate.

I read about the formula for ADSL here:
https://networkengineering.stackexchange.com/a/43794

But, maybe it is a bit different for VDSL? I did simply it a bit, as in his example, the poster assumes a 'desired downstream in Kilobits per second: 8096 (ADSL)'. His calculation is SNR - Attenuation = SNR Margin.

The two answers there contradict each other as to the definition of SNR margin. The first states it's the difference between the SNR required to hit a specific sync rate and the actual SNR. The second includes attenuation as you state.

The first stackexchange answer, which was accepted, corresponds with information on the Kitz Wiki:

https://wiki.kitz.co.uk/index.php?title=Decibels:

It should be noted that ADSL modems and routers generally report noise margin (or SNR margin), not SNR. The manufacturer specifies a SNR at which the error count reaches some threshold beyond which the connection would not be acceptably stable; the noise margin is the amount by which the measured SNR exceeds this base level of SNR. So, for example, if the minimum acceptable SNR was set at 6 dB, and the measured SNR was 10 dB, then the modem/router would report a noise margin of 4 dB.

Therefore I don't think the calculation subtracting attenuation from the SNR is correct (or at least misses out a few steps). Still, I was a bit unclear on the details of SNR vs SNR margin, so looking at this meant I learned something today - which is always good.

So far as I know, the basic princples of how ADSL and VDSL operate are the same. VDSL is able to use higher frequencies as the shorter copper line length means those signals won't get attenuated away (as much).

Maybe the only correct way to interpret this, is that a high 'SNR Margin' (or separate SNR margin figures for each band) is actually the best prediction of a high sync rate, assuming the 'SNR Margin' hasn't been set manually by the user.

Pretty much.

I'm curious, do you agree with my conclusion that the limitations on my FTTC connection sync speed, are the result of high attenuation on the D2 and D3 downstream frequency bands?

But, a D1 downstream of ~10DBs isn't particularly great either, according to this website:
https://www.speedguide.net/dsl_speed_calc.php

An attenuation of 5DBs on D1 would apparently provide a downstream sync rate of 78311kbps, even without the other VDSL2 bands, in most cases. 4DB should allow a sync rate of 80,000kbps, where a sync rate limit is imposed.

Possibly - the people on the Kitz (or ThinkBroadband) forums would be the ones to ask, as @Caged suggested. I've not been down the DSL stats rabbit hole for a long time. Your SNR graph is probably what I would expect to see in terms of shape - tapering off towards the higher frequencies but your attenuation is lower than mine yet you're getting a similar sync rate - so maybe your line is noisier? Do you know how far you are from your cabinet (as the cable runs)?

Ultimately, I suppose, what was your speed estimate when you signed up and are you getting that - if your line meets that your ISP is unlikely to do much. There's not much you can do your end other than making sure your modem (whether built into a router or seperate) is as close to the point where the phone line enters the building as possible and that you don't have any dodgy extension wiring.

 

[g67575]

My gut feeling is that the SNR is the main issue with my line (but a much lower attenuation would presumably allow a top sync rate of around 80,000kbps. Not achievable in practice, since the line is longer than 150m to the cabinet, probably more like 300-400m). I did a resync a bit earlier today and the SNR of the D3 band increased above 10DB (having dropped slightly before the resync), but also dropped the downstream sync speed (presumably to ensure sufficient SNR Margin).

I have a couple of telephone extensions wired to the FTTC master socket, I guess the other thing that might affect SNR is the use of a few homeplugs. Neither of these things I can change, my family wants these extensions kept, and we can't do away with homeplugs completely yet. I don't think it's the homeplugs to be honest, I've tried disconnecting them and resyncing but saw no improvement.

I seem to remember disconnecting the 'bell wires' on each extension + master socket a long time ago, so I'm sure that helped.

The other factor to take into account is my FTTC cabinet is ECI, and therefore doesn't support G.INP, which can give a nice boost to sync rates and reliability on many lines.

 

[g67575]

Sod it, just found that I might be getting a 1Gbps internet connection in my area at some point in 2022 (which is great news tbh ), but don't know if it's Virgin Media/DOCSIS or a FTTP connection yet.

The main weakness of VDSL2 has always been poor range though, with high attenuation beyond 150-200 meters.

 

[g67575]

.

 

[VersionMonkey]

Sod it, just found that I might be getting a 1Gbps internet connection in my area at some point in 2022 (which is great news tbh ), but don't know if it's Virgin Media/DOCSIS or a FTTP connection yet.

The main weakness of VDSL2 has always been poor range though, with high attenuation beyond 150-200 meters.

Either of those options would solve any and all problems with a VDSL line